dca7bb22b6
SWDEV-443796 - Fixed regression in DX11->OpenCL texture sharing case
...
Change-Id: Ia7d6cb08706ce9b15252f5412c9f9c57d53b2e6e
2024-02-05 05:08:11 +00:00
0567c3b720
SWDEV-301667 - Better log
...
Display queue base pointer in the log. This can be co-related with AQL
packets
Change-Id: I544f9b6db6ae01c85e57e4b3f0b3fffefcd7c2ed
2024-02-05 05:08:11 +00:00
ae0368d12d
SWDEV-422207 - Enable DEBUG_CLR_GRAPH_PACKET_CAPTURE environiment variable
...
Change-Id: I9bf72b9c1a56980352109bd4d42b54ecb2d1b8f9
2024-02-05 05:08:11 +00:00
370ff6ea3e
SWDEV-443538 - Fix cmake build warning
...
Use GLVND versin of GL
Change-Id: I1d6b0ec8672a232d017ad27a01803556992e8a49
2024-02-05 05:08:11 +00:00
9a6ddae7b2
SWDEV-301667 - Reset profiler correlation_id_
...
- The correlation_id had random junk values which we were inserting in
the dispatch AQL packet even when no profiler was attached but if we had
a valid timestamp.
- Also make sure we dont even write the reserved2 field in the AQL
packet if no profiler attached.
Change-Id: Icdb7493198c1bb5e2d786a97e027288660854cd7
2024-02-05 05:08:11 +00:00
842eda5e7f
SWDEV-440746 - Remove pending dispatch check for SDMA P2P
...
Change-Id: I7290345cfc60cd878fb39a06b03105441793c27b
2024-02-05 05:08:11 +00:00
dcd2fcff16
SWDEV-404889 - Switch to FreezeExecutable()
...
Debugger support requires FreezeExecutable() call inside the loader interface.
Change-Id: Iff01840a5980500d8d51da8d73c3f24b2960c44a
2024-02-05 05:08:11 +00:00
243b86dd23
SWDEV-368741 - Check for uninstalled hsa-amd-aqlprofile to avoid Segmentation Fault
...
Change-Id: I0c0c7a2ea7c35237332ddd7c2461e4d9ff8e2f1c
2024-02-05 05:08:11 +00:00
b53fcb4d02
SWDEV-440718 - Suffix hip bin name with _6.
...
Link to amd_comgr_2.dll
Change-Id: I5d21af0cc91f2082099273f75ebe1c3279b6365d
2024-02-05 05:08:11 +00:00
5e1de3cc26
SWDEV-433820 - Optimize queue initialization (ROCM)
...
Change-Id: I0619a9f66ae3d64a0f3fc36384f46adb302e725d
2024-02-05 05:08:11 +00:00
b5fadf9cd1
SWDEV-442421 - Fixed case where hipIpcGetMemHandle erroneously returns hipSuccess
...
Change-Id: Ie322a1a0165111d2ca129e68f49b54fe93aa088e
2024-02-05 05:08:11 +00:00
b366a7c992
SWDEV-437832 - Adding device property to check if the device is accelerator.
...
Change-Id: I8349e99c03422c268bbb60a8c143bd492d9cec09
2024-02-05 05:08:11 +00:00
0a055f874b
SWDEV-422207 - Added debug env to dump graph during Instantiation
...
Change-Id: Ibde2ae5b8d240f3986bcd168facc513a319c0f17
2024-02-05 05:08:11 +00:00
cc29df873a
SWDEV-439419 - Unlock host ptr when device ptr equals host ptr
...
hsa_amd_memory_lock_to_pool() and hsa_amd_memory_unlock()
should be called balanced.
Change-Id: I8b1549861bff752aabbb6399d717d1e346079a38
Signed-off-by: Lang Yu <Lang.Yu@amd.com >
2024-02-05 05:08:11 +00:00
5d322c027a
SWDEV-441201,SWDEV-441727 - Move BlitProgram creation after trap handler build
...
Change-Id: I89199ccc4bdf79940dc6693e07c03115c52bf613
2024-02-05 05:08:11 +00:00
28743d8dbd
SWDEV-432599 - Added null checks for rocMemory ptrs in copyMemory
...
Change-Id: I8ccda2c915dad22c4ea67556dfaab9ba47770069
2024-01-18 15:48:30 -05:00
77be355fd9
Revert "SWDEV-359287 - uniform_work_group_size is interger type is 1 for true and 0 for false. Lighting kernel's init sets uniform work group flag and hence not need to set it from options."
...
This reverts commit ea25e65834
2024-01-18 01:29:31 -05:00
ea25e65834
SWDEV-359287 - uniform_work_group_size is interger type is 1 for true and 0 for false. Lighting kernel's init sets uniform work group flag and hence not need to set it from options.
...
Change-Id: I8d3a9a6ed1c3bddaaa86f7cdc242a45e6a1d09f6
2024-01-17 04:42:52 -05:00
0cff14c9e1
SWDEV-441258 - remove full path for HIP LOG windows
...
Change-Id: Ibad6e9542c0cede38f5a114dcd352356ddedf019
2024-01-16 15:26:06 -05:00
13c6f56ca9
SWDEV-440746 - Enable hipExtAnyOrderLaunch extension for PAL
...
Extension allows to execute the kernels without a wait barrier and L1
invalidation.
Change-Id: I96c485204303f54a0240b93134f4560673e4bd17
2024-01-16 15:20:39 -05:00
0b0df605d4
SWDEV-301667 - Remove resetFenceDirty
...
Dont track the status of fence_dirty_ flag on the host, instead clear it
when we submit a barrier on the respective stream.
Change-Id: I4d98dbf20c81379c9c5da9f5b67629a8f9f6dfcd
2024-01-15 15:43:14 -05:00
31101c6219
SWDEV-440746 - Limit WG for compute P2P
...
Use only 16 workgroups for compute P2P copies.
That should be enough to utilize XGMI bandwidth.
Change-Id: I60dfe019279bb95f93c8874244c1738aad1896d8
2024-01-12 14:56:29 -05:00
5f7ebda6aa
Revert "Removed some options"
...
This reverts commit 15b8cf911a
2024-01-10 13:35:42 -05:00
d6b4892600
SWDEV-439242 - Fix perf regression
...
This issue is due to the fact that we set fence_dirty_ flag for every
kernel launch. Whenever any next HIP API is caused, the stream logic
assumes fence is dirty and queues a marker, when its not needed.
Change-Id: I10a49ee951daa92221aa6be75e2334849579a45d
2024-01-10 11:45:02 -05:00
eaa61fc740
SWDEV-301667 - Fix HSAIL compilation path for Blit
...
Recently some unused compiler options for HSAIL path were removed,
but it affected blit kernels compilation. Hence, remove those options.
Also delete assert for device to device copy in SDMA path for now.
Change-Id: Ib5d7f063af2ab4a3fc5d73d426e39c391b1011ac
2024-01-09 17:51:30 -05:00
dec1158d04
SWDEV-438532 - Enable wave limit for HSAIL
...
Luxmark still uses HSAIL path and one subtest can benefit from the wave limit.
Change-Id: I16c94e09cd6e2afd6341cb76bf2e9ab7b7713214
2024-01-09 17:00:50 -05:00
49037b9762
SWDEV-439241 - Keep direct map for MGPU
...
Single OCL context with multiple devices should use direct map
even with persistent memory, because cache coherency layer.
Change-Id: I02a48830440203a1dea176acd4dff4452b70f133
2024-01-08 18:14:15 -05:00
85c15d720d
SWDEV-438958 - Calculate persistent memory stored in the cache
...
- Make sure persistent memory from resource cache is properly adjusted
in free memory calculation.
Change-Id: I74ef68975ccde4694fb1cb904617c418e85dfc9f
2024-01-08 10:11:52 -05:00
f554a075c7
SWDEV-404921 - Add dynamic LDS size hidden kernel argument
...
Change-Id: If99019058e4bfdc988ef5fc4cef001b772380453
2024-01-06 08:46:55 +05:30
3f2f7252aa
SWDEV-365820 - Refactor build path
...
This is an initial change before we refactor the build/link paths for
kernel launches for HIP. This current change is needed as compiler was
setting some dump file which needed fs access which has slowdowns for
NFS mounted file systems
Change-Id: I828f9bb04d789b4f8c05c1ed08767f325efeb47c
2024-01-04 14:20:59 -05:00
755eb2962c
SWDEV-434846 - Limit the gpu single allocation percentage for all MI300 versions
...
Change-Id: I33dea3eaab249ce3f9a624d38267489f99cd530c
2024-01-03 23:47:44 -05:00
dfd4635f91
SWDEV-422207 - Tag captured kernel names for graphs
...
Change-Id: I9540daa4abf9c340541a681037e2dca4eec821ed
2024-01-03 11:50:05 -05:00
7d661bc7df
SWDEV-404889 - Enable debugger interface in PAL
...
Add GPU_DEBUG_ENABLE to control ttpm behavior. If enabled,
then HW will collect more debug info at some perf cost
Change-Id: Icee0686b903a7b1bd483710b9d611877cd43c6aa
2024-01-02 11:51:42 -05:00
dc8f66b86f
SWDEV-379212 - Handle template kernels with hiprtc lowered names
...
Change-Id: Ib8e6493a1f342f92a35031d5ee39b2e22132b56a
2024-01-02 03:28:28 -05:00
15b8cf911a
Removed some options
...
-xnack, -force-wgp-mode, -force-wave-size-32, -round-trip-spirv,
-fe-gen-spirv, -lower-pipe-builtins=0|1, -lower-atomics=0|1,
-set-lds=<value>, -set-scalar-registers=<value>,
-set-vector-registers=<value>, -limit-scalar-registers=<value>,
-limit-vector-registers=<value>, -sc-xnack-iommu,
-faa-for-barrier/-fno-a-for-barrier, -sc-dev-format, -verify-lwspir,
-verify-hwspir, -ffma-enable/-fno-fma-enable,
-fmad-enable/-fno-mad-enable, -fdisable-avx/-fno-disable-avx,
-fforce-llvm/-fno-force-llvm, -print-compile-phases,
-kernel-cache-enforce-miss, -kernel-cache-wipe, -kernel-cache,
-sc[=<filename>]/--load-sc-dll[=<filename>],
-be[=<filename>]/--load-be-dll[=<filename>],
-cg[=<filename>]/--load-cg-dll[=<filename>],
-link[=<filename>]/--load-link-dll[=<filename>],
-opt[=<filename>]/--load-opt-dll[=<filename>],
-fe[=<filename>]/--load-fe-dll[=<filename>],
-cl[=<filename>]/--load-cl-dll[=<filename>], -just-kernel=<kernel-name>,
-use-debugil, -fmulti-level-call/-fno-multi-level-call,
-fdebug-call/-fno-debug-call, -fmacro-call/-fno-macro-call,
-fstack-uav/-fno-stack-uav, -fdef-res-id/-fno-def-res-id,
-wokth=int/--waves-opt-kernel-threshold,
-ilkth=int/--inline-kernel-size-threshold,
-ilsth=int/--inline-size-threshold, -ilcth=int/--inline-cost-threshold,
-scopt=int/--sc-opt-level, -flib-no-inline/-fno-lib-no-inline,
-fuser-no-inline/-fno-user-no-inline,
-scras=int/--sc-si-opt-reg-alloc-strategy, -fsc-post-ra-sched,
-fsc-live-sched/-fno-sc-live-sched, -fsc-use-buffer-for-hsa-global,
-fsc-schedule-no-reorder, -fsc-min-reg-schedule,
-fsc-bias-schedule-to-minimize-insts,
-fsc-bias-schedule-to-minimize-regs, -fsc-disable-merge-memory,
-fsc-disable-loop-unroll, -fsc-use-mubuf/-fno-sc-use-mubuf,
-fsc-selective-inline/-fno-sc-selective-inline,
-fsc-keep-calls/-fno-sc-keep-calls, -slc=0|1/--simplifylibcall,
-stack-alignment=<n>, -fdiv2fmul=0|1, -prt-opt-liveness=0|1,
-liveness=0|1, -SRAE-threshold=<value>, -memcombine-max-vec-gen=<value>,
-small-global-objects, -fast-fmaf, -fast-fma, -bfo=0|1, -ebb=0|1, -aa,
-mem2reg=0|1, -licm=0|1, -unroll-allow-partial,
-unroll-threshold=<positive integer>, -unroll-count=<positive integer>,
-apt/--ap-threshold=<positive integer>, -srt/--sr-threshold=<positive
integer>, -fdebug-linker/-fno-debug-linker, -fbin-gpu64/-fno-bin-gpu64,
-fbin-disasm/-fno-bin-disasm, -fbin-bif30, -fbin-hsail/-fno-bin-hsail,
-fbin-amdil/-fno-bin-amdil, -fbin-spir/-fno-bin-spir, -fonly-bin-source,
-fper-pointer-uav/-fno-per-pointer-uav
Change-Id: I1af5b0a11b55bf75b727057143159fbbf1bb8f0e
2023-12-21 10:18:37 -05:00
748c6b3520
SWDEV-422207 - Fix simple graph test when DEBUG_CLR_GRAPH_PACKET_CAPTURE flag is enabled
...
- For new AccumulateCommand we enqueue nop barrier packet. So during stream sync we need system flush.
Change-Id: I0b97b626bcdae582ef95e7c95030d78df1fa5a54
2023-12-20 22:50:36 -05:00
e05923b139
SWDEV-413997 - Enable Virtual Mem support by default.
...
Change-Id: Ia3db3919701708cf95574692e1d47375ca99d7fd
2023-12-20 12:49:16 -05:00
adf9406a16
SWDEV-1 - Promote PAL verison to 843
...
Change-Id: I2b8d2fd1df61b376598da53bef530243f8d6e757
2023-12-19 10:25:32 -05:00
a6d480e098
SWDEV-436869 - Destroy host memory
...
Recent changes disabled system memory allocation
in the abstraciton layer. That requires memory
allocation/destruction in ROCR. Add destruction logic.
Change-Id: I68fe6b0a620ca743fe5850052ea0efa8bb7931c2
2023-12-15 16:39:41 -05:00
cfc07c88ee
SWDEV-436796 - Enable device memory for kernel arguments
...
Extra CPU read back will be performed before every submission to make sure
previous writes over PCIE reached GPU. HDP flush is done by CP.
Change-Id: I402d28ca26c8cee4a3920feb3599af8c285d0889
2023-12-15 13:11:50 -05:00
c8b3253a24
SWDEV-432575 - Disable direct map for persistent memory
...
Persistent memory should use direct access for write map and
indirect for read map.
Change-Id: I9fc84836d60088b24012ed25f7ef8c16e33796a3
2023-12-15 10:59:01 -05:00
9551d48e7f
SWDEV-411343 - Import VK buffer memory exported by name.
...
Change-Id: Idc7c840fcd6bf758ecc29ffd6558d24e5bd0a25b
2023-12-14 22:16:11 -05:00
ec010e4d2d
SWDEV-413997 - Fixing MGPU cases on PAL side by passing Global Context to virtual alloc.
...
Change-Id: I6614058d1456d199d710b12acd95160a79aa48c8
2023-12-14 11:34:32 -05:00
fb3dfcf889
SWDEV-436859 - Enable pitch for COPY_HOST_PTR
...
Original logic didn't use pitch because, abstraction layer had
a sysmem copy without pitch. Since extra sysmem copy was
disabled, the code has to accept pitch values from the app.
Change-Id: Ia9fba7b33ddff4e9109b4e63d0d6afa52f501c8f
2023-12-13 16:50:16 -05:00
b2102fe939
SWDEV-434846 - Correct the vgprs per simd for MI300
...
Change-Id: Id4862da7611f64392bfc1538fb644801ec0a9e7f
2023-12-13 03:06:21 -05:00
12461dbd6a
SWDEV-431399 - use x86 intrinsics only x86 platforms
...
This was highlighted by github issue:
https://github.com/ROCm-Developer-Tools/clr/issues/32
On RISC-V platforms this header might not be present.
Change-Id: I5f0959a7b281c760802e76541d50693d1a3902be
2023-12-12 19:10:50 -05:00
b3171d08e6
SWDEV-434241 - Replace size_t with uint64_t
...
The kernel accepts uint64_t, but with 32bit OCL build size_t was 32 bit
Change-Id: I6fe37d2e5e69c7bd62d7b1bd4cace758758b3482
2023-12-08 16:56:29 -05:00
3daf8b334a
SWDEV-430911 - Force SDMA only if explicitly specified & pick appropriate engine.
...
Change-Id: Ib34fa6b2782f74b753899fa8fddff646dc60c8ce
2023-12-04 21:58:47 -05:00
01217d0383
SWDEV-435015 - Disable sysmem allocation in abstraction layer
...
Change-Id: Idefdabd2997d618b962c42545ae67768c628c5d6
2023-12-04 11:40:28 -05:00
44761fe89b
SWDEV-434298 - Add destination offset
...
The end pointer in copy buffer requires destination offset
Change-Id: I01f2967144f741761fd2ce3244fd8d04564d986f
2023-11-29 16:33:29 -05:00
Ioannis Assiouras
Saleel Kudchadker
Anusha GodavarthySurya
Alex Xie
German Andryeyev
cadolphe
Ajay
kjayapra-amd
Lang Yu
Jaydeepkumar Patel
pghafari
Konstantin Zhuravlyov
ksankisa
Satyanvesh Dittakavi
Konstantin Zhuravlyov
Jatin Chaudhary