OpenCores
URL https://opencores.org/ocsvn/zipcpu/zipcpu/trunk

Subversion Repositories zipcpu

[/] [zipcpu/] [trunk/] [doc/] [src/] [spec.tex] - Diff between revs 33 and 36

Go to most recent revision | Show entire file | Details | Blame | View Log

Rev 33 Rev 36
Line 46... Line 46...
\documentclass{gqtekspec}
\documentclass{gqtekspec}
\project{Zip CPU}
\project{Zip CPU}
\title{Specification}
\title{Specification}
\author{Dan Gisselquist, Ph.D.}
\author{Dan Gisselquist, Ph.D.}
\email{dgisselq (at) opencores.org}
\email{dgisselq (at) opencores.org}
\revision{Rev.~0.3}
\revision{Rev.~0.4}
 
\definecolor{webred}{rgb}{0.2,0,0}
 
\definecolor{webgreen}{rgb}{0,0.2,0}
 
\usepackage[dvips,ps2pdf,colorlinks=true,
 
        anchorcolor=black,pagecolor=webgreen,pdfpagelabels,hypertexnames,
 
        pdfauthor={Dan Gisselquist},
 
        pdfsubject={Zip CPU}]{hyperref}
\begin{document}
\begin{document}
\pagestyle{gqtekspecplain}
\pagestyle{gqtekspecplain}
\titlepage
\titlepage
\begin{license}
\begin{license}
Copyright (C) \theyear\today, Gisselquist Technology, LLC
Copyright (C) \theyear\today, Gisselquist Technology, LLC
Line 68... Line 74...
You should have received a copy of the GNU General Public License along
You should have received a copy of the GNU General Public License along
with this program.  If not, see \hbox{<http://www.gnu.org/licenses/>} for a
with this program.  If not, see \hbox{<http://www.gnu.org/licenses/>} for a
copy.
copy.
\end{license}
\end{license}
\begin{revisionhistory}
\begin{revisionhistory}
 
0.4 & 9/19/2015 & Gisselquist & Added DMA controller, improved stall information, and self--assessment info.\\\hline
0.3 & 8/22/2015 & Gisselquist & First completed draft\\\hline
0.3 & 8/22/2015 & Gisselquist & First completed draft\\\hline
0.2 & 8/19/2015 & Gisselquist & Still Draft, more complete \\\hline
0.2 & 8/19/2015 & Gisselquist & Still Draft, more complete \\\hline
0.1 & 8/17/2015 & Gisselquist & Incomplete First Draft \\\hline
0.1 & 8/17/2015 & Gisselquist & Incomplete First Draft \\\hline
\end{revisionhistory}
\end{revisionhistory}
% Revision History
% Revision History
Line 87... Line 94...
 
 
The easiest, most obvious answer is the simple one: Because I can.
The easiest, most obvious answer is the simple one: Because I can.
 
 
There's more to it, though. There's a lot that I would like to do with a
There's more to it, though. There's a lot that I would like to do with a
processor, and I want to be able to do it in a vendor independent fashion.
processor, and I want to be able to do it in a vendor independent fashion.
I would like to be able to generate Verilog code that can run equivalently
First, I would like to be able to place this processor inside an FPGA.  Without
on both Xilinx and Altera chips, and that can be easily ported from one
paying royalties, ARM is out of the question.  I would then like to be able to
manufacturer's chipsets to another. Even more, before purchasing a chip or a
generate Verilog code, both for the processor and the system it sits within,
board, I would like to know that my soft core works. I would like to build a test
that can run equivalently on both Xilinx and Altera chips, and that can be
bench to test components with, and Verilator is my chosen test bench. This
easily ported from one manufacturer's chipsets to another. Even more, before
forces me to use all Verilog, and it prevents me from using any proprietary
purchasing a chip or a board, I would like to know that my soft core works. I
cores. For this reason, Microblaze and Nios are out of the question.
would like to build a test bench to test components with, and Verilator is my
 
chosen test bench. This forces me to use all Verilog, and it prevents me from
 
using any proprietary cores. For this reason, Microblaze and Nios are out of
 
the question.
 
 
Why not OpenRISC? That's a hard question. The OpenRISC team has done some
Why not OpenRISC? That's a hard question. The OpenRISC team has done some
wonderful work on an amazing processor, and I'll have to admit that I am
wonderful work on an amazing processor, and I'll have to admit that I am
envious of what they've accomplished. I would like to port binutils to the
envious of what they've accomplished. I would like to port binutils to the
Zip CPU, as I would like to port GCC and GDB. They are way ahead of me. The
Zip CPU, as I would like to port GCC and GDB. They are way ahead of me. The
Line 120... Line 130...
\chapter{Introduction}
\chapter{Introduction}
\pagenumbering{arabic}
\pagenumbering{arabic}
\setcounter{page}{1}
\setcounter{page}{1}
 
 
 
 
The original goal of the ZIP CPU was to be a very simple CPU.   You might
The original goal of the Zip CPU was to be a very simple CPU.   You might
think of it as a poor man's alternative to the OpenRISC architecture.
think of it as a poor man's alternative to the OpenRISC architecture.
For this reason, all instructions have been designed to be as simple as
For this reason, all instructions have been designed to be as simple as
possible, and are all designed to be executed in one instruction cycle per
possible, and are all designed to be executed in one instruction cycle per
instruction, barring pipeline stalls.  Indeed, even the bus has been simplified
instruction, barring pipeline stalls.  Indeed, even the bus has been simplified
to a constant 32-bit width, with no option for more or less.  This has
to a constant 32-bit width, with no option for more or less.  This has
Line 295... Line 305...
stored.
stored.
 
 
\section{Simplified Bus}
\section{Simplified Bus}
The bus architecture of the Zip CPU is that of a simplified WISHBONE bus.
The bus architecture of the Zip CPU is that of a simplified WISHBONE bus.
It has been simplified in this fashion: all operations are 32--bit operations.
It has been simplified in this fashion: all operations are 32--bit operations.
The bus is neither little endian nor bit endian.  For this reason, all words
The bus is neither little endian nor big endian.  For this reason, all words
are 32--bits.  All instructions are also 32--bits wide.  Everything has been
are 32--bits.  All instructions are also 32--bits wide.  Everything has been
built around the 32--bit word.
built around the 32--bit word.
 
 
\section{Register Set}
\section{Register Set}
The Zip CPU supports two sets of sixteen 32-bit registers, a supervisor
The Zip CPU supports two sets of sixteen 32-bit registers, a supervisor
Line 316... Line 326...
noted as (SP)--although there is nothing special about this register other
noted as (SP)--although there is nothing special about this register other
than this convention.
than this convention.
The CPU can access both register sets via move instructions from the
The CPU can access both register sets via move instructions from the
supervisor state, whereas the user state can only access the user registers.
supervisor state, whereas the user state can only access the user registers.
 
 
The status register is special, and bears further mention.  The lower
The status register is special, and bears further mention.  As shown in
10 bits of the status register form a set of CPU state and condition codes.
Fig.~\ref{tbl:cc-register},
Writes to other bits of this register are preserved.
\begin{table}\begin{center}
 
\begin{bitlist}
 
31\ldots 11 & R/W & Reserved for future uses\\\hline
 
10 & R & (Reserved for) Bus-Error Flag\\\hline
 
9 & R & Trap, or user interrupt, Flag.  Cleared on return to userspace.\\\hline
 
8 & R & (Reserved for) Illegal Instruction Flag\\\hline
 
7 & R/W & Break--Enable\\\hline
 
6 & R/W & Step\\\hline
 
5 & R/W & Global Interrupt Enable (GIE)\\\hline
 
4 & R/W & Sleep.  When GIE is also set, the CPU waits for an interrupt.\\\hline
 
3 & R/W & Overflow\\\hline
 
2 & R/W & Negative.  The sign bit was set as a result of the last ALU instruction.\\\hline
 
1 & R/W & Carry\\\hline
 
0 & R/W & Zero.  The last ALU operation produced a zero.\\\hline
 
\end{bitlist}
 
\caption{Condition Code Register Bit Assignment}\label{tbl:cc-register}
 
\end{center}\end{table}
 
the lower 11~bits of the status register form
 
a set of CPU state and condition codes.  Writes to other bits of this register
 
are preserved.
 
 
Of the condition codes, the bottom four bits are the current flags:
Of the condition codes, the bottom four bits are the current flags:
                Zero (Z),
                Zero (Z),
                Carry (C),
                Carry (C),
                Negative (N),
                Negative (N),
Line 369... Line 398...
%
%
 
 
This functionality was added to enable an external debugger to
This functionality was added to enable an external debugger to
        set and manage breakpoints.
        set and manage breakpoints.
 
 
The ninth bit is reserved for a floating point enable bit.  When set, the
The ninth bit is reserved for an illegal instruction bit.  When the CPU
arithmetic for the next instruction will be sent to a floating point unit.
tries to execute either a non-existant instruction, or an instruction from
Such a unit may later be added as an extension to the Zip CPU.  If the
an address that produces a bus error, the CPU will (once implemented) switch
CPU does not support floating point instructions, this bit will never be set.
to supervisor mode while setting this bit.  The bit will automatically be
The instruction set could also be simply extended to allow other data types
cleared upon any return to user mode.
in this fashion, such as two by 16--bit vector operations or four by 8--bit
 
vector operations.
 
 
 
The tenth bit is a trap bit.  It is set whenever the user requests a soft
The tenth bit is a trap bit.  It is set whenever the user requests a soft
interrupt, and cleared on any return to userspace command.  This allows the
interrupt, and cleared on any return to userspace command.  This allows the
supervisor, in supervisor mode, to determine whether it got to supervisor
supervisor, in supervisor mode, to determine whether it got to supervisor
mode from a trap or from an external interrupt or both.
mode from a trap or from an external interrupt or both.
Line 402... Line 429...
\end{tabular}
\end{tabular}
\caption{Condition Code / Status Register Bits}\label{tbl:ccbits}
\caption{Condition Code / Status Register Bits}\label{tbl:ccbits}
\end{center}\end{table}
\end{center}\end{table}
 
 
\section{Conditional Instructions}
\section{Conditional Instructions}
Most, although not quite all, instructions are conditionally executed.  From
Most, although not quite all, instructions may be conditionally executed.  From
the four condition code flags, eight conditions are defined.  These are shown
the four condition code flags, eight conditions are defined.  These are shown
in Tbl.~\ref{tbl:conditions}.
in Tbl.~\ref{tbl:conditions}.
\begin{table}
\begin{table}
\begin{center}
\begin{center}
\begin{tabular}{l|l|l}
\begin{tabular}{l|l|l}
Line 422... Line 449...
\end{tabular}
\end{tabular}
\caption{Conditions for conditional operand execution}\label{tbl:conditions}
\caption{Conditions for conditional operand execution}\label{tbl:conditions}
\end{center}
\end{center}
\end{table}
\end{table}
There is no condition code for less than or equal, not C or not V.  Sorry,
There is no condition code for less than or equal, not C or not V.  Sorry,
I ran out of space in 3--bits.  Using these conditions will take an extra
I ran out of space in 3--bits.  Conditioning on a non--supported condition
instruction and a pipeline stall.  (Ex: \hbox{\em (Stall)}; \hbox{\tt TST \$4,CC;} \hbox{\tt STO.NZ R0,(R1)})
is still possible, but it will take an extra instruction and a pipeline stall.  (Ex: \hbox{\em (Stall)}; \hbox{\tt TST \$4,CC;} \hbox{\tt STO.NZ R0,(R1)})
 
 
 
Conditionally executed ALU instructions will not further adjust the
 
condition codes.
 
 
\section{Operand B}
\section{Operand B}
Many instruction forms have a 21-bit source ``Operand B'' associated with them.
Many instruction forms have a 21-bit source ``Operand B'' associated with them.
This Operand B is either equal to a register plus a signed immediate offset,
This Operand B is either equal to a register plus a signed immediate offset,
or an immediate offset by itself.  This value is encoded as shown in
or an immediate offset by itself.  This value is encoded as shown in
Line 446... Line 476...
multiply?  Likewise, why have a 16--bit immediate when adding to a logical
multiply?  Likewise, why have a 16--bit immediate when adding to a logical
or arithmetic shift?  In these cases, the extra bits are reserved for future
or arithmetic shift?  In these cases, the extra bits are reserved for future
instruction possibilities.
instruction possibilities.
 
 
\section{Address Modes}
\section{Address Modes}
The ZIP CPU supports two addressing modes: register plus immediate, and
The Zip CPU supports two addressing modes: register plus immediate, and
immediate address.  Addresses are therefore encoded in the same fashion as
immediate address.  Addresses are therefore encoded in the same fashion as
Operand B's, shown above.
Operand B's, shown above.
 
 
A lot of long hard thought was put into whether to allow pre/post increment
A lot of long hard thought was put into whether to allow pre/post increment
and decrement addressing modes.  Finding no way to use these operators without
and decrement addressing modes.  Finding no way to use these operators without
Line 482... Line 512...
compiler or assembler know how to compile a MOV instruction without knowing
compiler or assembler know how to compile a MOV instruction without knowing
the mode of the CPU at the time?  For this reason, the compiler will assume
the mode of the CPU at the time?  For this reason, the compiler will assume
all MOV registers are supervisor registers, and display them as normal.
all MOV registers are supervisor registers, and display them as normal.
Anything with the user bit set will be treated as a user register.  The CPU
Anything with the user bit set will be treated as a user register.  The CPU
will quietly ignore the supervisor bits while in user mode, and anything
will quietly ignore the supervisor bits while in user mode, and anything
marked as a user register will always be valid.  (Did I just say that in the
marked as a user register will always be valid.
last paragraph?)
 
 
 
\section{Multiply Operations}
\section{Multiply Operations}
The Zip CPU supports two Multiply operations, a
The Zip CPU supports two Multiply operations, a 16x16 bit signed multiply
16x16 bit signed multiply (MPYS) and the same but unsigned (MPYU).  In both
({\tt MPYS}) and a 16x16 bit unsigned multiply ({\tt MPYU}).  In both
cases, the operand is a register plus a 16-bit immediate, subject to the
cases, the operand is a register plus a 16-bit immediate, subject to the
rule that the register cannot be the PC or CC registers.  The PC register
rule that the register cannot be the PC or CC registers.  The PC register
field has been stolen to create a multiply by immediate instruction.  The
field has been stolen to create a multiply by immediate instruction.  The
CC register field is reserved.
CC register field is reserved.
 
 
\section{Floating Point}
\section{Floating Point}
The ZIP CPU does not support floating point operations.  However, the
The Zip CPU does not (yet) support floating point operations.  However, the
instruction set reserves two possibilities for future floating point
instruction set reserves two possibilities for future floating point
operations.
operations.
 
 
The first floating point operation hole in the instruction set involves
The first floating point operation hole in the instruction set involves
setting the floating point bit in the CC register.  The next instruction
setting a proposed (but non-existent) floating point bit in the CC register.
will simply interpret its operands as floating point instructions.
The next instruction
 
would then simply interpret its operands as floating point instructions.
Not all instructions, however, have floating point equivalents.  Further, the
Not all instructions, however, have floating point equivalents.  Further, the
immediate fields do not apply in floating point mode, and must be set to
immediate fields do not apply in floating point mode, and must be set to
zero.  Not all instructions make sense as floating point operations.
zero.  Not all instructions make sense as floating point operations.
Therefore, only the CMP, SUB, ADD, and MPY instructions may be issued as
Therefore, only the CMP, SUB, ADD, and MPY instructions may be issued as
floating point instructions.  Other instructions allow the examining of the
floating point instructions.  Other instructions allow the examining of the
floating point bit in the CC register.  In all cases, the floating point bit
floating point bit in the CC register.  In all cases, the floating point bit
is cleared one instruction after it is set.
is cleared one instruction after it is set.
 
 
The other possibility for floating point operations involves exploiting the
The other possibility for floating point operations involves exploiting the
hole in the instruction set that the NOOP and BREAK instructions reside within.
hole in the instruction set that the NOOP and BREAK instructions reside within.
These two instructions use 24--bits of address space.  A simple adjustment
These two instructions use 24--bits of address space, when only a single bit
to this space could create instructions with 4--bit register addresses for
is necessary.  A simple adjustment to this space could create instructions
each register, a 3--bit field for conditional execution, and a 2--bit field
with 4--bit register addresses for each register, a 3--bit field for
for which operation.  In this fashion, such a floating point capability would
conditional execution, and a 2--bit field for which operation.
only fill 13--bits of the 24--bit field, still leaving lots of room for
In this fashion, such a floating point capability would only fill 13--bits of
expansion.
the 24--bit field, still leaving lots of room for expansion.
 
 
In both cases, the Zip CPU would support 32--bit single precision floats
In both cases, the Zip CPU would support 32--bit single precision floats
only.
only, since other choices would complicate the pipeline.
 
 
The current architecture does not support a floating point not-implemented
The current architecture does not support a floating point not-implemented
interrupt.  Any soft floating point emulation must be done deliberately.
interrupt.  Any soft floating point emulation must be done deliberately.
 
 
\section{Native Instructions}
\section{Native Instructions}
The instruction set for the Zip CPU is summarized in
The instruction set for the Zip CPU is summarized in
Tbl.~\ref{tbl:zip-instructions}.
Tbl.~\ref{tbl:zip-instructions}.
\begin{table}\begin{center}
\begin{table}\begin{center}
\begin{tabular}{|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|c|}\hline
\begin{tabular}{|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|l|c|}\hline
 
\rowcolor[gray]{0.85}
Op Code & \multicolumn{8}{c|}{31\ldots24} & \multicolumn{8}{c|}{23\ldots 16}
Op Code & \multicolumn{8}{c|}{31\ldots24} & \multicolumn{8}{c|}{23\ldots 16}
        & \multicolumn{8}{c|}{15\ldots 8} & \multicolumn{8}{c|}{7\ldots 0}
        & \multicolumn{8}{c|}{15\ldots 8} & \multicolumn{8}{c|}{7\ldots 0}
        & Sets CC? \\\hline
        & Sets CC? \\\hline\hline
CMP(Sub) & \multicolumn{4}{l|}{4'h0}
CMP(Sub) & \multicolumn{4}{l|}{4'h0}
                & \multicolumn{4}{l|}{D. Reg}
                & \multicolumn{4}{l|}{D. Reg}
                & \multicolumn{3}{l|}{Cond.}
                & \multicolumn{3}{l|}{Cond.}
                & \multicolumn{21}{l|}{Operand B}
                & \multicolumn{21}{l|}{Operand B}
                & Yes \\\hline
                & Yes \\\hline
Line 562... Line 593...
                & \\\hline
                & \\\hline
BREAK & \multicolumn{4}{l|}{4'h4}
BREAK & \multicolumn{4}{l|}{4'h4}
                & \multicolumn{4}{l|}{4'he}
                & \multicolumn{4}{l|}{4'he}
                & \multicolumn{24}{l|}{24'h01}
                & \multicolumn{24}{l|}{24'h01}
                & \\\hline
                & \\\hline
{\em Rsrd} & \multicolumn{4}{l|}{4'h4}
{\em Reserved} & \multicolumn{4}{l|}{4'h4}
                & \multicolumn{4}{l|}{4'he}
                & \multicolumn{4}{l|}{4'he}
                & \multicolumn{24}{l|}{24'bits, but not 0 or 1.}
                & \multicolumn{24}{l|}{24'bits, but not 0 or 1.}
                & \\\hline
                & \\\hline
LODIHI & \multicolumn{4}{l|}{4'h4}
LODIHI & \multicolumn{4}{l|}{4'h4}
                & \multicolumn{4}{l|}{4'hf}
                & \multicolumn{4}{l|}{4'hf}
Line 665... Line 696...
\caption{Zip CPU Instruction Set}\label{tbl:zip-instructions}
\caption{Zip CPU Instruction Set}\label{tbl:zip-instructions}
\end{center}\end{table}
\end{center}\end{table}
 
 
As you can see, there's lots of room for instruction set expansion.  The
As you can see, there's lots of room for instruction set expansion.  The
NOOP and BREAK instructions are the only instructions within one particular
NOOP and BREAK instructions are the only instructions within one particular
24--bit hole.  This spaces are reserved for future enhancements.  For example,
24--bit hole.  The rest of this space is reserved for future enhancements.
floating point operations, consisting of a 3-bit floating point operation,
 
two 4-bit registers, no immediate offset, and a 3-bit condition would fit
 
nicely into 14--bits of this address space--making it so that the floating
 
point bit in the CC register need not be used.
 
 
 
\section{Derived Instructions}
\section{Derived Instructions}
The ZIP CPU supports many other common instructions, but not all of them
The Zip CPU supports many other common instructions, but not all of them
are single cycle instructions.  The derived instruction tables,
are single cycle instructions.  The derived instruction tables,
Tbls.~\ref{tbl:derived-1}, \ref{tbl:derived-2}, and~\ref{tbl:derived-3},
Tbls.~\ref{tbl:derived-1}, \ref{tbl:derived-2}, \ref{tbl:derived-3}
 
and~\ref{tbl:derived-4},
help to capture some of how these other instructions may be implemented on
help to capture some of how these other instructions may be implemented on
the ZIP CPU.  Many of these instructions will have assembly equivalents,
the Zip CPU.  Many of these instructions will have assembly equivalents,
such as the branch instructions, to facilitate working with the CPU.
such as the branch instructions, to facilitate working with the CPU.
\begin{table}\begin{center}
\begin{table}\begin{center}
\begin{tabular}{p{1.4in}p{1.5in}p{3in}}\\\hline
\begin{tabular}{p{1.4in}p{1.5in}p{3in}}\\\hline
Mapped & Actual  & Notes \\\hline
Mapped & Actual  & Notes \\\hline
 
ABS Rx
 
        & \parbox[t]{1.5in}{TST -1,Rx\\NEG.LT Rx}
 
        & Absolute value, depends upon derived NEG.\\\hline
\parbox[t]{1.4in}{ADD Ra,Rx\\ADDC Rb,Ry}
\parbox[t]{1.4in}{ADD Ra,Rx\\ADDC Rb,Ry}
        & \parbox[t]{1.5in}{Add Ra,Rx\\ADD.C \$1,Ry\\Add Rb,Ry}
        & \parbox[t]{1.5in}{Add Ra,Rx\\ADD.C \$1,Ry\\Add Rb,Ry}
        & Add with carry \\\hline
        & Add with carry \\\hline
BRA.Cond +/-\$Addr
BRA.Cond +/-\$Addr
        & \hbox{MOV.cond \$Addr+PC,PC}
        & \hbox{MOV.cond \$Addr+PC,PC}
Line 728... Line 759...
        MOV \$3+PC,R0 \\
        MOV \$3+PC,R0 \\
        STO R0,1(SP) \\
        STO R0,1(SP) \\
        MOV \$Addr+PC,PC \\
        MOV \$Addr+PC,PC \\
        ADD \$1,SP}
        ADD \$1,SP}
        & Jump to Subroutine. Note the required cleanup instruction after
        & Jump to Subroutine. Note the required cleanup instruction after
        returning. \\\hline
        returning.  This could easily be turned into a three instruction
 
        operand, removing the preliminary stack instruction before and
 
        the cleanup after, by adjusting how any stack frame was built for
 
        this routine to include space at the top of the stack for the PC.
 
        \\\hline
JSR PC+\$Addr
JSR PC+\$Addr
        & \parbox[t]{1.5in}{MOV \$3+PC,R12 \\ MOV \$addr+PC,PC}
        & \parbox[t]{1.5in}{MOV \$3+PC,R12 \\ MOV \$addr+PC,PC}
        &This is the high speed
        &This is the high speed
        version of a subroutine call, necessitating a register to hold the
        version of a subroutine call, necessitating a register to hold the
        last PC address.  In its favor, this method doesn't suffer the
        last PC address.  In its favor, this method doesn't suffer the
Line 787... Line 822...
        LDIHI.C \$8000h,Rz \\
        LDIHI.C \$8000h,Rz \\
        LSR \$1,Rx \\
        LSR \$1,Rx \\
        OR Rz,Rx}
        OR Rz,Rx}
        & Logical shift right with carry \\\hline
        & Logical shift right with carry \\\hline
NEG Rx & \parbox[t]{1.5in}{XOR \$-1,Rx \\ ADD \$1,Rx} & \\\hline
NEG Rx & \parbox[t]{1.5in}{XOR \$-1,Rx \\ ADD \$1,Rx} & \\\hline
 
NEG.C Rx & \parbox[t]{1.5in}{MOV.C \$-1+Rx,Rx\\XOR.C \$-1,Rx} & \\\hline
NOOP & NOOP & While there are many
NOOP & NOOP & While there are many
        operations that do nothing, such as MOV Rx,Rx, or OR \$0,Rx, these
        operations that do nothing, such as MOV Rx,Rx, or OR \$0,Rx, these
        operations have consequences in that they might stall the bus if
        operations have consequences in that they might stall the bus if
        Rx isn't ready yet.  For this reason, we have a dedicated NOOP
        Rx isn't ready yet.  For this reason, we have a dedicated NOOP
        instruction. \\\hline
        instruction. \\\hline
Line 800... Line 836...
        & Note
        & Note
        that for interrupt purposes, one can never depend upon the value at
        that for interrupt purposes, one can never depend upon the value at
        (SP).  Hence you read from it, then increment it, lest having
        (SP).  Hence you read from it, then increment it, lest having
        incremented it first something then comes along and writes to that
        incremented it first something then comes along and writes to that
        value before you can read the result. \\\hline
        value before you can read the result. \\\hline
 
\end{tabular}
 
\caption{Derived Instructions, continued}\label{tbl:derived-2}
 
\end{center}\end{table}
 
\begin{table}\begin{center}
 
\begin{tabular}{p{1.4in}p{1.5in}p{3in}}\\\hline
PUSH Rx
PUSH Rx
        & \parbox[t]{1.5in}{SUB \$1,SP \\
        & \parbox[t]{1.5in}{SUB \$1,SP \\
        STO Rx,\$1(SP)}
        STO Rx,\$1(SP)}
        & \\\hline
        & \\\hline
 
PUSH Rx-Ry
 
        & \parbox[t]{1.5in}{SUB \$n,SP \\
 
        STO Rx,\$n(SP)
 
        \ldots \\
 
        STO Ry,\$1(SP)}
 
        & Multiple pushes at once only need the single subtract from the
 
        stack pointer.  This derived instruction is analogous to a similar one
 
        on the Motoroloa 68k architecture, although the Zip Assembler
 
        does not support this instruction (yet).\\\hline
RESET
RESET
        & \parbox[t]{1in}{STO \$1,\$watchdog(R12)\\NOOP\\NOOP}
        & \parbox[t]{1in}{STO \$1,\$watchdog(R12)\\NOOP\\NOOP}
        & \parbox[t]{3in}{This depends upon the peripheral base address being
        & \parbox[t]{3in}{This depends upon the peripheral base address being
        in R12.
        in R12.
 
 
        Another opportunity might be to jump to the reset address from within
        Another opportunity might be to jump to the reset address from within
        supervisor mode.}\\\hline
        supervisor mode.}\\\hline
RET & \parbox[t]{1.5in}{LOD \$-1(SP),PC}
RET & \parbox[t]{1.5in}{LOD \$1(SP),PC}
        & Note that this depends upon the calling context to clean up the
        & Note that this depends upon the calling context to clean up the
        stack, as outlined for the JSR instruction.  \\\hline
        stack, as outlined for the JSR instruction.  \\\hline
\end{tabular}
 
\caption{Derived Instructions, continued}\label{tbl:derived-2}
 
\end{center}\end{table}
 
\begin{table}\begin{center}
 
\begin{tabular}{p{1.4in}p{1.5in}p{3in}}\\\hline
 
RET & MOV R12,PC
RET & MOV R12,PC
        & This is the high(er) speed version, that doesn't touch the stack.
        & This is the high(er) speed version, that doesn't touch the stack.
        As such, it doesn't suffer a stall on memory read/write to the stack.
        As such, it doesn't suffer a stall on memory read/write to the stack.
        \\\hline
        \\\hline
STEP Rr,Rt
STEP Rr,Rt
Line 857... Line 902...
        XOR Rx,Ry \\
        XOR Rx,Ry \\
        XOR Ry,Rx}
        XOR Ry,Rx}
        & While no extra registers are needed, this example
        & While no extra registers are needed, this example
        does take 3-clocks. \\\hline
        does take 3-clocks. \\\hline
TRAP \#X
TRAP \#X
        & LDILO \$x,CC
        & \parbox[t]{1.5in}{LDI \$x,R0 \\ AND ~\$GIE,CC }
        & This approach uses the unused bits of the CC register as a TRAP
        & This works because whenever a user lowers the \$GIE flag, it sets
        address.  The user will need to make certain
        a TRAP bit within the CC register.  Therefore, upon entering the
        that the SLEEP and GIE bits are not set in \$x.  LDI would also work,
        supervisor state, the CPU only need check this bit to know that it
        however using LDILO permits the use of conditional traps.  (i.e.,
        got there via a TRAP.  The trap could be made conditional by making
        trap if the zero flag is set.)  Should you wish to trap off of a
        the LDI and the AND conditional.  In that case, the assembler would
        register value, you could equivalently load \$x into the register and
        quietly turn the LDI instruction into an LDILO and LDIHI pair,
        then MOV it into the CC register. \\\hline
        but the effectt would be the same. \\\hline
 
\end{tabular}
 
\caption{Derived Instructions, continued}\label{tbl:derived-3}
 
\end{center}\end{table}
 
\begin{table}\begin{center}
 
\begin{tabular}{p{1.4in}p{1.5in}p{3in}}\\\hline
TST Rx
TST Rx
        & TST \$-1,Rx
        & TST \$-1,Rx
        & Set the condition codes based upon Rx.  Could also do a CMP \$0,Rx,
        & Set the condition codes based upon Rx.  Could also do a CMP \$0,Rx,
        ADD \$0,Rx, SUB \$0,Rx, etc, AND \$-1,Rx, etc.  The TST and CMP
        ADD \$0,Rx, SUB \$0,Rx, etc, AND \$-1,Rx, etc.  The TST and CMP
        approaches won't stall future pipeline stages looking for the value
        approaches won't stall future pipeline stages looking for the value
Line 876... Line 926...
WAIT
WAIT
        & Or \$SLEEP,CC
        & Or \$SLEEP,CC
        & Wait 'til interrupt.  In an interrupts disabled context, this
        & Wait 'til interrupt.  In an interrupts disabled context, this
        becomes a HALT instruction.
        becomes a HALT instruction.
\end{tabular}
\end{tabular}
\caption{Derived Instructions, continued}\label{tbl:derived-3}
\caption{Derived Instructions, continued}\label{tbl:derived-4}
\end{center}\end{table}
\end{center}\end{table}
\iffalse
 
\fi
 
\section{Pipeline Stages}
\section{Pipeline Stages}
As mentioned in the introduction, and highlighted in Fig.~\ref{fig:cpu},
As mentioned in the introduction, and highlighted in Fig.~\ref{fig:cpu},
the Zip CPU supports a five stage pipeline.
the Zip CPU supports a five stage pipeline.
\begin{enumerate}
\begin{enumerate}
\item {\bf Prefetch}: Read instruction from memory (cache if possible).  This
\item {\bf Prefetch}: Reads instruction from memory and into a cache, if so
 
        configured.  This
        stage is actually pipelined itself, and so it will stall if the PC
        stage is actually pipelined itself, and so it will stall if the PC
        ever changes.  Stalls are also created here if the instruction isn't
        ever changes.  Stalls are also created here if the instruction isn't
        in the prefetch cache.
        in the prefetch cache.
\item {\bf Decode}: Decode instruction into op code, register(s) to read, and
 
        immediate offset.  This stage also determines whether the flags will
        The Zip CPU supports one of two prefetch methods, depending upon a flag
 
        set at build time within the {\tt zipcpu.v} file.  The simplest is a
 
        non--cached implementation of a prefetch.  This implementation is
 
        fairly small, and ideal for
 
        users of the Zip CPU who need the extra space on the FPGA fabric.
 
        However, because this non--cached version has no cache, the maximum
 
        number of instructions per clock is limited to about one per five.
 
 
 
        The second prefetch module is a pipelined prefetch with a cache.  This
 
        module tries to keep the instruction address within a window of valid
 
        instruction addresses.  While effective, it is not a traditional
 
        cache implementation.  One unique feature of this cache implementation,
 
        however, is that it can be cleared in a single clock.  A disappointing
 
        feature, though, was that it needs an extra internal pipeline stage
 
        to be implemented.
 
 
 
\item {\bf Decode}: Decodes an instruction into op code, register(s) to read,
 
        and immediate offset.  This stage also determines whether the flags will
        be set or whether the result will be written back.
        be set or whether the result will be written back.
\item {\bf Read Operands}: Read registers and apply any immediate values to
\item {\bf Read Operands}: Read registers and apply any immediate values to
        them.  There is no means of detecting or flagging arithmetic overflow
        them.  There is no means of detecting or flagging arithmetic overflow
        or carry when adding the immediate to the operand.  This stage will
        or carry when adding the immediate to the operand.  This stage will
        stall if any source operand is pending.
        stall if any source operand is pending.
\item Split into two tracks: An {\bf ALU} which will accomplish a simple
\item Split into two tracks: An {\bf ALU} which will accomplish a simple
        instruction, and the {\bf MemOps} stage which accomplishes memory
        instruction, and the {\bf MemOps} stage which handles {\tt LOD} (load)
        read/write.
        and {\tt STO} (store) instructions.
        \begin{itemize}
        \begin{itemize}
        \item Loads stall instructions that access the register until it is
        \item Loads will stall the entire pipeline until complete.
                written to the register set.
        \item Condition codes are available upon completion of the ALU stage
        \item Condition codes are available upon completion
        \item Issuing an instruction to the memory unit while the memory unit
        \item Issuing an instruction to the memory while the memory is busy will
                is busy will stall the entire pipeline.  If the bus deadlocks,
                stall the entire pipeline.  If the bus deadlocks, only a reset
                only a reset will release the CPU.  (Watchdog timer, anyone?)
                will release the CPU.  (Watchdog timer, anyone?)
 
        \item The Zip CPU currently has no means of reading and acting on any
        \item The Zip CPU currently has no means of reading and acting on any
        error conditions on the bus.
        error conditions on the bus.
        \end{itemize}
        \end{itemize}
\item {\bf Write-Back}: Conditionally write back the result to the register
\item {\bf Write-Back}: Conditionally write back the result to the register
        set, applying the condition.  This routine is bi-re-entrant: either the
        set, applying the condition.  This routine is bi-entrant: either the
        memory or the simple instruction may request a register write.
        memory or the simple instruction may request a register write.
\end{enumerate}
\end{enumerate}
 
 
The Zip CPU does not support out of order execution.  Therefore, if the memory
The Zip CPU does not support out of order execution.  Therefore, if the memory
unit stalls, every other instruction stalls.  Memory stores, however, can take
unit stalls, every other instruction stalls.  Memory stores, however, can take
place concurrently with ALU operations, although memory reads cannot.
place concurrently with ALU operations, although memory reads (loads) cannot.
 
 
\iffalse
 
 
 
\section{Pipeline Logic}
 
How the CPU handles some instruction combinations can be telling when
 
determining what happens in the pipeline.  The following lists some examples:
 
\begin{itemize}
 
\item {\bf Delayed Branching}
 
 
 
        I had originally hoped to implement delayed branching.   My goal
 
        was that the compiler would handle any pipeline stall conditions so
 
        that the pipeline logic could be simpler within the CPU.  I ran into
 
        two problems with this.
 
 
 
        The first problem has to deal with debug mode.  When the debugger
 
        single steps an instruction, that instruction goes to completion.
 
        This means that if the instruction moves a value to the PC register,
 
        the PC register would now contain that value, indicating that the
 
        next instruction would be on the other side of the branch.  There's
 
        just no easy way around this: the entire CPU state must be captured
 
        by the registers, to include the program counter.  What value should
 
        the program counter be equal to?  The branch?  Fine.  The address
 
        you are branching to?  Fine.  The address of the delay slot?  Problem.
 
 
 
        The second problem with delayed branching is the idea of suspending
 
        processing for an interrupt.  Which address should the CPU return
 
        to upon completing the interrupt processing?  The branch?  Good.  The
 
        address after the branch?  Also good.  The address of the delay slot?
 
        Not so good.
 
 
 
        If you then add into this mess the idea that, if the CPU is running
 
        from a really slow memory such as the flash, the delay slot may never
 
        be filled before the branch is determined, then this makes even less
 
        sense.
 
 
 
        For all of these reasons, this CPU does not support delayed branching.
 
 
 
\item {\bf Register Result:} {\tt MOV R0,R1; MOV R1,R2 }
 
 
 
        What value does R2 get, the value of R1 before the first move or the
 
        value of R0?  The Zip CPU has been optimized so that neither of these
 
        instructions require a pipeline stall--unless an immediate were to
 
        be added to R1 in the second instruction.
 
 
 
        The ZIP CPU architecture requires that R2 must equal R0 at the end of
 
        this operation.  Even better, such combinations do not (normally)
 
        stall the pipeline.
 
 
 
\item {\bf Condition Codes Result:} {\tt CMP R0,R1;} {\tt MOV.EQ \$x,PC}
 
 
 
        At issue is the same item as above, save that the CMP instruction
 
        updates the flags that the MOV instruction depends upon.
 
 
 
        The Zip CPU architecture requires that condition codes must be updated
 
        and available immediately for the next instruction without stalling the
 
        pipeline.
 
 
 
\item {\bf Condition Codes Register Result:} {\tt CMP R0,R1; MOV CC,R2}
 
 
 
        At issue is the
 
        fact that the logic supporting the CC register is more complicated than
 
        the logic supporting any other register.
 
 
 
        The ZIP CPU will stall for a cycle cycle on this instruction.
 
\item {\bf Condition Codes Register Operand:} {\tt MOV R0,R1; MOV CC,R2}
 
 
 
        Unlike the previous case, this move prior to reading the {\tt CC}
 
        register does not impact the {\tt CC} register.  Therefore, this
 
        does not stall the bus, whereas the previous one would.
 
\end{itemize}
 
 
 
As I've studied  this, I find several approaches to handling pipeline
 
        issues.  These approaches (and their consequences) are listed below.
 
 
 
\begin{itemize}
 
\item {\bf All issued instructions complete, stages stall individually}
 
 
 
        What about a slow pre-fetch?
 
 
 
        Nominally, this works well: any issued instruction
 
        just runs to completion.  If there are four issued instructions in the
 
        pipeline, with the writeback instruction being a write-to-PC
 
        instruction, the other three instructions naturally finish.
 
 
 
        This approach fails when reading instructions from the flash,
 
        since such reads require N clocks to clocks to complete.  Thus
 
        there may be only one instruction in the pipeline if reading from flash,
 
        or a full pipeline if reading from cache.  Each of these approaches
 
        would produce a different response.
 
 
 
        For this reason, the Zip CPU works off of a different basis: All
 
        instructions that enter either the ALU or the memory unit will
 
        complete.  Stages still stall individually.
 
 
 
\item {\bf Issued instructions may be canceled}
 
 
 
        The problem here is that
 
        memory operations cannot be canceled: even reads may have side effects
 
        on peripherals that cannot be canceled later.  Further, in the case of
 
        an interrupt, it's difficult to know what to cancel.  What happens in
 
        a \hbox{\tt MOV.C \$x,PC} followed by a \hbox{\tt MOV \$y,PC}
 
        instruction?  Which get canceled?
 
 
 
        Because it isn't clear what would need to be canceled, the Zip CPU
 
        will not permit this combination.  A MOV to the PC register will be
 
        followed by a stall, and possibly many stalls, so that the second
 
        move to PC will never be executed.
 
 
 
\item {\bf All issued instructions complete.}
 
 
 
        In this example, we try all issued instructions complete, but the
 
        entire pipeline stalls if one stage is not filled.  In this approach,
 
        though, we again struggle with the problems associated with
 
        delayed branching.  Upon attempting to restart the processor, where
 
        do you restart it from?
 
 
 
\item {\bf Memory instructions must complete}
 
 
 
        All instructions that enter into the memory module {\em must}
 
        complete.  Issued instructions from the prefetch, decode, or operand
 
        read stages may or may not complete.  Jumps into code must be valid,
 
        so that interrupt returns may be valid.  All instructions entering the
 
        ALU complete.
 
 
 
        This looks to be the simplest approach.
 
        While the logic may be difficult, this appears to be the only
 
        re-entrant approach.
 
 
 
        A {\tt new\_pc} flag will be high anytime the PC changes in an
 
        unpredictable way (i.e., it doesn't increment).  This includes jumps
 
        as well as interrupts and interrupt returns.  Whenever this flag may
 
        go high, memory operations and ALU operations will stall until the
 
        result is known.  When the flag does go high, anything in the prefetch,
 
        decode, and read-op stage will be invalidated.
 
 
 
\end{itemize}
 
\fi
 
 
 
\section{Pipeline Stalls}
\section{Pipeline Stalls}
The processing pipeline can and will stall for a variety of reasons.  Some of
The processing pipeline can and will stall for a variety of reasons.  Some of
these are obvious, some less so.  These reasons are listed below:
these are obvious, some less so.  These reasons are listed below:
\begin{itemize}
\begin{itemize}
\item When the prefetch cache is exhausted
\item When the prefetch cache is exhausted
 
 
This should be obvious.  If the prefetch cache doesn't have the instruction
This reason should be obvious.  If the prefetch cache doesn't have the
in memory, the entire pipeline must stall until enough of the prefetch cache
instruction in memory, the entire pipeline must stall until enough of the
is loaded to support the next instruction.
prefetch cache is loaded to support the next instruction.
 
 
\item While waiting for the pipeline to load following any taken branch, jump,
\item While waiting for the pipeline to load following any taken branch, jump,
        return from interrupt or switch to interrupt context (6 clocks)
        return from interrupt or switch to interrupt context (5 stall cycles)
 
 
If the PC suddenly changes, the pipeline is subsequently cleared and needs to
If the PC suddenly changes, the pipeline is subsequently cleared and needs to
be reloaded.  Given that there are five stages to the pipeline, that accounts
be reloaded.  Given that there are five stages to the pipeline, that accounts
for five of the six delay clocks.  The last clock is lost in the prefetch
for four of the five stalls.  The stall cycle is lost in the pipelined prefetch
stage which needs at least one clock with a valid PC before it can produce
stage which needs at least one clock with a valid PC before it can produce
a new output.  Hence, six clocks will always be lost anytime the pipeline needs
a new output.
to be cleared.
 
 
The Zip CPU handles {\tt MOV \$X(PC),PC}, {\tt ADD \$X,PC}, and
 
{\tt LDI \$X,PC} instructions specially, however.  These instructions, when
 
not conditioned on the flags, can execute with only 3~stall cycles.
 
 
\item When reading from a prior register while also adding an immediate offset
\item When reading from a prior register while also adding an immediate offset
\begin{enumerate}
\begin{enumerate}
\item\ {\tt OPCODE ?,RA}
\item\ {\tt OPCODE ?,RA}
\item\ {\em (stall)}
\item\ {\em (stall)}
Line 1086... Line 1017...
Since the addition of the immediate register within OpB decoding gets applied
Since the addition of the immediate register within OpB decoding gets applied
during the read operand stage so that it can be nicely settled before the ALU,
during the read operand stage so that it can be nicely settled before the ALU,
any instruction that will write back an operand must be separated from the
any instruction that will write back an operand must be separated from the
opcode that will read and apply an immediate offset by one instruction.  The
opcode that will read and apply an immediate offset by one instruction.  The
good news is that this stall can easily be mitigated by proper scheduling.
good news is that this stall can easily be mitigated by proper scheduling.
 
That is, any instruction that does not add an immediate to {\tt RA} may be
 
scheduled into the stall slot.
 
 
\item When writing to the CC or PC Register
\item When any write to either the CC or PC Register is followed by a memory
 
        operation
\begin{enumerate}
\begin{enumerate}
\item\ {\tt OPCODE RA,PC} {\em Ex: a branch opcode}
\item\ {\tt OPCODE RA,PC} {\em Ex: a branch opcode}
\item\ {\em (stall, even if jump not taken)}
\item\ {\em (stall, even if jump not taken)}
\item\ {\tt OPCODE RA,RB}
\item\ {\tt LOD \$X(RA),RB}
\end{enumerate}
\end{enumerate}
Since branches take place in the writeback stage, the Zip CPU will stall the
Since branches take place in the writeback stage, the Zip CPU will stall the
pipeline for one clock anytime there may be a possible jump.  This prevents
pipeline for one clock anytime there may be a possible jump.  This prevents
an instruction from executing a memory access after the jump but before the
an instruction from executing a memory access after the jump but before the
jump is recognized.
jump is recognized.
 
 
This stall cannot be mitigated through scheduling.
This stall may be mitigated by shuffling the operations immediately following
 
a potential branch so that an ALU operation follows the branch instead of a
 
memory operation.
 
 
\item When reading from the CC register after setting the flags
\item When reading from the CC register after setting the flags
\begin{enumerate}
\begin{enumerate}
\item\ {\tt ALUOP RA,RB}
\item\ {\tt ALUOP RA,RB} {\em Ex: a compare opcode}
\item\ {\em (stall}
\item\ {\em (stall)}
\item\ {\tt TST sys.ccv,CC}
\item\ {\tt TST sys.ccv,CC}
\item\ {\tt BZ somewhere}
\item\ {\tt BZ somewhere}
\end{enumerate}
\end{enumerate}
 
 
The reason for this stall is simply performance.  Many of the flags are
The reason for this stall is simply performance.  Many of the flags are
Line 1120... Line 1056...
This stall may be eliminated via proper scheduling, by placing an instruction
This stall may be eliminated via proper scheduling, by placing an instruction
that does not set flags in between the ALU operation and the instruction
that does not set flags in between the ALU operation and the instruction
that references the CC register.  For example, {\tt MOV \$addr+PC,uPC}
that references the CC register.  For example, {\tt MOV \$addr+PC,uPC}
followed by an {\tt RTU} ({\tt OR \$GIE,CC}) instruction will not incur
followed by an {\tt RTU} ({\tt OR \$GIE,CC}) instruction will not incur
this stall, whereas an {\tt OR \$BREAKEN,CC} followed by an {\tt OR \$STEP,CC}
this stall, whereas an {\tt OR \$BREAKEN,CC} followed by an {\tt OR \$STEP,CC}
will incur the stall.
will incur the stall, while a {\tt LDI \$BREAKEN|\$STEP,CC} will not.
 
 
\item When waiting for a memory read operation to complete
\item When waiting for a memory read operation to complete
\begin{enumerate}
\begin{enumerate}
\item\ {\tt LOD address,RA}
\item\ {\tt LOD address,RA}
\item\ {\em (multiple stalls, bus dependent, 7 clocks best)}
\item\ {\em (multiple stalls, bus dependent, 4 clocks best)}
\item\ {\tt OPCODE I+RA,RB}
\item\ {\tt OPCODE I+RA,RB}
\end{enumerate}
\end{enumerate}
 
 
Remember, the ZIP CPU does not support out of order execution.  Therefore,
Remember, the Zip CPU does not support out of order execution.  Therefore,
anytime the memory unit becomes busy both the memory unit and the ALU must
anytime the memory unit becomes busy both the memory unit and the ALU must
stall until the memory unit is cleared.  This is especially true of a load
stall until the memory unit is cleared.  This is especially true of a load
instruction, which must still write its operand back to the register file.
instruction, which must still write its operand back to the register file.
Store instructions are different, since they can be busy with no impact on
Store instructions are different, since they can be busy with no impact on
later ALU write back operations.  Hence, only loads stall the pipeline.
later ALU write back operations.  Hence, only loads stall the pipeline.
Line 1144... Line 1080...
will be busy, and unable to do anything else.
will be busy, and unable to do anything else.
 
 
\item Memory operation followed by a memory operation
\item Memory operation followed by a memory operation
\begin{enumerate}
\begin{enumerate}
\item\ {\tt STO address,RA}
\item\ {\tt STO address,RA}
\item\ {\em (multiple stalls, bus dependent, 7 clocks best)}
\item\ {\em (multiple stalls, bus dependent, 4 clocks best)}
\item\ {\tt LOD address,RB}
\item\ {\tt LOD address,RB}
\item\ {\em (multiple stalls, bus dependent, 7 clocks best)}
\item\ {\em (multiple stalls, bus dependent, 4 clocks best)}
\end{enumerate}
\end{enumerate}
 
 
In this case, the LOD instruction cannot start until the STALL is finished.
In this case, the LOD instruction cannot start until the STO is finished.
With proper scheduling, it is possible to do something in the ALU while the
With proper scheduling, it is possible to do something in the ALU while the
STO is busy, but otherwise this pipeline will stall waiting for it to complete.
memory unit is busy with the STO instruction, but otherwise this pipeline will
 
stall waiting for it to complete.
 
 
Note that even though the Wishbone bus can support pipelined accesses at
Note that even though the Wishbone bus can support pipelined accesses at
one access per clock, only the prefetch stage can take advantage of this.
one access per clock, only the prefetch stage can take advantage of this.
Load and Store instructions are stuck at one wishbone cycle per instruction.
Load and Store instructions are stuck at one wishbone cycle per instruction.
 
 
 
\item When waiting for a conditional memory read operation to complete
 
\begin{enumerate}
 
\item\ {\tt LOD.Z address,RA}
 
\item\ {\em (multiple stalls, bus dependent, 7 clocks best)}
 
\item\ {\tt OPCODE I+RA,RB}
 
\end{enumerate}
 
 
 
In this case, the Zip CPU doesn't warn the prefetch cache to get off the bus
 
two cycles before using the bus, so there's a potential for an extra three
 
cycle cost due to bus contention between the prefetch and the CPU.
 
 
 
This is true for both the LOD and the STO instructions, with the exception that
 
the STO instruction will continue in parallel with any ALU instructions that
 
follow it.
 
 
\end{itemize}
\end{itemize}
 
 
 
 
\chapter{Peripherals}\label{chap:periph}
\chapter{Peripherals}\label{chap:periph}
 
 
Line 1277... Line 1230...
O/S must also keep track of values written to the Jiffies register.  Thus,
O/S must also keep track of values written to the Jiffies register.  Thus,
when an `alarm' trips, it should be removed from the list of alarms, the list
when an `alarm' trips, it should be removed from the list of alarms, the list
should be sorted, and the next alarm in terms of Jiffies should be written
should be sorted, and the next alarm in terms of Jiffies should be written
to the register.
to the register.
 
 
\section{Manual Cache}
\section{Direct Memory Access Controller}
 
 
 
The Direct Memory Access (DMA) controller can be used to either move memory
 
from one location to another, to read from a peripheral into memory, or to
 
write from a peripheral into memory all without CPU intervention.  Further,
 
since the DMA controller can issue (and does issue) pipeline wishbone accesses,
 
any DMA memory move will by nature be faster than a corresponding program
 
accomplishing the same move.  To put this to numbers, it may take a program
 
18~clocks per word transferred, whereas this DMA controller can move one
 
word in two clocks--provided it has bus access.  (The CPU gets priority over the
 
bus.)
 
 
 
When copying memory from one location to another, the DMA controller will
 
copy in units of a given transfer length--up to 1024 words at a time.  It will
 
read that transfer length into its internal buffer, and then write to the
 
destination address from that buffer.  If the CPU interrupts a DMA transfer,
 
it will release the bus, let the CPU complete whatever it needs to do, and then
 
restart its transfer by writing the contents of its internal buffer and then
 
re-entering its read cycle again.
 
 
 
When coupled with a peripheral, the DMA controller can be configured to start
 
a memory copy on an interrupt line going high.  Further, the controller can be
 
configured to issue reads from (or two) the same address instead of incrementing
 
the address at each clock.  The DMA completes once the total number of items
 
specified (not the transfer length) have been transferred.
 
 
 
In each case, once the transfer is complete and the DMA unit returns to
 
idle, the DMA will issue an interrupt.
 
 
The manual cache is an experimental setting that may not remain with the Zip
 
CPU for very long.  It is designed to facilitate running from FLASH or ROM
 
memory, although the pipeline prefetch cache really makes this need obsolete.
 
The manual
 
cache works by copying data from a wishbone address (range) into the cache
 
register, and then by making that memory available as memory to the Zip System.
 
It is a {\em manual cache} because the processor must first specify what
 
memory to copy, and then once copied the processor can only access the cache
 
memory by the cache memory location.  There is no transparency.  It is perhaps
 
best described as a combination DMA controller and local memory.
 
 
 
Worse, this cache is likely going to be removed from the ZipSystem.  Having used
 
the ZipSystem now for some time, I have yet to find a need or use for the manual
 
cache.  I will likely replace this peripheral with a proper DMA controller.
 
 
 
\chapter{Operation}\label{chap:ops}
\chapter{Operation}\label{chap:ops}
 
 
The Zip CPU, and even the Zip System, is not a System on a Chip (SoC).  It
The Zip CPU, and even the Zip System, is not a System on a Chip (SoC).  It
needs to be connected to its operational environment in order to be used.
needs to be connected to its operational environment in order to be used.
Line 1320... Line 1286...
\end{enumerate}
\end{enumerate}
If you have enabled your CPU to start automatically, then upon power up the
If you have enabled your CPU to start automatically, then upon power up the
CPU will immediately start executing your instructions.
CPU will immediately start executing your instructions.
 
 
This is, however, not how I have used the Zip CPU.  I have instead used the
This is, however, not how I have used the Zip CPU.  I have instead used the
ZIP CPU in a more controlled environment.  For me, the CPU starts in a
Zip CPU in a more controlled environment.  For me, the CPU starts in a
halted state, and waits to be told to start.  Further, the RESET address is a
halted state, and waits to be told to start.  Further, the RESET address is a
location in RAM.  After bringing up the board I am using, and further the
location in RAM.  After bringing up the board I am using, and further the
bus that is on it, the RAM memory is then loaded externally with the program
bus that is on it, the RAM memory is then loaded externally with the program
I wish the Zip System to run.  Once the RAM is loaded, I release the CPU.
I wish the Zip System to run.  Once the RAM is loaded, I release the CPU.
The CPU then runs until its halt condition, at which point its task is
The CPU then runs until its halt condition, at which point its task is
complete.
complete.
 
 
Eventually, I intend to place an operating system onto the ZipSystem, I'm
Eventually, I intend to place an operating system onto the ZipSystem, I'm
just not there yet.
just not there yet.
 
 
 
The rest of this chapter examines some common programming constructs, and
 
how they might be applied to the Zip System.
 
 
 
\section{Example: Idle Task}
 
One task every operating system needs is the idle task, the task that takes
 
place when nothing else can run.  On the Zip CPU, this task is quite simple,
 
and it is shown in assemble in Tbl.~\ref{tbl:idle-asm}.
 
\begin{table}\begin{center}
 
\begin{tabular}{ll}
 
{\tt idle\_task:} \\
 
&        {\em ; Wait for the next interrupt, then switch to supervisor task} \\
 
&        {\tt WAIT} \\
 
&        {\em ; When we come back, it's because the supervisor wishes to} \\
 
&        {\em ; wait for an interrupt again, so go back to the top.} \\
 
&        {\tt BRA idle\_task} \\
 
\end{tabular}
 
\caption{Example Idle Loop}\label{tbl:idle-asm}
 
\end{center}\end{table}
 
When this task runs, the CPU will fill up all of the pipeline stages up the
 
ALU.  The {\tt WAIT} instruction, upon leaving the ALU, places the CPU into
 
a sleep state where nothing more moves.  Sure, there may be some more settling,
 
the pipe cache continue to read until full, other instructions may issue until
 
the pipeline fills, but then everything will stall.  Then, once an interrupt
 
takes place, control passes to the supervisor task to handle the interrupt.
 
When control passes back to this task, it will be on the next instruction.
 
Since that next instruction sends us back to the top of the task, the idle
 
task thus does nothing but wait for an interrupt.
 
 
 
This should be the lowest priority task, the task that runs when nothing else
 
can.  It will help lower the FPGA power usage overall---at least its dynamic
 
power usage.
 
 
 
\section{Example: Memory Copy}
 
One common operation is that of a memory move or copy.  Consider the C code
 
shown in Tbl.~\ref{tbl:memcp-c}.
 
\begin{table}\begin{center}
 
\parbox{4in}{\begin{tabbing}
 
{\tt void} \= {\tt memcp(void *dest, void *src, int len) \{} \\
 
        \> {\tt for(int i=0; i<len; i++)} \\
 
        \> \hspace{0.2in} {\tt *dest++ = *src++;} \\
 
\}
 
\end{tabbing}}
 
\caption{Example Memory Copy code in C}\label{tbl:memcp-c}
 
\end{center}\end{table}
 
This same code can be translated in Zip Assembly as shown in
 
Tbl.~\ref{tbl:memcp-asm}.
 
\begin{table}\begin{center}
 
\begin{tabular}{ll}
 
memcp: \\
 
&        {\em ; R0 = *dest, R1 = *src, R2 = LEN} \\
 
&        {\em ; The following will operate in 17 clocks per word minus one clock} \\
 
&        {\tt CMP 0,R2} \\
 
&        {\tt LOD.Z -1(SP),PC} {\em ; A conditional return }\\
 
&        {\em ; (One stall on potentially writing to PC)} \\
 
&        {\tt LOD (R1),R3} \\
 
&        {\em ; (4 stalls, cannot be scheduled away)} \\
 
&        {\tt STO R3,(R2)} {\em ; (4 schedulable stalls, has no impact now)} \\
 
&        {\tt ADD 1,R1} \\
 
&        {\tt SUB 1,R2} \\
 
&        {\tt BNZ loop} \\
 
&        {\em ; (5 stalls, if branch taken, to clear and refill the pipeline)} \\
 
&        {\tt RET} \\
 
\end{tabular}
 
\caption{Example Memory Copy code in Zip Assembly}\label{tbl:memcp-asm}
 
\end{center}\end{table}
 
This example points out several things associated with the Zip CPU.  First,
 
a straightforward implementation of a for loop is not the fastest loop
 
structure.  For this reason, we have placed the test to continue at the
 
end.  Second, all pointers are {\tt void} pointers to arbitrary 32--bit
 
data types.  The Zip CPU does not have explicit support for smaller or larger
 
data types, and so this memory copy cannot be applied at a byte level.
 
Third, we've optimized the conditional jump to a return instruction into a
 
conditional return instruction.
 
 
 
\section{Context Switch}
 
 
 
Fundamental to any multiprocessing system is the ability to switch from one
 
task to the next.  In the ZipSystem, this is accomplished in one of a couple
 
ways.  The first step is that an interrupt happens.  Anytime an interrupt
 
happens, the CPU needs to execute the following tasks in supervisor mode:
 
\begin{enumerate}
 
\item Check for a trap instruction.  That  is, if the user task requested a
 
        trap, we may not wish to adjust the context, check interrupts, or call
 
        the scheduler.  Tbl.~\ref{tbl:trap-check}
 
\begin{table}\begin{center}
 
\begin{tabular}{ll}
 
{\tt return\_to\_user:} \\
 
&       {\em; The instruction before the context switch processing must} \\
 
&       {\em; be the RTU instruction that enacted user mode in the first} \\
 
&       {\em; place.  We show it here just for reference.} \\
 
&       {\tt RTU} \\
 
{\tt trap\_check:} \\
 
&       {\tt MOV uCC,R0} \\
 
&       {\tt TST \$TRAP,R0} \\
 
&       {\tt BNZ swap\_out} \\
 
&       {; \em Do something here to execute the trap} \\
 
&       {; \em Don't need to call the scheduler, so we can just return} \\
 
&       {\tt BRA return\_to\_user} \\
 
\end{tabular}
 
\caption{Checking for whether the user issued a TRAP instruction}\label{tbl:trap-check}
 
\end{center}\end{table}
 
        shows the rudiments of this code, while showing nothing of how the
 
        actual trap would be implemented.
 
 
 
You may also wish to note that the instruction before the first instruction
 
in our context swap {\em must be} a return to userspace instruction.
 
Remember, the supervisor process is re--entered where it left off.  This is
 
different from many other processors that enter interrupt mode at some vector
 
or other.  In this case, we always enter supervisor mode right where we last
 
left.\footnote{The one exception to this rule is upon reset where supervisor
 
mode is entered at a pre--programmed wishbone memory address.}
 
 
 
\item Capture user counters.  If the operating system is keeping track of
 
        system usage via the accounting counters, those counters need to be
 
        copied and accumulated into some master counter at this point.
 
 
 
\item Preserve the old context.  This involves pushing all the user registers
 
        onto the user stack and then copying the resulting stack address
 
        into the tasks task structure, as shown in Tbl.~\ref{tbl:context-out}.
 
\begin{table}\begin{center}
 
\begin{tabular}{ll}
 
{\tt swap\_out:} \\
 
&        {\tt MOV -15(uSP),R1} \\
 
&        {\tt STO R1,stack(R12)} \\
 
&        {\tt MOV uPC,R0} \\
 
&        {\tt STO R0,15(R1)} \\
 
&        {\tt MOV uCC,R0} \\
 
&        {\tt STO R0,14(R1)} \\
 
&       {\em ; We can skip storing the stack, uSP, since it'll be stored}\\
 
&       {\em ; elsewhere (in the task structure) }\\
 
&        {\tt MOV uR13,R0} \\
 
&        {\tt STO R0,13(R1)} \\
 
        & \ldots {\em ; Need to repeat for all user registers} \\
 
&        {\tt MOV uR0,R0} \\
 
&        {\tt STO R0,1(R1)} \\
 
\end{tabular}
 
\caption{Example Storing User Task Context}\label{tbl:context-out}
 
\end{center}\end{table}
 
For the sake of discussion, we assume the supervisor maintains a
 
pointer to the current task's structure in supervisor register
 
{\tt R12}, and that {\tt stack} is an offset to the beginning of this
 
structure indicating where the stack pointer is to be kept within it.
 
 
 
        For those who are still interested, the full code for this context
 
        save can be found as an assembler macro within the assembler
 
        include file, {\tt sys.i}.
 
 
 
\item Reset the watchdog timer.  If you are using the watchdog timer, it should
 
        be reset on a context swap, to know that things are still working.
 
        Example code for this is shown in Tbl.~\ref{tbl:reset-watchdog}.
 
\begin{table}\begin{center}
 
\begin{tabular}{ll}
 
\multicolumn{2}{l}{{\tt `define WATCHDOG\_ADDRESS 32'hc000\_0002}}\\
 
\multicolumn{2}{l}{{\tt `define WATCHDOG\_TICKS 32'd1\_000\_000} {; \em = 10 ms}}\\
 
&       {\tt LDI WATCHDOG\_ADDRESS,R0} \\
 
&       {\tt LDI WATCHDOG\_TICKS,R1} \\
 
&       {\tt STO R1,(R0)}
 
\end{tabular}
 
\caption{Example Watchdog Reset}\label{tbl:reset-watchdog}
 
\end{center}\end{table}
 
 
 
\item Interrupt handling.  An interrupt handler within the Zip System is nothing
 
        more than a task.  At context swap time, the supervisor needs to
 
        disable all of the interrupts that have tripped, and then enable
 
        all of the tasks that would deal with each of these interrupts.
 
        These can be user tasks, run at higher priority than any other user
 
        tasks.  Either way, they will need to re--enable their own interrupt
 
        themselves, if the interrupt is still relevant.
 
 
 
        An example of this master interrut handling is shown in
 
        Tbl.~\ref{tbl:pre-handler}.
 
\begin{table}\begin{center}
 
\begin{tabular}{ll}
 
{\tt pre\_handler:} \\
 
&       {\tt LDI PIC\_ADDRESS,R0 } \\
 
&       {\em ; Start by grabbing the interrupt state from the interrupt}\\
 
&       {\em ; controller.  We'll store this into the register R7 so that }\\
 
&       {\em ; we can keep and preserve this information for the scheduler}\\
 
&       {\em ; to use later. }\\
 
&       {\tt LOD (R0),R1} \\
 
&       {\tt MOV R1,R7 } \\
 
&       {\em ; As a next step, we need to acknowledge and disable all active}\\
 
&       {\em ; interrupts. We'll start by calculating all of our active}\\
 
&       {\em ; interrupts.}\\
 
&       {\tt AND 0x07fff,R1 } \\
 
&       {\em ; Put the active interrupts into the upper half of R1} \\
 
&       {\tt ROL 16,R1 } \\
 
&       {\tt LDILO 0x0ffff,R1   } \\
 
&       {\tt AND R7,R1}\\
 
&       {\em ; Acknowledge and disable active interrupts}\\
 
&       {\em ; This also disables all interrupts from the controller, so}\\
 
&       {\em ; we'll need to re-enable interrupts in general shortly } \\
 
&       {\tt STO R1,(R0) } \\
 
&       {\em ; We leave our active interrupt mask in R7 so the scheduler can}\\
 
&       {\em ; release any tasks that depended upon them. } \\
 
\end{tabular}
 
\caption{Example checking for active interrupts}\label{tbl:pre-handler}
 
\end{center}\end{table}
 
 
 
\item Calling the scheduler.  This needs to be done to pick the next task
 
        to switch to.  It may be an interrupt handler, or it may  be a normal
 
        user task.  From a priority standpoint, it would make sense that the
 
        interrupt handlers all have a higher priority than the user tasks,
 
        and that once they have been called the user tasks may then be called
 
        again.  If no task is ready to run, run the idle task to wait for an
 
        interrupt.
 
 
 
        This suggests a minimum of four task priorities:
 
        \begin{enumerate}
 
        \item Interrupt handlers, executed with their interrupts disabled
 
        \item Device drivers, executed with interrupts re-enabled
 
        \item User tasks
 
        \item The idle task, executed when nothing else is able to execute
 
        \end{enumerate}
 
 
 
        For our purposes here, we'll just assume that a pointer to the current
 
        task is maintained in {\tt R12}, that a {\tt JSR scheduler} is
 
        called, and that the next current task is likewise placed into
 
        {\tt R12}.
 
 
 
\item Restore the new tasks context.  Given that the scheduler has returned a
 
        task that can be run at this time, the stack pointer needs to be
 
        pulled out of the tasks task structure, placed into the user
 
        register, and then the rest of the user registers need to be popped
 
        back off of the stack to run this task.  An example of this is
 
        shown in Tbl.~\ref{tbl:context-in},
 
\begin{table}\begin{center}
 
\begin{tabular}{ll}
 
{\tt swap\_in:} \\
 
&       {\tt LOD stack(R12),R1} \\
 
&       {\tt MOV 15(R1),uSP} \\
 
&       {\tt LOD 15(R1),R0} \\
 
&       {\tt MOV R0,uPC} \\
 
&       {\tt LOD 14(R1),R0} \\
 
&       {\tt MOV R0,uCC} \\
 
&       {\tt LOD 13(R1),R0} \\
 
&       {\tt MOV R0,uR12} \\
 
        & \ldots {\em ; Need to repeat for all user registers} \\
 
&       {\tt LOD 1(R1),R0} \\
 
&       {\tt MOV R0,uR0} \\
 
&       {\tt BRA return\_to\_user} \\
 
\end{tabular}
 
\caption{Example Restoring User Task Context}\label{tbl:context-in}
 
\end{center}\end{table}
 
        assuming as before that the task
 
        pointer is found in supervisor register {\tt R12}.
 
        As with storing the user context, the full code associated with
 
        restoring the user context can be found in the assembler include
 
        file, {\tt sys.i}.
 
 
 
\item Clear the userspace accounting registers.  In order to keep track of
 
        per process system usage, these registers need to be cleared before
 
        reactivating the userspace process.  That way, upon the next
 
        interrupt, we'll know how many clocks the userspace program has
 
        encountered, and how many instructions it was able to issue in
 
        those many clocks.
 
 
 
\item Jump back to the instruction just before saving the last tasks context,
 
        because that location in memory contains the return from interrupt
 
        command that we are going to need to execute, in order to guarantee
 
        that we return back here again.
 
\end{enumerate}
 
 
\chapter{Registers}\label{chap:regs}
\chapter{Registers}\label{chap:regs}
 
 
The ZipSystem registers fall into two categories, ZipSystem internal registers
The ZipSystem registers fall into two categories, ZipSystem internal registers
accessed via the ZipCPU shown in Tbl.~\ref{tbl:zpregs},
accessed via the ZipCPU shown in Tbl.~\ref{tbl:zpregs},
\begin{table}[htbp]
\begin{table}[htbp]
\begin{center}\begin{reglist}
\begin{center}\begin{reglist}
PIC   & \scalebox{0.8}{\tt 0xc0000000} & 32 & R/W & Primary Interrupt Controller \\\hline
PIC   & \scalebox{0.8}{\tt 0xc0000000} & 32 & R/W & Primary Interrupt Controller \\\hline
WDT   & \scalebox{0.8}{\tt 0xc0000001} & 32 & R/W & Watchdog Timer \\\hline
WDT   & \scalebox{0.8}{\tt 0xc0000001} & 32 & R/W & Watchdog Timer \\\hline
CCHE  & \scalebox{0.8}{\tt 0xc0000002} & 32 & R/W & Manual Cache Controller \\\hline
  & \scalebox{0.8}{\tt 0xc0000002} & 32 & R/W & {\em (Reserved for future use)} \\\hline
CTRIC & \scalebox{0.8}{\tt 0xc0000003} & 32 & R/W & Secondary Interrupt Controller \\\hline
CTRIC & \scalebox{0.8}{\tt 0xc0000003} & 32 & R/W & Secondary Interrupt Controller \\\hline
TMRA  & \scalebox{0.8}{\tt 0xc0000004} & 32 & R/W & Timer A\\\hline
TMRA  & \scalebox{0.8}{\tt 0xc0000004} & 32 & R/W & Timer A\\\hline
TMRB  & \scalebox{0.8}{\tt 0xc0000005} & 32 & R/W & Timer B\\\hline
TMRB  & \scalebox{0.8}{\tt 0xc0000005} & 32 & R/W & Timer B\\\hline
TMRC  & \scalebox{0.8}{\tt 0xc0000006} & 32 & R/W & Timer C\\\hline
TMRC  & \scalebox{0.8}{\tt 0xc0000006} & 32 & R/W & Timer C\\\hline
JIFF  & \scalebox{0.8}{\tt 0xc0000007} & 32 & R/W & Jiffies \\\hline
JIFF  & \scalebox{0.8}{\tt 0xc0000007} & 32 & R/W & Jiffies \\\hline
Line 1354... Line 1582...
MICNT  & \scalebox{0.8}{\tt 0xc000000b} & 32 & R/W & Master Instruction Counter\\\hline
MICNT  & \scalebox{0.8}{\tt 0xc000000b} & 32 & R/W & Master Instruction Counter\\\hline
UTASK  & \scalebox{0.8}{\tt 0xc000000c} & 32 & R/W & User Task Clock Counter \\\hline
UTASK  & \scalebox{0.8}{\tt 0xc000000c} & 32 & R/W & User Task Clock Counter \\\hline
UMSTL  & \scalebox{0.8}{\tt 0xc000000d} & 32 & R/W & User Stall Counter \\\hline
UMSTL  & \scalebox{0.8}{\tt 0xc000000d} & 32 & R/W & User Stall Counter \\\hline
UPSTL  & \scalebox{0.8}{\tt 0xc000000e} & 32 & R/W & User Pre--Fetch Stall Counter \\\hline
UPSTL  & \scalebox{0.8}{\tt 0xc000000e} & 32 & R/W & User Pre--Fetch Stall Counter \\\hline
UICNT  & \scalebox{0.8}{\tt 0xc000000f} & 32 & R/W & User Instruction Counter\\\hline
UICNT  & \scalebox{0.8}{\tt 0xc000000f} & 32 & R/W & User Instruction Counter\\\hline
 
DMACTRL  & \scalebox{0.8}{\tt 0xc0000010} & 32 & R/W & DMA Control Register\\\hline
 
DMALEN  & \scalebox{0.8}{\tt 0xc0000011} & 32 & R/W & DMA total transfer length\\\hline
 
DMASRC  & \scalebox{0.8}{\tt 0xc0000012} & 32 & R/W & DMA source address\\\hline
 
DMADST  & \scalebox{0.8}{\tt 0xc0000013} & 32 & R/W & DMA destination address\\\hline
% Cache  & \scalebox{0.8}{\tt 0xc0100000} & & & Base address of the Cache memory\\\hline
% Cache  & \scalebox{0.8}{\tt 0xc0100000} & & & Base address of the Cache memory\\\hline
\end{reglist}
\end{reglist}
\caption{Zip System Internal/Peripheral Registers}\label{tbl:zpregs}
\caption{Zip System Internal/Peripheral Registers}\label{tbl:zpregs}
\end{center}\end{table}
\end{center}\end{table}
and the two debug registers shown in Tbl.~\ref{tbl:dbgregs}.
and the two debug registers shown in Tbl.~\ref{tbl:dbgregs}.
Line 1489... Line 1721...
the fact.  Second, whenever activating a user task, the Operating System will
the fact.  Second, whenever activating a user task, the Operating System will
set the four user counters to zero.  When the user task has completed, the
set the four user counters to zero.  When the user task has completed, the
Operating System will read the timers back off, to determine how much of the
Operating System will read the timers back off, to determine how much of the
CPU the process had consumed.
CPU the process had consumed.
 
 
 
The final peripheral to discuss is the DMA controller.  This controller
 
has four registers.  Of these four, the length, source and destination address
 
registers should need no further explanation.  They are full 32--bit registers
 
specifying the entire transfer length, the starting address to read from, and
 
the starting address to write to.  The registers can be written to when the
 
DMA is idle, and read at any time.  The control register, however, will need
 
some more explanation.
 
 
 
The bit allocation of the control register is shown in Tbl.~\ref{tbl:dmacbits}.
 
\begin{table}\begin{center}
 
\begin{bitlist}
 
31 & R & DMA Active\\\hline
 
30 & R & Wishbone error, transaction aborted (cleared on any write)\\\hline
 
29 & R/W & Set to '1' to prevent the controller from incrementing the source address, '0' for normal memory copy. \\\hline
 
28 & R/W & Set to '0' to prevent the controller from incrementing the
 
        destination address, '0' for normal memory copy. \\\hline
 
27 \ldots 16 & W & The DMA Key.  Write a 12'hfed to these bits to start the
 
        activate any DMA transfer.  \\\hline
 
27 & R & Always reads '0', to force the deliberate writing of the key. \\\hline
 
26 \ldots 16 & R & Indicates the number of items in the transfer buffer that
 
        have yet to be written. \\\hline
 
15 & R/W & Set to '1' to trigger on an interrupt, or '0' to start immediately
 
        upon receiving a valid key.\\\hline
 
14\ldots 10 & R/W & Select among one of 32~possible interrupt lines.\\\hline
 
9\ldots 0 & R/W & Intermediate transfer length minus one.  Thus, to transfer
 
        one item at a time set this value to 0. To transfer 1024 at a time,
 
        set it to 1024.\\\hline
 
\end{bitlist}
 
\caption{DMA Control Register Bits}\label{tbl:dmacbits}
 
\end{center}\end{table}
 
This control register has been designed so that the common case of memory
 
access need only set the key and the transfer length.  Hence, writing a
 
\hbox{32'h0fed03ff} to the control register will start any memory transfer.
 
On the other hand, if you wished to read from a serial port (constant address)
 
and put the result into a buffer every time a word was available, you
 
might wish to write \hbox{32'h2fed8000}--this assumes, of course, that you
 
have a serial port wired to the zero bit of this interrupt control.  (The
 
DMA controller does not use the interrupt controller, and cannot clear
 
interrupts.)  As a third example, if you wished to write to an external
 
FIFO anytime it was less than half full (had fewer than 512 items), and
 
interrupt line 2 indicated this condition, you might wish to issue a
 
\hbox{32'h1fed8dff} to this port.
 
 
\section{Debug Port Registers}
\section{Debug Port Registers}
Accessing the Zip System via the debug port isn't as straight forward as
Accessing the Zip System via the debug port isn't as straight forward as
accessing the system via the wishbone bus.  The debug port itself has been
accessing the system via the wishbone bus.  The debug port itself has been
reduced to two addresses, as outlined earlier in Tbl.~\ref{tbl:dbgregs}.
reduced to two addresses, as outlined earlier in Tbl.~\ref{tbl:dbgregs}.
Access to the Zip System begins with the Debug Control register, shown in
Access to the Zip System begins with the Debug Control register, shown in
Line 1503... Line 1778...
13 & R & CPU GIE setting\\\hline
13 & R & CPU GIE setting\\\hline
12 & R & CPU is sleeping\\\hline
12 & R & CPU is sleeping\\\hline
11 & W & Command clear PF cache\\\hline
11 & W & Command clear PF cache\\\hline
10 & R/W & Command HALT, Set to '1' to halt the CPU\\\hline
10 & R/W & Command HALT, Set to '1' to halt the CPU\\\hline
9 & R & Stall Status, '1' if CPU is busy\\\hline
9 & R & Stall Status, '1' if CPU is busy\\\hline
8 & R/W & Step Command, set to '1' to step the CPU\\\hline
8 & R/W & Step Command, set to '1' to step the CPU, also sets the halt bit\\\hline
7 & R & Interrupt Request \\\hline
7 & R & Interrupt Request \\\hline
6 & R/W & Command RESET \\\hline
6 & R/W & Command RESET \\\hline
5\ldots 0 & R/W & Debug Register Address \\\hline
5\ldots 0 & R/W & Debug Register Address \\\hline
\end{bitlist}
\end{bitlist}
\caption{Debug Control Register Bits}\label{tbl:dbgctrl}
\caption{Debug Control Register Bits}\label{tbl:dbgctrl}
Line 1555... Line 1830...
UICNT & 47 & 32 & R/W & User instruction counter\\\hline
UICNT & 47 & 32 & R/W & User instruction counter\\\hline
\end{reglist}
\end{reglist}
\caption{Debug Register Addresses}\label{tbl:dbgaddrs}
\caption{Debug Register Addresses}\label{tbl:dbgaddrs}
\end{center}\end{table}
\end{center}\end{table}
Primarily, these ``registers'' include access to the entire CPU register
Primarily, these ``registers'' include access to the entire CPU register
set, as well as the 16~internal peripherals.  To read one of these registers
set, as well as the internal peripherals.  To read one of these registers
once the address is set, simply issue a read from the data port.  To write
once the address is set, simply issue a read from the data port.  To write
one of these registers or peripheral ports, simply write to the data port
one of these registers or peripheral ports, simply write to the data port
after setting the proper address.
after setting the proper address.
 
 
In this manner, all of the CPU's internal state may be read and adjusted.
In this manner, all of the CPU's internal state may be read and adjusted.
Line 1644... Line 1919...
 
 
\chapter{Clocks}\label{chap:clocks}
\chapter{Clocks}\label{chap:clocks}
 
 
This core is based upon the Basys--3 development board sold by Digilent.
This core is based upon the Basys--3 development board sold by Digilent.
The Basys--3 development board contains one external 100~MHz clock, which is
The Basys--3 development board contains one external 100~MHz clock, which is
sufficient to run the ZIP CPU core.
sufficient to run the Zip CPU core.
\begin{table}[htbp]
\begin{table}[htbp]
\begin{center}
\begin{center}
\begin{clocklist}
\begin{clocklist}
i\_clk & External & 100~MHz & 100~MHz & System clock.\\\hline
i\_clk & External & 100~MHz & 100~MHz & System clock.\\\hline
\end{clocklist}
\end{clocklist}
Line 1710... Line 1985...
only supports one such interrupt line by default.  For us, this line is the
only supports one such interrupt line by default.  For us, this line is the
output of another interrupt controller, but that's a board specific setup
output of another interrupt controller, but that's a board specific setup
detail.  Finally, the Zip System produces one external interrupt whenever
detail.  Finally, the Zip System produces one external interrupt whenever
the CPU halts to wait for the debugger.
the CPU halts to wait for the debugger.
 
 
 
\chapter{Initial Assessment}\label{chap:assessment}
 
 
 
Having now worked with the Zip CPU for a while, it is worth offering an
 
honest assessment of how well it works and how well it was designed. At the
 
end of this assessment, I will propose some changes that may take place in a
 
later version of this Zip CPU to make it better.
 
 
 
\section{The Good}
 
\begin{itemize}
 
\item The Zip CPU is light weight and fully featured as it exists today. For
 
        anyone who wishes to build a general purpose CPU and then to
 
        experiment with building and adding particular features, the Zip CPU
 
        makes a good starting point--it is fairly simple. Modifications should
 
        be simple enough.
 
\item As an estimate of the ``weight'' of this implementation, the CPU has
 
        cost me less than 150 hours to implement from its inception.
 
\item The Zip CPU was designed to be an implementable soft core that could be
 
        placed within an FPGA, controlling actions internal to the FPGA. It
 
        fits this role rather nicely. It does not fit the role of a system on
 
        a chip very well, but then it was never intended to be a system on a
 
        chip but rather a system within a chip.
 
\item The extremely simplified instruction set of the Zip CPU was a good
 
        choice. Although it does not have many of the commonly used
 
        instructions, PUSH, POP, JSR, and RET among them, the simplified
 
        instruction set has demonstrated an amazing versatility. I will contend
 
        therefore and for anyone who will listen, that this instruction set
 
        offers a full and complete capability for whatever a user might wish
 
        to do with two exceptions: bytewise character access and accelerated
 
        floating-point support.
 
\item This simplified instruction set is easy to decode.
 
\item The simplified bus transactions (32-bit words only) were also very easy
 
        to implement.
 
\item The novel approach of having a single interrupt vector, which just
 
        brings the CPU back to the instruction it left off at within the last
 
        interrupt context doesn't appear to have been that much of a problem.
 
        If most modern systems handle interrupt vectoring in software anyway,
 
        why maintain hardware support for it?
 
\item My goal of a high rate of instructions per clock may not be the proper
 
        measure. For example, if instructions are being read from a SPI flash
 
        device, such as is common among FPGA implementations, these same
 
        instructions may suffer stalls of between 64 and 128 cycles per
 
        instruction just to read the instruction from the flash. Executing the
 
        instruction in a single clock cycle is no longer the appropriate
 
        measure. At the same time, it should be possible to use the DMA
 
        peripheral to copy instructions from the FLASH to a temporary memory
 
        location, after which they may be executed at a single instruction
 
        cycle per access again.
 
\end{itemize}
 
 
 
\section{The Not so Good}
 
\begin{itemize}
 
\item While one of the stated goals was to use a small amount of logic,
 
        3k~LUTs isn't that impressively small. Indeed, it's really much
 
        too expensive when compared against other 8 and 16-bit CPUs that have
 
        less than 1k LUTs.
 
 
 
        Still, \ldots it's not bad, it's just not astonishingly good.
 
 
 
\item The fact that the instruction width equals the bus width means that the
 
        instruction fetch cycle will always be interfering with any load or
 
        store memory operation, with the only exception being if the
 
        instruction is already in the cache.  {\em This has become the
 
        fundamental limit on the speed and performance of the CPU!}
 
        Those familiar with the Von--Neumann approach of sharing a bus
 
        between data and instructions will not be surprised by this assessment.
 
 
 
        This could be fixed in one of three ways: the instruction set
 
        architecture could be modified to handle Very Long Instruction Words
 
        (VLIW) so that each 32--bit word would encode two or more instructions,
 
        the instruction fetch bus width could be increased from 32--bits to
 
        64--bits or more, or the instruction bus could be separated from the
 
        data bus.  Any and all of these approaches would increase the overall
 
        LUT count.
 
 
 
\item The (non-existant) floating point unit was an after-thought, isn't even
 
        built as a potential option, and most likely won't support the full
 
        IEEE standard set of FPU instructions--even for single point precision.
 
        This (non-existant) capability would benefit the most from an
 
        out-of-order execution capability, which the Zip CPU does not have.
 
 
 
        Still, sharing FPU registers with the main register set was a good
 
        idea and worth preserving, as it simplifies context swapping.
 
 
 
        Perhaps this really isn't a problem, but rather a feature.  By not
 
        implementing FPU instructions, the Zip CPU maintains a lower LUT count
 
        than it would have if it did implement these instructions.
 
 
 
\item The CPU has no character support. This is both good and bad.
 
        Realistically, the CPU works just fine without it. Characters can be
 
        supported as subsets of 32-bit words without any problem. Practically,
 
        though, it will make compiling non-Zip CPU code difficult--especially
 
        anything that assumes sizeof(int)=4*sizeof(char), or that tries to
 
        create unions with characters and integers and then attempts to
 
        reference the address of the characters within that union.
 
 
 
\item The Zip CPU does not support a data cache. One can still be built
 
        externally, but this is a limitation of the CPU proper as built.
 
        Further, under the theory of the Zip CPU design (that of an embedded
 
        soft-core processor within an FPGA, where any ``address'' may reference
 
        either memory or a peripheral that may have side-effects), any data
 
        cache would need to be based upon an initial knowledge of whether or
 
        not it is supporting memory (cachable) or peripherals. This knowledge
 
        must exist somewhere, and that somewhere is currently (and by design)
 
        external to the CPU.
 
 
 
        This may also be written off as a ``feature'' of the Zip CPU, since
 
        the addition of a data cache can greatly increase the LUT count of
 
        a soft core.
 
 
 
\item Many other instruction sets offer three operand instructions, whereas
 
        the Zip CPU only offers two operand instructions. This means that it
 
        takes the Zip CPU more instructions to do many of the same operations.
 
        The good part of this is that it gives the Zip CPU a greater amount of
 
        flexibility in its immediate operand mode, although that increased
 
        flexibility isn't necessarily as valuable as one might like.
 
 
 
\item The Zip CPU does not currently detect and trap on either illegal
 
        instructions or bus errors.  Attempts to access non--existent
 
        memory quietly return erroneous results, rather than halting the
 
        process (user mode) or halting or resetting the CPU (supervisor mode).
 
 
 
\item The Zip CPU doesn't support out of order execution. I suppose it could
 
        be modified to do so, but then it would no longer be the ``simple''
 
        and low LUT count CPU it was designed to be. The two primary results
 
        are that 1) loads may unnecessarily stall the CPU, even if other
 
        things could be done while waiting for the load to complete, 2)
 
        bus errors on stores will never be caught at the point of the error,
 
        and 3) branch prediction becomes more difficult.
 
 
 
\item Although switching to an interrupt context in the Zip CPU design doesn't
 
        require a tremendous swapping of registers, in reality it still
 
        does--since any task swap still requires saving and restoring all
 
        16~user registers. That's a lot of memory movement just to service
 
        an interrupt.
 
 
 
\item The Zip CPU is by no means generic: it will never handle addresses
 
        larger than 32-bits (16GB) without a complete and total redesign.
 
        This may limit its utility as a generic CPU in the future, although
 
        as an embedded CPU within an FPGA this isn't really much of a limit
 
        or restriction.
 
 
 
\item While the Zip CPU has its own assembler, it has no linker and does not
 
        (yet) support a compiler. The standard C library is an even longer
 
        shot. My dream of having binutils and gcc support has not been
 
        realized and at this rate may not be realized. (I've been intimidated
 
        by the challenge everytime I've looked through those codes.)
 
 
 
\item While the Wishbone Bus (B4) supports a pipelined mode with single cycle
 
        execution, the Zip CPU is unable to exploit this parallelism. Instead,
 
        apart from the DMA and the pipelined prefetch, all loads and stores
 
        are single wishbone bus operations requiring a minimum of 3 clocks.
 
        (In practice, this has turned into 7-clocks.)
 
 
 
\iffalse
 
\item There is no control over whether or not an instruction sets the
 
        condition codes--certain instructions always set the condition codes,
 
        other instructions never set them. This effectively limits conditional
 
        instructions to a single instruction only (with two or more
 
        instructions as an exception), as the first instruction that sets
 
        condition codes will break the condition code chain.
 
 
 
        {\em (A proposed change below address this.)}
 
 
 
\item Using the CC register as a trap address was a bad idea--it limits the CC
 
        registers ability to be used in future expansion, such as by adding
 
        exception indication flags: bus error, floating point exception, etc.
 
 
 
        {\em (This can be changed by a different O/S implementation of the trap
 
        instruction.)}
 
\item The current implementation suffers from too many stalls on any
 
        branch--even if the branch is known early on.
 
 
 
        {\em (This is addressed in proposals below.)}
 
        % Addressed, 20150918
 
 
 
\item In a similar fashion, a switch to interrupt context forces the pipeline
 
        to be cleared, whereas it might make more sense to just continue
 
        executing the instructions already in the pipeline while the prefetch
 
        stage is working on switching to the interrupt context.
 
 
 
        {\em (Also addressed in proposals below.)}
 
        % This should happen so rarely that it is not really a problem
 
\fi
 
 
 
\end{itemize}
 
 
 
\section{The Next Generation}
 
This section could also be labeled as my ``To do'' list.
 
 
 
Given the feedback listed above, perhaps its time to consider what changes could be made to improve the Zip CPU in the future. I offer the following as proposals:
 
 
 
\begin{itemize}
 
\item {\bf Remove the low LUT goal.} It wasn't really achieved, and the
 
        proposals below will only increase the amount of logic the Zip CPU
 
        requires.  While I expect that the Zip CPU will always be somewhat
 
        of a light weight, it will never be the smallest kid on the block.
 
 
 
        I'm actually struggling with this idea.  The whole goal of the Zip
 
        CPU was to be light weight.  Wouldn't it make more sense to create and
 
        maintain options whereby it would remain lightweight?  For example, if
 
        the process accounting registers are anything but light weight, why
 
        keep them?  Why not instead make some compile flags that just turn them
 
        off, keeping the CPU lightweight?  The same holds for the prefetch
 
        cache.
 
 
 
\iffalse
 
\item {\bf Adjust the Zip CPU so that conditional instructions do not set
 
        flags}, although they may explicitly set condition codes if writing
 
        to the CC register.
 
 
 
        This is a simple change to the core, and may show up in new releases.
 
        % Fixed, 20150918
 
\fi
 
 
 
\item The `{\tt .V}' condition was never used in any code other than my test
 
        code.  Suggest changing it to a `{\tt .LE}' condition, which seems
 
        to be more useful.
 
 
 
\iffalse
 
\item Add in an {\bf unpredictable branch delay slot}, so that on any branch
 
        the delay slot may or may not be executed before the branch.
 
        Instructions that do not depend upon the branch, and that should be
 
        executed were the branch not taken, could be placed into the delay
 
        slot. Thus, if the branch isn't taken, we wouldn't suffer the stall,
 
        whereas it wouldn't affect the timing of the branch if taken. It would
 
        just do something irrelevant.
 
 
 
        % Changes made, 20150918, make this option no longer relevant
 
 
 
\item {\bf Re-engineer Branch Processing.}  There's no reason why a {\tt BRA}
 
        instruction should create five stall cycles.  The decode stage, plus
 
        the prefetch engine, should be able to drop this number of stalls via
 
        better branch handling.
 
 
 
        Indeed, this could turn into a simple means of branch prediction:
 
        if {\tt BRA} suffered a single stall only, whereas {\tt BRA.C}
 
        suffered five stalls, then {\tt BRA.!C} followed by {\tt BRA} would
 
        be faster than a {\tt BRA.C} instruction.  This would then allow a
 
        compiler to do explicit branch optimizations.
 
 
 
        Of course, longer branches using {\tt ADD X,PC} would still not be
 
        optimized.
 
 
 
        % DONE: 20150918 -- to include the ADD X,PC instructions
 
 
 
\item {\bf Request bus access for Load/Store two cycles earlier.}  The problem
 
        here is the contention for the bus between the memory unit and the
 
        prefetch unit.  Currently, the memory unit must ask the prefetch
 
        unit to release the bus if it is in the middle of a bus cycle.  At this
 
        point, the prefetch drops the {\tt STB} line on the next clock and must
 
        then wait for the last {\tt ACK} before releasing the bus.  If the
 
        request takes one clock, dropping the strobe line the next, waiting
 
        for an acknowledgement takes another, and then the bus must be idle
 
        for one cycle before starting again, these extra cycles add up.
 
        It should be possible to tell the prefetch stage to give up the bus
 
        as soon as the decoder knows the instruction will need the bus.
 
        Indeed, if done in the decode stage, this might drop the seven cycle
 
        access down by two cycles.
 
 
 
        % FIXED: 20150918
 
\fi
 
 
 
\item {\bf Consider a more traditional Instruction Cache.}  The current
 
        pipelined instruction cache just reads a window of memory into
 
        its cache.  If the CPU leaves that window, the entire cache is
 
        invalidated.  A more traditional cache, however, might allow
 
        common subroutines to stay within the cache without invalidating the
 
        entire cache structure.
 
 
 
\iffalse
 
\item {\bf Very Long Instruction Word (VLIW).}  Now, to speed up operation, I
 
        propose that the Zip CPU instruction set be modified towards a Very
 
        Long Instruction Word (VLIW) implementation. In this implementation,
 
        an instruction word may contain either one or two separate
 
        instructions. The first instruction would take up the high order bits,
 
        the second optional instruction the lower 16-bits. Further, I propose
 
        that any of the ALU instructions (SUB through LSR) automatically have
 
        a second instruction whenever their operand `B' is a register, and use
 
        the full 20-bit immediate if not. This will effectively eliminate the
 
        register plus immediate mode for all of these instructions.
 
 
 
        This is the minimal required change to increase the number of
 
        instructions per clock cycle. Other changes would need to take place
 
        as well to support this. These include:
 
        \begin{itemize}
 
        \item Instruction words containing two instructions would take two
 
                clocks to complete, while requiring only a single cycle
 
                instruction fetch.
 
        \item Instructions preceded by a label in the asseembler must always
 
                start in the high order word.
 
        \item VLIW's, once started, must always execute to completion. The
 
                upper word may set the PC, the lower word may not. Regardless
 
                of whether the upper word sets the PC, the lower word must
 
                still be guaranteed to complete before the PC changes. On any
 
                switch to (or from) interrupt context, both instructions must
 
                complete or none of the instructions in the word shall
 
                complete prior to the switch.
 
        \item STEP commands and BREAK instructions will only take place after
 
                the entire word is executed.
 
        \end{itemize}
 
 
 
        If done well, the assembler should be able to handle these changes
 
        with the biggest impacts to the user being increased performance and
 
        a loss of the register plus immediate ALU modes. (These weren't really
 
        relevant for the XOR, OR, AND, etc. operations anyway.) Machine code
 
        compatibility will not be maintained.
 
 
 
        A proposed secondary instruction set might consist of: a four bit
 
        operand (any of the prior instructions would be supported, with some
 
        exceptions such as moves to and from user registers while in
 
        supervisor mode not being supported), a 4-bit register result (PC not
 
        allowed), a 3-bit conditional (identical to the conditional for the
 
        upper word), a single bit for whether or not an immediate is present
 
        or not, followed by either a 4-bit register or a 4-bit signed
 
        immediate. The multiply instruction would steal the immediate flag to
 
        be used as a sign indication, forcing both operands to be registers
 
        without any immediate offsets.
 
 
 
        {\em Initial conversion of several library functions to this secondary
 
        instruction set has demonstrated little to no gain.   The problem was
 
        that the new instruction set was made by joining a rarely used
 
        instruction (ALU with register and not immediate) with a more common
 
        instruction.  The utility was then limited by the utility of the rare
 
        instrction, which limited the impact of the entire approach.  }
 
\else
 
\item {\bf Very Long Instruction Word (VLIW).}  The goal here would be to
 
        create a new instruction set whereby two instructions would be encoded
 
        in each 32--bit word.  While this may speed up
 
        CPU operation, it would necessitate an instruction redesign.
 
\fi
 
 
 
\end{itemize}
 
 
 
 
% Appendices
% Appendices
% Index
% Index
\end{document}
\end{document}
 
 
 
 

powered by: WebSVN 2.1.0

© copyright 1999-2024 OpenCores.org, equivalent to Oliscience, all rights reserved. OpenCores®, registered trademark.