| Age | Commit message (Collapse) | Author |
|
The block containing predicate ret instruction should add the
consecutive block to its successor_ids set. next_addr should be assigned
with current instruction address add instruction size instead of 1.
Signed-off-by: Mengchi Zhang <[email protected]>
|
|
support for CUDA 8.0 PTX
|
|
|
|
|
|
|
|
|
|
|
|
shfl instruction implemented
|
|
Initial support of CUDA Dynamic Parallelism on GPGPUSim
|
|
|
|
memory mapping, turn off by default
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
shader, should use hw_cta_id to store shared mem info
|
|
|
|
|
|
|
|
|
|
distributor directly
|
|
|
|
|
|
|
|
|
|
|
|
child kernels in CDP. BUG: .weak .entry and .weak .global directives in ptx file. BUG: empty_protected() for stream manager causes deadlock, change to empty()
|
|
|
|
|
|
|
|
call.uni in reconvergence
|
|
|
|
|
|
Kernel launch to stream not yet implemented
|
|
|
|
cudaGetParameterBufferV2 and call cudaLaunchDeviceV2
|
|
|
|
|
|
|
|
|
|
Restored madp instruction
|
|
|
|
Additional support for CUDA 7.5
|
|
fix a silly typo
|