1 |
3 |
ericw |
/*
|
2 |
|
|
--------------------------------------------------------------------------------
|
3 |
|
|
|
4 |
|
|
Module : core.v
|
5 |
|
|
|
6 |
|
|
--------------------------------------------------------------------------------
|
7 |
|
|
|
8 |
|
|
Function:
|
9 |
|
|
- Processor core with 8 stage pipeline, 8 threads, and 4 stacks per thread.
|
10 |
|
|
|
11 |
|
|
Instantiates:
|
12 |
|
|
- control_ring.v
|
13 |
|
|
- data_ring.v
|
14 |
|
|
- reg_set.v
|
15 |
|
|
- reg_set_shim.v
|
16 |
|
|
- dp_ram_infer.v
|
17 |
|
|
|
18 |
|
|
|
19 |
|
|
--------------------
|
20 |
|
|
- Revision History -
|
21 |
|
|
--------------------
|
22 |
|
|
|
23 |
|
|
v01.10 - 2013-07-06
|
24 |
|
|
- Shuffled opcode encoding again:
|
25 |
|
|
- Unused jmp_iz never houses rd_i ops.
|
26 |
|
|
- Unused jmp_i never houses wr_i ops.
|
27 |
|
|
- jmp_i always immediate field reduced to 5 bits for consistency,
|
28 |
|
|
which leaves an unused 32 code gap.
|
29 |
|
|
- Edited boot code tests to work with this new encoding.
|
30 |
|
|
|
31 |
|
|
v01.09 (cont) - 2013-06-25
|
32 |
|
|
- Passes boot code verification for:
|
33 |
|
|
- All I/O ops.
|
34 |
|
|
|
35 |
|
|
v01.09 - 2013-06-24
|
36 |
|
|
- First public release.
|
37 |
|
|
- Shuffled opcode encoding again:
|
38 |
|
|
- Unused never jmp_i fits read/write perfectly.
|
39 |
|
|
- Unused always jmp_i zero comparison allows for 1 bit immediate field
|
40 |
|
|
expansion of jmp_i(gle/z).
|
41 |
|
|
- Removed all skip instructions (redundant w/ jmp_i).
|
42 |
|
|
- All the above allows add_i immediate field expansion to 6 bits.
|
43 |
|
|
- Opcode encoding is now more straightforward with fewer immediate field sizes.
|
44 |
|
|
- Worked on pc_ring.v and pointer_ring.v, now use simple loops rather than
|
45 |
|
|
generate, the logic produced better tracks the module paramters.
|
46 |
|
|
- Removed "skp" control bit from pc_ring.v, op_decode.v, control_ring.v.
|
47 |
|
|
- Removed intr_ack_o from core I/O.
|
48 |
|
|
- Renamed "register_set*" => "reg_set*".
|
49 |
|
|
- (Considering adding a leading zero count opcode, it doesn't take many LEs.)
|
50 |
|
|
- Seeing a write to thread 7 stack 0 memory at startup, don't think it
|
51 |
|
|
means anything due to clearing.
|
52 |
|
|
- Passes boot code verification for:
|
53 |
|
|
- Interrupts.
|
54 |
|
|
- Stack ops, depth, error reporting (all threads).
|
55 |
|
|
- All branch / conditional ops.
|
56 |
|
|
- All logical / arithmetic / shift ops.
|
57 |
|
|
|
58 |
|
|
v01.08 - 2013-06-14
|
59 |
|
|
- Added parameter options for stack pointer error protections & brought them
|
60 |
|
|
to the top level.
|
61 |
|
|
- The unsigned restoring division subroutine works!
|
62 |
|
|
|
63 |
|
|
v01.07 - 2013-06-11
|
64 |
|
|
- Changed opcodes to make swapping of (A?B) operands cover all logical needs:
|
65 |
|
|
- op_skp_ul => op_skp_l
|
66 |
|
|
- op_skp_ge => op_skp_uge
|
67 |
|
|
- op_jmp_iul => op_jmp_il
|
68 |
|
|
- op_jmp_ige => op_jmp_iuge
|
69 |
|
|
- Changed verification boot code tests to reflect above (passes).
|
70 |
|
|
|
71 |
|
|
v01.06 - 2013-06-10
|
72 |
|
|
- Fixed latency of test fields in op_decode.v (were only registered 1x, s/b 2x).
|
73 |
|
|
- Fixed subtle immediate address offset sign issue in pc_ring.v.
|
74 |
|
|
- Fixed subtle immediate data sign issue in stacks_mux.v.
|
75 |
|
|
- Minor style edits to visually align port names @ vector_sr.v instances.
|
76 |
|
|
- The log2 subroutine works!
|
77 |
|
|
|
78 |
|
|
v01.05 - 2013-06-07
|
79 |
|
|
- Put the skip instructions back (for convenience & clarity).
|
80 |
|
|
- Changes to op_decode.v, separate immediate data and address decodes,
|
81 |
|
|
misc. edits to improve speed.
|
82 |
|
|
- Renamed "op_codes.h" => "op_encode.h".
|
83 |
|
|
- Lots of minor edits at the higher levels.
|
84 |
|
|
- Added "copyright.txt" to directory.
|
85 |
|
|
|
86 |
|
|
v01.04 - 2013-06-06
|
87 |
|
|
- Added op_jmp_i (A?B) instructions.
|
88 |
|
|
- Removed all skip instructions (redundant).
|
89 |
|
|
|
90 |
|
|
v01.03 - 2013-06-04
|
91 |
|
|
- Changed op_jmp_i to be conditional (A?0).
|
92 |
|
|
- Renamed "addr_regs.h" => "register_set_addr.h".
|
93 |
|
|
- New boot code does log2.
|
94 |
|
|
|
95 |
|
|
v01.02 (cont) - 2013-05-23
|
96 |
|
|
- Old boot code file renamed: "boot_code.h" => "boot_code_00.h".
|
97 |
|
|
- New boot code file tests all op_codes and gives final report.
|
98 |
|
|
|
99 |
|
|
v01.02 - 2013-05-22
|
100 |
|
|
- Memory writes now work, the fix was to swap pc/op_code ROM side A with
|
101 |
|
|
data RW side B of main memory "dp_ram_infer.v". It seems side A is
|
102 |
|
|
incorrectly used as the master mode for both ports.
|
103 |
|
|
- Renamed "alu_input_mux.v" => "stacks_mux.v".
|
104 |
|
|
- Removed enable from "register_set.v" and "register_set_shim.v".
|
105 |
|
|
- Monkeyed with "register_set_shim.v" a bit.
|
106 |
|
|
- Removed async resets from "dp_ram_infer.v" and "dq_ram_infer.v".
|
107 |
|
|
- Passes boot code tests 0, 1, 2, 3, 4.
|
108 |
|
|
|
109 |
|
|
v01.01 - 2013-05-22
|
110 |
|
|
- Added clear and interrupt BASE and SPAN parameters.
|
111 |
|
|
- Because BASE is a simple MSB concatenation, for it to be effective
|
112 |
|
|
make BASE >= 2^(THRD_W+SPAN). BASE LSBs below this point are ignored.
|
113 |
|
|
- Individual clear and interrupt address interspacing = 2^SPAN.
|
114 |
|
|
- Example:
|
115 |
|
|
CLR_BASE ='h0 positions thread 0 clear @ 0.
|
116 |
|
|
CLR_SPAN =2 positions thread 1 clear @ 4, thread 2 @ 8, etc.
|
117 |
|
|
INTR_BASE='h20 positions thread 0 interrupt @ 'd32.
|
118 |
|
|
INTR_SPAN=2 positions thread 1 interrupt @ 'd36, thread 2 @ 'd40, etc.
|
119 |
|
|
- Moved most core top port parameters to localparam.
|
120 |
|
|
- Modified boot code tests accordingly.
|
121 |
|
|
- Passes boot code tests 0, 1, 2.
|
122 |
|
|
- Memory writes still don't work!
|
123 |
|
|
|
124 |
|
|
v01.00 - 2013-05-21 - born
|
125 |
|
|
- EP3C5E144C8: ~180MHz, ~1785 LEs (34% of logic).
|
126 |
|
|
- Fixed immediate value bug (was only registered 1x in op_decode.v, s/b 2x).
|
127 |
|
|
- Passes boot code tests 0, 1, 2.
|
128 |
|
|
- Memory writes don't work!
|
129 |
|
|
|
130 |
|
|
--------------------------------------------------------------------------------
|
131 |
|
|
*/
|
132 |
|
|
|
133 |
|
|
module core
|
134 |
|
|
#(
|
135 |
|
|
parameter integer DATA_W = 32, // data width
|
136 |
|
|
parameter integer THREADS = 8, // threads
|
137 |
|
|
parameter [DATA_W/4-1:0] VER_MAJ = 'h01, // core version
|
138 |
|
|
parameter [DATA_W/4-1:0] VER_MIN = 'h10 // core version
|
139 |
|
|
)
|
140 |
|
|
(
|
141 |
|
|
// clocks & resets
|
142 |
|
|
input wire clk_i, // clock
|
143 |
|
|
input wire rst_i, // async. reset, active high
|
144 |
|
|
//
|
145 |
|
|
input wire [THREADS-1:0] intr_req_i, // event request, active high
|
146 |
|
|
//
|
147 |
|
|
input wire [DATA_W-1:0] io_i, // gpio
|
148 |
|
|
output wire [DATA_W-1:0] io_o
|
149 |
|
|
);
|
150 |
|
|
|
151 |
|
|
|
152 |
|
|
/*
|
153 |
|
|
----------------------
|
154 |
|
|
-- internal signals --
|
155 |
|
|
----------------------
|
156 |
|
|
*/
|
157 |
|
|
`include "functions.h" // for clog2()
|
158 |
|
|
//
|
159 |
|
|
localparam integer ADDR_W = 16; // address width
|
160 |
|
|
localparam integer PNTR_W = 5; // stack pointer width
|
161 |
|
|
localparam integer MEM_ADDR_W = 13; // main memory width
|
162 |
|
|
localparam [ADDR_W-1:0] CLR_BASE = 'h0; // clear address base (concat)
|
163 |
|
|
localparam integer CLR_SPAN = 2; // clear address span (2^n)
|
164 |
|
|
localparam [ADDR_W-1:0] INTR_BASE = 'h20; // interrupt address base (concat)
|
165 |
|
|
localparam integer INTR_SPAN = 2; // interrupt address span (2^n)
|
166 |
|
|
localparam integer THRD_W = clog2( THREADS );
|
167 |
|
|
localparam integer STACKS = 4; // number of stacks
|
168 |
|
|
localparam integer STK_W = clog2( STACKS );
|
169 |
|
|
localparam integer IM_DATA_W = 8; // immediate data width
|
170 |
|
|
localparam integer IM_ADDR_W = 5; // immediate address width
|
171 |
|
|
localparam integer LG_W = 2;
|
172 |
|
|
localparam integer OP_CODE_W = DATA_W/2;
|
173 |
|
|
localparam integer REG_ADDR_W = 4;
|
174 |
|
|
localparam integer IM_RW_W = 4;
|
175 |
|
|
localparam integer POP_PROT = 1; // 1=error protection, 0=none
|
176 |
|
|
localparam integer PUSH_PROT = 1; // 1=error protection, 0=none
|
177 |
|
|
//
|
178 |
|
|
wire [THREADS-1:0] clr_req;
|
179 |
|
|
wire [THREADS-1:0] intr_en;
|
180 |
|
|
wire [OP_CODE_W-1:0] op_code;
|
181 |
|
|
wire op_code_er;
|
182 |
|
|
wire [STK_W-1:0] a_sel, b_sel;
|
183 |
|
|
wire imda, sgn, ext;
|
184 |
|
|
wire [LG_W-1:0] lg;
|
185 |
|
|
wire add, sub, mul, shl, cpy, dm, rtn, rd, wr;
|
186 |
|
|
wire stk_clr;
|
187 |
|
|
wire [STACKS-1:0] pop, push, pop_er, push_er;
|
188 |
|
|
wire [DATA_W-1:0] a, b;
|
189 |
|
|
wire [IM_DATA_W-1:0] im_data;
|
190 |
|
|
wire [IM_ADDR_W-1:0] im_addr;
|
191 |
|
|
wire nez, ne, ltz, lt;
|
192 |
|
|
wire [THRD_W-1:0] thrd_0, thrd_2, thrd_3, thrd_6;
|
193 |
|
|
wire [ADDR_W-1:0] pc_1, pc_3, pc_4;
|
194 |
|
|
wire [DATA_W/2-1:0] dm_rd_data;
|
195 |
|
|
wire [DATA_W/2-1:0] rd_data, wr_data;
|
196 |
|
|
wire [ADDR_W-1:0] addr;
|
197 |
|
|
wire regs_wr, regs_rd, dm_wr;
|
198 |
|
|
|
199 |
|
|
|
200 |
|
|
/*
|
201 |
|
|
================
|
202 |
|
|
== code start ==
|
203 |
|
|
================
|
204 |
|
|
*/
|
205 |
|
|
|
206 |
|
|
|
207 |
|
|
// the control ring
|
208 |
|
|
control_ring
|
209 |
|
|
#(
|
210 |
|
|
.DATA_W ( DATA_W ),
|
211 |
|
|
.ADDR_W ( ADDR_W ),
|
212 |
|
|
.THREADS ( THREADS ),
|
213 |
|
|
.THRD_W ( THRD_W ),
|
214 |
|
|
.STACKS ( STACKS ),
|
215 |
|
|
.STK_W ( STK_W ),
|
216 |
|
|
.IM_DATA_W ( IM_DATA_W ),
|
217 |
|
|
.IM_ADDR_W ( IM_ADDR_W ),
|
218 |
|
|
.OP_CODE_W ( OP_CODE_W ),
|
219 |
|
|
.LG_W ( LG_W ),
|
220 |
|
|
.CLR_BASE ( CLR_BASE ),
|
221 |
|
|
.CLR_SPAN ( CLR_SPAN ),
|
222 |
|
|
.INTR_BASE ( INTR_BASE ),
|
223 |
|
|
.INTR_SPAN ( INTR_SPAN )
|
224 |
|
|
)
|
225 |
|
|
control_ring
|
226 |
|
|
(
|
227 |
|
|
.clk_i ( clk_i ),
|
228 |
|
|
.rst_i ( rst_i ),
|
229 |
|
|
.clr_req_i ( clr_req ),
|
230 |
|
|
.clr_ack_o ( ), // unused
|
231 |
|
|
.intr_en_i ( intr_en ),
|
232 |
|
|
.intr_req_i ( intr_req_i ),
|
233 |
|
|
.intr_ack_o ( ), // unused
|
234 |
|
|
.op_code_i ( op_code ),
|
235 |
|
|
.op_code_er_o ( op_code_er ),
|
236 |
|
|
.b_lo_i ( b[DATA_W/2-1:0] ),
|
237 |
|
|
.im_data_o ( im_data ),
|
238 |
|
|
.a_sel_o ( a_sel ),
|
239 |
|
|
.b_sel_o ( b_sel ),
|
240 |
|
|
.imda_o ( imda ),
|
241 |
|
|
.sgn_o ( sgn ),
|
242 |
|
|
.ext_o ( ext ),
|
243 |
|
|
.lg_o ( lg ),
|
244 |
|
|
.add_o ( add ),
|
245 |
|
|
.sub_o ( sub ),
|
246 |
|
|
.mul_o ( mul ),
|
247 |
|
|
.shl_o ( shl ),
|
248 |
|
|
.cpy_o ( cpy ),
|
249 |
|
|
.dm_o ( dm ),
|
250 |
|
|
.rtn_o ( rtn ),
|
251 |
|
|
.rd_o ( rd ),
|
252 |
|
|
.wr_o ( wr ),
|
253 |
|
|
.stk_clr_o ( stk_clr ),
|
254 |
|
|
.pop_o ( pop ),
|
255 |
|
|
.push_o ( push ),
|
256 |
|
|
.nez_i ( nez ),
|
257 |
|
|
.ne_i ( ne ),
|
258 |
|
|
.ltz_i ( ltz ),
|
259 |
|
|
.lt_i ( lt ),
|
260 |
|
|
.thrd_0_o ( thrd_0 ),
|
261 |
|
|
.thrd_2_o ( thrd_2 ),
|
262 |
|
|
.thrd_3_o ( thrd_3 ),
|
263 |
|
|
.thrd_6_o ( thrd_6 ),
|
264 |
|
|
.im_addr_o ( im_addr ),
|
265 |
|
|
.pc_1_o ( pc_1 ),
|
266 |
|
|
.pc_3_o ( pc_3 ),
|
267 |
|
|
.pc_4_o ( pc_4 )
|
268 |
|
|
);
|
269 |
|
|
|
270 |
|
|
|
271 |
|
|
// the data ring
|
272 |
|
|
data_ring
|
273 |
|
|
#(
|
274 |
|
|
.DATA_W ( DATA_W ),
|
275 |
|
|
.ADDR_W ( ADDR_W ),
|
276 |
|
|
.THREADS ( THREADS ),
|
277 |
|
|
.THRD_W ( THRD_W ),
|
278 |
|
|
.STACKS ( STACKS ),
|
279 |
|
|
.STK_W ( STK_W ),
|
280 |
|
|
.PNTR_W ( PNTR_W ),
|
281 |
|
|
.IM_DATA_W ( IM_DATA_W ),
|
282 |
|
|
.LG_W ( LG_W ),
|
283 |
|
|
.POP_PROT ( POP_PROT ),
|
284 |
|
|
.PUSH_PROT ( PUSH_PROT )
|
285 |
|
|
)
|
286 |
|
|
data_ring
|
287 |
|
|
(
|
288 |
|
|
.clk_i ( clk_i ),
|
289 |
|
|
.rst_i ( rst_i ),
|
290 |
|
|
.a_sel_i ( a_sel ),
|
291 |
|
|
.b_sel_i ( b_sel ),
|
292 |
|
|
.imda_i ( imda ),
|
293 |
|
|
.sgn_i ( sgn ),
|
294 |
|
|
.ext_i ( ext ),
|
295 |
|
|
.lg_i ( lg ),
|
296 |
|
|
.add_i ( add ),
|
297 |
|
|
.sub_i ( sub ),
|
298 |
|
|
.mul_i ( mul ),
|
299 |
|
|
.shl_i ( shl ),
|
300 |
|
|
.cpy_i ( cpy ),
|
301 |
|
|
.dm_i ( dm ),
|
302 |
|
|
.rtn_i ( rtn ),
|
303 |
|
|
.stk_clr_i ( stk_clr ),
|
304 |
|
|
.pop_i ( pop ),
|
305 |
|
|
.push_i ( push ),
|
306 |
|
|
.thrd_6_i ( thrd_6 ),
|
307 |
|
|
.im_data_i ( im_data ),
|
308 |
|
|
.dm_data_i ( rd_data ),
|
309 |
|
|
.pc_3_i ( pc_3 ),
|
310 |
|
|
.a_o ( a ),
|
311 |
|
|
.b_o ( b ),
|
312 |
|
|
.nez_o ( nez ),
|
313 |
|
|
.ne_o ( ne ),
|
314 |
|
|
.ltz_o ( ltz ),
|
315 |
|
|
.lt_o ( lt ),
|
316 |
|
|
.pop_er_o ( pop_er ),
|
317 |
|
|
.push_er_o ( push_er )
|
318 |
|
|
);
|
319 |
|
|
|
320 |
|
|
|
321 |
|
|
// shim for memory and register set access
|
322 |
|
|
reg_set_shim
|
323 |
|
|
#(
|
324 |
|
|
.REGS_IN ( 1 ),
|
325 |
|
|
.REGS_OUT ( 1 ),
|
326 |
|
|
.DATA_W ( DATA_W ),
|
327 |
|
|
.ADDR_W ( ADDR_W ),
|
328 |
|
|
.REG_ADDR_W ( REG_ADDR_W ),
|
329 |
|
|
.IM_ADDR_W ( IM_RW_W )
|
330 |
|
|
)
|
331 |
|
|
reg_set_shim
|
332 |
|
|
(
|
333 |
|
|
.clk_i ( clk_i ),
|
334 |
|
|
.rst_i ( rst_i ),
|
335 |
|
|
.a_i ( a ),
|
336 |
|
|
.ext_i ( ext ),
|
337 |
|
|
.wr_data_o ( wr_data ),
|
338 |
|
|
.b_lo_i ( b[DATA_W/2-1:0] ),
|
339 |
|
|
.im_addr_i ( im_addr[IM_RW_W-1:0] ),
|
340 |
|
|
.pc_1_i ( pc_1 ),
|
341 |
|
|
.addr_o ( addr ),
|
342 |
|
|
.wr_i ( wr ),
|
343 |
|
|
.rd_i ( rd ),
|
344 |
|
|
.regs_wr_o ( regs_wr ),
|
345 |
|
|
.regs_rd_o ( regs_rd ),
|
346 |
|
|
.dm_wr_o ( dm_wr )
|
347 |
|
|
);
|
348 |
|
|
|
349 |
|
|
|
350 |
|
|
// internal register set
|
351 |
|
|
reg_set
|
352 |
|
|
#(
|
353 |
|
|
.REGS_IN ( 1 ),
|
354 |
|
|
.REGS_OUT ( 1 ),
|
355 |
|
|
.DATA_W ( DATA_W/2 ),
|
356 |
|
|
.ADDR_W ( REG_ADDR_W ),
|
357 |
|
|
.THREADS ( THREADS ),
|
358 |
|
|
.THRD_W ( THRD_W ),
|
359 |
|
|
.STACKS ( STACKS ),
|
360 |
|
|
.STK_W ( STK_W ),
|
361 |
|
|
.VER_MAJ ( VER_MAJ ),
|
362 |
|
|
.VER_MIN ( VER_MIN )
|
363 |
|
|
)
|
364 |
|
|
reg_set
|
365 |
|
|
(
|
366 |
|
|
.clk_i ( clk_i ),
|
367 |
|
|
.rst_i ( rst_i ),
|
368 |
|
|
.addr_i ( addr[REG_ADDR_W-1:0] ),
|
369 |
|
|
.wr_i ( regs_wr ),
|
370 |
|
|
.rd_i ( regs_rd ),
|
371 |
|
|
.data_i ( wr_data ),
|
372 |
|
|
.data_o ( rd_data ),
|
373 |
|
|
.dm_data_i ( dm_rd_data ),
|
374 |
|
|
.clr_req_o ( clr_req ),
|
375 |
|
|
.intr_en_o ( intr_en ),
|
376 |
|
|
.thrd_0_i ( thrd_0 ),
|
377 |
|
|
.op_code_er_i ( op_code_er ),
|
378 |
|
|
.thrd_2_i ( thrd_2 ),
|
379 |
|
|
.pop_er_i ( pop_er ),
|
380 |
|
|
.thrd_3_i ( thrd_3 ),
|
381 |
|
|
.push_er_i ( push_er ),
|
382 |
|
|
.io_lo_i ( io_i[DATA_W/2-1:0] ),
|
383 |
|
|
.io_hi_i ( io_i[DATA_W-1:DATA_W/2] ),
|
384 |
|
|
.io_lo_o ( io_o[DATA_W/2-1:0] ),
|
385 |
|
|
.io_hi_o ( io_o[DATA_W-1:DATA_W/2] )
|
386 |
|
|
);
|
387 |
|
|
|
388 |
|
|
|
389 |
|
|
// instruction and data memory
|
390 |
|
|
dp_ram_infer
|
391 |
|
|
#(
|
392 |
|
|
.REG_A_OUT ( 1 ),
|
393 |
|
|
.REG_B_OUT ( 1 ),
|
394 |
|
|
.DATA_W ( OP_CODE_W ),
|
395 |
|
|
.ADDR_W ( MEM_ADDR_W ),
|
396 |
|
|
.RD_MODE ( "WR_DATA" ) // functional don't care
|
397 |
|
|
)
|
398 |
|
|
main_mem
|
399 |
|
|
(
|
400 |
|
|
.a_clk_i ( clk_i ),
|
401 |
|
|
.a_addr_i ( addr ),
|
402 |
|
|
.a_wr_i ( dm_wr ),
|
403 |
|
|
.a_data_i ( wr_data ),
|
404 |
|
|
.a_data_o ( dm_rd_data ),
|
405 |
|
|
.b_clk_i ( clk_i ),
|
406 |
|
|
.b_addr_i ( pc_4 ),
|
407 |
|
|
.b_wr_i ( 1'b0 ), // unused
|
408 |
|
|
.b_data_i ( ), // unused
|
409 |
|
|
.b_data_o ( op_code )
|
410 |
|
|
);
|
411 |
|
|
|
412 |
|
|
|
413 |
|
|
endmodule
|