Files
OSACA/benchmarks/cmp-r32_r32-TP.S
2017-10-12 18:55:45 +02:00

143 lines
2.8 KiB
ArmAsm

#define INSTR cmp
#define NINST 64
#define N edi
#define i r8d
.intel_syntax noprefix
.globl ninst
.data
ninst:
.long NINST
.align 32
PI:
.long 0xf01b866e, 0x400921f9, 0xf01b866e, 0x400921f9, 0xf01b866e, 0x400921f9, 0xf01b866e, 0x400921f9, 0xf01b866e, 0x400921f9, 0xf01b866e, 0x400921f9, 0xf01b866e, 0x400921f9, 0xf01b866e, 0x400921f9
.text
.globl latency
.type latency, @function
.align 32
latency:
push rbp
mov rbp, rsp
xor i, i
test N, N
jle done
# create DP 1.0
vpcmpeqw xmm0, xmm0, xmm0 # all ones
vpsllq xmm0, xmm0, 54 # logical left shift: 11111110..0 (54=64-(10-1))
vpsrlq xmm0, xmm0, 2 # logical right shift: 1 bit for sign; leading mantissa bit is zero
push rax
push rbx
push rcx
push rdx
push r9
push r10
push r11
push r12
push r13
push r14
push r15
xor rax, rax
xor rbx, rbx
xor rcx, rcx
xor rdx, rdx
xor r9, r9
xor r10, r10
xor r11, r11
xor r12, r12
xor r13, r13
xor r14, r14
xor r15, r15
# copy DP 1.0
vmovq rax, xmm0
vmovq rbx, xmm0
# Create DP 2.0
add rbx, rax
# Create DP 0.5
div rax
movq rcx, rax
vmovq rax, xmm0
loop:
inc i
INSTR edx, eax
INSTR r9d, ebx
INSTR r10d, ecx
INSTR r11d, eax
INSTR r12d, ebx
INSTR r13d, ecx
INSTR r14d, eax
INSTR r15d, ebx
INSTR edx, ecx
INSTR r9d, eax
INSTR r10d, ebx
INSTR r11d, ecx
INSTR r12d, eax
INSTR r13d, ebx
INSTR r14d, ecx
INSTR r15d, eax
INSTR edx, ebx
INSTR r9d, ecx
INSTR r10d, eax
INSTR r11d, ebx
INSTR r12d, ecx
INSTR r13d, eax
INSTR r14d, ebx
INSTR r15d, ecx
INSTR edx, eax
INSTR r9d, ebx
INSTR r10d, ecx
INSTR r11d, eax
INSTR r12d, ebx
INSTR r13d, ecx
INSTR r14d, eax
INSTR r15d, ebx
INSTR edx, ecx
INSTR r9d, eax
INSTR r10d, ebx
INSTR r11d, ecx
INSTR r12d, eax
INSTR r13d, ebx
INSTR r14d, ecx
INSTR r15d, eax
INSTR edx, ebx
INSTR r9d, ecx
INSTR r10d, eax
INSTR r11d, ebx
INSTR r12d, ecx
INSTR r13d, eax
INSTR r14d, ebx
INSTR r15d, ecx
INSTR edx, eax
INSTR r9d, ebx
INSTR r10d, ecx
INSTR r11d, eax
INSTR r12d, ebx
INSTR r13d, ecx
INSTR r14d, eax
INSTR r15d, ebx
INSTR edx, ecx
INSTR r9d, eax
INSTR r10d, ebx
INSTR r11d, ecx
INSTR r12d, eax
INSTR r13d, ebx
INSTR r14d, ecx
INSTR r15d, eax
cmp i, N
jl loop
pop r15
pop r14
pop r13
pop r12
pop r11
pop r10
pop r9
pop rdx
pop rcx
pop rbx
pop rax
done:
mov rsp, rbp
pop rbp
ret
.size latency, .-latency