Block or Report
Block or report pfultz2
Report abuse
Contact GitHub support about this user’s behavior. Learn more about reporting abuse.
Report abusePopular repositories
1,252 contributions in the last year
Less
More
Contribution activity
March 2022
Created 16 commits in 3 repositories
Created 1 repository
- pfultz2/cppcheck-addon Python
Created a pull request in ROCmSoftwarePlatform/AMDMIGraphX that received 5 comments
Refactor runtime compiled kernels to use the same compile_ops pipeline
This adds the infrastructure so we can compile everything in parallel, whereas before only pointwise kernels were compiled in parallel. This will a…
+506
−432
•
5
comments
Opened 15 other pull requests in 2 repositories
ROCmSoftwarePlatform/AMDMIGraphX
4
open
7
merged
1
closed
- Improve handling of string literals in value class
- Extend lifetimes in C++ API
- Use ifdef instead of comment for the auto-generated method declarations for type erased classes
- Make get_context experimental
- Remove borrowed lifetime from operators that are no longer borrowing their lifetime
- Use ccache for runtime compilation
- Add python API to construct shape class
- Add iterators to kernels tensor_view and fix roialign to work with non-standard shape
- Boost the max number of workgroups for pointwise ops
- Update to clang format 10
- Add initial experimental custom op
- Free space on github jobs
Reviewed 25 pull requests in 5 repositories
ROCmSoftwarePlatform/AMDMIGraphX
16 pull requests
- Improve handling of string literals in value class
- Lp normalization op
- upgrade docker images to ROCm 5.0.2
- scatter operator refactoring to include reduction
- Python binding for shape : Fix constructor for the shape and enable tests
- Complete GPU implementation of CumSum op
- Expose APIs for the MIGraphX program
- Update gitignore to other comparable ROCm projects
- Increase max groups in kernel
- Add initial experimental custom op
- Expose context in C++ API
- Improve print ins
-
Use
add_common_opfor handling types and broadcast in Clip Onnx parsing - Use fp32 compute_type when calling rocBLAS API
- isnan operator
- Mode as enum for pooling and roi_align
RadeonOpenCompute/rocm-cmake
4 pull requests
danmar/cppcheck
3 pull requests
ROCmSoftwarePlatform/MIOpen
1 pull request
ROCmSoftwarePlatform/half
1 pull request
Created an issue in ROCmSoftwarePlatform/AMDMIGraphX that received 1 comment
Add stream syncing
First we need to extend the context and add two methods:
struct context
{ // New methods to add: void wait_for(any_ptr queue); void finish_on(any_p…
1
comment