ckProfiler and device-level XDL GEMM operator (#48)

* add DeviceGemmXdl

* update script

* fix naming issue

* fix comment

* output HostTensorDescriptor

* rename

* padded GEMM for fwd v4r4r4 nhwc

* refactor

* refactor

* refactor

* adding ckProfiler

* adding ckProfiler

* refactor

* fix tuning parameter bug

* add more gemm instances

* add more fp16 GEMM instances

* fix profiler driver

* fix bug in tuning parameter

* add fp32 gemm instances

* small fix

* refactor

* rename

* refactor gemm profiler; adding DeviceConv and conv profiler

* refactor

* fix

* add conv profiler

* refactor

* adding more GEMM and Conv instance

* Create README.md

Add build instruction for ckProfiler

* Create README.md

Add Readme for gemm_xdl example

* Update README.md

Remove build instruction from top most folder

* Update README.md

* clean up

[ROCm/composable_kernel commit: e823d518cb]
This commit is contained in:
Chao Liu
2021-11-14 11:28:32 -06:00
committed by GitHub
parent 1f6ca26819
commit 2f5ccb68f5
60 changed files with 9797 additions and 1123 deletions

View File

@@ -8,11 +8,11 @@ MY_PROJECT_INSTALL=../install.dir
cmake \
-D CMAKE_INSTALL_PREFIX=${MY_PROJECT_INSTALL} \
-D HALF_INCLUDE_DIR="/root/workspace/external/half/include" \
-D BUILD_DEV=ON \
-D BUILD_DEV=OFF \
-D CMAKE_BUILD_TYPE=Release \
-D CMAKE_CXX_FLAGS="-DCK_AMD_GPU_GFX908 -O3 --amdgpu-target=gfx908 -mllvm --amdgpu-spill-vgpr-to-agpr=0 -gline-tables-only -save-temps=$PWD" \
-D CMAKE_CXX_FLAGS="-DCK_AMD_GPU_GFX908 --amdgpu-target=gfx908 -O3 -ftemplate-backtrace-limit=0 -mllvm --amdgpu-spill-vgpr-to-agpr=0 -gline-tables-only -save-temps=$PWD" \
-D CMAKE_CXX_COMPILER=/opt/rocm/bin/hipcc \
-D CMAKE_PREFIX_PATH=/opt/rocm \
-D CMAKE_VERBOSE_MAKEFILE:BOOL=ON \
${MY_PROJECT_SOURCE}

71
script/conv_driver.sh Executable file
View File

@@ -0,0 +1,71 @@
#!/bin/bash
## GPU visibility
export HIP_VISIBLE_DEVICES=0
make -j conv_fwd_driver_offline
#make -j conv_bwd_driver_offline
#make -j conv_wrw_driver_offline
DRIVER="./host/driver_offline/conv_fwd_driver_offline"
#DRIVER="./host/driver_offline/conv_bwd_driver_offline"
#DRIVER="./host/driver_offline/conv_wrw_driver_offline"
LAYOUT=$1
ALGO=$2
VERIFY=$3
INIT=$4
LOG=$5
REPEAT=$6
DESIRED_GRID_SIZE=$7
######### layout algo verify init log repeat N__ K___ C___ Y X Hi_ Wi__ Strides Dilations LeftPads RightPads Desired_grid_size__
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 192 3 3 71 71 2 2 1 1 1 1 1 1 $DESIRED_GRID_SIZE
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 192 3 3 71 71 2 2 1 1 1 1 1 1 $DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 1024 1 7 17 17 1 1 1 1 0 3 0 3 $DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 256 3 3 14 14 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 128 3 3 14 14 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 512 3 3 7 7 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
$DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 192 3 3 35 35 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 256 3 3 30 30 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 512 3 3 16 16 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
$DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 2048 1024 1 1 14 14 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 1024 1 1 14 14 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 2048 1 1 7 7 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
$DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 256 3 3 14 14 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 256 1 1 14 14 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
$DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 32 256 3 3 1 1 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 32 256 1 1 1 1 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
$DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 64 1 1 2 2 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 128 1 1 2 2 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
# Resnet50
######### layout algo verify init log repeat N__ K___ C___ Y X Hi_ Wi__ Strides Dilations LeftPads RightPads Desired_grid_size__
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 2048 1024 1 1 14 14 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 1024 1 1 14 14 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 1024 1 1 14 14 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 128 128 3 3 28 28 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 128 1 1 28 28 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 128 128 3 3 58 58 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 2048 1 1 7 7 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 1024 256 1 1 14 14 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 256 3 3 14 14 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 256 3 3 30 30 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 128 256 1 1 56 56 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 256 1 1 56 56 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 64 256 1 1 56 56 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 512 3 3 16 16 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 1024 512 1 1 28 28 2 2 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 128 512 1 1 28 28 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 512 1 1 28 28 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 2048 512 1 1 7 7 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 512 3 3 7 7 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 64 1 1 56 56 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 64 64 1 1 56 56 1 1 1 1 0 0 0 0 $DESIRED_GRID_SIZE
##DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 64 64 3 3 56 56 1 1 1 1 1 1 1 1 $DESIRED_GRID_SIZE

20
script/example_gemm_xdl.sh Executable file
View File

@@ -0,0 +1,20 @@
#!/bin/bash
## GPU visibility
export HIP_VISIBLE_DEVICES=1
make -j gemm_xdl
DRIVER="./example/gemm_xdl"
VERIFY=$1
INIT=$2
LOG=$3
REPEAT=$4
######### verify init log repeat M___ N___ K___ StrideA StrideB StrideC
#$DRIVER $VERIFY $INIT $LOG $REPEAT 960 1024 1024 1024 1024 1024
#$DRIVER $VERIFY $INIT $LOG $REPEAT 1024 1024 1024 1024 1024 1024
#$DRIVER $VERIFY $INIT $LOG $REPEAT 1920 2048 2048 2048 2048 2048
$DRIVER $VERIFY $INIT $LOG $REPEAT 3840 4096 4096 4096 4096 4096
#$DRIVER $VERIFY $INIT $LOG $REPEAT 7680 8192 8192 8192 8192 8192

25
script/gemm_driver.sh Executable file
View File

@@ -0,0 +1,25 @@
#!/bin/bash
## GPU visibility
export HIP_VISIBLE_DEVICES=0
make -j gemm_driver_offline
DRIVER="./host/driver_offline/gemm_driver_offline"
LAYOUT=$1
ALGO=$2
VERIFY=$3
INIT=$4
LOG=$5
REPEAT=$6
M01=$7
N01=$8
######### layout algo verify init log repeat M___ N___ K___ M01_ N01_
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 960 1024 1024 $M01 $N01
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 1024 1024 1024 $M01 $N01
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 1920 2048 2048 $M01 $N01
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 3840 4096 4096 $M01 $N01
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 7680 8192 8192 $M01 $N01

View File

@@ -1,137 +0,0 @@
#!/bin/bash
## GPU visibility
export ROCR_VISIBLE_DEVICE=0
export GPU_DEVICE_ORDINAL=0
make -j conv_fwd_driver_offline
#make -j conv_bwd_driver_offline
#make -j conv_wrw_driver_offline
#make -j gemm_driver_offline
DRIVER="./host/driver_offline/conv_fwd_driver_offline"
LAYOUT=$1
ALGO=$2
VERIFY=$3
INIT=$4
LOG=$5
REPEAT=$6
#M01=$7
#N01=$8
KBATCH=$7
######### layout algo verify init log repeat N__ K___ C___ Y X Hi_ Wi__ Strides Dilations LeftPads RightPads
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 192 3 3 71 71 2 2 1 1 1 1 1 1
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 192 3 3 71 71 2 2 1 1 1 1 1 1
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 1024 1 7 17 17 1 1 1 1 0 3 0 3
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 256 3 3 14 14 1 1 1 1 1 1 1 1
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 128 3 3 14 14 1 1 1 1 1 1 1 1
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 512 3 3 7 7 1 1 1 1 1 1 1 1
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 192 3 3 35 35 2 2 1 1 0 0 0 0
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 256 3 3 30 30 2 2 1 1 0 0 0 0
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 512 3 3 16 16 2 2 1 1 0 0 0 0
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 2048 1024 1 1 14 14 2 2 1 1 0 0 0 0
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 1024 1 1 14 14 1 1 1 1 0 0 0 0
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 2048 1 1 7 7 1 1 1 1 0 0 0 0
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 256 3 3 14 14 1 1 1 1 1 1 1 1
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 128 3 3 14 14 1 1 1 1 1 1 1 1
######### layout algo verify init log repeat M___ N___ K___
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 960 1024 1024 $M01 $N01
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 1920 2048 2048 $M01 $N01
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 3840 4096 4096 $M01 $N01
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 7680 8192 8192 $M01 $N01
# Resnet50
######### layout algo verify init log repeat N__ K___ C___ Y X Hi_ Wi__ Strides Dilations LeftPads RightPads
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 2048 1024 1 1 14 14 2 2 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 1024 1 1 14 14 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 1024 1 1 14 14 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 128 128 3 3 28 28 1 1 1 1 1 1 1 1
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 128 1 1 28 28 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 128 128 3 3 58 58 2 2 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 2048 1 1 7 7 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 1024 256 1 1 14 14 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 256 3 3 14 14 1 1 1 1 1 1 1 1
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 256 3 3 30 30 2 2 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 128 256 1 1 56 56 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 256 1 1 56 56 2 2 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 64 256 1 1 56 56 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 512 3 3 16 16 2 2 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 1024 512 1 1 28 28 2 2 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 128 512 1 1 28 28 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 512 1 1 28 28 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 2048 512 1 1 7 7 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 512 512 3 3 7 7 1 1 1 1 1 1 1 1
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 256 64 1 1 56 56 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 64 64 1 1 56 56 1 1 1 1 0 0 0 0
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 256 64 64 3 3 56 56 1 1 1 1 1 1 1 1
# 256x128x32 c64
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 2048 1024 1 1 14 14 2 2 1 1 0 0 0 0 7
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 1024 1 1 14 14 1 1 1 1 0 0 0 0 56
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 1024 1 1 14 14 1 1 1 1 0 0 0 0 56
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 128 3 3 28 28 1 1 1 1 1 1 1 1 $KBATCH
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 128 1 1 28 28 1 1 1 1 0 0 0 0 224
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 128 3 3 58 58 2 2 1 1 0 0 0 0 $KBATCH
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 2048 1 1 7 7 1 1 1 1 0 0 0 0 14
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 1024 256 1 1 14 14 1 1 1 1 0 0 0 0 56
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 256 3 3 14 14 1 1 1 1 1 1 1 1 28
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 256 3 3 30 30 2 2 1 1 0 0 0 0 28
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 256 1 1 56 56 1 1 1 1 0 0 0 0 $KBATCH
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 256 1 1 56 56 2 2 1 1 0 0 0 0 224
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 64 256 1 1 56 56 1 1 1 1 0 0 0 0 $KBATCH
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 512 3 3 16 16 2 2 1 1 0 0 0 0 7
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 1024 512 1 1 28 28 2 2 1 1 0 0 0 0 56
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 512 1 1 28 28 1 1 1 1 0 0 0 0 $KBATCH
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 512 1 1 28 28 1 1 1 1 0 0 0 0 224
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 2048 512 1 1 7 7 1 1 1 1 0 0 0 0 14
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 512 3 3 7 7 1 1 1 1 1 1 1 1 7
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 64 1 1 56 56 1 1 1 1 0 0 0 0 $KBATCH
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 64 64 1 1 56 56 1 1 1 1 0 0 0 0 $KBATCH
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 64 64 3 3 56 56 1 1 1 1 1 1 1 1 $KBATCH
# 128x128x32 c64
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 2048 1024 1 1 14 14 2 2 1 1 0 0 0 0 7
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 1024 1 1 14 14 1 1 1 1 0 0 0 0 56
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 1024 1 1 14 14 1 1 1 1 0 0 0 0 28
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 128 3 3 28 28 1 1 1 1 1 1 1 1 112
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 128 1 1 28 28 1 1 1 1 0 0 0 0 224
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 128 3 3 58 58 2 2 1 1 0 0 0 0 112
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 2048 1 1 7 7 1 1 1 1 0 0 0 0 14
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 1024 256 1 1 14 14 1 1 1 1 0 0 0 0 56
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 256 3 3 14 14 1 1 1 1 1 1 1 1 28
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 256 3 3 30 30 2 2 1 1 0 0 0 0 28
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 256 1 1 56 56 1 1 1 1 0 0 0 0 448
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 256 1 1 56 56 2 2 1 1 0 0 0 0 224
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 64 256 1 1 56 56 1 1 1 1 0 0 0 0 $KBATCH
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 512 3 3 16 16 2 2 1 1 0 0 0 0 7
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 1024 512 1 1 28 28 2 2 1 1 0 0 0 0 28
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 128 512 1 1 28 28 1 1 1 1 0 0 0 0 224
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 512 1 1 28 28 1 1 1 1 0 0 0 0 112
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 2048 512 1 1 7 7 1 1 1 1 0 0 0 0 14
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 512 512 3 3 7 7 1 1 1 1 1 1 1 1 7
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 64 1 1 56 56 1 1 1 1 0 0 0 0 $KBATCH
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 64 64 1 1 56 56 1 1 1 1 0 0 0 0 $KBATCH
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 64 64 3 3 56 56 1 1 1 1 1 1 1 1 $KBATCH
# 128x64x32 c64
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 64 1 1 56 56 1 1 1 1 0 0 0 0 112
# 64x128x32 c64
$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 64 256 1 1 56 56 1 1 1 1 0 0 0 0 $KBATCH
# 64x64x32 c32
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 64 256 1 1 56 56 1 1 1 1 0 0 0 0 112
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 256 64 1 1 56 56 1 1 1 1 0 0 0 0 112
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 64 64 1 1 56 56 1 1 1 1 0 0 0 0 448
#$DRIVER $LAYOUT $ALGO $VERIFY $INIT $LOG $REPEAT 128 64 64 3 3 56 56 1 1 1 1 1 1 1 1 448