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
|
;; Scheduling description for the Xilinx PowerPC 405 APU Floating Point Unit.
;; Copyright (C) 2008-2015 Free Software Foundation, Inc.
;; Contributed by Michael Eager (eager@eagercon.com).
;;
;; 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/>.
;;----------------------------------------------------
;; Xilinx APU FPU Pipeline Description
;;
;; - attr 'type' and 'fp_type' should definitely
;; be cleaned up at some point in the future.
;; ddiv,sdiv,dmul,smul etc are quite confusing.
;; Should use consistent fp* attrs. 'fp_type'
;; should also go away, leaving us only with 'fp'
;;
;;----------------------------------------------------
;; -------------------------------------------------------------------------
;; Latencies
;; Latest latency figures (all in FCB cycles). PowerPC to FPU frequency ratio
;; assumed to be 1/2. (most common deployment)
;; Add 2 PPC cycles for (register file access + wb) and 2 PPC cycles
;; for issue (from PPC)
;; SP DP
;; Loads: 4 6
;; Stores: 1 2 (from availability of data)
;; Move/Abs/Neg: 1 1
;; Add/Subtract: 5 7
;; Multiply: 4 11
;; Multiply-add: 10 19
;; Convert (any): 4 6
;; Divide/Sqrt: 27 56
;; Compares: 1 2
;;
;; bypasses needed for forwarding capability of the FPU.
;; Add this at some future time.
;; -------------------------------------------------------------------------
(define_automaton "Xfpu")
(define_cpu_unit "Xfpu_issue,Xfpu_addsub,Xfpu_mul,Xfpu_div,Xfpu_sqrt" "Xfpu")
(define_insn_reservation "fp-default" 2
(and (and
(eq_attr "type" "fp")
(eq_attr "fp_type" "fp_default"))
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*2")
(define_insn_reservation "fp-compare" 6
(and (eq_attr "type" "fpcompare") ;; Inconsistent naming
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*2,Xfpu_addsub")
(define_insn_reservation "fp-addsub-s" 14
(and (and
(eq_attr "type" "fp")
(eq_attr "fp_type" "fp_addsub_s"))
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*2,Xfpu_addsub")
(define_insn_reservation "fp-addsub-d" 18
(and (and
(eq_attr "type" "fp")
(eq_attr "fp_type" "fp_addsub_d"))
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*2,Xfpu_addsub")
(define_insn_reservation "fp-mul-s" 12
(and (and
(eq_attr "type" "fp")
(eq_attr "fp_type" "fp_mul_s"))
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*2,Xfpu_mul")
(define_insn_reservation "fp-mul-d" 16 ;; Actually 28. Long latencies are killing the automaton formation. Need to figure out why.
(and (and
(eq_attr "type" "fp")
(eq_attr "fp_type" "fp_mul_d"))
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*2,Xfpu_mul")
(define_insn_reservation "fp-div-s" 24 ;; Actually 34
(and (eq_attr "type" "sdiv") ;; Inconsistent attr naming
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*2,Xfpu_div*10") ;; Unpipelined
(define_insn_reservation "fp-div-d" 34 ;; Actually 116
(and (eq_attr "type" "ddiv")
(eq_attr "cpu" "ppc405")) ;; Inconsistent attr naming
"Xfpu_issue*2,Xfpu_div*10") ;; Unpipelined
(define_insn_reservation "fp-maddsub-s" 24
(and (and
(eq_attr "type" "fp")
(eq_attr "fp_type" "fp_maddsub_s"))
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*2,Xfpu_mul,nothing*7,Xfpu_addsub")
(define_insn_reservation "fp-maddsub-d" 34 ;; Actually 42
(and (and
(eq_attr "type" "dmul") ;; Inconsistent attr naming
(eq_attr "fp_type" "fp_maddsub_d"))
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*2,Xfpu_mul,nothing*7,Xfpu_addsub")
(define_insn_reservation "fp-load" 10 ;; FIXME. Is double/single precision the same ?
(and (eq_attr "type" "fpload")
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*10")
(define_insn_reservation "fp-store" 4
(and (eq_attr "type" "fpstore")
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*4")
(define_insn_reservation "fp-sqrt-s" 24 ;; Actually 56
(and (eq_attr "type" "ssqrt")
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*2,Xfpu_sqrt*10") ;; Unpipelined
(define_insn_reservation "fp-sqrt-d" 34 ;; Actually 116
(and (eq_attr "type" "dsqrt")
(eq_attr "cpu" "ppc405"))
"Xfpu_issue*2,Xfpu_sqrt*10") ;; Unpipelined
|
