;; ARM 1136J[F]-S Pipeline Description
|
;; ARM 1136J[F]-S Pipeline Description
|
;; Copyright (C) 2003, 2007 Free Software Foundation, Inc.
|
;; Copyright (C) 2003, 2007 Free Software Foundation, Inc.
|
;; Written by CodeSourcery, LLC.
|
;; Written by CodeSourcery, LLC.
|
;;
|
;;
|
;; This file is part of GCC.
|
;; This file is part of GCC.
|
;;
|
;;
|
;; GCC is free software; you can redistribute it and/or modify it
|
;; GCC is free software; you can redistribute it and/or modify it
|
;; under the terms of the GNU General Public License as published by
|
;; under the terms of the GNU General Public License as published by
|
;; the Free Software Foundation; either version 3, or (at your option)
|
;; the Free Software Foundation; either version 3, or (at your option)
|
;; any later version.
|
;; any later version.
|
;;
|
;;
|
;; GCC is distributed in the hope that it will be useful, but
|
;; GCC is distributed in the hope that it will be useful, but
|
;; WITHOUT ANY WARRANTY; without even the implied warranty of
|
;; WITHOUT ANY WARRANTY; without even the implied warranty of
|
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
;; General Public License for more details.
|
;; General Public License for more details.
|
;;
|
;;
|
;; You should have received a copy of the GNU General Public License
|
;; You should have received a copy of the GNU General Public License
|
;; along with GCC; see the file COPYING3. If not see
|
;; along with GCC; see the file COPYING3. If not see
|
;; . */
|
;; . */
|
|
|
;; These descriptions are based on the information contained in the
|
;; These descriptions are based on the information contained in the
|
;; ARM1136JF-S Technical Reference Manual, Copyright (c) 2003 ARM
|
;; ARM1136JF-S Technical Reference Manual, Copyright (c) 2003 ARM
|
;; Limited.
|
;; Limited.
|
;;
|
;;
|
|
|
;; This automaton provides a pipeline description for the ARM
|
;; This automaton provides a pipeline description for the ARM
|
;; 1136J-S and 1136JF-S cores.
|
;; 1136J-S and 1136JF-S cores.
|
;;
|
;;
|
;; The model given here assumes that the condition for all conditional
|
;; The model given here assumes that the condition for all conditional
|
;; instructions is "true", i.e., that all of the instructions are
|
;; instructions is "true", i.e., that all of the instructions are
|
;; actually executed.
|
;; actually executed.
|
|
|
(define_automaton "arm1136jfs")
|
(define_automaton "arm1136jfs")
|
|
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;; Pipelines
|
;; Pipelines
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
|
|
;; There are three distinct pipelines (page 1-26 and following):
|
;; There are three distinct pipelines (page 1-26 and following):
|
;;
|
;;
|
;; - A 4-stage decode pipeline, shared by all three. It has fetch (1),
|
;; - A 4-stage decode pipeline, shared by all three. It has fetch (1),
|
;; fetch (2), decode, and issue stages. Since this is always involved,
|
;; fetch (2), decode, and issue stages. Since this is always involved,
|
;; we do not model it in the scheduler.
|
;; we do not model it in the scheduler.
|
;;
|
;;
|
;; - A 4-stage ALU pipeline. It has shifter, ALU (main integer operations),
|
;; - A 4-stage ALU pipeline. It has shifter, ALU (main integer operations),
|
;; and saturation stages. The fourth stage is writeback; see below.
|
;; and saturation stages. The fourth stage is writeback; see below.
|
;;
|
;;
|
;; - A 4-stage multiply-accumulate pipeline. It has three stages, called
|
;; - A 4-stage multiply-accumulate pipeline. It has three stages, called
|
;; MAC1 through MAC3, and a fourth writeback stage.
|
;; MAC1 through MAC3, and a fourth writeback stage.
|
;;
|
;;
|
;; The 4th-stage writeback is shared between the ALU and MAC pipelines,
|
;; The 4th-stage writeback is shared between the ALU and MAC pipelines,
|
;; which operate in lockstep. Results from either pipeline will be
|
;; which operate in lockstep. Results from either pipeline will be
|
;; moved into the writeback stage. Because the two pipelines operate
|
;; moved into the writeback stage. Because the two pipelines operate
|
;; in lockstep, we schedule them as a single "execute" pipeline.
|
;; in lockstep, we schedule them as a single "execute" pipeline.
|
;;
|
;;
|
;; - A 4-stage LSU pipeline. It has address generation, data cache (1),
|
;; - A 4-stage LSU pipeline. It has address generation, data cache (1),
|
;; data cache (2), and writeback stages. (Note that this pipeline,
|
;; data cache (2), and writeback stages. (Note that this pipeline,
|
;; including the writeback stage, is independent from the ALU & LSU pipes.)
|
;; including the writeback stage, is independent from the ALU & LSU pipes.)
|
|
|
(define_cpu_unit "e_1,e_2,e_3,e_wb" "arm1136jfs") ; ALU and MAC
|
(define_cpu_unit "e_1,e_2,e_3,e_wb" "arm1136jfs") ; ALU and MAC
|
; e_1 = Sh/Mac1, e_2 = ALU/Mac2, e_3 = SAT/Mac3
|
; e_1 = Sh/Mac1, e_2 = ALU/Mac2, e_3 = SAT/Mac3
|
(define_cpu_unit "l_a,l_dc1,l_dc2,l_wb" "arm1136jfs") ; Load/Store
|
(define_cpu_unit "l_a,l_dc1,l_dc2,l_wb" "arm1136jfs") ; Load/Store
|
|
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;; ALU Instructions
|
;; ALU Instructions
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
|
|
;; ALU instructions require eight cycles to execute, and use the ALU
|
;; ALU instructions require eight cycles to execute, and use the ALU
|
;; pipeline in each of the eight stages. The results are available
|
;; pipeline in each of the eight stages. The results are available
|
;; after the alu stage has finished.
|
;; after the alu stage has finished.
|
;;
|
;;
|
;; If the destination register is the PC, the pipelines are stalled
|
;; If the destination register is the PC, the pipelines are stalled
|
;; for several cycles. That case is not modelled here.
|
;; for several cycles. That case is not modelled here.
|
|
|
;; ALU operations with no shifted operand
|
;; ALU operations with no shifted operand
|
(define_insn_reservation "11_alu_op" 2
|
(define_insn_reservation "11_alu_op" 2
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "alu"))
|
(eq_attr "type" "alu"))
|
"e_1,e_2,e_3,e_wb")
|
"e_1,e_2,e_3,e_wb")
|
|
|
;; ALU operations with a shift-by-constant operand
|
;; ALU operations with a shift-by-constant operand
|
(define_insn_reservation "11_alu_shift_op" 2
|
(define_insn_reservation "11_alu_shift_op" 2
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "alu_shift"))
|
(eq_attr "type" "alu_shift"))
|
"e_1,e_2,e_3,e_wb")
|
"e_1,e_2,e_3,e_wb")
|
|
|
;; ALU operations with a shift-by-register operand
|
;; ALU operations with a shift-by-register operand
|
;; These really stall in the decoder, in order to read
|
;; These really stall in the decoder, in order to read
|
;; the shift value in a second cycle. Pretend we take two cycles in
|
;; the shift value in a second cycle. Pretend we take two cycles in
|
;; the shift stage.
|
;; the shift stage.
|
(define_insn_reservation "11_alu_shift_reg_op" 3
|
(define_insn_reservation "11_alu_shift_reg_op" 3
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "alu_shift_reg"))
|
(eq_attr "type" "alu_shift_reg"))
|
"e_1*2,e_2,e_3,e_wb")
|
"e_1*2,e_2,e_3,e_wb")
|
|
|
;; alu_ops can start sooner, if there is no shifter dependency
|
;; alu_ops can start sooner, if there is no shifter dependency
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
"11_alu_op")
|
"11_alu_op")
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
"11_alu_shift_op"
|
"11_alu_shift_op"
|
"arm_no_early_alu_shift_value_dep")
|
"arm_no_early_alu_shift_value_dep")
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
"11_alu_shift_reg_op"
|
"11_alu_shift_reg_op"
|
"arm_no_early_alu_shift_dep")
|
"arm_no_early_alu_shift_dep")
|
(define_bypass 2 "11_alu_shift_reg_op"
|
(define_bypass 2 "11_alu_shift_reg_op"
|
"11_alu_op")
|
"11_alu_op")
|
(define_bypass 2 "11_alu_shift_reg_op"
|
(define_bypass 2 "11_alu_shift_reg_op"
|
"11_alu_shift_op"
|
"11_alu_shift_op"
|
"arm_no_early_alu_shift_value_dep")
|
"arm_no_early_alu_shift_value_dep")
|
(define_bypass 2 "11_alu_shift_reg_op"
|
(define_bypass 2 "11_alu_shift_reg_op"
|
"11_alu_shift_reg_op"
|
"11_alu_shift_reg_op"
|
"arm_no_early_alu_shift_dep")
|
"arm_no_early_alu_shift_dep")
|
|
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"arm_no_early_mul_dep")
|
"arm_no_early_mul_dep")
|
(define_bypass 2 "11_alu_shift_reg_op"
|
(define_bypass 2 "11_alu_shift_reg_op"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"arm_no_early_mul_dep")
|
"arm_no_early_mul_dep")
|
|
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;; Multiplication Instructions
|
;; Multiplication Instructions
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
|
|
;; Multiplication instructions loop in the first two execute stages until
|
;; Multiplication instructions loop in the first two execute stages until
|
;; the instruction has been passed through the multiplier array enough
|
;; the instruction has been passed through the multiplier array enough
|
;; times.
|
;; times.
|
|
|
;; Multiply and multiply-accumulate results are available after four stages.
|
;; Multiply and multiply-accumulate results are available after four stages.
|
(define_insn_reservation "11_mult1" 4
|
(define_insn_reservation "11_mult1" 4
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "insn" "mul,mla"))
|
(eq_attr "insn" "mul,mla"))
|
"e_1*2,e_2,e_3,e_wb")
|
"e_1*2,e_2,e_3,e_wb")
|
|
|
;; The *S variants set the condition flags, which requires three more cycles.
|
;; The *S variants set the condition flags, which requires three more cycles.
|
(define_insn_reservation "11_mult2" 4
|
(define_insn_reservation "11_mult2" 4
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "insn" "muls,mlas"))
|
(eq_attr "insn" "muls,mlas"))
|
"e_1*2,e_2,e_3,e_wb")
|
"e_1*2,e_2,e_3,e_wb")
|
|
|
(define_bypass 3 "11_mult1,11_mult2"
|
(define_bypass 3 "11_mult1,11_mult2"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"arm_no_early_mul_dep")
|
"arm_no_early_mul_dep")
|
(define_bypass 3 "11_mult1,11_mult2"
|
(define_bypass 3 "11_mult1,11_mult2"
|
"11_alu_op")
|
"11_alu_op")
|
(define_bypass 3 "11_mult1,11_mult2"
|
(define_bypass 3 "11_mult1,11_mult2"
|
"11_alu_shift_op"
|
"11_alu_shift_op"
|
"arm_no_early_alu_shift_value_dep")
|
"arm_no_early_alu_shift_value_dep")
|
(define_bypass 3 "11_mult1,11_mult2"
|
(define_bypass 3 "11_mult1,11_mult2"
|
"11_alu_shift_reg_op"
|
"11_alu_shift_reg_op"
|
"arm_no_early_alu_shift_dep")
|
"arm_no_early_alu_shift_dep")
|
(define_bypass 3 "11_mult1,11_mult2"
|
(define_bypass 3 "11_mult1,11_mult2"
|
"11_store1"
|
"11_store1"
|
"arm_no_early_store_addr_dep")
|
"arm_no_early_store_addr_dep")
|
|
|
;; Signed and unsigned multiply long results are available across two cycles;
|
;; Signed and unsigned multiply long results are available across two cycles;
|
;; the less significant word is available one cycle before the more significant
|
;; the less significant word is available one cycle before the more significant
|
;; word. Here we conservatively wait until both are available, which is
|
;; word. Here we conservatively wait until both are available, which is
|
;; after three iterations and the memory cycle. The same is also true of
|
;; after three iterations and the memory cycle. The same is also true of
|
;; the two multiply-accumulate instructions.
|
;; the two multiply-accumulate instructions.
|
(define_insn_reservation "11_mult3" 5
|
(define_insn_reservation "11_mult3" 5
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "insn" "smull,umull,smlal,umlal"))
|
(eq_attr "insn" "smull,umull,smlal,umlal"))
|
"e_1*3,e_2,e_3,e_wb*2")
|
"e_1*3,e_2,e_3,e_wb*2")
|
|
|
;; The *S variants set the condition flags, which requires three more cycles.
|
;; The *S variants set the condition flags, which requires three more cycles.
|
(define_insn_reservation "11_mult4" 5
|
(define_insn_reservation "11_mult4" 5
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "insn" "smulls,umulls,smlals,umlals"))
|
(eq_attr "insn" "smulls,umulls,smlals,umlals"))
|
"e_1*3,e_2,e_3,e_wb*2")
|
"e_1*3,e_2,e_3,e_wb*2")
|
|
|
(define_bypass 4 "11_mult3,11_mult4"
|
(define_bypass 4 "11_mult3,11_mult4"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"arm_no_early_mul_dep")
|
"arm_no_early_mul_dep")
|
(define_bypass 4 "11_mult3,11_mult4"
|
(define_bypass 4 "11_mult3,11_mult4"
|
"11_alu_op")
|
"11_alu_op")
|
(define_bypass 4 "11_mult3,11_mult4"
|
(define_bypass 4 "11_mult3,11_mult4"
|
"11_alu_shift_op"
|
"11_alu_shift_op"
|
"arm_no_early_alu_shift_value_dep")
|
"arm_no_early_alu_shift_value_dep")
|
(define_bypass 4 "11_mult3,11_mult4"
|
(define_bypass 4 "11_mult3,11_mult4"
|
"11_alu_shift_reg_op"
|
"11_alu_shift_reg_op"
|
"arm_no_early_alu_shift_dep")
|
"arm_no_early_alu_shift_dep")
|
(define_bypass 4 "11_mult3,11_mult4"
|
(define_bypass 4 "11_mult3,11_mult4"
|
"11_store1"
|
"11_store1"
|
"arm_no_early_store_addr_dep")
|
"arm_no_early_store_addr_dep")
|
|
|
;; Various 16x16->32 multiplies and multiply-accumulates, using combinations
|
;; Various 16x16->32 multiplies and multiply-accumulates, using combinations
|
;; of high and low halves of the argument registers. They take a single
|
;; of high and low halves of the argument registers. They take a single
|
;; pass through the pipeline and make the result available after three
|
;; pass through the pipeline and make the result available after three
|
;; cycles.
|
;; cycles.
|
(define_insn_reservation "11_mult5" 3
|
(define_insn_reservation "11_mult5" 3
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "insn" "smulxy,smlaxy,smulwy,smlawy,smuad,smuadx,smlad,smladx,smusd,smusdx,smlsd,smlsdx"))
|
(eq_attr "insn" "smulxy,smlaxy,smulwy,smlawy,smuad,smuadx,smlad,smladx,smusd,smusdx,smlsd,smlsdx"))
|
"e_1,e_2,e_3,e_wb")
|
"e_1,e_2,e_3,e_wb")
|
|
|
(define_bypass 2 "11_mult5"
|
(define_bypass 2 "11_mult5"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"arm_no_early_mul_dep")
|
"arm_no_early_mul_dep")
|
(define_bypass 2 "11_mult5"
|
(define_bypass 2 "11_mult5"
|
"11_alu_op")
|
"11_alu_op")
|
(define_bypass 2 "11_mult5"
|
(define_bypass 2 "11_mult5"
|
"11_alu_shift_op"
|
"11_alu_shift_op"
|
"arm_no_early_alu_shift_value_dep")
|
"arm_no_early_alu_shift_value_dep")
|
(define_bypass 2 "11_mult5"
|
(define_bypass 2 "11_mult5"
|
"11_alu_shift_reg_op"
|
"11_alu_shift_reg_op"
|
"arm_no_early_alu_shift_dep")
|
"arm_no_early_alu_shift_dep")
|
(define_bypass 2 "11_mult5"
|
(define_bypass 2 "11_mult5"
|
"11_store1"
|
"11_store1"
|
"arm_no_early_store_addr_dep")
|
"arm_no_early_store_addr_dep")
|
|
|
;; The same idea, then the 32-bit result is added to a 64-bit quantity.
|
;; The same idea, then the 32-bit result is added to a 64-bit quantity.
|
(define_insn_reservation "11_mult6" 4
|
(define_insn_reservation "11_mult6" 4
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "insn" "smlalxy"))
|
(eq_attr "insn" "smlalxy"))
|
"e_1*2,e_2,e_3,e_wb*2")
|
"e_1*2,e_2,e_3,e_wb*2")
|
|
|
;; Signed 32x32 multiply, then the most significant 32 bits are extracted
|
;; Signed 32x32 multiply, then the most significant 32 bits are extracted
|
;; and are available after the memory stage.
|
;; and are available after the memory stage.
|
(define_insn_reservation "11_mult7" 4
|
(define_insn_reservation "11_mult7" 4
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "insn" "smmul,smmulr"))
|
(eq_attr "insn" "smmul,smmulr"))
|
"e_1*2,e_2,e_3,e_wb")
|
"e_1*2,e_2,e_3,e_wb")
|
|
|
(define_bypass 3 "11_mult6,11_mult7"
|
(define_bypass 3 "11_mult6,11_mult7"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"arm_no_early_mul_dep")
|
"arm_no_early_mul_dep")
|
(define_bypass 3 "11_mult6,11_mult7"
|
(define_bypass 3 "11_mult6,11_mult7"
|
"11_alu_op")
|
"11_alu_op")
|
(define_bypass 3 "11_mult6,11_mult7"
|
(define_bypass 3 "11_mult6,11_mult7"
|
"11_alu_shift_op"
|
"11_alu_shift_op"
|
"arm_no_early_alu_shift_value_dep")
|
"arm_no_early_alu_shift_value_dep")
|
(define_bypass 3 "11_mult6,11_mult7"
|
(define_bypass 3 "11_mult6,11_mult7"
|
"11_alu_shift_reg_op"
|
"11_alu_shift_reg_op"
|
"arm_no_early_alu_shift_dep")
|
"arm_no_early_alu_shift_dep")
|
(define_bypass 3 "11_mult6,11_mult7"
|
(define_bypass 3 "11_mult6,11_mult7"
|
"11_store1"
|
"11_store1"
|
"arm_no_early_store_addr_dep")
|
"arm_no_early_store_addr_dep")
|
|
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;; Branch Instructions
|
;; Branch Instructions
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
|
|
;; These vary greatly depending on their arguments and the results of
|
;; These vary greatly depending on their arguments and the results of
|
;; stat prediction. Cycle count ranges from zero (unconditional branch,
|
;; stat prediction. Cycle count ranges from zero (unconditional branch,
|
;; folded dynamic prediction) to seven (incorrect predictions, etc). We
|
;; folded dynamic prediction) to seven (incorrect predictions, etc). We
|
;; assume an optimal case for now, because the cost of a cache miss
|
;; assume an optimal case for now, because the cost of a cache miss
|
;; overwhelms the cost of everything else anyhow.
|
;; overwhelms the cost of everything else anyhow.
|
|
|
(define_insn_reservation "11_branches" 0
|
(define_insn_reservation "11_branches" 0
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "branch"))
|
(eq_attr "type" "branch"))
|
"nothing")
|
"nothing")
|
|
|
;; Call latencies are not predictable. A semi-arbitrary very large
|
;; Call latencies are not predictable. A semi-arbitrary very large
|
;; number is used as "positive infinity" so that everything should be
|
;; number is used as "positive infinity" so that everything should be
|
;; finished by the time of return.
|
;; finished by the time of return.
|
(define_insn_reservation "11_call" 32
|
(define_insn_reservation "11_call" 32
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "call"))
|
(eq_attr "type" "call"))
|
"nothing")
|
"nothing")
|
|
|
;; Branches are predicted. A correctly predicted branch will be no
|
;; Branches are predicted. A correctly predicted branch will be no
|
;; cost, but we're conservative here, and use the timings a
|
;; cost, but we're conservative here, and use the timings a
|
;; late-register would give us.
|
;; late-register would give us.
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
"11_branches")
|
"11_branches")
|
(define_bypass 2 "11_alu_shift_reg_op"
|
(define_bypass 2 "11_alu_shift_reg_op"
|
"11_branches")
|
"11_branches")
|
(define_bypass 2 "11_load1,11_load2"
|
(define_bypass 2 "11_load1,11_load2"
|
"11_branches")
|
"11_branches")
|
(define_bypass 3 "11_load34"
|
(define_bypass 3 "11_load34"
|
"11_branches")
|
"11_branches")
|
|
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;; Load/Store Instructions
|
;; Load/Store Instructions
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
|
|
|
;; The models for load/store instructions do not accurately describe
|
;; The models for load/store instructions do not accurately describe
|
;; the difference between operations with a base register writeback.
|
;; the difference between operations with a base register writeback.
|
;; These models assume that all memory references hit in dcache. Also,
|
;; These models assume that all memory references hit in dcache. Also,
|
;; if the PC is one of the registers involved, there are additional stalls
|
;; if the PC is one of the registers involved, there are additional stalls
|
;; not modelled here. Addressing modes are also not modelled.
|
;; not modelled here. Addressing modes are also not modelled.
|
|
|
(define_insn_reservation "11_load1" 3
|
(define_insn_reservation "11_load1" 3
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "load1"))
|
(eq_attr "type" "load1"))
|
"l_a+e_1,l_dc1,l_dc2,l_wb")
|
"l_a+e_1,l_dc1,l_dc2,l_wb")
|
|
|
;; Load byte results are not available until the writeback stage, where
|
;; Load byte results are not available until the writeback stage, where
|
;; the correct byte is extracted.
|
;; the correct byte is extracted.
|
|
|
(define_insn_reservation "11_loadb" 4
|
(define_insn_reservation "11_loadb" 4
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "load_byte"))
|
(eq_attr "type" "load_byte"))
|
"l_a+e_1,l_dc1,l_dc2,l_wb")
|
"l_a+e_1,l_dc1,l_dc2,l_wb")
|
|
|
(define_insn_reservation "11_store1" 0
|
(define_insn_reservation "11_store1" 0
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "store1"))
|
(eq_attr "type" "store1"))
|
"l_a+e_1,l_dc1,l_dc2,l_wb")
|
"l_a+e_1,l_dc1,l_dc2,l_wb")
|
|
|
;; Load/store double words into adjacent registers. The timing and
|
;; Load/store double words into adjacent registers. The timing and
|
;; latencies are different depending on whether the address is 64-bit
|
;; latencies are different depending on whether the address is 64-bit
|
;; aligned. This model assumes that it is.
|
;; aligned. This model assumes that it is.
|
(define_insn_reservation "11_load2" 3
|
(define_insn_reservation "11_load2" 3
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "load2"))
|
(eq_attr "type" "load2"))
|
"l_a+e_1,l_dc1,l_dc2,l_wb")
|
"l_a+e_1,l_dc1,l_dc2,l_wb")
|
|
|
(define_insn_reservation "11_store2" 0
|
(define_insn_reservation "11_store2" 0
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "store2"))
|
(eq_attr "type" "store2"))
|
"l_a+e_1,l_dc1,l_dc2,l_wb")
|
"l_a+e_1,l_dc1,l_dc2,l_wb")
|
|
|
;; Load/store multiple registers. Two registers are stored per cycle.
|
;; Load/store multiple registers. Two registers are stored per cycle.
|
;; Actual timing depends on how many registers are affected, so we
|
;; Actual timing depends on how many registers are affected, so we
|
;; optimistically schedule a low latency.
|
;; optimistically schedule a low latency.
|
(define_insn_reservation "11_load34" 4
|
(define_insn_reservation "11_load34" 4
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "load3,load4"))
|
(eq_attr "type" "load3,load4"))
|
"l_a+e_1,l_dc1*2,l_dc2,l_wb")
|
"l_a+e_1,l_dc1*2,l_dc2,l_wb")
|
|
|
(define_insn_reservation "11_store34" 0
|
(define_insn_reservation "11_store34" 0
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(and (eq_attr "tune" "arm1136js,arm1136jfs")
|
(eq_attr "type" "store3,store4"))
|
(eq_attr "type" "store3,store4"))
|
"l_a+e_1,l_dc1*2,l_dc2,l_wb")
|
"l_a+e_1,l_dc1*2,l_dc2,l_wb")
|
|
|
;; A store can start immediately after an alu op, if that alu op does
|
;; A store can start immediately after an alu op, if that alu op does
|
;; not provide part of the address to access.
|
;; not provide part of the address to access.
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
(define_bypass 1 "11_alu_op,11_alu_shift_op"
|
"11_store1"
|
"11_store1"
|
"arm_no_early_store_addr_dep")
|
"arm_no_early_store_addr_dep")
|
(define_bypass 2 "11_alu_shift_reg_op"
|
(define_bypass 2 "11_alu_shift_reg_op"
|
"11_store1"
|
"11_store1"
|
"arm_no_early_store_addr_dep")
|
"arm_no_early_store_addr_dep")
|
|
|
;; An alu op can start sooner after a load, if that alu op does not
|
;; An alu op can start sooner after a load, if that alu op does not
|
;; have an early register dependency on the load
|
;; have an early register dependency on the load
|
(define_bypass 2 "11_load1"
|
(define_bypass 2 "11_load1"
|
"11_alu_op")
|
"11_alu_op")
|
(define_bypass 2 "11_load1"
|
(define_bypass 2 "11_load1"
|
"11_alu_shift_op"
|
"11_alu_shift_op"
|
"arm_no_early_alu_shift_value_dep")
|
"arm_no_early_alu_shift_value_dep")
|
(define_bypass 2 "11_load1"
|
(define_bypass 2 "11_load1"
|
"11_alu_shift_reg_op"
|
"11_alu_shift_reg_op"
|
"arm_no_early_alu_shift_dep")
|
"arm_no_early_alu_shift_dep")
|
|
|
(define_bypass 3 "11_loadb"
|
(define_bypass 3 "11_loadb"
|
"11_alu_op")
|
"11_alu_op")
|
(define_bypass 3 "11_loadb"
|
(define_bypass 3 "11_loadb"
|
"11_alu_shift_op"
|
"11_alu_shift_op"
|
"arm_no_early_alu_shift_value_dep")
|
"arm_no_early_alu_shift_value_dep")
|
(define_bypass 3 "11_loadb"
|
(define_bypass 3 "11_loadb"
|
"11_alu_shift_reg_op"
|
"11_alu_shift_reg_op"
|
"arm_no_early_alu_shift_dep")
|
"arm_no_early_alu_shift_dep")
|
|
|
;; A mul op can start sooner after a load, if that mul op does not
|
;; A mul op can start sooner after a load, if that mul op does not
|
;; have an early multiply dependency
|
;; have an early multiply dependency
|
(define_bypass 2 "11_load1"
|
(define_bypass 2 "11_load1"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"arm_no_early_mul_dep")
|
"arm_no_early_mul_dep")
|
(define_bypass 3 "11_load34"
|
(define_bypass 3 "11_load34"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"arm_no_early_mul_dep")
|
"arm_no_early_mul_dep")
|
(define_bypass 3 "11_loadb"
|
(define_bypass 3 "11_loadb"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"11_mult1,11_mult2,11_mult3,11_mult4,11_mult5,11_mult6,11_mult7"
|
"arm_no_early_mul_dep")
|
"arm_no_early_mul_dep")
|
|
|
;; A store can start sooner after a load, if that load does not
|
;; A store can start sooner after a load, if that load does not
|
;; produce part of the address to access
|
;; produce part of the address to access
|
(define_bypass 2 "11_load1"
|
(define_bypass 2 "11_load1"
|
"11_store1"
|
"11_store1"
|
"arm_no_early_store_addr_dep")
|
"arm_no_early_store_addr_dep")
|
(define_bypass 3 "11_loadb"
|
(define_bypass 3 "11_loadb"
|
"11_store1"
|
"11_store1"
|
"arm_no_early_store_addr_dep")
|
"arm_no_early_store_addr_dep")
|
|
|
