-
Notifications
You must be signed in to change notification settings - Fork 27.7k
Tests for Vec256 classes #15676
Copy link
Copy link
Closed
Labels
featureA request for a proper, new feature.A request for a proper, new feature.module: cpuCPU specific problem (e.g., perf, algorithm)CPU specific problem (e.g., perf, algorithm)module: testsIssues related to tests (not the torch.testing module)Issues related to tests (not the torch.testing module)module: vectorizationRelated to SIMD vectorization, e.g., Vec256Related to SIMD vectorization, e.g., Vec256triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module
Description
Metadata
Metadata
Assignees
Labels
featureA request for a proper, new feature.A request for a proper, new feature.module: cpuCPU specific problem (e.g., perf, algorithm)CPU specific problem (e.g., perf, algorithm)module: testsIssues related to tests (not the torch.testing module)Issues related to tests (not the torch.testing module)module: vectorizationRelated to SIMD vectorization, e.g., Vec256Related to SIMD vectorization, e.g., Vec256triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module
🚀 Feature
We should have tests in C++ for the Vec256 classes.
Motivation
Currently, the Vec256 code is only tested indirectly through operators that use it. This has led to insufficient test coverage and a number of recent bugs in Vec256 and the operators that use it. We should add sufficient test infrastructure so that we can easily add test cases for new functions on Vec256. It's important that the tests cover the different data types (float, double, int32, int64, etc.) and instruction sets (generic, avx, avx2).
This will require changes to CMake since the current test cases don't enable the AVX/AVX2 instruction sets.
cc @VitalyFedyunin