branch: master
Commits on master
- 6c7df14 enforce UOps.CONST arg has python type based on dtype (#3952) 2 years ago
- 91f3326 hotfix: increase recursion limit 2 years ago
- 68ca4d4 split to schedule.py (#3949) 2 years ago
- da07f31 hotfix: remove bf16 test entirely 2 years ago
- 0d5845f hotfix: jit is flaky on mac 2 years ago
- 150ea2e create engine folder and move code (#3948) 2 years ago
- 629cbc5 only abstractions 2 (#3947) 2 years ago
- 77589bc rename Scalar to ConstType and cast_scalar to as_const (#3946) 2 years ago
- d6d902a wtf (#3944) 2 years ago
- 5530b0c fuzz_linearizer: reduce debug verbosity and make easier for CI usage (#3942) 2 years ago
- 8df6587 hotfix 97.3 for beautiful_mnist (#3941) 2 years ago
- b1e3817 correctly handle Tensor.rand whwn default_float = bf16 (#3940) 2 years ago
- f6ff76b check only upcast int amount in upcasted_axis (#3938) 2 years ago
- e2d6f76 _alloc and _free with options (#3934) 2 years ago
- 739f47e check on cuEventSynchronize (#3933) 2 years ago
- 778d17f intel matmul (#3830) 2 years ago
- ef53767 bf16 support in metal (#3929) 2 years ago
- 72d617a opencl on OSX does not support fp16 extension (#3931) 2 years ago
- cb6e7b5 examples: Fix parameter bandwidth accounting for quantized LLama (#3930) 2 years ago
- 4ecd578 #include <tgmath.h> in ops_clang (#3927) 2 years ago
- 514c432 Fix issues with pointer provenance in load/store through ALU (#3916) 2 years ago
- d651835 verify beautiful_mnist.py eval acc and put into benchmark ci (#3926) 2 years ago
- dc50802 clean up clang src header (#3925) 2 years ago
- 2080325 output_buffer isn't used anymore (#3919) 2 years ago
- f2a9ea4 lru allocator for copyin host buffers (#3918) 2 years ago
- e0e234b hotfix, str compare version for cuda 2 years ago
- 715850a Fix sm89 PTX=1 compilation (#3915) 2 years ago
- 83f39a8 env var to change default float (#3902) 2 years ago
- 03899a7 increase atol on reset train 2 years ago
- d8fafca assign regression (#3907) 2 years ago