rocm-5.7.0

ROCm release v5.7.0

rocm-5.6.1

ROCm release v5.6.1

rocm-5.6.0

ROCm release v5.6.0

rocm-5.5.1

ROCm release v5.5.1

rocm-5.5.0

ROCm release v5.5.0

rocm-5.4.4

ROCm release v5.4.4

rocm-5.4.3

ROCm release v5.4.3

rocm-5.4.2

ROCm release v5.4.2

rocm-5.4.1

ROCm release v5.4.1

rocm-5.4.0

ROCm release v5.4.0

788ce62 Use minimum block size of 64 threads (#1427)
4d471bd Add JIT pad (#1411) (#1441)
360b180 Updates for RC1 (#1425)
83784c5 memset fix (#1414)
01d0ecf Fix rank 2 batch norm (#1412)
32f6388 Refactor dynamic padding mode (#1387)
be309bf Rewrite TF batch norm; remove batch_norm_inference (#1371)
4f3cc41 Simplify unit algebraic ops (#1281)
f7d987b Stream sync Changset (#1358)
a9a4740 Fast softmax (#1290)
c9ffb38 Add output_alias and runs_on_offload_target flags for the custom ops (#1309)
e19f78a Use find_2.0 API for the convolution (#1346)
c2842c1 Fix invalid program in debug mode from find_splits (#1390)
70e6396 Add compute_fp32 flag for quant_gemm tests (#1360)
4011819 Add onnx mod operator gpu cpu (#1306)
c00f820 Rewrite ONNX parse batch norm (#1362)
492c4a6 Use larger vector size instead of preloading for broadcasted inputs (#1389)
66bbff1 Upgrade cppcheck to 2.9 (#1400)
94bc41d check concurrency on PR level with one running and one pending performance tests (#1401)
1b575b5 update codecov version (#1402)
8ea8473 Remove unused device functions (#1394)
d9578ba Parameterize epsilon for layernorm kernel (#1367)
9a70050 Multibroadcast find_mul_conv (#1384)
97a1ed2 Improve layernorm and reductions performance (#1348)
34c08db Disabled concurrency, queue added to perf-test.yml (#1386)
10f37f4 Fix typo for add_sigmoid (#1385)
255fb11 Update deprecated Pybind constructor (#1382)
e1e36cd [mlir] Replaced find_library with find_package to locate MLIR static library (#1373)
333860c Reduce problem size of unbatched_gemm tests (#1383)
4b76dd0 Fix split_reshape for slice len of 1 (#1379)
7662d9c Implement concat using jit compilation (#1356)
827baee expose underlying migraphx::argument data pointer in pybind (#1376)
a10a8ef Use rocblas_gemm_ex for batched gemms with broadcasted B (#1354)
d78bcdf Bump version to 2.4 (#1375)
ed2c73a Remove unused headers (#1363)
f266705 Fix TF literal parsing for relu6 (#1370)
60aa0e4 Fix accuracy bug when vectorizing slices (#1364)
d37a4df Enable cppcheck rule for 'not', 'or' keywords (#1361)
794a433 Add pass to rewrite gelu as fast gelu (#1299)
ed7973d Insert contiguous for reshape as necessary (#1351)
349635c Show kernel time when using gpu-driver (#1289)
8752875 Improvements to handling and add constant passed to dot operator (#1280)
af7f22d Fix test suite compile in Ubuntu 22.04 (#1353)
1704bb0 fix bug size_t -> std::size_t (#1350)
fa3c21f Dynamic ref NMS (#1288)
79e15ca Update is_supported (#1334)
b691abd Enable tidy for fpga backend (#1347)
3c133f8 Remove print (#1345)
ac507c6 Fix json strings in driver models (#1341)
8045f7c pybind updates for torch_migraphx library (#1323)
7c8f269 run performance benchmarks on types (#1343)
1784584 Add jit layernorm fusion (#1301)
18e4a2c Improve horizontal fusion of contiguous (#1292)
0e17a72 Fix softmax accuracy issues (#1342)
cb53687 formatting (#1339)
bab9502 Remove prints (#1338)
55cb7d3 Enable switching to bare pointer ABI for MLIR (#1333)
7ecb2de onnxruntime renamed master to main (#1336)