derek/gem5
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

dev-amdgpu: Implement GPU clock MMIOs

Browse Source 
The ROCr runtime uses a combination of HSA signal timestamps and
hardware MMIOs to calculate profiling times. At the beginning of an
application a timestamp is read from the GPU using MMIOs. The clock
MMIOs reside in the GFX MMIO region, so a new AMDGPUGfx class is added
to handle these MMIOs.

The timestamp value is expected to be in nanoseconds, so we simply use
the gem5 tick converted to ns.

Change-Id: I7d1cba40d5042a7f7a81fd4d132402dc11b71bd4
This commit is contained in:
Matthew Poremba
2023-10-06 13:02:31 -05:00
parent 6a4b2bb096
commit 75a7f30dfb
5 changed files with 157 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
src/dev/amdgpu/SConscript
View File

@@ -39,6 +39,7 @@ SimObject('AMDGPU.py', sim_objects=['AMDGPUDevice', 'AMDGPUInterruptHandler',
tags='x86 isa')
Source('amdgpu_device.cc', tags='x86 isa')
Source('amdgpu_gfx.cc', tags='x86 isa')
Source('amdgpu_nbio.cc', tags='x86 isa')
Source('amdgpu_vm.cc', tags='x86 isa')
Source('interrupt_handler.cc', tags='x86 isa')