| Age | Commit message (Collapse) | Author |
|---|
|
|
|
https://github.rcac.purdue.edu/abdallm/gpgpu-sim_distribution into dev-purdue-integration
|
|
|
|
|
|
|
|
|
|
|
|
|
|
github.rcac.purdue.edu:jain156/gpgpu-sim_distribution into jain156-dev-purdue-integration
|
|
|
|
|
|
Signed-off-by: Mengchi Zhang <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
uniform
|
|
|
|
dev-purdue-integration
|
|
causes ridiculous output file sizes
|
|
|
|
|
|
|
|
https://github.rcac.purdue.edu/abdallm/gpgpu-sim_distribution into dev-purdue-integration
|
|
|
|
|
|
|
|
|
|
|
|
Added shmem_divergence_hist and warp_inst_classification
|
|
Adding the global load and store divergence statistics.
The previous histogram has been modified to look at all memory requests, but that is not meaningful - so it can as well be removed.
|
|
Adding gpuwattch power stats to the output.
Enabled inst and mem breakdown stats in config file.
|
|
Replaced the mem div stats with mem div histogram
|
|
Checking in the data footprint stats as reported in the micro paper.
The implementation uses gpgpu pointer, due to which I don't feel fine pushing to mainline.
May be I can do the stats check and increment in l2cache.cc through a public function
to avoid exposing gpgpu pointer to public
|
|
Added Memory Access breakdown statistics.
Divided INST stats into INST type and INST Mem accesses.
|
|
Added Memory divergence stats
Data shows same divergence characteristics across PTX and PTX+
|
|
|
|
|
|
|
|
interaction
|
|
|
|
|
|
|
|
truncating the address here fixes the issue and we start propoerly hitting in the L2
|
