mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-14 18:17:44 +00:00
GEMM+Bias+ReLU+Add (#76)
* tweak conv for odd C
* update script
* clean up elementwise op
* fix build
* clean up
* added example for gemm+bias+relu+add
* added example for gemm+bias+relu
* add profiler for gemm_s_shuffle; re-org files
* add profiler
* fix build
* clean up
* clean up
* clean up
* fix build
[ROCm/composable_kernel commit: 823657ed12]
This commit is contained in:
46
script/conv2d_fwd.sh
Executable file
46
script/conv2d_fwd.sh
Executable file
@@ -0,0 +1,46 @@
|
||||
#!/bin/bash
|
||||
|
||||
## GPU visibility
|
||||
export HIP_VISIBLE_DEVICES=0
|
||||
|
||||
make -j $1
|
||||
|
||||
DRIVER=example/$1
|
||||
VERIFY=$2
|
||||
INIT=$3
|
||||
REPEAT=$4
|
||||
|
||||
# test
|
||||
######## verify init repeat N__ K___ C___ Y X Hi__ Wi__ Strides Dilations LeftPads RightPads Desired_grid_size__
|
||||
$DRIVER $VERIFY $INIT $REPEAT 128 256 192 3 3 71 71 2 2 1 1 1 1 1 1 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 128 256 64 1 1 1 1 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 256 64 3 7 7 230 230 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 128 512 512 3 3 7 7 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 256 64 3 7 7 224 224 2 2 1 1 3 3 3 3
|
||||
|
||||
N=$5
|
||||
|
||||
# Resnet50
|
||||
######## verify init repeat N__ K___ C___ Y X Hi__ Wi__ Strides Dilations LeftPads RightPads Desired_grid_size__
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 2048 1024 1 1 14 14 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 256 1024 1 1 14 14 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 1024 1 1 14 14 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 128 128 3 3 28 28 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 128 1 1 28 28 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 128 128 3 3 58 58 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 2048 1 1 7 7 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 1024 256 1 1 14 14 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 256 256 3 3 14 14 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 256 256 3 3 30 30 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 128 256 1 1 56 56 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 256 1 1 56 56 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 64 256 1 1 56 56 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 512 3 3 16 16 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 1024 512 1 1 28 28 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 128 512 1 1 28 28 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 256 512 1 1 28 28 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 2048 512 1 1 7 7 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 512 3 3 7 7 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 256 64 1 1 56 56 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 64 64 1 1 56 56 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 64 64 3 3 56 56 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
|
||||
20
script/gemm.sh
Executable file
20
script/gemm.sh
Executable file
@@ -0,0 +1,20 @@
|
||||
#!/bin/bash
|
||||
|
||||
## GPU visibility
|
||||
export HIP_VISIBLE_DEVICES=0
|
||||
|
||||
make -j $1
|
||||
|
||||
DRIVER=example/$1
|
||||
VERIFY=$2
|
||||
INIT=$3
|
||||
REPEAT=$4
|
||||
|
||||
######## verify init repeat M___ N___ K___ StrideA StrideB StrideC StrideC1
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 256 256 256 256 256 256 256
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 960 1024 1024 1024 1024 1024 1024
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 1920 2048 2048 2048 2048 2048 2048
|
||||
$DRIVER $VERIFY $INIT $REPEAT 3840 4096 4096 4096 4096 4096 4096
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 7680 8192 8192 8192 8192 8192 8192
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 1024 1024 1024 1024 1024 1024 1024
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 2048 2048 2048 2048 2048 2048 2048
|
||||
46
script/pool2d_fwd.sh
Executable file
46
script/pool2d_fwd.sh
Executable file
@@ -0,0 +1,46 @@
|
||||
#!/bin/bash
|
||||
|
||||
## GPU visibility
|
||||
export HIP_VISIBLE_DEVICES=0
|
||||
|
||||
make -j $1
|
||||
|
||||
DRIVER=example/$1
|
||||
VERIFY=$2
|
||||
INIT=$3
|
||||
REPEAT=$4
|
||||
|
||||
# test
|
||||
######## verify init repeat N__ C___ Y X Hi__ Wi__ Strides LeftPads RightPads
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 128 192 3 3 71 71 2 2 1 1 1 1
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 128 64 1 1 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT 256 3 7 7 230 230 2 2 0 0 0 0
|
||||
$DRIVER $VERIFY $INIT $REPEAT 256 1024 14 14 14 14 1 1 0 0 0 0
|
||||
|
||||
N=$5
|
||||
|
||||
# Resnet50
|
||||
######## verify init repeat N__ C___ Y X Hi__ Wi__ Strides LeftPads RightPads
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 1024 1 1 14 14 2 2 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 1024 1 1 14 14 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 1024 1 1 14 14 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 128 3 3 28 28 1 1 1 1 1 1
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 128 1 1 28 28 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 128 3 3 58 58 2 2 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 2048 1 1 7 7 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 256 1 1 14 14 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 256 3 3 14 14 1 1 1 1 1 1
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 256 3 3 30 30 2 2 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 256 1 1 56 56 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 256 1 1 56 56 2 2 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 256 1 1 56 56 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 3 3 16 16 2 2 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 1 1 28 28 2 2 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 1 1 28 28 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 1 1 28 28 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 1 1 7 7 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 512 3 3 7 7 1 1 1 1 1 1
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 64 1 1 56 56 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 64 1 1 56 56 1 1 0 0 0 0
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 64 3 3 56 56 1 1 1 1 1 1
|
||||
#$DRIVER $VERIFY $INIT $REPEAT $N 3 7 7 230 230 2 2 0 0 0 0
|
||||
@@ -19,11 +19,89 @@ REPEAT=$9
|
||||
|
||||
# test
|
||||
######## op datatype in_layout wei_layout out_layout verify init log repeat N__ K___ C___ Y X Hi__ Wi__ Strides Dilations LeftPads RightPads Desired_grid_size__
|
||||
$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT 128 256 192 3 3 71 71 2 2 1 1 1 1 1 1 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT 128 256 192 3 3 71 71 2 2 1 1 1 1 1 1 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT 128 256 256 3 3 30 30 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT 128 256 256 3 3 28 28 2 2 1 1 1 1 1 1 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT 128 1024 256 1 1 14 14 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
|
||||
N=${10}
|
||||
|
||||
# Resnet50 from Bing
|
||||
######## op datatype in_layout wei_layout out_layout verify init log repeat N__ K___ C___ Y X Hi__ Wi__ Strides Dilations LeftPads RightPads
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 1024 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 1024 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 128 3 3 28 28 1 1 1 1 1 1 1 1
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 128 1 1 28 28 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 128 3 3 56 56 2 2 1 1 1 1 1 1
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 2048 1 1 7 7 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 1024 256 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 256 3 3 14 14 1 1 1 1 1 1 1 1
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 256 3 3 28 28 2 2 1 1 1 1 1 1
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 256 1 1 56 56 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 256 1 1 56 56 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 512 3 3 14 14 2 2 1 1 1 1 1 1
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 512 1 1 28 28 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 512 1 1 28 28 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 2048 512 1 1 7 7 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 512 3 3 7 7 1 1 1 1 1 1 1 1
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 64 1 1 56 56 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 64 1 1 56 56 1 1 1 1 0 0 0 0
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 64 3 3 56 56 1 1 1 1 1 1 1 1
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 8 7 7 224 224 2 2 1 1 3 3 3 3
|
||||
|
||||
|
||||
# Resnet50 from Bing
|
||||
#################### op____________________ datatype in_layout wei_layout out_layout verify init log repeat N__ K___ C___ Y X Hi__ Wi__ Strides Dilations LeftPads RightPads
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 3 7 7 224 224 2 2 1 1 3 3 3 3
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 64 1 1 56 56 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 64 3 3 56 56 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 64 1 1 56 56 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 256 1 1 56 56 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 64 3 3 56 56 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 64 1 1 56 56 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 256 1 1 56 56 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 64 3 3 56 56 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 64 1 1 56 56 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 256 1 1 56 56 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 128 3 3 56 56 2 2 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 128 1 1 28 28 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 512 1 1 28 28 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 128 3 3 28 28 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 128 1 1 28 28 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 512 1 1 28 28 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 128 3 3 28 28 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 128 1 1 28 28 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 512 1 1 28 28 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 128 128 3 3 28 28 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 128 1 1 28 28 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 512 1 1 28 28 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 256 3 3 28 28 2 2 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 1024 256 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 1024 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 256 3 3 14 14 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 1024 256 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 1024 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 256 3 3 14 14 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 1024 256 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 1024 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 256 3 3 14 14 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 1024 256 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 1024 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 256 3 3 14 14 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 1024 256 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 1024 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 256 3 3 14 14 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 1024 256 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 1024 1 1 14 14 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 512 3 3 14 14 2 2 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 2048 512 1 1 7 7 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 2048 1 1 7 7 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 512 3 3 7 7 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 2048 512 1 1 7 7 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 2048 1 1 7 7 1 1 1 1 0 0 0 0
|
||||
#profiler/ckProfiler conv_fwd_bias_relu $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 512 512 3 3 7 7 1 1 1 1 1 1 1 1
|
||||
#profiler/ckProfiler conv_fwd_bias_relu_add $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 2048 512 1 1 7 7 1 1 1 1 0 0 0 0
|
||||
|
||||
#N=${10}
|
||||
|
||||
# Resnet50
|
||||
######## op datatype in_layout wei_layout out_layout verify init log repeat N__ K___ C___ Y X Hi__ Wi__ Strides Dilations LeftPads RightPads Desired_grid_size__
|
||||
@@ -49,6 +127,7 @@ REPEAT=$9
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 256 64 1 1 56 56 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 64 1 1 56 56 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 64 3 3 56 56 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT $N 64 3 7 7 230 230 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
|
||||
|
||||
# SSD
|
||||
######## op datatype in_layout wei_layout out_layout verify init log repeat N__ K___ C___ Y X Hi__ Wi__ Strides Dilations LeftPads RightPads Desired_grid_size__
|
||||
@@ -96,5 +175,3 @@ REPEAT=$9
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT 120 510 512 3 3 10 10 1 1 1 1 1 1 1 1
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT 120 510 256 3 3 5 5 1 1 1 1 1 1 1 1
|
||||
#$DRIVER $OP $DATATYPE $IN_LAYOUT $WEI_LAYOUT $OUT_LAYOUT $VERIFY $INIT $LOG $REPEAT 120 340 256 3 3 3 3 1 1 1 1 1 1 1 1
|
||||
|
||||
|
||||
|
||||
Reference in New Issue
Block a user