Introduction
Introduction Statistics Contact Development Disclaimer Help
tAdd support for CUDA compute capabilities 4.x, 6.2, and 7.x - sphere - GPU-bas…
git clone git://src.adamsgaard.dk/sphere
Log
Files
Refs
LICENSE
---
commit 116df3bcf276dec7fca4fc44a8b91c39db229f66
parent 167ae1fe9a881f423cdffd2f3737ce6e9de69f78
Author: Anders Damsgaard <[email protected]>
Date: Sat, 27 Apr 2019 14:27:26 +0200
Add support for CUDA compute capabilities 4.x, 6.2, and 7.x
Diffstat:
M src/device.cu | 7 +++++++
1 file changed, 7 insertions(+), 0 deletions(-)
---
diff --git a/src/device.cu b/src/device.cu
t@@ -31,6 +31,7 @@
// Returns the number of cores per streaming multiprocessor, which is
// a function of the device compute capability
+// https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#compute-ca…
int cudaCoresPerSM(int major, int minor)
{
if (major == 1)
t@@ -41,12 +42,18 @@ int cudaCoresPerSM(int major, int minor)
return 48;
else if (major == 3)
return 192;
+ else if (major == 4)
+ return 128;
else if (major == 5)
return 128;
else if (major == 6 && minor == 0)
return 64;
else if (major == 6 && minor == 1)
return 128;
+ else if (major == 6 && minor == 2)
+ return 128;
+ else if (major == 7)
+ return 32;
else
printf("Error in cudaCoresPerSM Device compute capability value "
"(%d.%d) not recognized.", major, minor);
You are viewing proxied material from mx1.adamsgaard.dk. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.