From e9ef037549e19a1e8ec6c40155227c2dd7cf0cd6 Mon Sep 17 00:00:00 2001 From: Cedric Nugteren Date: Sat, 4 Mar 2017 15:24:55 +0100 Subject: [PATCH] Added tuning results for the Radeon HD6750M GPU (Apple OpenCL) --- README.md | 1 + src/database/kernels/copy.hpp | 6 ++++-- src/database/kernels/pad.hpp | 6 ++++-- src/database/kernels/padtranspose.hpp | 2 ++ src/database/kernels/transpose.hpp | 6 ++++-- src/database/kernels/xaxpy.hpp | 2 ++ src/database/kernels/xdot.hpp | 2 ++ src/database/kernels/xgemm.hpp | 4 +++- src/database/kernels/xgemm_direct.hpp | 6 ++++-- src/database/kernels/xgemv.hpp | 2 ++ src/database/kernels/xgemv_fast.hpp | 2 ++ src/database/kernels/xgemv_fast_rot.hpp | 4 +++- src/database/kernels/xger.hpp | 6 ++++-- 13 files changed, 37 insertions(+), 12 deletions(-) diff --git a/README.md b/README.md index 91cc4023..93a9b35f 100644 --- a/README.md +++ b/README.md @@ -127,6 +127,7 @@ The CLBlast library is already tuned for the most commonly used OpenCL devices a - Tesla K40m * AMD GPUs: - AMD Radeon R9 M370X Compute Engine + - ATI Radeon HD 6750M - Ellesmere - Hawaii - Oland diff --git a/src/database/kernels/copy.hpp b/src/database/kernels/copy.hpp index b2aa736f..bd316233 100644 --- a/src/database/kernels/copy.hpp +++ b/src/database/kernels/copy.hpp @@ -45,6 +45,7 @@ const Database::DatabaseEntry CopySingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"COPY_DIMX",32}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",1} } }, + { "ATI Radeon HD 6750M", { {"COPY_DIMX",16}, {"COPY_DIMY",8}, {"COPY_VW",2}, {"COPY_WPT",1} } }, { "Ellesmere", { {"COPY_DIMX",8}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",8} } }, { "Hawaii", { {"COPY_DIMX",32}, {"COPY_DIMY",8}, {"COPY_VW",2}, {"COPY_WPT",2} } }, { "Oland", { {"COPY_DIMX",32}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",2} } }, @@ -111,7 +112,7 @@ const Database::DatabaseEntry CopySingle = { }, { // Default kDeviceTypeAll, "default", { - { "default", { {"COPY_DIMX",32}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",4} } }, + { "default", { {"COPY_DIMX",32}, {"COPY_DIMY",8}, {"COPY_VW",2}, {"COPY_WPT",2} } }, } }, } @@ -124,6 +125,7 @@ const Database::DatabaseEntry CopyComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"COPY_DIMX",32}, {"COPY_DIMY",8}, {"COPY_VW",1}, {"COPY_WPT",1} } }, + { "ATI Radeon HD 6750M", { {"COPY_DIMX",8}, {"COPY_DIMY",8}, {"COPY_VW",1}, {"COPY_WPT",1} } }, { "Ellesmere", { {"COPY_DIMX",16}, {"COPY_DIMY",16}, {"COPY_VW",1}, {"COPY_WPT",4} } }, { "Hawaii", { {"COPY_DIMX",32}, {"COPY_DIMY",8}, {"COPY_VW",1}, {"COPY_WPT",2} } }, { "Oland", { {"COPY_DIMX",8}, {"COPY_DIMY",16}, {"COPY_VW",1}, {"COPY_WPT",1} } }, @@ -182,7 +184,7 @@ const Database::DatabaseEntry CopyComplexSingle = { }, { // Default kDeviceTypeAll, "default", { - { "default", { {"COPY_DIMX",32}, {"COPY_DIMY",8}, {"COPY_VW",1}, {"COPY_WPT",1} } }, + { "default", { {"COPY_DIMX",16}, {"COPY_DIMY",8}, {"COPY_VW",1}, {"COPY_WPT",2} } }, } }, } diff --git a/src/database/kernels/pad.hpp b/src/database/kernels/pad.hpp index e4005527..84845349 100644 --- a/src/database/kernels/pad.hpp +++ b/src/database/kernels/pad.hpp @@ -45,6 +45,7 @@ const Database::DatabaseEntry PadSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",1} } }, + { "ATI Radeon HD 6750M", { {"PAD_DIMX",8}, {"PAD_DIMY",16}, {"PAD_WPTX",2}, {"PAD_WPTY",1} } }, { "Ellesmere", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",2}, {"PAD_WPTY",2} } }, { "Hawaii", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",4} } }, { "Oland", { {"PAD_DIMX",8}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",2} } }, @@ -52,7 +53,7 @@ const Database::DatabaseEntry PadSingle = { { "Tahiti", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",2} } }, { "Tonga", { {"PAD_DIMX",16}, {"PAD_DIMY",16}, {"PAD_WPTX",2}, {"PAD_WPTY",2} } }, { "Turks", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",2}, {"PAD_WPTY",1} } }, - { "default", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",2}, {"PAD_WPTY",1} } }, + { "default", { {"PAD_DIMX",8}, {"PAD_DIMY",16}, {"PAD_WPTX",1}, {"PAD_WPTY",2} } }, } }, { // ARM GPUs @@ -124,6 +125,7 @@ const Database::DatabaseEntry PadComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",1} } }, + { "ATI Radeon HD 6750M", { {"PAD_DIMX",16}, {"PAD_DIMY",8}, {"PAD_WPTX",2}, {"PAD_WPTY",1} } }, { "Ellesmere", { {"PAD_DIMX",16}, {"PAD_DIMY",16}, {"PAD_WPTX",2}, {"PAD_WPTY",4} } }, { "Hawaii", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",2} } }, { "Oland", { {"PAD_DIMX",8}, {"PAD_DIMY",32}, {"PAD_WPTX",1}, {"PAD_WPTY",1} } }, @@ -131,7 +133,7 @@ const Database::DatabaseEntry PadComplexSingle = { { "Tahiti", { {"PAD_DIMX",16}, {"PAD_DIMY",16}, {"PAD_WPTX",1}, {"PAD_WPTY",1} } }, { "Tonga", { {"PAD_DIMX",16}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",2} } }, { "Turks", { {"PAD_DIMX",16}, {"PAD_DIMY",8}, {"PAD_WPTX",4}, {"PAD_WPTY",4} } }, - { "default", { {"PAD_DIMX",16}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",2} } }, + { "default", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",2} } }, } }, { // ARM GPUs diff --git a/src/database/kernels/padtranspose.hpp b/src/database/kernels/padtranspose.hpp index 92aa4f7b..325df121 100644 --- a/src/database/kernels/padtranspose.hpp +++ b/src/database/kernels/padtranspose.hpp @@ -45,6 +45,7 @@ const Database::DatabaseEntry PadtransposeSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"PADTRA_PAD",0}, {"PADTRA_TILE",16}, {"PADTRA_WPT",4} } }, + { "ATI Radeon HD 6750M", { {"PADTRA_PAD",1}, {"PADTRA_TILE",16}, {"PADTRA_WPT",1} } }, { "Ellesmere", { {"PADTRA_PAD",1}, {"PADTRA_TILE",8}, {"PADTRA_WPT",4} } }, { "Hawaii", { {"PADTRA_PAD",1}, {"PADTRA_TILE",16}, {"PADTRA_WPT",4} } }, { "Oland", { {"PADTRA_PAD",0}, {"PADTRA_TILE",16}, {"PADTRA_WPT",4} } }, @@ -124,6 +125,7 @@ const Database::DatabaseEntry PadtransposeComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"PADTRA_PAD",0}, {"PADTRA_TILE",16}, {"PADTRA_WPT",4} } }, + { "ATI Radeon HD 6750M", { {"PADTRA_PAD",1}, {"PADTRA_TILE",16}, {"PADTRA_WPT",1} } }, { "Ellesmere", { {"PADTRA_PAD",0}, {"PADTRA_TILE",8}, {"PADTRA_WPT",4} } }, { "Hawaii", { {"PADTRA_PAD",0}, {"PADTRA_TILE",16}, {"PADTRA_WPT",2} } }, { "Oland", { {"PADTRA_PAD",0}, {"PADTRA_TILE",8}, {"PADTRA_WPT",4} } }, diff --git a/src/database/kernels/transpose.hpp b/src/database/kernels/transpose.hpp index 0ee2d83b..ebe1db8f 100644 --- a/src/database/kernels/transpose.hpp +++ b/src/database/kernels/transpose.hpp @@ -45,6 +45,7 @@ const Database::DatabaseEntry TransposeSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"TRA_DIM",4}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",8} } }, + { "ATI Radeon HD 6750M", { {"TRA_DIM",8}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",2} } }, { "Ellesmere", { {"TRA_DIM",16}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",4} } }, { "Hawaii", { {"TRA_DIM",4}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",8} } }, { "Oland", { {"TRA_DIM",8}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",4} } }, @@ -52,7 +53,7 @@ const Database::DatabaseEntry TransposeSingle = { { "Tahiti", { {"TRA_DIM",4}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",4} } }, { "Tonga", { {"TRA_DIM",8}, {"TRA_PAD",1}, {"TRA_SHUFFLE",1}, {"TRA_WPT",2} } }, { "Turks", { {"TRA_DIM",8}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",2} } }, - { "default", { {"TRA_DIM",8}, {"TRA_PAD",1}, {"TRA_SHUFFLE",1}, {"TRA_WPT",2} } }, + { "default", { {"TRA_DIM",8}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",2} } }, } }, { // ARM GPUs @@ -124,6 +125,7 @@ const Database::DatabaseEntry TransposeComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"TRA_DIM",4}, {"TRA_PAD",1}, {"TRA_SHUFFLE",1}, {"TRA_WPT",4} } }, + { "ATI Radeon HD 6750M", { {"TRA_DIM",16}, {"TRA_PAD",1}, {"TRA_SHUFFLE",1}, {"TRA_WPT",1} } }, { "Ellesmere", { {"TRA_DIM",4}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",4} } }, { "Hawaii", { {"TRA_DIM",16}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",1} } }, { "Oland", { {"TRA_DIM",4}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",2} } }, @@ -131,7 +133,7 @@ const Database::DatabaseEntry TransposeComplexSingle = { { "Tahiti", { {"TRA_DIM",16}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",1} } }, { "Tonga", { {"TRA_DIM",16}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",1} } }, { "Turks", { {"TRA_DIM",8}, {"TRA_PAD",1}, {"TRA_SHUFFLE",1}, {"TRA_WPT",4} } }, - { "default", { {"TRA_DIM",4}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",2} } }, + { "default", { {"TRA_DIM",8}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",1} } }, } }, { // ARM GPUs diff --git a/src/database/kernels/xaxpy.hpp b/src/database/kernels/xaxpy.hpp index 4b747f0a..18f63c4c 100644 --- a/src/database/kernels/xaxpy.hpp +++ b/src/database/kernels/xaxpy.hpp @@ -45,6 +45,7 @@ const Database::DatabaseEntry XaxpySingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"VW",1}, {"WGS",128}, {"WPT",1} } }, + { "ATI Radeon HD 6750M", { {"VW",1}, {"WGS",256}, {"WPT",2} } }, { "Ellesmere", { {"VW",1}, {"WGS",64}, {"WPT",4} } }, { "Hawaii", { {"VW",2}, {"WGS",64}, {"WPT",2} } }, { "Oland", { {"VW",1}, {"WGS",128}, {"WPT",1} } }, @@ -124,6 +125,7 @@ const Database::DatabaseEntry XaxpyComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"VW",2}, {"WGS",64}, {"WPT",8} } }, + { "ATI Radeon HD 6750M", { {"VW",1}, {"WGS",64}, {"WPT",1} } }, { "Ellesmere", { {"VW",2}, {"WGS",256}, {"WPT",1} } }, { "Hawaii", { {"VW",1}, {"WGS",128}, {"WPT",2} } }, { "Oland", { {"VW",1}, {"WGS",128}, {"WPT",1} } }, diff --git a/src/database/kernels/xdot.hpp b/src/database/kernels/xdot.hpp index ea154b6e..c3f86cf9 100644 --- a/src/database/kernels/xdot.hpp +++ b/src/database/kernels/xdot.hpp @@ -45,6 +45,7 @@ const Database::DatabaseEntry XdotSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"WGS1",128}, {"WGS2",32} } }, + { "ATI Radeon HD 6750M", { {"WGS1",256}, {"WGS2",32} } }, { "Ellesmere", { {"WGS1",128}, {"WGS2",32} } }, { "Oland", { {"WGS1",256}, {"WGS2",32} } }, { "Pitcairn", { {"WGS1",128}, {"WGS2",32} } }, @@ -106,6 +107,7 @@ const Database::DatabaseEntry XdotComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"WGS1",64}, {"WGS2",32} } }, + { "ATI Radeon HD 6750M", { {"WGS1",256}, {"WGS2",256} } }, { "Ellesmere", { {"WGS1",256}, {"WGS2",32} } }, { "Oland", { {"WGS1",128}, {"WGS2",32} } }, { "Pitcairn", { {"WGS1",256}, {"WGS2",32} } }, diff --git a/src/database/kernels/xgemm.hpp b/src/database/kernels/xgemm.hpp index 751f403b..768aa817 100644 --- a/src/database/kernels/xgemm.hpp +++ b/src/database/kernels/xgemm.hpp @@ -44,6 +44,7 @@ const Database::DatabaseEntry XgemmSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"KWG",32}, {"KWI",2}, {"MDIMA",16}, {"MDIMC",16}, {"MWG",64}, {"NDIMB",8}, {"NDIMC",16}, {"NWG",128}, {"SA",0}, {"SB",0}, {"STRM",0}, {"STRN",0}, {"VWM",2}, {"VWN",8} } }, + { "ATI Radeon HD 6750M", { {"KWG",32}, {"KWI",2}, {"MDIMA",8}, {"MDIMC",16}, {"MWG",128}, {"NDIMB",8}, {"NDIMC",8}, {"NWG",128}, {"SA",0}, {"SB",0}, {"STRM",1}, {"STRN",1}, {"VWM",8}, {"VWN",8} } }, { "Ellesmere", { {"KWG",32}, {"KWI",2}, {"MDIMA",8}, {"MDIMC",8}, {"MWG",16}, {"NDIMB",16}, {"NDIMC",16}, {"NWG",64}, {"SA",1}, {"SB",1}, {"STRM",0}, {"STRN",0}, {"VWM",1}, {"VWN",2} } }, { "Hawaii", { {"KWG",16}, {"KWI",2}, {"MDIMA",16}, {"MDIMC",32}, {"MWG",128}, {"NDIMB",32}, {"NDIMC",8}, {"NWG",64}, {"SA",1}, {"SB",1}, {"STRM",1}, {"STRN",1}, {"VWM",4}, {"VWN",2} } }, { "Oland", { {"KWG",16}, {"KWI",2}, {"MDIMA",32}, {"MDIMC",16}, {"MWG",64}, {"NDIMB",32}, {"NDIMC",16}, {"NWG",128}, {"SA",1}, {"SB",1}, {"STRM",1}, {"STRN",0}, {"VWM",2}, {"VWN",4} } }, @@ -123,6 +124,7 @@ const Database::DatabaseEntry XgemmComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"KWG",32}, {"KWI",2}, {"MDIMA",32}, {"MDIMC",32}, {"MWG",64}, {"NDIMB",8}, {"NDIMC",8}, {"NWG",64}, {"SA",0}, {"SB",0}, {"STRM",1}, {"STRN",1}, {"VWM",2}, {"VWN",8} } }, + { "ATI Radeon HD 6750M", { {"KWG",32}, {"KWI",2}, {"MDIMA",8}, {"MDIMC",8}, {"MWG",32}, {"NDIMB",16}, {"NDIMC",16}, {"NWG",64}, {"SA",1}, {"SB",1}, {"STRM",0}, {"STRN",0}, {"VWM",4}, {"VWN",4} } }, { "Ellesmere", { {"KWG",32}, {"KWI",2}, {"MDIMA",16}, {"MDIMC",16}, {"MWG",32}, {"NDIMB",8}, {"NDIMC",8}, {"NWG",32}, {"SA",1}, {"SB",1}, {"STRM",0}, {"STRN",0}, {"VWM",1}, {"VWN",4} } }, { "Hawaii", { {"KWG",32}, {"KWI",2}, {"MDIMA",32}, {"MDIMC",8}, {"MWG",32}, {"NDIMB",8}, {"NDIMC",16}, {"NWG",32}, {"SA",1}, {"SB",0}, {"STRM",1}, {"STRN",0}, {"VWM",1}, {"VWN",1} } }, { "Oland", { {"KWG",32}, {"KWI",2}, {"MDIMA",16}, {"MDIMC",8}, {"MWG",32}, {"NDIMB",32}, {"NDIMC",32}, {"NWG",128}, {"SA",1}, {"SB",0}, {"STRM",0}, {"STRN",1}, {"VWM",2}, {"VWN",4} } }, @@ -130,7 +132,7 @@ const Database::DatabaseEntry XgemmComplexSingle = { { "Tahiti", { {"KWG",16}, {"KWI",2}, {"MDIMA",8}, {"MDIMC",8}, {"MWG",32}, {"NDIMB",8}, {"NDIMC",16}, {"NWG",32}, {"SA",1}, {"SB",0}, {"STRM",0}, {"STRN",1}, {"VWM",2}, {"VWN",1} } }, { "Tonga", { {"KWG",16}, {"KWI",2}, {"MDIMA",32}, {"MDIMC",8}, {"MWG",64}, {"NDIMB",16}, {"NDIMC",32}, {"NWG",64}, {"SA",1}, {"SB",1}, {"STRM",1}, {"STRN",0}, {"VWM",2}, {"VWN",1} } }, { "Turks", { {"KWG",16}, {"KWI",2}, {"MDIMA",8}, {"MDIMC",8}, {"MWG",32}, {"NDIMB",32}, {"NDIMC",8}, {"NWG",32}, {"SA",0}, {"SB",1}, {"STRM",0}, {"STRN",0}, {"VWM",2}, {"VWN",1} } }, - { "default", { {"KWG",32}, {"KWI",2}, {"MDIMA",32}, {"MDIMC",32}, {"MWG",32}, {"NDIMB",8}, {"NDIMC",8}, {"NWG",32}, {"SA",1}, {"SB",1}, {"STRM",0}, {"STRN",0}, {"VWM",1}, {"VWN",2} } }, + { "default", { {"KWG",32}, {"KWI",2}, {"MDIMA",8}, {"MDIMC",8}, {"MWG",32}, {"NDIMB",8}, {"NDIMC",8}, {"NWG",32}, {"SA",0}, {"SB",0}, {"STRM",0}, {"STRN",0}, {"VWM",4}, {"VWN",4} } }, } }, { // ARM GPUs diff --git a/src/database/kernels/xgemm_direct.hpp b/src/database/kernels/xgemm_direct.hpp index 4984475f..081589f8 100644 --- a/src/database/kernels/xgemm_direct.hpp +++ b/src/database/kernels/xgemm_direct.hpp @@ -44,10 +44,11 @@ const Database::DatabaseEntry XgemmDirectSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",8}, {"NDIMCD",8}, {"PADA",1}, {"PADB",1}, {"VWMD",2}, {"VWND",2}, {"WGD",32} } }, + { "ATI Radeon HD 6750M", { {"KWID",8}, {"MDIMAD",8}, {"MDIMCD",16}, {"NDIMBD",8}, {"NDIMCD",8}, {"PADA",1}, {"PADB",0}, {"VWMD",2}, {"VWND",2}, {"WGD",32} } }, { "Ellesmere", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",32}, {"NDIMCD",32}, {"PADA",1}, {"PADB",1}, {"VWMD",2}, {"VWND",1}, {"WGD",32} } }, { "Tonga", { {"KWID",16}, {"MDIMAD",16}, {"MDIMCD",16}, {"NDIMBD",32}, {"NDIMCD",8}, {"PADA",0}, {"PADB",1}, {"VWMD",1}, {"VWND",1}, {"WGD",32} } }, { "Turks", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",8}, {"NDIMCD",8}, {"PADA",1}, {"PADB",1}, {"VWMD",1}, {"VWND",1}, {"WGD",16} } }, - { "default", { {"KWID",2}, {"MDIMAD",16}, {"MDIMCD",16}, {"NDIMBD",16}, {"NDIMCD",16}, {"PADA",1}, {"PADB",1}, {"VWMD",2}, {"VWND",2}, {"WGD",32} } }, + { "default", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",8}, {"NDIMCD",8}, {"PADA",1}, {"PADB",1}, {"VWMD",1}, {"VWND",2}, {"WGD",16} } }, } }, { // Intel CPUs @@ -75,7 +76,7 @@ const Database::DatabaseEntry XgemmDirectSingle = { }, { // Default kDeviceTypeAll, "default", { - { "default", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",8}, {"NDIMCD",8}, {"PADA",1}, {"PADB",1}, {"VWMD",4}, {"VWND",4}, {"WGD",32} } }, + { "default", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",8}, {"NDIMCD",8}, {"PADA",1}, {"PADB",1}, {"VWMD",1}, {"VWND",2}, {"WGD",32} } }, } }, } @@ -88,6 +89,7 @@ const Database::DatabaseEntry XgemmDirectComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"KWID",2}, {"MDIMAD",16}, {"MDIMCD",16}, {"NDIMBD",16}, {"NDIMCD",16}, {"PADA",1}, {"PADB",1}, {"VWMD",1}, {"VWND",1}, {"WGD",16} } }, + { "ATI Radeon HD 6750M", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",8}, {"NDIMCD",8}, {"PADA",1}, {"PADB",1}, {"VWMD",1}, {"VWND",1}, {"WGD",8} } }, { "Tonga", { {"KWID",2}, {"MDIMAD",16}, {"MDIMCD",16}, {"NDIMBD",16}, {"NDIMCD",16}, {"PADA",1}, {"PADB",1}, {"VWMD",2}, {"VWND",2}, {"WGD",32} } }, { "Turks", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",8}, {"NDIMCD",8}, {"PADA",1}, {"PADB",1}, {"VWMD",2}, {"VWND",2}, {"WGD",16} } }, { "default", { {"KWID",2}, {"MDIMAD",16}, {"MDIMCD",16}, {"NDIMBD",16}, {"NDIMCD",16}, {"PADA",1}, {"PADB",1}, {"VWMD",2}, {"VWND",2}, {"WGD",32} } }, diff --git a/src/database/kernels/xgemv.hpp b/src/database/kernels/xgemv.hpp index 05f47554..773c6df7 100644 --- a/src/database/kernels/xgemv.hpp +++ b/src/database/kernels/xgemv.hpp @@ -45,6 +45,7 @@ const Database::DatabaseEntry XgemvSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"WGS1",128}, {"WPT1",1} } }, + { "ATI Radeon HD 6750M", { {"WGS1",32}, {"WPT1",1} } }, { "Ellesmere", { {"WGS1",256}, {"WPT1",1} } }, { "Hawaii", { {"WGS1",128}, {"WPT1",1} } }, { "Oland", { {"WGS1",128}, {"WPT1",1} } }, @@ -117,6 +118,7 @@ const Database::DatabaseEntry XgemvComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"WGS1",64}, {"WPT1",1} } }, + { "ATI Radeon HD 6750M", { {"WGS1",64}, {"WPT1",1} } }, { "Ellesmere", { {"WGS1",32}, {"WPT1",1} } }, { "Hawaii", { {"WGS1",64}, {"WPT1",1} } }, { "Oland", { {"WGS1",64}, {"WPT1",1} } }, diff --git a/src/database/kernels/xgemv_fast.hpp b/src/database/kernels/xgemv_fast.hpp index efc64d0c..e52979b2 100644 --- a/src/database/kernels/xgemv_fast.hpp +++ b/src/database/kernels/xgemv_fast.hpp @@ -45,6 +45,7 @@ const Database::DatabaseEntry XgemvFastSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"VW2",1}, {"WGS2",128}, {"WPT2",1} } }, + { "ATI Radeon HD 6750M", { {"VW2",2}, {"WGS2",64}, {"WPT2",2} } }, { "Ellesmere", { {"VW2",1}, {"WGS2",64}, {"WPT2",1} } }, { "Hawaii", { {"VW2",1}, {"WGS2",64}, {"WPT2",1} } }, { "Oland", { {"VW2",1}, {"WGS2",64}, {"WPT2",1} } }, @@ -117,6 +118,7 @@ const Database::DatabaseEntry XgemvFastComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"VW2",2}, {"WGS2",256}, {"WPT2",2} } }, + { "ATI Radeon HD 6750M", { {"VW2",1}, {"WGS2",128}, {"WPT2",1} } }, { "Ellesmere", { {"VW2",1}, {"WGS2",64}, {"WPT2",1} } }, { "Hawaii", { {"VW2",1}, {"WGS2",64}, {"WPT2",1} } }, { "Oland", { {"VW2",1}, {"WGS2",64}, {"WPT2",1} } }, diff --git a/src/database/kernels/xgemv_fast_rot.hpp b/src/database/kernels/xgemv_fast_rot.hpp index aff83ff2..8ae36211 100644 --- a/src/database/kernels/xgemv_fast_rot.hpp +++ b/src/database/kernels/xgemv_fast_rot.hpp @@ -44,6 +44,7 @@ const Database::DatabaseEntry XgemvFastRotSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"VW3",8}, {"WGS3",64}, {"WPT3",32} } }, + { "ATI Radeon HD 6750M", { {"VW3",8}, {"WGS3",128}, {"WPT3",16} } }, { "Ellesmere", { {"VW3",8}, {"WGS3",32}, {"WPT3",32} } }, { "Tonga", { {"VW3",8}, {"WGS3",128}, {"WPT3",32} } }, { "Turks", { {"VW3",8}, {"WGS3",128}, {"WPT3",16} } }, @@ -93,6 +94,7 @@ const Database::DatabaseEntry XgemvFastRotComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"VW3",8}, {"WGS3",16}, {"WPT3",16} } }, + { "ATI Radeon HD 6750M", { {"VW3",8}, {"WGS3",32}, {"WPT3",8} } }, { "Ellesmere", { {"VW3",2}, {"WGS3",32}, {"WPT3",16} } }, { "Tonga", { {"VW3",4}, {"WGS3",32}, {"WPT3",32} } }, { "Turks", { {"VW3",4}, {"WGS3",32}, {"WPT3",8} } }, @@ -119,7 +121,7 @@ const Database::DatabaseEntry XgemvFastRotComplexSingle = { }, { // Default kDeviceTypeAll, "default", { - { "default", { {"VW3",4}, {"WGS3",64}, {"WPT3",16} } }, + { "default", { {"VW3",4}, {"WGS3",16}, {"WPT3",16} } }, } }, } diff --git a/src/database/kernels/xger.hpp b/src/database/kernels/xger.hpp index ab528800..70832f6f 100644 --- a/src/database/kernels/xger.hpp +++ b/src/database/kernels/xger.hpp @@ -45,6 +45,7 @@ const Database::DatabaseEntry XgerSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"WGS1",256}, {"WGS2",1}, {"WPT",1} } }, + { "ATI Radeon HD 6750M", { {"WGS1",16}, {"WGS2",16}, {"WPT",4} } }, { "Ellesmere", { {"WGS1",64}, {"WGS2",4}, {"WPT",2} } }, { "Hawaii", { {"WGS1",64}, {"WGS2",2}, {"WPT",1} } }, { "Oland", { {"WGS1",32}, {"WGS2",4}, {"WPT",2} } }, @@ -52,7 +53,7 @@ const Database::DatabaseEntry XgerSingle = { { "Tahiti", { {"WGS1",256}, {"WGS2",1}, {"WPT",1} } }, { "Tonga", { {"WGS1",256}, {"WGS2",1}, {"WPT",2} } }, { "Turks", { {"WGS1",64}, {"WGS2",4}, {"WPT",2} } }, - { "default", { {"WGS1",32}, {"WGS2",8}, {"WPT",2} } }, + { "default", { {"WGS1",16}, {"WGS2",16}, {"WPT",1} } }, } }, { // ARM GPUs @@ -99,7 +100,7 @@ const Database::DatabaseEntry XgerSingle = { }, { // Default kDeviceTypeAll, "default", { - { "default", { {"WGS1",32}, {"WGS2",8}, {"WPT",2} } }, + { "default", { {"WGS1",32}, {"WGS2",4}, {"WPT",2} } }, } }, } @@ -112,6 +113,7 @@ const Database::DatabaseEntry XgerComplexSingle = { { // AMD GPUs kDeviceTypeGPU, "AMD", { { "AMD Radeon R9 M370X Compute Engine", { {"WGS1",64}, {"WGS2",4}, {"WPT",1} } }, + { "ATI Radeon HD 6750M", { {"WGS1",16}, {"WGS2",16}, {"WPT",1} } }, { "Ellesmere", { {"WGS1",16}, {"WGS2",8}, {"WPT",2} } }, { "Hawaii", { {"WGS1",64}, {"WGS2",1}, {"WPT",2} } }, { "Oland", { {"WGS1",4}, {"WGS2",8}, {"WPT",1} } },