summaryrefslogtreecommitdiff
path: root/src/database/kernels/xconvgemm/xconvgemm_32.hpp
diff options
context:
space:
mode:
authorCedric Nugteren <web@cedricnugteren.nl>2021-08-20 08:18:36 +0200
committerGitHub <noreply@github.com>2021-08-20 08:18:36 +0200
commit93d6070e276b574d28f257f25e7f11bbea77507a (patch)
tree40b3a873ad52c1ce46c8a8bfc1559af965cca9a7 /src/database/kernels/xconvgemm/xconvgemm_32.hpp
parent6dbd6d96bc6898affef0ff873d1e3ac41bfa223b (diff)
parent2eaabeed100225aca81db469e50a8871636727f8 (diff)
Merge pull request #423 from CNugteren/new_tuning_results
New tuning results for 1 Intel CPU and 5 NVIDIA GPUs
Diffstat (limited to 'src/database/kernels/xconvgemm/xconvgemm_32.hpp')
-rw-r--r--src/database/kernels/xconvgemm/xconvgemm_32.hpp31
1 files changed, 30 insertions, 1 deletions
diff --git a/src/database/kernels/xconvgemm/xconvgemm_32.hpp b/src/database/kernels/xconvgemm/xconvgemm_32.hpp
index 216a9f42..0cb0acfe 100644
--- a/src/database/kernels/xconvgemm/xconvgemm_32.hpp
+++ b/src/database/kernels/xconvgemm/xconvgemm_32.hpp
@@ -12,6 +12,14 @@ namespace database {
const DatabaseEntry XconvgemmSingle = {
"Xconvgemm", Precision::kSingle, {"KWID", "MDIMAD", "MDIMCD", "NDIMBD", "NDIMCD", "PADA", "PADB", "VWMD", "VWND", "WGD"}, {
+ { // Intel CPUs
+ kDeviceTypeCPU, "Intel", {
+ { "default", {
+ { Name{"Intel(R) Core(TM) i9-9980HK CPU @ 2.40GHz "}, Params{ 1, 16, 32, 32, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ { kDeviceNameDefault , Params{ 1, 16, 32, 32, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ } },
+ }
+ },
{ // Intel GPUs
kDeviceTypeGPU, "Intel", {
{ "default", {
@@ -21,10 +29,31 @@ const DatabaseEntry XconvgemmSingle = {
} },
}
},
+ { // NVIDIA GPUs
+ kDeviceTypeGPU, "NVIDIA", {
+ { "SM7.0", {
+ { Name{"Quadro GV100 "}, Params{ 1, 8, 32, 16, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ { Name{"Tesla V100-PCIE-16GB "}, Params{ 1, 8, 32, 32, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ { kDeviceNameDefault , Params{ 1, 8, 32, 32, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ } },
+ { "SM7.5", {
+ { Name{"Quadro T2000 "}, Params{ 1, 32, 32, 32, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ { Name{"Tesla T4 "}, Params{ 1, 32, 32, 32, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ { kDeviceNameDefault , Params{ 1, 32, 32, 32, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ } },
+ { "SM8.0", {
+ { Name{"A100-PCIE-40GB "}, Params{ 1, 8, 32, 32, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ { kDeviceNameDefault , Params{ 1, 8, 32, 32, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ } },
+ { "default", {
+ { kDeviceNameDefault , Params{ 1, 8, 32, 32, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ } },
+ }
+ },
{ // Default
kDeviceTypeAll, "default", {
{ "default", {
- { kDeviceNameDefault , Params{ 1, 8, 16, 16, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
+ { kDeviceNameDefault , Params{ 1, 16, 32, 32, 8, 0, 0, 1, 1, 32, 0, 0, 0, 0, 0, 0 } },
} },
}
},