- The member of the hipMemAllocationProp struct is named requestedHandleTypes. However, in HIP, it is named requestedHandleType. I have decided not to follow CUDA's naming convention, as it would require a lot of changes across HIP repositories (clr, hip, hip-tests).
- Fix hipMemGetAllocationPropertiesFromHandle to update out parameter properly.
Change-Id: I8417a2e73b0ee9fa5ca0d36546e75cfce82cbf3e
The hipGraph will use VMM by default when allocating memory.
However, the handle of Phy mem has been added to Memobj by default.
Since the Memobj will track the whole address range from handle to
handle + size, this needs the system to reserve the whole address
range. If the system range have not reserved by the system, then it
will have the potential issue that clr finds the Memobj incorrectly.
This patch removes the handle from the Memobj to fix this potential
issue.
Change-Id: I2da38e6b2d11d0d48e1afe66c46899500c290624
* Adding tool pc sampling support
Fixing merge issue
tool support on SDKupdates
link amd-comgr
Sanitizer failure fix
fix format
Addressing review comments
misc fix
Adding dispatch id to the CSV output
AddingCHANGELOG
[ROCProfV3][PC Sampling] Initial ROCProfV3 PC sampling tests for JSON and CSV formats (#17)
ROCProfV3 initial tests for JSON and CSV output.
Simple kernels that simplify the verification of samples to instruction decoding
has been introduced.
removing option to enable pc sampling explicitly
Adding documentation
no pc-sampling option in tests anymore
Addressing review comments
Updating docs
an option for choosing whether all units must be sampled
try ignoring PC sampling tests (#36)
* run pc-sampling tests on MI2xx runners
* use v_fmac_f32 instead of s_nop 0 in tests
* fixing docs
[ROCm/rocprofiler-sdk commit: 50b185b9ac]
* Adding tool pc sampling support
Fixing merge issue
tool support on SDKupdates
link amd-comgr
Sanitizer failure fix
fix format
Addressing review comments
misc fix
Adding dispatch id to the CSV output
AddingCHANGELOG
[ROCProfV3][PC Sampling] Initial ROCProfV3 PC sampling tests for JSON and CSV formats (#17)
ROCProfV3 initial tests for JSON and CSV output.
Simple kernels that simplify the verification of samples to instruction decoding
has been introduced.
removing option to enable pc sampling explicitly
Adding documentation
no pc-sampling option in tests anymore
Addressing review comments
Updating docs
an option for choosing whether all units must be sampled
try ignoring PC sampling tests (#36)
* run pc-sampling tests on MI2xx runners
* use v_fmac_f32 instead of s_nop 0 in tests
* fixing docs
Starting from Python version 3.12, a backslash-character pair that
is not a valid escape sequence generates a SyntaxWarning, instead
of DeprecationWarning. Using raw strings fixes the warnings.
Change-Id: I6a9f35e2e2e5d9d09410dfb5964fdc19cf0bfa4a
Starting from Python version 3.12, a backslash-character pair that
is not a valid escape sequence generates a SyntaxWarning, instead
of DeprecationWarning. Using raw strings fixes the warnings.
Change-Id: I6a9f35e2e2e5d9d09410dfb5964fdc19cf0bfa4a
[ROCm/rocprofiler commit: 01f01adedf]
When all 64-bits of the signal value are 0, we can skip polling for that
signal.
We need to keep signals as 64-bit numbers as part of the spec. But most
users of ROCr do will never set the signal value to more than 32-bits.
When the dependent-signals are less than 32-bits, avoid adding extra
SDMA poll packet as this adds latency to the SDMA copies.
Change-Id: I37dca65fe3f060dc7164f49b98cb1985023663c4
* CI: PC Sampling to be included in code coverage
* Update continuous_integration.yml
* Update continuous_integration.yml
* Update continuous_integration.yml
* Update continuous_integration.yml
[ROCm/rocprofiler-sdk commit: eedee0fb96]
* CI: PC Sampling to be included in code coverage
* Update continuous_integration.yml
* Update continuous_integration.yml
* Update continuous_integration.yml
* Update continuous_integration.yml
- Refactor blit code and clean ASAN instrumentation
- Use unified function for rocr copy
- Enable shader copy path for unpinned writeBuffer/readBuffer paths
- Set GPU_FORCE_BLIT_COPY_SIZE=16 which means we will use BLIT copy for
pinned copies or unpinned H2D/D2H copies < 16KB
Change-Id: I42045cca79234b340dbf53dafb93044199736ae4
* Fix use-after-free for ATT code objects
* Formatting
* Temporary fix for last kernel
---------
Co-authored-by: Giovanni Baraldi <gbaraldi@amd.com>
[ROCm/rocprofiler-sdk commit: 21c577ba60]
* Fix for tests. Fix for merge conflict in buffered_output.
* Formatting
---------
Co-authored-by: Giovanni Baraldi <gbaraldi@amd.com>
[ROCm/rocprofiler-sdk commit: 4a80ecbfb8]
* _sync tests that use warp match functions (__match_any_sync and __match_all_sync) were disabled for nvidia because those functions are supported only on devices with compute capability 7.x or higher. In this solution, tests are always build (with specifying appropriate compile flag) and in in the runtime it is decided weather they should be skipped or not.
Change-Id: Ifa9e55f3b47aa3a00027cb986cd6ae46aed45ebd
Bhardwaj, Gopesh
Marko Arandjelovic
Ioannis Assiouras
Branislav Brzak
Aidan Belton-Schure
Marko Arandjelovic
Baraldi, Giovanni
Shane Xiao
Nagaraj, Sriraksha
Jason Tang
David Yat Sin
Elwazir, Ammar
David Galiffi
Saleel Kudchadker
German Andryeyev
Indic, Vladimir
Jatin Chaudhary
Jaydeep Patel
Jimbo Xie
Anusha GodavarthySurya
Chris Freehill
Choudhary, Rahul
Mallya, Ameya Keshava
Vladana Stojiljkovic
Satyanvesh Dittakavi