Skip to content

GitLab

Commit ba66e90e authored by peastman's avatar peastman
Browse files

Merge pull request #658 from peastman/980 

Workaround to support GTX 980
parents 68411b8d 942d4d7a
Hide whitespace changes
Inline Side-by-side
Showing with 5 additions and 0 deletions
platforms/cuda/src/CudaContext.cpp
View file @ ba66e90e
...@@ -136,6 +136,11 @@ CudaContext::CudaContext(const System& system, int deviceIndex, bool useBlocking ...@@ -136,6 +136,11 @@ CudaContext::CudaContext(const System& system, int deviceIndex, bool useBlocking
this->deviceIndex = deviceIndex; this->deviceIndex = deviceIndex;
int major, minor; int major, minor;
CHECK_RESULT(cuDeviceComputeCapability(&major, &minor, device)); CHECK_RESULT(cuDeviceComputeCapability(&major, &minor, device));
// This is a workaround to support GTX 980 with CUDA 6.5. It reports its compute capability
// as 5.2, but the compiler doesn't support anything beyond 5.0. We can remove this once
// CUDA 7.0 is released.
if (major == 5)
minor = 0;
gpuArchitecture = intToString(major)+intToString(minor); gpuArchitecture = intToString(major)+intToString(minor);
computeCapability = major+0.1*minor; computeCapability = major+0.1*minor;
if ((useDoublePrecision || useMixedPrecision) && computeCapability < 1.3) if ((useDoublePrecision || useMixedPrecision) && computeCapability < 1.3)
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment