Age | Commit message (Expand) | Author | Files | Lines |
2021-03-03 | [mlir] Move cuda tests | Christian Sigg | 13 | -665/+0 |
2021-02-24 | [mlir] Specify cuda-runner pass pipeline as command line options. | Christian Sigg | 12 | -1/+28 |
2021-02-03 | [mlir] Add gpu async integration test. | Christian Sigg | 1 | -0/+66 |
2021-01-28 | [mlir] NFC: split --shared-libs option into multiple lines. | Christian Sigg | 11 | -11/+22 |
2021-01-21 | Revert [mlir] Link mlir_runner_utils statically into cuda/rocm-runtime-wrappe... | Christian Sigg | 11 | -11/+11 |
2021-01-20 | Fix cuda-runner tests. | Christian Sigg | 11 | -11/+11 |
2021-01-20 | [mlir] Link mlir_runner_utils statically into cuda/rocm-runtime-wrappers. | Christian Sigg | 11 | -11/+44 |
2020-11-16 | [MLIR] Extend Symbol verification to reject public symbol declarations. | Rahul Joshi | 11 | -11/+11 |
2020-08-10 | [MLIR] Adding gpu.host_register op and lower it to a runtime call. | Christian Sigg | 11 | -31/+20 |
2020-08-02 | Remove debug flags from test (NFC) | Mehdi Amini | 1 | -1/+1 |
2020-07-27 | [MLIR] NFC: Rename mcuMemHostRegister* to mgpuMemHostRegister* to make it con... | Christian Sigg | 11 | -29/+29 |
2020-06-10 | [MLIR][Standard] Make the `dim` operation index an operand. | Frederik Gossen | 5 | -9/+15 |
2020-05-19 | Unrank mcuMemHostRegister tensor argument. | Christian Sigg | 11 | -71/+58 |
2020-04-16 | [MLIR] Allow for multiple gpu modules during translation. | Stephan Herhut | 1 | -0/+28 |
2020-03-23 | Fix all-reduce int tests by host-registering memrefs. | Christian Sigg | 6 | -12/+54 |
2020-03-19 | [MLIR] Add test for multiple gpu.all_reduce in the same kernel when lowering ... | Valentin Clement | 1 | -0/+64 |
2020-03-11 | [MLIR] Add `and`, `or`, `xor`, `min`, `max` too gpu.all_reduce and the nvvm l... | Valentin Clement | 5 | -0/+292 |
2020-03-11 | Revert "[MLIR] Add `and`, `or`, `xor`, `min`, `max` too gpu.all_reduce and th... | Stephan Herhut | 5 | -292/+0 |
2020-03-10 | [MLIR] Add `and`, `or`, `xor`, `min`, `max` too gpu.all_reduce and the nvvm l... | Stephan Herhut | 5 | -0/+292 |
2020-02-10 | [mlir] use unpacked memref descriptors at function boundaries | Alex Zinenko | 1 | -3/+4 |
2020-02-03 | [MLIR] Make gpu.launch implicitly capture uses of values defined above. | Stephan Herhut | 4 | -12/+8 |
2020-01-30 | Add 'gpu.terminator' operation. | Stephan Herhut | 4 | -4/+4 |
2019-12-20 | Add gpu.shuffle op. | Christian Sigg | 1 | -0/+32 |
2019-12-05 | Add UnrankedMemRef Type | nmostafa | 2 | -4/+6 |
2019-11-26 | Automated rollback of commit d60133f89bb08341718bb3132b19bc891f7d4f4d | Nicolas Vasilache | 3 | -3/+3 |
2019-11-25 | Changing directory shortcut for CPU/GPU runner utils. | Christian Sigg | 3 | -3/+3 |
2019-11-21 | Change CUDA tests to use print_memref. | Christian Sigg | 3 | -18/+18 |
2019-11-19 | Make type and rank explicit in mcuMemHostRegister function. | Christian Sigg | 3 | -9/+6 |
2019-10-16 | Support custom accumulator provided as region to gpu.all_reduce. | Christian Sigg | 2 | -1/+30 |
2019-10-07 | Support reduction of partial warps. | Christian Sigg | 1 | -3/+3 |
2019-10-01 | Change all_reduce lowering to support 2D and 3D blocks. | Christian Sigg | 1 | -12/+18 |
2019-09-26 | Add AllReduceOp to GPU dialect with lowering to NVVM. | Christian Sigg | 1 | -0/+25 |
2019-08-20 | JitRunner: support entry functions returning void | Alex Zinenko | 1 | -1/+1 |
2019-07-04 | Add an mlir-cuda-runner tool. | Stephan Herhut | 2 | -0/+32 |