Added "train_set" file. A large dataset to observe the speedups betwe…

…en CPU and GPU execution times.
MKLab-ITI · Oct 4, 2013 · 16ad23b · 16ad23b
1 parent 091f78a
commit 16ad23b
Show file tree

Hide file tree

Showing 9 changed files with 30,004 additions and 810 deletions.
diff --git a/binaries/linux/README-GPU b/binaries/linux/README-GPU
@@ -6,6 +6,8 @@ The use of the new executable is exactly the same as with the original one.
 This binary was built with the CUBLAS API version 2 which is compatible with SDKs from 4.0 and up.
 
 To test the binary "svm-train-gpu" you can run the easy.py script which is located in the "tools" folder.
+To observe speed improvements between CPU and GPU execution we provide a custom relatively large dataset (train_set) which can be used as an input to easy.py.
+
 
 FEATURES
 

diff --git a/binaries/linux/heart_scale b/binaries/linux/heart_scale
diff --git a/binaries/linux/train_set b/binaries/linux/train_set
diff --git a/binaries/windows/x64/README-GPU b/binaries/windows/x64/README-GPU
@@ -6,6 +6,7 @@ The use of the new executable is exactly the same as with the original one.
 This executable was built with the CUBLAS API version 2 which is compatible with SDKs from 4.0 and up.
 
 To test the executable "svm-train-gpu" you can run the easy.py script which is located in the "tools" folder.
+To observe speed improvements between CPU and GPU execution we provide a custom relatively large dataset (train_set) which can be used as an input to easy.py.
 
 FEATURES
 

diff --git a/binaries/windows/x64/heart_scale b/binaries/windows/x64/heart_scale
diff --git a/binaries/windows/x64/train_set b/binaries/windows/x64/train_set
diff --git a/binaries/windows/x86/README-GPU b/binaries/windows/x86/README-GPU
@@ -6,6 +6,7 @@ The use of the new executable is exactly the same as with the original one.
 This executable was built with the CUBLAS API version 2 which is compatible with SDKs from 4.0 and up.
 
 To test the executable "svm-train-gpu" you can run the easy.py script which is located in the "tools" folder.
+To observe speed improvements between CPU and GPU execution we provide a custom relatively large dataset (train_set) which can be used as an input to easy.py.
 
 FEATURES
 

diff --git a/binaries/windows/x86/heart_scale b/binaries/windows/x86/heart_scale
diff --git a/binaries/windows/x86/train_set b/binaries/windows/x86/train_set