Searched hist:11345 (Results 1 - 5 of 5) sorted by relevance

/gem5/tests/quick/se/04.gpu/ref/x86/linux/gpu-ruby-GPU_RfO/
H A Dconfig.ini11369:0c1ae495b5e4 Thu Feb 18 10:42:00 EST 2016 Steve Reinhardt <steve.reinhardt@amd.com> stats: update gpu-ruby-GPU_RfO stats

Output changed way back in this cset:

changeset: 11345:b6a66a90e0a1
user: John Kalamatianos <john.kalamatianos@amd.com>
summary: gpu: fix bugs with MemFence, Flat Instrs and Resource utilization
H A Dsimout11369:0c1ae495b5e4 Thu Feb 18 10:42:00 EST 2016 Steve Reinhardt <steve.reinhardt@amd.com> stats: update gpu-ruby-GPU_RfO stats

Output changed way back in this cset:

changeset: 11345:b6a66a90e0a1
user: John Kalamatianos <john.kalamatianos@amd.com>
summary: gpu: fix bugs with MemFence, Flat Instrs and Resource utilization
H A Dstats.txt11369:0c1ae495b5e4 Thu Feb 18 10:42:00 EST 2016 Steve Reinhardt <steve.reinhardt@amd.com> stats: update gpu-ruby-GPU_RfO stats

Output changed way back in this cset:

changeset: 11345:b6a66a90e0a1
user: John Kalamatianos <john.kalamatianos@amd.com>
summary: gpu: fix bugs with MemFence, Flat Instrs and Resource utilization
/gem5/src/gpu-compute/
H A Dwavefront.cc11345:b6a66a90e0a1 Thu Feb 18 10:42:00 EST 2016 John Kalamatianos <john.kalamatianos@amd.com> gpu: fix bugs with MemFence, Flat Instrs and Resource utilization

Both Memory Fence is now flagged as Global Memory only to avoid resource
oversubscribing.
Flat instructions now check for Shared Memory resource busy to avoid
oversubscribing resources.
All WaitClass resources now use cycles (not ticks) to register the number
of pipe stages between Scoreboard and Execute to be consistent with
instruction scheduling logic which always used clock cycles.
H A Dcompute_unit.cc11345:b6a66a90e0a1 Thu Feb 18 10:42:00 EST 2016 John Kalamatianos <john.kalamatianos@amd.com> gpu: fix bugs with MemFence, Flat Instrs and Resource utilization

Both Memory Fence is now flagged as Global Memory only to avoid resource
oversubscribing.
Flat instructions now check for Shared Memory resource busy to avoid
oversubscribing resources.
All WaitClass resources now use cycles (not ticks) to register the number
of pipe stages between Scoreboard and Execute to be consistent with
instruction scheduling logic which always used clock cycles.

Completed in 13 milliseconds