diff options
author | Cedric Nugteren <web@cedricnugteren.nl> | 2016-05-22 16:46:05 +0200 |
---|---|---|
committer | Cedric Nugteren <web@cedricnugteren.nl> | 2016-05-22 16:46:05 +0200 |
commit | f0cb3fdc81031625370d58da77f7cbe73fc130a7 (patch) | |
tree | 512a3209ab023f1168db3e16ada84f2782b3e1d4 /include/internal/database | |
parent | c8ff3f143fe94c87b23fd1bf36c1a4f91d305f01 (diff) |
Fixed tuning results for half-precision; added first results for the xGER kernels
Diffstat (limited to 'include/internal/database')
-rw-r--r-- | include/internal/database/copy.h | 6 | ||||
-rw-r--r-- | include/internal/database/pad.h | 6 | ||||
-rw-r--r-- | include/internal/database/padtranspose.h | 6 | ||||
-rw-r--r-- | include/internal/database/transpose.h | 6 | ||||
-rw-r--r-- | include/internal/database/xaxpy.h | 6 | ||||
-rw-r--r-- | include/internal/database/xgemv.h | 6 | ||||
-rw-r--r-- | include/internal/database/xger.h | 18 |
7 files changed, 36 insertions, 18 deletions
diff --git a/include/internal/database/copy.h b/include/internal/database/copy.h index b4695868..63f8e814 100644 --- a/include/internal/database/copy.h +++ b/include/internal/database/copy.h @@ -18,13 +18,13 @@ const Database::DatabaseEntry Database::CopyHalf = { "Copy", Precision::kHalf, { { // Intel GPUs kDeviceTypeGPU, "Intel", { - { "Intel(R) HD Graphics Skylake ULT GT2", { {"COPY_DIMX",32}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",4} } }, - { "default", { {"COPY_DIMX",32}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",4} } }, + { "Intel(R) HD Graphics Skylake ULT GT2", { {"COPY_DIMX",8}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",1} } }, + { "default", { {"COPY_DIMX",8}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",1} } }, } }, { // Default kDeviceTypeAll, "default", { - { "default", { {"COPY_DIMX",32}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",4} } }, + { "default", { {"COPY_DIMX",8}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",1} } }, } }, } diff --git a/include/internal/database/pad.h b/include/internal/database/pad.h index aae7631b..d0a85e7c 100644 --- a/include/internal/database/pad.h +++ b/include/internal/database/pad.h @@ -18,13 +18,13 @@ const Database::DatabaseEntry Database::PadHalf = { "Pad", Precision::kHalf, { { // Intel GPUs kDeviceTypeGPU, "Intel", { - { "Intel(R) HD Graphics Skylake ULT GT2", { {"PAD_DIMX",16}, {"PAD_DIMY",8}, {"PAD_WPTX",4}, {"PAD_WPTY",4} } }, - { "default", { {"PAD_DIMX",16}, {"PAD_DIMY",8}, {"PAD_WPTX",4}, {"PAD_WPTY",4} } }, + { "Intel(R) HD Graphics Skylake ULT GT2", { {"PAD_DIMX",8}, {"PAD_DIMY",8}, {"PAD_WPTX",2}, {"PAD_WPTY",1} } }, + { "default", { {"PAD_DIMX",8}, {"PAD_DIMY",8}, {"PAD_WPTX",2}, {"PAD_WPTY",1} } }, } }, { // Default kDeviceTypeAll, "default", { - { "default", { {"PAD_DIMX",16}, {"PAD_DIMY",8}, {"PAD_WPTX",4}, {"PAD_WPTY",4} } }, + { "default", { {"PAD_DIMX",8}, {"PAD_DIMY",8}, {"PAD_WPTX",2}, {"PAD_WPTY",1} } }, } }, } diff --git a/include/internal/database/padtranspose.h b/include/internal/database/padtranspose.h index 2d8d47f2..0eb3b528 100644 --- a/include/internal/database/padtranspose.h +++ b/include/internal/database/padtranspose.h @@ -18,13 +18,13 @@ const Database::DatabaseEntry Database::PadtransposeHalf = { "Padtranspose", Precision::kHalf, { { // Intel GPUs kDeviceTypeGPU, "Intel", { - { "Intel(R) HD Graphics Skylake ULT GT2", { {"PADTRA_PAD",0}, {"PADTRA_TILE",8}, {"PADTRA_WPT",2} } }, - { "default", { {"PADTRA_PAD",0}, {"PADTRA_TILE",8}, {"PADTRA_WPT",2} } }, + { "Intel(R) HD Graphics Skylake ULT GT2", { {"PADTRA_PAD",0}, {"PADTRA_TILE",8}, {"PADTRA_WPT",1} } }, + { "default", { {"PADTRA_PAD",0}, {"PADTRA_TILE",8}, {"PADTRA_WPT",1} } }, } }, { // Default kDeviceTypeAll, "default", { - { "default", { {"PADTRA_PAD",0}, {"PADTRA_TILE",8}, {"PADTRA_WPT",2} } }, + { "default", { {"PADTRA_PAD",0}, {"PADTRA_TILE",8}, {"PADTRA_WPT",1} } }, } }, } diff --git a/include/internal/database/transpose.h b/include/internal/database/transpose.h index 8e5ace67..d7bdd90a 100644 --- a/include/internal/database/transpose.h +++ b/include/internal/database/transpose.h @@ -18,13 +18,13 @@ const Database::DatabaseEntry Database::TransposeHalf = { "Transpose", Precision::kHalf, { { // Intel GPUs kDeviceTypeGPU, "Intel", { - { "Intel(R) HD Graphics Skylake ULT GT2", { {"TRA_DIM",8}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",8} } }, - { "default", { {"TRA_DIM",8}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",8} } }, + { "Intel(R) HD Graphics Skylake ULT GT2", { {"TRA_DIM",16}, {"TRA_PAD",0}, {"TRA_SHUFFLE",0}, {"TRA_WPT",4} } }, + { "default", { {"TRA_DIM",16}, {"TRA_PAD",0}, {"TRA_SHUFFLE",0}, {"TRA_WPT",4} } }, } }, { // Default kDeviceTypeAll, "default", { - { "default", { {"TRA_DIM",8}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",8} } }, + { "default", { {"TRA_DIM",16}, {"TRA_PAD",0}, {"TRA_SHUFFLE",0}, {"TRA_WPT",4} } }, } }, } diff --git a/include/internal/database/xaxpy.h b/include/internal/database/xaxpy.h index 6c5e478b..72e6a43c 100644 --- a/include/internal/database/xaxpy.h +++ b/include/internal/database/xaxpy.h @@ -18,13 +18,13 @@ const Database::DatabaseEntry Database::XaxpyHalf = { "Xaxpy", Precision::kHalf, { { // Intel GPUs kDeviceTypeGPU, "Intel", { - { "Intel(R) HD Graphics Skylake ULT GT2", { {"VW",8}, {"WGS",512}, {"WPT",1} } }, - { "default", { {"VW",8}, {"WGS",512}, {"WPT",1} } }, + { "Intel(R) HD Graphics Skylake ULT GT2", { {"VW",4}, {"WGS",512}, {"WPT",8} } }, + { "default", { {"VW",4}, {"WGS",512}, {"WPT",8} } }, } }, { // Default kDeviceTypeAll, "default", { - { "default", { {"VW",8}, {"WGS",512}, {"WPT",1} } }, + { "default", { {"VW",4}, {"WGS",512}, {"WPT",8} } }, } }, } diff --git a/include/internal/database/xgemv.h b/include/internal/database/xgemv.h index c6e9d600..0d11f5ee 100644 --- a/include/internal/database/xgemv.h +++ b/include/internal/database/xgemv.h @@ -18,13 +18,13 @@ const Database::DatabaseEntry Database::XgemvHalf = { "Xgemv", Precision::kHalf, { { // Intel GPUs kDeviceTypeGPU, "Intel", { - { "Intel(R) HD Graphics Skylake ULT GT2", { {"WGS1",256}, {"WPT1",1}, {"VW2",2}, {"WGS2",64}, {"WPT2",2}, {"VW3",4}, {"WGS3",64}, {"WPT3",4} } }, - { "default", { {"WGS1",256}, {"WPT1",1}, {"VW2",2}, {"WGS2",64}, {"WPT2",2}, {"VW3",4}, {"WGS3",64}, {"WPT3",4} } }, + { "Intel(R) HD Graphics Skylake ULT GT2", { {"WGS1",128}, {"WPT1",1}, {"VW2",2}, {"WGS2",128}, {"WPT2",2}, {"VW3",1}, {"WGS3",64}, {"WPT3",1} } }, + { "default", { {"WGS1",128}, {"WPT1",1}, {"VW2",2}, {"WGS2",128}, {"WPT2",2}, {"VW3",1}, {"WGS3",64}, {"WPT3",1} } }, } }, { // Default kDeviceTypeAll, "default", { - { "default", { {"WGS1",256}, {"WPT1",1}, {"VW2",2}, {"WGS2",64}, {"WPT2",2}, {"VW3",4}, {"WGS3",64}, {"WPT3",4} } }, + { "default", { {"WGS1",128}, {"WPT1",1}, {"VW2",2}, {"WGS2",128}, {"WPT2",2}, {"VW3",1}, {"WGS3",64}, {"WPT3",1} } }, } }, } diff --git a/include/internal/database/xger.h b/include/internal/database/xger.h index dae857cd..81b8b98a 100644 --- a/include/internal/database/xger.h +++ b/include/internal/database/xger.h @@ -14,6 +14,24 @@ namespace clblast { // ================================================================================================= +const Database::DatabaseEntry Database::XgerHalf = { + "Xger", Precision::kHalf, { + { // Intel GPUs + kDeviceTypeGPU, "Intel", { + { "Intel(R) HD Graphics Skylake ULT GT2", { {"WGS1",64}, {"WGS2",1}, {"WPT",1} } }, + { "default", { {"WGS1",64}, {"WGS2",1}, {"WPT",1} } }, + } + }, + { // Default + kDeviceTypeAll, "default", { + { "default", { {"WGS1",64}, {"WGS2",1}, {"WPT",1} } }, + } + }, + } +}; + +// ================================================================================================= + const Database::DatabaseEntry Database::XgerSingle = { "Xger", Precision::kSingle, { { // AMD GPUs |