diff options
author | Cedric Nugteren <web@cedricnugteren.nl> | 2017-09-14 21:27:33 +0200 |
---|---|---|
committer | Cedric Nugteren <web@cedricnugteren.nl> | 2017-09-14 21:27:33 +0200 |
commit | 0d13d814c21f1fdb3b0781b84bf12914aec5225a (patch) | |
tree | f4ed5bfe6b13a5e865b1973ca571df87f1af4bc2 /src/database/kernels/xgemv_fast_rot | |
parent | 14a61d24258498811ac0dfd11cad6087a231fa2a (diff) |
Added architecture layer in the tuning database for better performance on unseen devices
Diffstat (limited to 'src/database/kernels/xgemv_fast_rot')
5 files changed, 192 insertions, 80 deletions
diff --git a/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_16.hpp b/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_16.hpp index 8d516141..15d1b4fc 100644 --- a/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_16.hpp +++ b/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_16.hpp @@ -14,19 +14,28 @@ const DatabaseEntry XgemvFastRotHalf = { "XgemvFastRot", Precision::kHalf, {"VW3", "WGS3", "WPT3"}, { { // AMD GPUs kDeviceTypeGPU, "AMD", { - { "Ellesmere", { 8, 32, 32 } }, - { "default", { 8, 32, 32 } }, + { "Ellesmere", { + { "AMD Radeon RX 480", { 8, 32, 32 } }, + { "default", { 8, 32, 32 } }, + } }, + { "default", { + { "default", { 8, 32, 32 } }, + } }, } }, { // Intel GPUs kDeviceTypeGPU, "Intel", { - { "Intel(R) HD Graphics Skylake ULT GT2", { 8, 128, 32 } }, - { "default", { 8, 128, 32 } }, + { "default", { + { "Intel(R) HD Graphics Skylake ULT GT2", { 8, 128, 32 } }, + { "default", { 8, 128, 32 } }, + } }, } }, { // Default kDeviceTypeAll, "default", { - { "default", { 8, 128, 32 } }, + { "default", { + { "default", { 8, 128, 32 } }, + } }, } }, } diff --git a/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_32.hpp b/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_32.hpp index 71b8b355..66e3eecf 100644 --- a/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_32.hpp +++ b/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_32.hpp @@ -14,54 +14,93 @@ const DatabaseEntry XgemvFastRotSingle = { "XgemvFastRot", Precision::kSingle, {"VW3", "WGS3", "WPT3"}, { { // AMD GPUs kDeviceTypeGPU, "AMD", { - { "AMD Radeon R9 M370X Compute Engine", { 8, 64, 32 } }, - { "ATI Radeon HD 6750M", { 8, 128, 16 } }, - { "Ellesmere", { 8, 32, 32 } }, - { "Fiji", { 4, 32, 16 } }, - { "Tonga", { 8, 128, 32 } }, - { "Turks", { 8, 128, 16 } }, - { "default", { 8, 32, 32 } }, + { "Ellesmere", { + { "AMD Radeon RX 480", { 8, 32, 32 } }, + { "default", { 8, 32, 32 } }, + } }, + { "Fiji", { + { "AMD Radeon R9 Fury X", { 4, 32, 16 } }, + { "AMD Radeon R9 M370X Compute Engine", { 8, 64, 32 } }, + { "default", { 8, 32, 32 } }, + } }, + { "Tonga", { + { "AMD Radeon R9 380", { 8, 128, 32 } }, + { "default", { 8, 128, 32 } }, + } }, + { "Turks", { + { "AMD Radeon HD 6770M", { 8, 128, 16 } }, + { "default", { 8, 128, 16 } }, + } }, + { "Vancouver", { + { "ATI Radeon HD 6750M", { 8, 128, 16 } }, + { "default", { 8, 128, 16 } }, + } }, + { "default", { + { "default", { 8, 32, 32 } }, + } }, } }, { // Intel CPUs kDeviceTypeCPU, "Intel", { - { "Intel(R) Core(TM) i7-2670QM CPU @ 2.20GHz", { 4, 32, 32 } }, - { "Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz", { 8, 128, 8 } }, - { "Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz", { 4, 32, 32 } }, - { "Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz", { 8, 16, 8 } }, - { "default", { 8, 32, 32 } }, + { "default", { + { "Intel(R) Core(TM) i7-2670QM CPU @ 2.20GHz", { 4, 32, 32 } }, + { "Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz", { 8, 128, 8 } }, + { "Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz", { 4, 32, 32 } }, + { "Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz", { 8, 16, 8 } }, + { "default", { 8, 32, 32 } }, + } }, } }, { // Intel GPUs kDeviceTypeGPU, "Intel", { - { "Intel(R) HD Graphics 5500 BroadWell U-Processor GT2", { 8, 64, 32 } }, - { "Intel(R) HD Graphics Haswell Ultrabook GT2 Mobile", { 4, 64, 16 } }, - { "Intel(R) HD Graphics IvyBridge M GT2", { 2, 32, 16 } }, - { "Intel(R) HD Graphics Skylake ULT GT2", { 4, 64, 16 } }, - { "Iris Pro", { 4, 16, 16 } }, - { "default", { 4, 64, 16 } }, + { "default", { + { "Intel(R) HD Graphics 5500 BroadWell U-Processor GT2", { 8, 64, 32 } }, + { "Intel(R) HD Graphics Haswell Ultrabook GT2 Mobile", { 4, 64, 16 } }, + { "Intel(R) HD Graphics IvyBridge M GT2", { 2, 32, 16 } }, + { "Intel(R) HD Graphics Skylake ULT GT2", { 4, 64, 16 } }, + { "Iris Pro", { 4, 16, 16 } }, + { "default", { 4, 64, 16 } }, + } }, } }, { // NVIDIA GPUs kDeviceTypeGPU, "NVIDIA", { - { "GeForce GT 650M", { 8, 32, 16 } }, - { "GeForce GTX 1080", { 8, 32, 32 } }, - { "GeForce GTX 750 Ti", { 8, 32, 32 } }, - { "GeForce GTX TITAN", { 1, 16, 16 } }, - { "GeForce GTX TITAN Black", { 4, 128, 16 } }, - { "TITAN X (Pascal)", { 8, 64, 32 } }, - { "default", { 8, 32, 32 } }, + { "SM3.0", { + { "GeForce GT 650M", { 8, 32, 16 } }, + { "default", { 8, 32, 16 } }, + } }, + { "SM3.5", { + { "GeForce GTX TITAN", { 1, 16, 16 } }, + { "GeForce GTX TITAN Black", { 4, 128, 16 } }, + { "default", { 1, 16, 16 } }, + } }, + { "SM5.0", { + { "GeForce GTX 750 Ti", { 8, 32, 32 } }, + { "default", { 8, 32, 32 } }, + } }, + { "SM6.1", { + { "GeForce GTX 1080", { 8, 32, 32 } }, + { "TITAN X (Pascal)", { 8, 64, 32 } }, + { "default", { 8, 64, 32 } }, + } }, + { "default", { + { "default", { 8, 32, 32 } }, + } }, } }, { // QUALCOMM GPUs kDeviceTypeGPU, "QUALCOMM", { - { "QUALCOMM Adreno(TM)", { 4, 64, 16 } }, - { "default", { 4, 64, 16 } }, + { "default", { + { "QUALCOMM Adreno(TM)", { 4, 64, 16 } }, + { "default", { 4, 64, 16 } }, + } }, } }, { // Default kDeviceTypeAll, "default", { - { "default", { 8, 32, 32 } }, + { "default", { + { "default", { 8, 32, 32 } }, + } }, } }, } diff --git a/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_3232.hpp b/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_3232.hpp index 4fd88fc4..445e7161 100644 --- a/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_3232.hpp +++ b/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_3232.hpp @@ -14,37 +14,60 @@ const DatabaseEntry XgemvFastRotComplexSingle = { "XgemvFastRot", Precision::kComplexSingle, {"VW3", "WGS3", "WPT3"}, { { // AMD GPUs kDeviceTypeGPU, "AMD", { - { "AMD Radeon R9 M370X Compute Engine", { 8, 16, 16 } }, - { "ATI Radeon HD 6750M", { 8, 32, 8 } }, - { "Ellesmere", { 2, 32, 16 } }, - { "Fiji", { 4, 32, 32 } }, - { "Tonga", { 4, 32, 32 } }, - { "Turks", { 4, 32, 8 } }, - { "default", { 8, 16, 16 } }, + { "Ellesmere", { + { "AMD Radeon RX 480", { 2, 32, 16 } }, + { "default", { 2, 32, 16 } }, + } }, + { "Fiji", { + { "AMD Radeon R9 Fury X", { 4, 32, 32 } }, + { "AMD Radeon R9 M370X Compute Engine", { 8, 16, 16 } }, + { "default", { 8, 16, 16 } }, + } }, + { "Tonga", { + { "AMD Radeon R9 380", { 4, 32, 32 } }, + { "default", { 4, 32, 32 } }, + } }, + { "Turks", { + { "AMD Radeon HD 6770M", { 4, 32, 8 } }, + { "default", { 4, 32, 8 } }, + } }, + { "Vancouver", { + { "ATI Radeon HD 6750M", { 8, 32, 8 } }, + { "default", { 8, 32, 8 } }, + } }, + { "default", { + { "default", { 8, 16, 16 } }, + } }, } }, { // Intel CPUs kDeviceTypeCPU, "Intel", { - { "Intel(R) Core(TM) i7-2670QM CPU @ 2.20GHz", { 4, 32, 32 } }, - { "Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz", { 8, 32, 8 } }, - { "Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz", { 4, 32, 32 } }, - { "Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz", { 4, 16, 16 } }, - { "default", { 4, 32, 32 } }, + { "default", { + { "Intel(R) Core(TM) i7-2670QM CPU @ 2.20GHz", { 4, 32, 32 } }, + { "Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz", { 8, 32, 8 } }, + { "Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz", { 4, 32, 32 } }, + { "Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz", { 4, 16, 16 } }, + { "default", { 4, 32, 32 } }, + } }, } }, { // Intel GPUs kDeviceTypeGPU, "Intel", { - { "Intel(R) HD Graphics 5500 BroadWell U-Processor GT2", { 2, 16, 16 } }, - { "Intel(R) HD Graphics Haswell Ultrabook GT2 Mobile", { 4, 128, 8 } }, - { "Intel(R) HD Graphics IvyBridge M GT2", { 4, 32, 8 } }, - { "Intel(R) HD Graphics Skylake ULT GT2", { 4, 64, 16 } }, - { "Iris Pro", { 4, 16, 16 } }, - { "default", { 2, 32, 8 } }, + { "default", { + { "Intel(R) HD Graphics 5500 BroadWell U-Processor GT2", { 2, 16, 16 } }, + { "Intel(R) HD Graphics Haswell Ultrabook GT2 Mobile", { 4, 128, 8 } }, + { "Intel(R) HD Graphics IvyBridge M GT2", { 4, 32, 8 } }, + { "Intel(R) HD Graphics Skylake ULT GT2", { 4, 64, 16 } }, + { "Iris Pro", { 4, 16, 16 } }, + { "default", { 2, 32, 8 } }, + } }, } }, { // Default kDeviceTypeAll, "default", { - { "default", { 4, 16, 16 } }, + { "default", { + { "default", { 4, 16, 16 } }, + } }, } }, } diff --git a/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_64.hpp b/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_64.hpp index 66299b56..842c5bae 100644 --- a/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_64.hpp +++ b/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_64.hpp @@ -14,35 +14,61 @@ const DatabaseEntry XgemvFastRotDouble = { "XgemvFastRot", Precision::kDouble, {"VW3", "WGS3", "WPT3"}, { { // AMD GPUs kDeviceTypeGPU, "AMD", { - { "AMD Radeon R9 M370X Compute Engine", { 4, 16, 16 } }, - { "Ellesmere", { 4, 16, 16 } }, - { "Fiji", { 4, 32, 32 } }, - { "Tonga", { 4, 16, 16 } }, - { "default", { 4, 16, 16 } }, + { "Ellesmere", { + { "AMD Radeon RX 480", { 4, 16, 16 } }, + { "default", { 4, 16, 16 } }, + } }, + { "Fiji", { + { "AMD Radeon R9 Fury X", { 4, 32, 32 } }, + { "AMD Radeon R9 M370X Compute Engine", { 4, 16, 16 } }, + { "default", { 4, 16, 16 } }, + } }, + { "Tonga", { + { "AMD Radeon R9 380", { 4, 16, 16 } }, + { "default", { 4, 16, 16 } }, + } }, + { "default", { + { "default", { 4, 16, 16 } }, + } }, } }, { // Intel CPUs kDeviceTypeCPU, "Intel", { - { "Intel(R) Core(TM) i7-2670QM CPU @ 2.20GHz", { 4, 32, 32 } }, - { "Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz", { 8, 16, 8 } }, - { "Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz", { 4, 32, 32 } }, - { "Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz", { 8, 16, 8 } }, - { "default", { 8, 32, 32 } }, + { "default", { + { "Intel(R) Core(TM) i7-2670QM CPU @ 2.20GHz", { 4, 32, 32 } }, + { "Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz", { 8, 16, 8 } }, + { "Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz", { 4, 32, 32 } }, + { "Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz", { 8, 16, 8 } }, + { "default", { 8, 32, 32 } }, + } }, } }, { // NVIDIA GPUs kDeviceTypeGPU, "NVIDIA", { - { "GeForce GTX 1080", { 8, 32, 32 } }, - { "GeForce GTX 750 Ti", { 4, 32, 16 } }, - { "GeForce GTX TITAN", { 1, 16, 16 } }, - { "GeForce GTX TITAN Black", { 1, 16, 16 } }, - { "TITAN X (Pascal)", { 8, 32, 32 } }, - { "default", { 4, 32, 16 } }, + { "SM3.5", { + { "GeForce GTX TITAN", { 1, 16, 16 } }, + { "GeForce GTX TITAN Black", { 1, 16, 16 } }, + { "default", { 1, 16, 16 } }, + } }, + { "SM5.0", { + { "GeForce GTX 750 Ti", { 4, 32, 16 } }, + { "default", { 4, 32, 16 } }, + } }, + { "SM6.1", { + { "GeForce GTX 1080", { 8, 32, 32 } }, + { "TITAN X (Pascal)", { 8, 32, 32 } }, + { "default", { 8, 32, 32 } }, + } }, + { "default", { + { "default", { 4, 32, 16 } }, + } }, } }, { // Default kDeviceTypeAll, "default", { - { "default", { 4, 16, 16 } }, + { "default", { + { "default", { 4, 16, 16 } }, + } }, } }, } diff --git a/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_6464.hpp b/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_6464.hpp index bc1964ff..84340b4c 100644 --- a/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_6464.hpp +++ b/src/database/kernels/xgemv_fast_rot/xgemv_fast_rot_6464.hpp @@ -14,25 +14,40 @@ const DatabaseEntry XgemvFastRotComplexDouble = { "XgemvFastRot", Precision::kComplexDouble, {"VW3", "WGS3", "WPT3"}, { { // AMD GPUs kDeviceTypeGPU, "AMD", { - { "AMD Radeon R9 M370X Compute Engine", { 4, 32, 16 } }, - { "Ellesmere", { 4, 16, 16 } }, - { "Fiji", { 4, 32, 8 } }, - { "Tonga", { 4, 16, 8 } }, - { "default", { 8, 32, 16 } }, + { "Ellesmere", { + { "AMD Radeon RX 480", { 4, 16, 16 } }, + { "default", { 4, 16, 16 } }, + } }, + { "Fiji", { + { "AMD Radeon R9 Fury X", { 4, 32, 8 } }, + { "AMD Radeon R9 M370X Compute Engine", { 4, 32, 16 } }, + { "default", { 4, 32, 8 } }, + } }, + { "Tonga", { + { "AMD Radeon R9 380", { 4, 16, 8 } }, + { "default", { 4, 16, 8 } }, + } }, + { "default", { + { "default", { 8, 32, 16 } }, + } }, } }, { // Intel CPUs kDeviceTypeCPU, "Intel", { - { "Intel(R) Core(TM) i7-2670QM CPU @ 2.20GHz", { 2, 16, 16 } }, - { "Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz", { 4, 64, 4 } }, - { "Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz", { 2, 16, 16 } }, - { "Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz", { 8, 16, 16 } }, - { "default", { 8, 16, 16 } }, + { "default", { + { "Intel(R) Core(TM) i7-2670QM CPU @ 2.20GHz", { 2, 16, 16 } }, + { "Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz", { 4, 64, 4 } }, + { "Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz", { 2, 16, 16 } }, + { "Intel(R) Core(TM) i7-5930K CPU @ 3.50GHz", { 8, 16, 16 } }, + { "default", { 8, 16, 16 } }, + } }, } }, { // Default kDeviceTypeAll, "default", { - { "default", { 4, 16, 16 } }, + { "default", { + { "default", { 4, 16, 16 } }, + } }, } }, } |