Debian Codegraph @

llvm/test/tools/llvm-mca/AArch64/Exynos/float-recp.s

raw
1 # NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
2 # RUN: llvm-mca -march=aarch64 -mcpu=exynos-m3 -resource-pressure=false < %s | FileCheck %s -check-prefixes=ALL,M3
3 # RUN: llvm-mca -march=aarch64 -mcpu=exynos-m4 -resource-pressure=false < %s | FileCheck %s -check-prefixes=ALL,M4
4 # RUN: llvm-mca -march=aarch64 -mcpu=exynos-m5 -resource-pressure=false < %s | FileCheck %s -check-prefixes=ALL,M5
5
6 fmov s31, #1.00000000
7 fdiv s30, s31, s30
8
9 # Newton series for 1 / x.
10 frecpe s1, s0
11 frecps s2, s0, s1
12 fmul s1, s1, s2
13 frecps s0, s0, s1
14 fmul s0, s1, s0
15
16 # ALL: Iterations: 100
17 # ALL-NEXT: Instructions: 700
18
19 # M3-NEXT: Total Cycles: 1803
20 # M4-NEXT: Total Cycles: 1703
21 # M5-NEXT: Total Cycles: 1703
22
23 # ALL-NEXT: Total uOps: 700
24
25 # ALL: Dispatch Width: 6
26
27 # M3-NEXT: uOps Per Cycle: 0.39
28 # M3-NEXT: IPC: 0.39
29 # M3-NEXT: Block RThroughput: 2.0
30
31 # M4-NEXT: uOps Per Cycle: 0.41
32 # M4-NEXT: IPC: 0.41
33 # M4-NEXT: Block RThroughput: 1.5
34
35 # M5-NEXT: uOps Per Cycle: 0.41
36 # M5-NEXT: IPC: 0.41
37 # M5-NEXT: Block RThroughput: 1.3
38
39 # ALL: Instruction Info:
40 # ALL-NEXT: [1]: #uOps
41 # ALL-NEXT: [2]: Latency
42 # ALL-NEXT: [3]: RThroughput
43 # ALL-NEXT: [4]: MayLoad
44 # ALL-NEXT: [5]: MayStore
45 # ALL-NEXT: [6]: HasSideEffects (U)
46
47 # ALL: [1] [2] [3] [4] [5] [6] Instructions:
48 # ALL-NEXT: 1 1 0.33 fmov s31, #1.00000000
49
50 # M3-NEXT: 1 7 2.00 fdiv s30, s31, s30
51 # M3-NEXT: 1 4 0.50 frecpe s1, s0
52
53 # M4-NEXT: 1 7 1.50 fdiv s30, s31, s30
54 # M4-NEXT: 1 3 0.50 frecpe s1, s0
55
56 # M5-NEXT: 1 7 1.00 fdiv s30, s31, s30
57 # M5-NEXT: 1 3 0.50 frecpe s1, s0
58
59 # ALL-NEXT: 1 4 0.33 frecps s2, s0, s1
60 # ALL-NEXT: 1 3 0.33 fmul s1, s1, s2
61 # ALL-NEXT: 1 4 0.33 frecps s0, s0, s1
62 # ALL-NEXT: 1 3 0.33 fmul s0, s1, s0