[gem5-users] Re: GCN3 - SLICC - GPU_VIPER-TCC.sm and GPU_TCP-TCP.sm Correctness

[AMD Public Use] Hi Sampad, I have not tested VIPER atomics, but I think we are relying on the backing store to actually perform the atomic. You are correct that DataBlock::atomicPartial calls the atomic functor through the WriteMask object; however, this is not what VIPER uses. If you look

[gem5-users] Re: GCN3 - SLICC - GPU_VIPER-TCC.sm and GPU_TCP-TCP.sm Correctness

That works for me. I am snooping messages in the Network Interface. As long as the message size is correct, all is well. Thanks a lot. Virus-free. www.avast.com

[gem5-users] Re: GCN3 - SLICC - GPU_VIPER-TCC.sm and GPU_TCP-TCP.sm Correctness

Hi Sampad, I believe this relates to the fact that the coherence protocol is not actually sending out the data, but instead gem5 uses the backing store to functionally read/write data. Essentially, in a real system, yes we would need to send the data, which is why the message size accounts for