summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--include/internal/database.h2
-rw-r--r--include/internal/database/xgemv.h18
-rw-r--r--src/database.cc2
3 files changed, 20 insertions, 2 deletions
diff --git a/include/internal/database.h b/include/internal/database.h
index d048e4f8..b3788666 100644
--- a/include/internal/database.h
+++ b/include/internal/database.h
@@ -69,7 +69,7 @@ class Database {
// The database consists of separate database entries, stored together in a vector
static const DatabaseEntry XaxpyHalf, XaxpySingle, XaxpyDouble, XaxpyComplexSingle, XaxpyComplexDouble;
static const DatabaseEntry XdotHalf, XdotSingle, XdotDouble, XdotComplexSingle, XdotComplexDouble;
- static const DatabaseEntry XgemvSingle, XgemvDouble, XgemvComplexSingle, XgemvComplexDouble;
+ static const DatabaseEntry XgemvHalf, XgemvSingle, XgemvDouble, XgemvComplexSingle, XgemvComplexDouble;
static const DatabaseEntry XgerSingle, XgerDouble, XgerComplexSingle, XgerComplexDouble;
static const DatabaseEntry XgemmSingle, XgemmDouble, XgemmComplexSingle, XgemmComplexDouble;
static const DatabaseEntry CopyHalf, CopySingle, CopyDouble, CopyComplexSingle, CopyComplexDouble;
diff --git a/include/internal/database/xgemv.h b/include/internal/database/xgemv.h
index bbbe62f6..c6e9d600 100644
--- a/include/internal/database/xgemv.h
+++ b/include/internal/database/xgemv.h
@@ -14,6 +14,24 @@
namespace clblast {
// =================================================================================================
+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} } },
+ }
+ },
+ { // Default
+ kDeviceTypeAll, "default", {
+ { "default", { {"WGS1",256}, {"WPT1",1}, {"VW2",2}, {"WGS2",64}, {"WPT2",2}, {"VW3",4}, {"WGS3",64}, {"WPT3",4} } },
+ }
+ },
+ }
+};
+
+// =================================================================================================
+
const Database::DatabaseEntry Database::XgemvSingle = {
"Xgemv", Precision::kSingle, {
{ // AMD GPUs
diff --git a/src/database.cc b/src/database.cc
index b46dde18..138dbd2d 100644
--- a/src/database.cc
+++ b/src/database.cc
@@ -31,7 +31,7 @@ namespace clblast {
const std::vector<Database::DatabaseEntry> Database::database = {
XaxpyHalf, XaxpySingle, XaxpyDouble, XaxpyComplexSingle, XaxpyComplexDouble,
XdotHalf, XdotSingle, XdotDouble, XdotComplexSingle, XdotComplexDouble,
- XgemvSingle, XgemvDouble, XgemvComplexSingle, XgemvComplexDouble,
+ XgemvHalf, XgemvSingle, XgemvDouble, XgemvComplexSingle, XgemvComplexDouble,
XgerSingle, XgerDouble, XgerComplexSingle, XgerComplexDouble,
XgemmSingle, XgemmDouble, XgemmComplexSingle, XgemmComplexDouble,
CopyHalf, CopySingle, CopyDouble, CopyComplexSingle, CopyComplexDouble,