Commit 55d982c3 authored by danyao12's avatar danyao12
Browse files

add benchmark_bwd_v3.sh

parent 63b152d6
#!/bin/sh
# TODO: run this script from CK root or build directory
EXE="$(find . -name tile_example_fmha_bwd -type f | head -n 1)"
VALID=0
for prec in "bf16" ; do
for perm in 0 ; do
for hdim in 128 ; do
nhead=$((2048 / $hdim)) # follow fav2 setup
$EXE -prec=$prec -b=4 -h=$nhead -d=$hdim -s=4096 -iperm=$perm -operm=$perm -kname=1 -bwd_v3=1 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
# $EXE -prec=$prec -b=4 -h=$nhead -d=$hdim -s=4096 -iperm=$perm -operm=$perm -kname=1 -mask=1 -bwd_v3=1 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
$EXE -prec=$prec -b=4 -h=$nhead -d=$hdim -s=4096 -iperm=$perm -operm=$perm -kname=1 -bwd_v3=1 -v3_atomic_fp32=0 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
# $EXE -prec=$prec -b=4 -h=$nhead -d=$hdim -s=4096 -iperm=$perm -operm=$perm -kname=1 -mask=1 -bwd_v3=1 -v3_atomic_fp32=0 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
$EXE -prec=$prec -b=2 -h=$nhead -d=$hdim -s=8192 -iperm=$perm -operm=$perm -kname=1 -bwd_v3=1 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
# $EXE -prec=$prec -b=2 -h=$nhead -d=$hdim -s=8192 -iperm=$perm -operm=$perm -kname=1 -mask=1 -bwd_v3=1 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
$EXE -prec=$prec -b=2 -h=$nhead -d=$hdim -s=8192 -iperm=$perm -operm=$perm -kname=1 -bwd_v3=1 -v3_atomic_fp32=0 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
# $EXE -prec=$prec -b=2 -h=$nhead -d=$hdim -s=8192 -iperm=$perm -operm=$perm -kname=1 -mask=1 -bwd_v3=1 -v3_atomic_fp32=0 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
$EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=16384 -iperm=$perm -operm=$perm -kname=1 -bwd_v3=1 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
# $EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=16384 -iperm=$perm -operm=$perm -kname=1 -mask=1 -bwd_v3=1 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
$EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=16384 -iperm=$perm -operm=$perm -kname=1 -bwd_v3=1 -v3_atomic_fp32=0 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
# $EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=16384 -iperm=$perm -operm=$perm -kname=1 -mask=1 -bwd_v3=1 -v3_atomic_fp32=0 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
$EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=32768 -iperm=$perm -operm=$perm -kname=1 -bwd_v3=1 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
# $EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=32768 -iperm=$perm -operm=$perm -kname=1 -mask=1 -bwd_v3=1 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
$EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=32768 -iperm=$perm -operm=$perm -kname=1 -bwd_v3=1 -v3_atomic_fp32=0 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
# $EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=32768 -iperm=$perm -operm=$perm -kname=1 -mask=1 -bwd_v3=1 -v3_atomic_fp32=0 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
$EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=65536 -iperm=$perm -operm=$perm -kname=1 -bwd_v3=1 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
# $EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=65536 -iperm=$perm -operm=$perm -kname=1 -mask=1 -bwd_v3=1 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
$EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=65536 -iperm=$perm -operm=$perm -kname=1 -bwd_v3=1 -v3_atomic_fp32=0 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
# $EXE -prec=$prec -b=1 -h=$nhead -d=$hdim -s=65536 -iperm=$perm -operm=$perm -kname=1 -mask=1 -bwd_v3=1 -v3_atomic_fp32=0 -v3_bf16_cvt=2 -v=$VALID ; sleep 3
done
done
done
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment