sync #13
Merged
sync #13
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Issue: ##6434 ### Brief Summary It contains four parts(`driver`, `context`, `device` and `caching_allocator`). The code is similar to `cuda/rhi`. However, there are still some differences between `amdgpu/rhi` and `cuda/rhi` #### context 1. The method of obtaining the hardware version 2. Context::launch #### driver 1. ROCm/hip internal functions #### cmake The current cmake compilation system is sufficient to support the Unit test in #6597 Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
…ace_usages_with() (#7001) Issue: #6933 ### Brief Summary `ImmediateIRModifier` aims at replacing `Stmt::replace_usages_with`, which visits the whole tree for a single replacement. `ImmediateIRModifier` is currently associated with a pass, visits the whole tree once at the beginning of that pass, and performs a single replacement with amortized constant time. It is now used in two most recent passes, `eliminate_immutable_local_vars` and `scalarize`. More passes can be modified to leverage it in the future. After this PR, the profiling result of the script in #6933 shows that the time of `eliminate_immutable_local_vars` reduces from `0.956 s` to `0.162 s`, the time of `scalarize` reduces from `3.510 s` to `0.478 s`, and the total time of `compile` reduces from `8.550 s` to `4.696 s`. Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
…rResourceSet & RasterResources (#6954) Issue: #6832 ### Brief Summary ResourceBinder is not split into two structures, one controls binding of shared-accessible resources, the other is dedicated to control binding of rasterizer states (vertex buffers, etc.) This makes the mapping onto Vulkan DescriptorSets easier, and cleans up the implementation of resource binding all over the place. In addition, these binding states are now no longer attached to the program and can be thus pre-filled / pre-allocated to achieve lower overhead. Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Fixes #6953 Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Device API has been updated.
#7020) …rResourceSet & RasterResources (#6954) This is a re-apply of #6954 Author: Bob Cao <[email protected]>
galeselee
force-pushed
the
amdgpu_codegen
branch
from
f64f5fb to
4e20707
Compare
galeselee
pushed a commit
that referenced
this pull request
Jan 16, 2023
Related Issue: taichi-dev#7140 ### Brief Summary On macOS, when test worker hard fails (abort, EXC_BAD_ACCESS, etc.), backward_cpp's signal handler will re-raise the signal and catch it afterwards, make it an infinite loop, at the moment the offending process can't be terminated easily (except a SIGKILL), eat CPU cycles and blocks test runner. ``` (lldb) bt * thread #1, queue = 'com.apple.main-thread', stop reason = signal SIGABRT * frame #0: 0x00000001a04f0e28 libsystem_kernel.dylib`__pthread_kill + 8 frame #1: 0x00000001a052343c libsystem_pthread.dylib`pthread_kill + 292 frame #2: 0x00000001a0402e10 libsystem_c.dylib`raise + 32 frame #3: 0x00000001283a0848 taichi_python.cpython-38-darwin.so`backward::SignalHandling::sig_handler(int, __siginfo*, void*) + 28 frame #4: 0x00000001a056ec44 libsystem_platform.dylib`_sigtramp + 56 frame #5: 0x00000001a052343c libsystem_pthread.dylib`pthread_kill + 292 frame #6: 0x00000001a0402e10 libsystem_c.dylib`raise + 32 frame #7: 0x00000001283a0848 taichi_python.cpython-38-darwin.so`backward::SignalHandling::sig_handler(int, __siginfo*, void*) + 28 frame #8: 0x00000001a056ec44 libsystem_platform.dylib`_sigtramp + 56 frame #9: 0x00000001a052343c libsystem_pthread.dylib`pthread_kill + 292 frame #10: 0x00000001a0402e10 libsystem_c.dylib`raise + 32 frame #11: 0x00000001283a0848 taichi_python.cpython-38-darwin.so`backward::SignalHandling::sig_handler(int, __siginfo*, void*) + 28 frame #12: 0x00000001a056ec44 libsystem_platform.dylib`_sigtramp + 56 frame #13: 0x00000001a052343c libsystem_pthread.dylib`pthread_kill + 292 frame #14: 0x00000001a0402e10 libsystem_c.dylib`raise + 32 frame #15: 0x00000001283a0848 taichi_python.cpython-38-darwin.so`backward::SignalHandling::sig_handler(int, __siginfo*, void*) + 28 frame #16: 0x00000001a056ec44 libsystem_platform.dylib`_sigtramp + 56 frame #17: 0x00000001a052343c libsystem_pthread.dylib`pthread_kill + 292 frame taichi-dev#18: 0x00000001a0402e10 libsystem_c.dylib`raise + 32 frame taichi-dev#19: 0x00000001283a0848 taichi_python.cpython-38-darwin.so`backward::SignalHandling::sig_handler(int, __siginfo*, void*) + 28 frame taichi-dev#20: 0x00000001a056ec44 libsystem_platform.dylib`_sigtramp + 56 frame taichi-dev#21: 0x00000001a052343c libsystem_pthread.dylib`pthread_kill + 292 frame taichi-dev#22: 0x00000001a0402e10 libsystem_c.dylib`raise + 32 frame taichi-dev#23: 0x00000001283a0848 taichi_python.cpython-38-darwin.so`backward::SignalHandling::sig_handler(int, __siginfo*, void*) + 28 frame taichi-dev#24: 0x00000001a056ec44 libsystem_platform.dylib`_sigtramp + 56 frame taichi-dev#25: 0x00000001a052343c libsystem_pthread.dylib`pthread_kill + 292 frame taichi-dev#26: 0x00000001a0402e10 libsystem_c.dylib`raise + 32 frame taichi-dev#27: 0x00000001283a0848 taichi_python.cpython-38-darwin.so`backward::SignalHandling::sig_handler(int, __siginfo*, void*) + 28 frame taichi-dev#28: 0x00000001a056ec44 libsystem_platform.dylib`_sigtramp + 56 frame taichi-dev#29: 0x00000001a052343c libsystem_pthread.dylib`pthread_kill + 292 frame taichi-dev#30: 0x00000001a0402e10 libsystem_c.dylib`raise + 32 frame taichi-dev#31: 0x00000001a056ec44 libsystem_platform.dylib`_sigtramp + 56 frame taichi-dev#32: 0x00000001a052343c libsystem_pthread.dylib`pthread_kill + 292 frame taichi-dev#33: 0x00000001a046b454 libsystem_c.dylib`abort + 124 frame taichi-dev#34: 0x0000000100194fc0 python`os_abort + 12 frame taichi-dev#35: 0x00000001000758a8 python`cfunction_vectorcall_NOARGS + 324 frame taichi-dev#36: 0x00000001001140f0 python`call_function + 460 frame taichi-dev#37: 0x000000010011086c python`_PyEval_EvalFrameDefault + 27176 frame taichi-dev#38: 0x00000001000287e4 python`function_code_fastcall + 128 frame taichi-dev#39: 0x0000000100028008 python`PyVectorcall_Call + 120 frame taichi-dev#40: 0x0000000100110b20 python`_PyEval_EvalFrameDefault + 27868 frame taichi-dev#41: 0x000000010010982c python`_PyEval_EvalCodeWithName + 3008 frame taichi-dev#42: 0x0000000100028948 python`_PyFunction_Vectorcall + 208 frame taichi-dev#43: 0x0000000100028008 python`PyVectorcall_Call + 120 ```
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Issue: #
Brief Summary