tAdd support for CUDA compute capabilities 4.x, 6.2, and 7.x - sphere - GPU-based 3D discrete element method algorithm with optional fluid coupling
(HTM) git clone git://src.adamsgaard.dk/sphere
(DIR) Log
(DIR) Files
(DIR) Refs
(DIR) LICENSE
---
(DIR) commit 116df3bcf276dec7fca4fc44a8b91c39db229f66
(DIR) parent 167ae1fe9a881f423cdffd2f3737ce6e9de69f78
(HTM) Author: Anders Damsgaard <anders@adamsgaard.dk>
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(-)
---
(DIR) 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-capabilities
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);