The scratch_backing_memory_byte_size is not used by CP, but it is
currently used by rocgdb. Putting the field back, but we need to find a
solution for alt_scratch_backing_memory_byte_size.
Also, completely disabling alternate scratch as we need some changes to
support debugger.
The negative queue tests generate an exception which triggers a coredump
generation. Disable RLIMIT so that the coredumps are not generated for
these tests.
Check for RLIMIT_CORE before collecting data for coredump. If the
current limit is 0, then we can return early without spending time
collecting coredump data.
A HSA_IMAGE_ENABLE_3D_SWIZZLE_DEBUG environment flag exists already to
enable/disable this. Default value is false (view3dAs2dArray = 1)
Enabling this flag will enable support for swizzles that do 3D
interleaving on GFX9, GF10 and GFX11. By default support for swizzles that
do 3D interleaving is disabled.
Use the core Driver object in the CPU agent to make it OS/driver
agnostic.
Implement the GetMemoryProperties() and GetCacheProperties methods
for the KFD driver.
Add support for these 2 new queries:
- HSA_AMD_AGENT_INFO_SCRATCH_LIMIT_MAX
Maximum amount of scratch memory allowed on this agent
- HSA_AMD_AGENT_INFO_SCRATCH_LIMIT_CURRENT
Current limit for scratch memory on this agent
Updating ROCr code to match new handshake protocol with CP FW for
asynchronous scratch reclaim.
Increase previous limits when scratch reclaim feature is available.
Environment variable HSA_HIGH_PRECISION_MODE can be used to control MFMA
precision
Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Change-Id: Ib78dd9dd8867025e090a3cca96ab6db4f65dea12
For debian use cases, package conflict is required to remove the
deprecated package during package upgrade Also removed the duplicate
setting of package obseletes in RPM usecase.
- When waiting on non-interrupt signals, do not uSleep. This causes
regressions compared to interrupt signal usage.
- Cleanup code.
Change-Id: I706bda0b13e64ffec0b607c1915d8380a2ce0dea
BUILD_SHARED_LIBS is a global flag so we don't need to set a default
option for it in both libhsakmt and hsa-runtime, only the top level
CMakeLists file. Also updated README to reflect that libhsakmt is
always built statically and gets linked to libhsa-runtime.
Change-Id: I1511f68a268032bec9758bc731d8074f33ec980f
Convert test to use multi-GPU framework.
Add mutex to fix intermixed log issue and annotate logging with
gpu node number.
Signed-off-by: David Belanger <david.belanger@amd.com>
Change-Id: Ic2beeadb1eb4b5a9a0710ac1dbd60b9bf1d84c33
"s_waitcnt 0" (deprecated in gfx12) is redundant here.
s_endpgm will wait for all outstanding instructions
to complete before executing.
Change-Id: Ia8b4dd0fd8dd713e7ba2cba9db85b7b12cee1dd4
Signed-off-by: Lang Yu <lang.yu@amd.com>
Since GFX950 can support page table fragment up to 18 without
performance loss. So set GFX950 default svm.alignment_order to 18.
Change-Id: Ibcdb7f041fb07a38e924c471beec261ea227ca1d
Signed-off-by: James Zhu <James.Zhu@amd.com>
This patch creates the blacklist for gfx950 by copying gfx942 but adding
KFDGWSTest.Semaphore as GWS support is completely removed from gfx950.
Change-Id: I5d7c17e57b8cfd9fae63780ecc9dd55662cfdade
Signed-off-by: Amber Lin <Amber.Lin@amd.com>
David Yat Sin
Jonathan Kim
Su, Daniel
Lao, Darren
David Belanger
Khatri, Shweta
Tony Gutierrez
James Zhu
Cheruvally, Aravindan
Longlong Yao
Adel Johar
Harish Kasiviswanathan
Ranjith Ramakrishnan
Saleel Kudchadker
Luna Nova
Choudhary, Rahul
sonadeem
Sv. Lockal
Lang Yu
Yiannis Papadopoulos
Min Zhou
Amber Lin