summaryrefslogtreecommitdiff
path: root/gcc/config/sh/sh4-300.md
blob: b41f1cc061a3c134b52da2cd8043743aab425323 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
;; DFA scheduling description for ST40-300.
;; Copyright (C) 2004-2015 Free Software Foundation, Inc.

;; This file is part of GCC.

;; GCC is free software; you can redistribute it and/or modify
;; it under the terms of the GNU General Public License as published by
;; the Free Software Foundation; either version 3, or (at your option)
;; any later version.

;; GCC is distributed in the hope that it will be useful,
;; but WITHOUT ANY WARRANTY; without even the implied warranty of
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
;; GNU General Public License for more details.

;; You should have received a copy of the GNU General Public License
;; along with GCC; see the file COPYING3.  If not see
;; <http://www.gnu.org/licenses/>.

;; Load and store instructions save a cycle if they are aligned on a
;; four byte boundary.  Using a function unit for stores encourages
;; gcc to separate load and store instructions by one instruction,
;; which makes it more likely that the linker will be able to word
;; align them when relaxing.

;; The following description models the ST40-300 pipeline using the DFA based
;; scheduler.

;; Two automata are defined to reduce number of states
;; which a single large automaton will have. (Factoring)

(define_automaton "sh4_300_inst_pipeline,sh4_300_fpu_pipe")

;; This unit is basically the decode unit of the processor.
;; Since SH4 is a dual issue machine,it is as if there are two
;; units so that any insn can be processed by either one
;; of the decoding unit.
(define_cpu_unit "sh4_300_pipe_01,sh4_300_pipe_02" "sh4_300_inst_pipeline")

;; The floating point units.
(define_cpu_unit "sh4_300_fpt,sh4_300_fpu,sh4_300_fds" "sh4_300_fpu_pipe")

;; integer multiplier unit
(define_cpu_unit "sh4_300_mul" "sh4_300_inst_pipeline")

;; LS unit
(define_cpu_unit "sh4_300_ls" "sh4_300_inst_pipeline")

;; The address calculator used for branch instructions.
;; This will be reserved after "issue" of branch instructions
;; and this is to make sure that no two branch instructions
;; can be issued in parallel.
(define_cpu_unit "sh4_300_br" "sh4_300_inst_pipeline")

;; ----------------------------------------------------
;; This reservation is to simplify the dual issue description.

(define_reservation  "sh4_300_issue"  "sh4_300_pipe_01|sh4_300_pipe_02")

(define_reservation "all" "sh4_300_pipe_01+sh4_300_pipe_02")

;;(define_insn_reservation "nil" 0 (eq_attr "type" "nil") "nothing")

;; MOV RM,RN / MOV #imm8,RN / STS PR,RN
(define_insn_reservation "sh4_300_mov" 0
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "move,movi8,prget"))
  "sh4_300_issue")

;; Fixed STS from MACL / MACH
(define_insn_reservation "sh4_300_mac_gp" 0
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "mac_gp"))
  "sh4_300_issue+sh4_300_mul")

;; Fixed LDS to MACL / MACH
(define_insn_reservation "sh4_300_gp_mac" 1
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "gp_mac"))
  "sh4_300_issue+sh4_300_mul")

;; Instructions without specific resource requirements with latency 1.
(define_insn_reservation "sh4_300_simple_arith" 1
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "mt_group,arith,dyn_shift,prset"))
  "sh4_300_issue")

;; Load and store instructions have no alignment peculiarities for the ST40-300,
;; but they use the load-store unit, which they share with the fmove type
;; insns (fldi[01]; fmov frn,frm; flds; fsts; fabs; fneg) .
;; Loads have a latency of three.

;; Load Store instructions.
(define_insn_reservation "sh4_300_load" 3
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "load,pcload,load_si,pcload_si,pload"))
  "sh4_300_issue+sh4_300_ls")

(define_insn_reservation "sh4_300_mac_load" 3
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "mem_mac"))
  "sh4_300_issue+sh4_300_ls+sh4_300_mul")

(define_insn_reservation "sh4_300_fload" 4
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "fload,pcfload"))
  "sh4_300_issue+sh4_300_ls+sh4_300_fpt")

;; sh_adjust_cost describes the reduced latency of the feeding insns of a store.
;; The latency of an auto-increment register is 1; the latency of the memory
;; output is not actually considered here anyway.
(define_insn_reservation "sh4_300_store" 1
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "store,pstore"))
  "sh4_300_issue+sh4_300_ls")

(define_insn_reservation "sh4_300_fstore" 1
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "fstore"))
  "sh4_300_issue+sh4_300_ls+sh4_300_fpt")

;; Fixed STS.L from MACL / MACH
(define_insn_reservation "sh4_300_mac_store" 1
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "mac_mem"))
  "sh4_300_issue+sh4_300_mul+sh4_300_ls")

(define_insn_reservation "sh4_300_gp_fpul" 2
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "gp_fpul"))
  "sh4_300_issue+sh4_300_fpt")

(define_insn_reservation "sh4_300_fpul_gp" 1
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "fpul_gp"))
  "sh4_300_issue+sh4_300_fpt")

;; Branch (BF,BF/S,BT,BT/S,BRA)
;; Branch Far (JMP,RTS,BRAF)
;; Group:	BR
;; When displacement is 0 for BF / BT, we have effectively conditional
;; execution of one instruction, without pipeline disruption.
;; Otherwise, the latency depends on prediction success.
;; We can't really do much with the latency, even if we could express it,
;; but the pairing restrictions are useful to take into account.
;; ??? If the branch is likely, and not paired with a preceding insn,
;; or likely and likely not predicted, we might want to fill the delay slot.
;; However, there appears to be no machinery to make the compiler
;; recognize these scenarios.
(define_insn_reservation "sh4_300_branch"  1
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "cbranch,jump,return,jump_ind"))
  "sh4_300_issue+sh4_300_br")

;; RTE
(define_insn_reservation "sh4_300_return_from_exp" 9
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "rte"))
  "sh4_300_pipe_01+sh4_300_pipe_02*9")

;; OCBP, OCBWB
;; Group:	CO
;; Latency: 	1-5
;; Issue Rate: 	1
;; cwb is used for the sequence
;;	ocbwb  @%0
;;	extu.w %0,%2
;;	or     %1,%2
;;	mov.l  %0,@%2
;; This description is likely inexact, but this pattern should not actually
;; appear when compiling for sh4-300; we should use isbi instead.
;; If a -mtune option is added later, we should use the icache array
;; dispatch method instead.
(define_insn_reservation "sh4_300_ocbwb"  3
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "cwb"))
  "all*3")

;; JSR,BSR,BSRF
;; Calls have a mandatory delay slot, which we'd like to fill with an insn
;; that can be paired with the call itself.
;; Scheduling runs before reorg, so we approximate this by saying that we
;; want the call to be paired with a preceding insn.
;; In most cases, the insn that loads the address of the call should have
;; a nonzero latency (mov rn,rm doesn't make sense since we could use rn
;; for the address then).  Thus, a preceding insn that can be paired with
;; a call should be eligible for the delay slot.
;;
;; calls introduce a longisch delay that is likely to flush the pipelines
;; of the caller's instructions.  Ordinary functions tend to end with a
;; load to restore a register (in the delay slot of rts), while sfuncs
;; tend to end with an EX or MT insn.  But that is not actually relevant,
;; since there are no instructions that contend for memory access early.
;; We could, of course, provide exact scheduling information for specific
;; sfuncs, if that should prove useful.
(define_insn_reservation "sh4_300_call" 16
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "call,sfunc"))
  "sh4_300_issue+sh4_300_br,all*15")

;; FMOV.S / FMOV.D
(define_insn_reservation "sh4_300_fmov" 1
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "fmove"))
  "sh4_300_issue+sh4_300_fpt")

;; LDS to FPSCR
(define_insn_reservation "sh4_300_fpscr_load" 8
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "gp_fpscr"))
  "sh4_300_issue+sh4_300_fpu+sh4_300_fpt")

;; LDS.L to FPSCR
(define_insn_reservation "sh4_300_fpscr_load_mem" 8
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type"  "mem_fpscr"))
  "sh4_300_issue+sh4_300_fpu+sh4_300_fpt+sh4_300_ls")


;; Fixed point multiplication (DMULS.L DMULU.L MUL.L MULS.W,MULU.W)
(define_insn_reservation "multi" 2
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "smpy,dmpy"))
  "sh4_300_issue+sh4_300_mul")

;; FPCHG, FRCHG, FSCHG
(define_insn_reservation "fpscr_toggle"  1
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "fpscr_toggle"))
  "sh4_300_issue+sh4_300_fpu+sh4_300_fpt")

;; FCMP/EQ, FCMP/GT
(define_insn_reservation "fp_cmp"  3
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "fp_cmp,dfp_cmp"))
  "sh4_300_issue+sh4_300_fpu")

;; Single precision floating point (FADD,FLOAT,FMAC,FMUL,FSUB,FTRC)
;; Double-precision floating-point (FADD,FCNVDS,FCNVSD,FLOAT,FSUB,FTRC)
(define_insn_reservation "fp_arith"  6
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "fp,ftrc_s,dfp_arith,dfp_conv"))
  "sh4_300_issue+sh4_300_fpu")

;; Single Precision FDIV/SQRT
(define_insn_reservation "fp_div" 19
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "fdiv"))
  "sh4_300_issue+sh4_300_fpu+sh4_300_fds,sh4_300_fds*15")

;; Double-precision floating-point FMUL
(define_insn_reservation "dfp_mul" 9
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "dfp_mul"))
  "sh4_300_issue+sh4_300_fpu,sh4_300_fpu*3")

;; Double precision FDIV/SQRT
(define_insn_reservation "dp_div" 35
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "dfdiv"))
  "sh4_300_issue+sh4_300_fpu+sh4_300_fds,sh4_300_fds*31")

;; ??? We don't really want these for sh4-300.
;; this pattern itself is likely to finish in 3 cycles, but also
;; to disrupt branch prediction for taken branches for the following
;; condbranch.
(define_insn_reservation "sh4_300_arith3" 5
  (and (eq_attr "pipe_model" "sh4_300")
       (eq_attr "type" "arith3"))
  "sh4_300_issue,all*4")

;; arith3b insns without brach redirection make use of the 0-offset 0-latency
;; branch feature, and thus schedule the same no matter if the branch is taken
;; or not.  If the branch is redirected, the taken branch might take longer,
;; but then, we don't have to take the next branch.
;; ??? should we suppress branch redirection for sh4-300 to improve branch
;; target hit rates?
(define_insn_reservation "arith3b" 2
  (and (eq_attr "pipe_model" "sh4")
       (eq_attr "type" "arith3"))
  "issue,all")