Displaying 2 results from an estimated 2 matches for "__shfl_sync".
2020 Sep 24
2
cuda __shfl_sync problem
...h
i generate IR similarly and i link it with the device IR
programmatically with Linker::linkModules(..)
Then after some analysis i use llc to get ptx:
llc device.bc --march=nvptx64 --mcpu=sm_52 --filetype=asm -o device.ptx
This works fine but the problem is that the instrumentation code uses
__shfl_sync() and ptxas gives me the following error:
ptxas device.ptx, line 1033; error : Feature 'shfl.sync' requires PTX
ISA .version 6.0 or later
Now according to
https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#warp-shuffle-functions,
__shfl_sync is supported by compute capa...
2020 Sep 25
2
cuda __shfl_sync problem
...programmatically with Linker::linkModules(..)
>>
>> Then after some analysis i use llc to get ptx:
>>
>> llc device.bc --march=nvptx64 --mcpu=sm_52 --filetype=asm -o device.ptx
>>
>> This works fine but the problem is that the instrumentation code uses
>> __shfl_sync() and ptxas gives me the following error:
>>
>> ptxas device.ptx, line 1033; error : Feature 'shfl.sync' requires
>> PTX ISA .version 6.0 or later
>>
>> Now according to
>> https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#warp-shuffle-...