summaryrefslogtreecommitdiff
path: root/src/database/kernels
diff options
context:
space:
mode:
authorCedric Nugteren <web@cedricnugteren.nl>2017-06-01 22:52:08 +0200
committerCedric Nugteren <web@cedricnugteren.nl>2017-06-01 22:52:08 +0200
commit33ed1e5a06461ca4e4ec4a1c02b9a8b19432c5d2 (patch)
treed373e0dccceb5a1b42aa56f31d1c8f4e88313c36 /src/database/kernels
parentf57e209aab73279e5024eaaa4df3b92d900fb7a4 (diff)
Added tuning results for GeForce GT 650M (thanks to bzcheeseman)
Diffstat (limited to 'src/database/kernels')
-rw-r--r--src/database/kernels/copy.hpp1
-rw-r--r--src/database/kernels/pad.hpp3
-rw-r--r--src/database/kernels/transpose.hpp1
-rw-r--r--src/database/kernels/xaxpy.hpp3
-rw-r--r--src/database/kernels/xdot.hpp1
-rw-r--r--src/database/kernels/xgemm.hpp1
-rw-r--r--src/database/kernels/xgemm_direct.hpp3
-rw-r--r--src/database/kernels/xgemv.hpp1
-rw-r--r--src/database/kernels/xgemv_fast.hpp1
-rw-r--r--src/database/kernels/xgemv_fast_rot.hpp1
-rw-r--r--src/database/kernels/xger.hpp1
11 files changed, 14 insertions, 3 deletions
diff --git a/src/database/kernels/copy.hpp b/src/database/kernels/copy.hpp
index fd8be5e6..28ed07a3 100644
--- a/src/database/kernels/copy.hpp
+++ b/src/database/kernels/copy.hpp
@@ -94,6 +94,7 @@ const Database::DatabaseEntry CopySingle = {
{ // NVIDIA GPUs
kDeviceTypeGPU, "NVIDIA", {
{ "GRID K520", { {"COPY_DIMX",16}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",1} } },
+ { "GeForce GT 650M", { {"COPY_DIMX",16}, {"COPY_DIMY",16}, {"COPY_VW",4}, {"COPY_WPT",2} } },
{ "GeForce GTX 1070", { {"COPY_DIMX",8}, {"COPY_DIMY",16}, {"COPY_VW",4}, {"COPY_WPT",1} } },
{ "GeForce GTX 1080", { {"COPY_DIMX",8}, {"COPY_DIMY",32}, {"COPY_VW",4}, {"COPY_WPT",1} } },
{ "GeForce GTX 480", { {"COPY_DIMX",8}, {"COPY_DIMY",8}, {"COPY_VW",4}, {"COPY_WPT",1} } },
diff --git a/src/database/kernels/pad.hpp b/src/database/kernels/pad.hpp
index 4336c687..a2a30713 100644
--- a/src/database/kernels/pad.hpp
+++ b/src/database/kernels/pad.hpp
@@ -94,6 +94,7 @@ const Database::DatabaseEntry PadSingle = {
{ // NVIDIA GPUs
kDeviceTypeGPU, "NVIDIA", {
{ "GRID K520", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",2}, {"PAD_WPTY",1} } },
+ { "GeForce GT 650M", { {"PAD_DIMX",32}, {"PAD_DIMY",16}, {"PAD_WPTX",2}, {"PAD_WPTY",2} } },
{ "GeForce GTX 1070", { {"PAD_DIMX",16}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",1} } },
{ "GeForce GTX 1080", { {"PAD_DIMX",16}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",1} } },
{ "GeForce GTX 480", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",4} } },
@@ -108,7 +109,7 @@ const Database::DatabaseEntry PadSingle = {
{ "TITAN X (Pascal)", { {"PAD_DIMX",16}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",2} } },
{ "Tesla K20m", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",2}, {"PAD_WPTY",1} } },
{ "Tesla K40m", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",1} } },
- { "default", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",1}, {"PAD_WPTY",4} } },
+ { "default", { {"PAD_DIMX",32}, {"PAD_DIMY",8}, {"PAD_WPTX",4}, {"PAD_WPTY",1} } },
}
},
{ // Default
diff --git a/src/database/kernels/transpose.hpp b/src/database/kernels/transpose.hpp
index 4020e903..9e44e61c 100644
--- a/src/database/kernels/transpose.hpp
+++ b/src/database/kernels/transpose.hpp
@@ -94,6 +94,7 @@ const Database::DatabaseEntry TransposeSingle = {
{ // NVIDIA GPUs
kDeviceTypeGPU, "NVIDIA", {
{ "GRID K520", { {"TRA_DIM",32}, {"TRA_PAD",1}, {"TRA_SHUFFLE",1}, {"TRA_WPT",2} } },
+ { "GeForce GT 650M", { {"TRA_DIM",8}, {"TRA_PAD",1}, {"TRA_SHUFFLE",0}, {"TRA_WPT",4} } },
{ "GeForce GTX 1070", { {"TRA_DIM",8}, {"TRA_PAD",0}, {"TRA_SHUFFLE",1}, {"TRA_WPT",4} } },
{ "GeForce GTX 1080", { {"TRA_DIM",4}, {"TRA_PAD",0}, {"TRA_SHUFFLE",0}, {"TRA_WPT",4} } },
{ "GeForce GTX 480", { {"TRA_DIM",16}, {"TRA_PAD",1}, {"TRA_SHUFFLE",0}, {"TRA_WPT",2} } },
diff --git a/src/database/kernels/xaxpy.hpp b/src/database/kernels/xaxpy.hpp
index bb384abf..be9659c0 100644
--- a/src/database/kernels/xaxpy.hpp
+++ b/src/database/kernels/xaxpy.hpp
@@ -94,6 +94,7 @@ const Database::DatabaseEntry XaxpySingle = {
{ // NVIDIA GPUs
kDeviceTypeGPU, "NVIDIA", {
{ "GRID K520", { {"VW",2}, {"WGS",64}, {"WPT",1} } },
+ { "GeForce GT 650M", { {"VW",2}, {"WGS",1024}, {"WPT",1} } },
{ "GeForce GTX 1070", { {"VW",1}, {"WGS",64}, {"WPT",4} } },
{ "GeForce GTX 1080", { {"VW",1}, {"WGS",256}, {"WPT",1} } },
{ "GeForce GTX 480", { {"VW",2}, {"WGS",128}, {"WPT",1} } },
@@ -108,7 +109,7 @@ const Database::DatabaseEntry XaxpySingle = {
{ "TITAN X (Pascal)", { {"VW",4}, {"WGS",128}, {"WPT",1} } },
{ "Tesla K20m", { {"VW",4}, {"WGS",128}, {"WPT",1} } },
{ "Tesla K40m", { {"VW",4}, {"WGS",128}, {"WPT",1} } },
- { "default", { {"VW",4}, {"WGS",256}, {"WPT",1} } },
+ { "default", { {"VW",4}, {"WGS",1024}, {"WPT",1} } },
}
},
{ // Default
diff --git a/src/database/kernels/xdot.hpp b/src/database/kernels/xdot.hpp
index 2555a74b..be3f3bbb 100644
--- a/src/database/kernels/xdot.hpp
+++ b/src/database/kernels/xdot.hpp
@@ -78,6 +78,7 @@ const Database::DatabaseEntry XdotSingle = {
{ // NVIDIA GPUs
kDeviceTypeGPU, "NVIDIA", {
{ "GRID K520", { {"WGS1",128}, {"WGS2",32} } },
+ { "GeForce GT 650M", { {"WGS1",128}, {"WGS2",64} } },
{ "GeForce GTX 1070", { {"WGS1",128}, {"WGS2",1024} } },
{ "GeForce GTX 1080", { {"WGS1",512}, {"WGS2",64} } },
{ "GeForce GTX 480", { {"WGS1",512}, {"WGS2",32} } },
diff --git a/src/database/kernels/xgemm.hpp b/src/database/kernels/xgemm.hpp
index bf296497..f41eedf0 100644
--- a/src/database/kernels/xgemm.hpp
+++ b/src/database/kernels/xgemm.hpp
@@ -93,6 +93,7 @@ const Database::DatabaseEntry XgemmSingle = {
{ // NVIDIA GPUs
kDeviceTypeGPU, "NVIDIA", {
{ "GRID K520", { {"KWG",16}, {"KWI",2}, {"MDIMA",16}, {"MDIMC",8}, {"MWG",32}, {"NDIMB",8}, {"NDIMC",16}, {"NWG",64}, {"SA",1}, {"SB",1}, {"STRM",1}, {"STRN",1}, {"VWM",2}, {"VWN",4} } },
+ { "GeForce GT 650M", { {"KWG",32}, {"KWI",2}, {"MDIMA",8}, {"MDIMC",8}, {"MWG",32}, {"NDIMB",32}, {"NDIMC",32}, {"NWG",64}, {"SA",1}, {"SB",1}, {"STRM",0}, {"STRN",0}, {"VWM",4}, {"VWN",2} } },
{ "GeForce GTX 1070", { {"KWG",16}, {"KWI",2}, {"MDIMA",32}, {"MDIMC",16}, {"MWG",128}, {"NDIMB",32}, {"NDIMC",8}, {"NWG",128}, {"SA",1}, {"SB",1}, {"STRM",1}, {"STRN",0}, {"VWM",4}, {"VWN",1} } },
{ "GeForce GTX 1080", { {"KWG",32}, {"KWI",2}, {"MDIMA",16}, {"MDIMC",8}, {"MWG",64}, {"NDIMB",8}, {"NDIMC",8}, {"NWG",64}, {"SA",1}, {"SB",1}, {"STRM",1}, {"STRN",1}, {"VWM",4}, {"VWN",8} } },
{ "GeForce GTX 480", { {"KWG",16}, {"KWI",2}, {"MDIMA",16}, {"MDIMC",8}, {"MWG",64}, {"NDIMB",32}, {"NDIMC",16}, {"NWG",64}, {"SA",1}, {"SB",1}, {"STRM",1}, {"STRN",1}, {"VWM",2}, {"VWN",2} } },
diff --git a/src/database/kernels/xgemm_direct.hpp b/src/database/kernels/xgemm_direct.hpp
index b2f67b26..a0959f3b 100644
--- a/src/database/kernels/xgemm_direct.hpp
+++ b/src/database/kernels/xgemm_direct.hpp
@@ -68,11 +68,12 @@ const Database::DatabaseEntry XgemmDirectSingle = {
},
{ // NVIDIA GPUs
kDeviceTypeGPU, "NVIDIA", {
+ { "GeForce GT 650M", { {"KWID",16}, {"MDIMAD",16}, {"MDIMCD",16}, {"NDIMBD",8}, {"NDIMCD",16}, {"PADA",1}, {"PADB",0}, {"VWMD",2}, {"VWND",2}, {"WGD",32} } },
{ "GeForce GTX 1080", { {"KWID",16}, {"MDIMAD",16}, {"MDIMCD",8}, {"NDIMBD",16}, {"NDIMCD",8}, {"PADA",1}, {"PADB",1}, {"VWMD",1}, {"VWND",1}, {"WGD",32} } },
{ "GeForce GTX 750 Ti", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",8}, {"NDIMCD",8}, {"PADA",1}, {"PADB",1}, {"VWMD",4}, {"VWND",2}, {"WGD",32} } },
{ "GeForce GTX TITAN Black", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",16}, {"NDIMCD",16}, {"PADA",1}, {"PADB",1}, {"VWMD",4}, {"VWND",2}, {"WGD",32} } },
{ "TITAN X (Pascal)", { {"KWID",8}, {"MDIMAD",32}, {"MDIMCD",8}, {"NDIMBD",8}, {"NDIMCD",16}, {"PADA",1}, {"PADB",1}, {"VWMD",1}, {"VWND",1}, {"WGD",32} } },
- { "default", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",8}, {"NDIMCD",8}, {"PADA",1}, {"PADB",1}, {"VWMD",4}, {"VWND",2}, {"WGD",32} } },
+ { "default", { {"KWID",2}, {"MDIMAD",8}, {"MDIMCD",8}, {"NDIMBD",16}, {"NDIMCD",16}, {"PADA",1}, {"PADB",1}, {"VWMD",4}, {"VWND",2}, {"WGD",32} } },
}
},
{ // Default
diff --git a/src/database/kernels/xgemv.hpp b/src/database/kernels/xgemv.hpp
index cb83d6f7..3e527c41 100644
--- a/src/database/kernels/xgemv.hpp
+++ b/src/database/kernels/xgemv.hpp
@@ -87,6 +87,7 @@ const Database::DatabaseEntry XgemvSingle = {
{ // NVIDIA GPUs
kDeviceTypeGPU, "NVIDIA", {
{ "GRID K520", { {"WGS1",256}, {"WPT1",1} } },
+ { "GeForce GT 650M", { {"WGS1",256}, {"WPT1",1} } },
{ "GeForce GTX 1070", { {"WGS1",128}, {"WPT1",1} } },
{ "GeForce GTX 1080", { {"WGS1",32}, {"WPT1",1} } },
{ "GeForce GTX 480", { {"WGS1",64}, {"WPT1",1} } },
diff --git a/src/database/kernels/xgemv_fast.hpp b/src/database/kernels/xgemv_fast.hpp
index 4b4acf8e..f44aed28 100644
--- a/src/database/kernels/xgemv_fast.hpp
+++ b/src/database/kernels/xgemv_fast.hpp
@@ -87,6 +87,7 @@ const Database::DatabaseEntry XgemvFastSingle = {
{ // NVIDIA GPUs
kDeviceTypeGPU, "NVIDIA", {
{ "GRID K520", { {"VW2",2}, {"WGS2",256}, {"WPT2",2} } },
+ { "GeForce GT 650M", { {"VW2",2}, {"WGS2",32}, {"WPT2",2} } },
{ "GeForce GTX 1070", { {"VW2",1}, {"WGS2",256}, {"WPT2",1} } },
{ "GeForce GTX 1080", { {"VW2",1}, {"WGS2",128}, {"WPT2",1} } },
{ "GeForce GTX 480", { {"VW2",1}, {"WGS2",128}, {"WPT2",1} } },
diff --git a/src/database/kernels/xgemv_fast_rot.hpp b/src/database/kernels/xgemv_fast_rot.hpp
index 8088fd65..87bc6d01 100644
--- a/src/database/kernels/xgemv_fast_rot.hpp
+++ b/src/database/kernels/xgemv_fast_rot.hpp
@@ -72,6 +72,7 @@ const Database::DatabaseEntry XgemvFastRotSingle = {
},
{ // NVIDIA GPUs
kDeviceTypeGPU, "NVIDIA", {
+ { "GeForce GT 650M", { {"VW3",8}, {"WGS3",32}, {"WPT3",16} } },
{ "GeForce GTX 1080", { {"VW3",8}, {"WGS3",32}, {"WPT3",32} } },
{ "GeForce GTX 750 Ti", { {"VW3",8}, {"WGS3",32}, {"WPT3",32} } },
{ "GeForce GTX TITAN", { {"VW3",1}, {"WGS3",16}, {"WPT3",16} } },
diff --git a/src/database/kernels/xger.hpp b/src/database/kernels/xger.hpp
index 95dac88b..e15da762 100644
--- a/src/database/kernels/xger.hpp
+++ b/src/database/kernels/xger.hpp
@@ -86,6 +86,7 @@ const Database::DatabaseEntry XgerSingle = {
{ // NVIDIA GPUs
kDeviceTypeGPU, "NVIDIA", {
{ "GRID K520", { {"WGS1",128}, {"WGS2",1}, {"WPT",2} } },
+ { "GeForce GT 650M", { {"WGS1",32}, {"WGS2",16}, {"WPT",4} } },
{ "GeForce GTX 1070", { {"WGS1",512}, {"WGS2",1}, {"WPT",1} } },
{ "GeForce GTX 1080", { {"WGS1",16}, {"WGS2",4}, {"WPT",1} } },
{ "GeForce GTX 480", { {"WGS1",256}, {"WGS2",1}, {"WPT",4} } },