gpu-compute: Fix LGKM decrementing for flat atomic insts
authorKyle Roarty <kyleroarty1716@gmail.com>
Fri, 15 Jan 2021 00:06:30 +0000 (18:06 -0600)
committerKyle Roarty <kyleroarty1716@gmail.com>
Fri, 15 Jan 2021 06:00:06 +0000 (06:00 +0000)
commita4657f1b84f5deeb1573b01989eb3f2818c13f43
tree2c5c7b572e23f3047763633fd31f801953edd157
parentfb952a124aac92aab520cb485f1295cbe211d271
gpu-compute: Fix LGKM decrementing for flat atomic insts

A prior commit (f6ec145fc0) fixed early LGKM decrementing for flat loads
and stores, but failed to address flat atomics.

Per the GCN3 ISA, LGKM count is decremented on flat atomics with return
when the data has been returned. This patch checks if the flat
instruction is an atomic with return, and decrements LGKM count if so.

Change-Id: I5c0c2c205a8b21327d4c42ba71c59842c15bd63b
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/39155
Reviewed-by: Matthew Poremba <matthew.poremba@amd.com>
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Matthew Poremba <matthew.poremba@amd.com>
Tested-by: kokoro <noreply+kokoro@google.com>
src/gpu-compute/global_memory_pipeline.cc