-
Notifications
You must be signed in to change notification settings - Fork 35
Insights: tenstorrent/tt-mlir
Overview
Could not load contribution data
Please try again later
33 Pull requests merged by 18 people
-
Introduce
Conv2dConfigin TTNN->EmitC#3721 merged
Jun 11, 2025 -
Add convenience getChipDesc method for accessing a specific chip id
#3713 merged
Jun 10, 2025 -
TTIR decomposition of ttir.prod op
#3694 merged
Jun 10, 2025 -
Set pykernel dep. to latest tt-mlir wheel release on tt-forge
#3730 merged
Jun 10, 2025 -
Ensure allocation pass failure when out of memory
#3712 merged
Jun 10, 2025 -
Use default github token
#3725 merged
Jun 10, 2025 -
Fixing various ttir-builder PCC failures
#3710 merged
Jun 10, 2025 -
Improved flatbuffer schema verification
#3714 merged
Jun 9, 2025 -
Fixing ttir-builder get_dimension_size golden PCC
#3707 merged
Jun 9, 2025 -
Fixing ttir-builder upsample2d
#3697 merged
Jun 9, 2025 -
Fixing ttir-builder PCC failures caused by dtype mismatch
#3709 merged
Jun 9, 2025 -
Prevent unintended typecast from uint32 to float32
#3675 merged
Jun 9, 2025 -
Add ShardSpec to EmitC
#3636 merged
Jun 9, 2025 -
add constraint APIs for MaximumOp and MinimumOp
#3673 merged
Jun 9, 2025 -
clean up #include order in test/unittests
#3653 merged
Jun 9, 2025 -
[TTNN] Support four-value padding for Conv2d
#3580 merged
Jun 9, 2025 -
ttir-builder and tests updates/clean-up
#3684 merged
Jun 6, 2025 -
D2M: Add sub and ceil ops and enable more golden tests
#3599 merged
Jun 6, 2025 -
Update PyKernel Docs< 10000 /li>
#3674 merged
Jun 6, 2025 -
Support tosa bitwise operations
#3687 merged
Jun 6, 2025 -
Uplift tt-metal to 489eea8b + Switched Pykernel demo to use TT-NN Wheel + Removed TT-NN Py Support
#3692 merged
Jun 5, 2025 -
[D2M] Add Lowering for Rsqrt, Neg, Cos
#3663 merged
Jun 5, 2025 -
Fix clang-19 build issues
#3686 merged
Jun 5, 2025 -
Changed the folding patterns of all_reduce
#3670 merged
Jun 5, 2025 -
Uplift third_party/tt-metal to 41ce500767a364f66034f6924837dabc133e8d4d 2025-06-04
#3669 merged
Jun 4, 2025 -
(nit): Update tt-explorer.md documentation
#3667 merged
Jun 4, 2025 -
Ease PyKernel and TTNN (py) Interface w/ Code Example
#3431 merged
Jun 4, 2025 -
feat: add constraint APIs for subtract op
#3638 merged
Jun 4, 2025 -
Enabled progress bars and significantly sped up workflow execution
#3454 merged
Jun 4, 2025 -
Added Nightly Uplift from Explorer
#3652 merged
Jun 4, 2025 -
update CODEOWNERS (/test/unittests/Allocation -> d2m)
#3659 merged
Jun 4, 2025 -
Update CODEOWNERS
#3656 merged
Jun 4, 2025 -
Improving TTIR dialect operation descriptions
#3529 merged
Jun 4, 2025
19 Pull requests opened by 15 people
-
[OpModel] support slice op
#3671 opened
Jun 4, 2025 -
Add fusing path to match/rewrite scatter op intoupdate/fill_cache
#3676 opened
Jun 4, 2025 -
#3495: Added gather op support for automatic parallelization pass with shardy.
#3677 opened
Jun 4, 2025 -
Uplift third_party/tt-metal to 10fec6086165fc1005e54981d5dc78b91289c077 2025-06-06
#3689 opened
Jun 5, 2025 -
Moved op by op infra to tools folder
#3691 opened
Jun 5, 2025 -
[OpModel] support convtranspose2d op
#3693 opened
Jun 5, 2025 -
Add ComputeKernelConfig support
#3696 opened
Jun 6, 2025 -
Fix consteval in canonical emitc path
#3716 opened
Jun 9, 2025 -
[OpModel] support to_memory_config op
#3722 opened
Jun 9, 2025 -
Builder doc corrections
#3724 opened
Jun 9, 2025 -
Adding an `ttnn::AllGatherOp` workaround
#3726 opened
Jun 10, 2025 -
Enable optimizer by default in tt-mlir
#3728 opened
Jun 10, 2025 -
Removing reshape workaround for dump_device
#3731 opened
Jun 10, 2025 -
Update rewriteOutput as per-result based checking
#3734 opened
Jun 10, 2025 -
Build tt-mlir with cmake 3.24
#3736 opened
Jun 10, 2025 -
Adding api to customize/constrain input tensors to ttir-builder
#3737 opened
Jun 10, 2025 -
Fix `conv*` Tests & General Builder API Cleanup
#3739 opened
Jun 10, 2025 -
Make tensor_cache thread safe, update invalid assert
#3740 opened
Jun 11, 2025 -
[CI Hotfix] Downgrade TTNN Wheel for PyKernel Test
#3741 opened
Jun 11, 2025
39 Issues closed by 15 people
-
Conv2dConfig conversion in TTNNToEmitC
#2852 closed
Jun 11, 2025 -
[TTNN] Decomposition of ttnn.prod op
#1861 closed
Jun 10, 2025 -
Test Lowering For Linear
#1764 closed
Jun 10, 2025 -
[Regression][Conv2d] Output Tensor mismatch
#3641 closed
Jun 10, 2025 -
[Regression2][Conv2d] Output Tensor mismatch
#3688 closed
Jun 10, 2025 -
test_embedding PCC fix
#3682 closed
Jun 10, 2025 -
test_argmax PCC fix
#3680 closed
Jun 10, 2025 -
test_cbrt golden pcc fix
#2658 closed
Jun 10, 2025 -
test_index golden pcc fix
#2604 closed
Jun 10, 2025 -
test_prod PCC fix
#3683 closed
Jun 9, 2025 -
Add constraints API for MaximumOp
#3639 closed
Jun 9, 2025 -
test_get_dimension_size PCC fix
#3681 closed
Jun 9, 2025 -
TTIRBuilder ttir.upsample error
#3678 closed
Jun 9, 2025 -
test_logical_not golden pcc fix
#1801 closed
Jun 9, 2025 -
test_logical_xor golden pcc fix
#1800 closed
Jun 9, 2025 -
test_lt golden pcc fix
#1226 closed
Jun 9, 2025 -
test_le golden pcc fix
#1225 closed
Jun 9, 2025 -
test_gt golden pcc fix
#1224 closed
Jun 9, 2025 -
test_ge golden pcc fix
#1223 closed
Jun 9, 2025 -
test_eq golden pcc fix
#1221 closed
Jun 9, 2025 -
test_typecast golden pcc fix
#2603 closed
Jun 9, 2025 -
test_softmax golden pcc fix
#3065 closed
Jun 9, 2025 -
[EmitC] Add ShardSpec conversion
#3591 closed
Jun 9, 2025 -
Support passing 4 values to padding for Conv2dOp
#2822 closed
Jun 9, 2025 -
d2m ttir.subtract
#3011 closed
Jun 6, 2025 -
d2m ttir.ceil
#3003 closed
Jun 6, 2025 -
Test Lowering For Repeat
#1762 closed
Jun 5, 2025 -
d2m ttir.rsqrt
#3008 closed
Jun 5, 2025 -
d2m ttir.neg
#3006 closed
Jun 5, 2025 -
d2m ttir.cos
#3001 closed
Jun 5, 2025 -
Create a fusing pass in TT-MLIR compiler
#1059 closed
Jun 5, 2025 -
Python binding for Generic OP
#1927 closed
Jun 5, 2025 -
Support Non Float Types In Runtime
#1775 closed
Jun 5, 2025 -
python ttir test generation to ttrt runtime error: DataType mismatch
#2654 closed
Jun 4, 2025 -
Simplify Barrier between TTNN (py) and PyKernel Ops
#3323 closed
Jun 4, 2025 -
Add constraints API for SubtractOp
#3556 closed
Jun 4, 2025 -
[Multi-device] No CCL ops lowered to TTNN for 2D mesh
#3650 closed
Jun 4, 2025 -
[Regression][Conv2d] Hang while running on device
#3644 closed
Jun 4, 2025 -
[Regression][Conv2d] - Runtimerror: Statically allocated circular buffers
#3643 closed
Jun 4, 2025
25 Issues opened by 12 people
-
[D2M] ttmlir-opt infinite loop/cycle when lowering tensors of rank >2 in ttir-to-ttmetal-pipeline
#3738 opened
Jun 10, 2025 -
Add Constraints Api for Sum Op
#3735 opened
Jun 10, 2025 -
[Multi-device] Fix bug in using tt-xla JAX 0.6.0
#3733 opened
Jun 10, 2025 -
`empty` Intermittently Failing Golden Tests
#3732 opened
Jun 10, 2025 -
[EmitC] Run optimized resnet thru ttnn-standalone with 1 and 2 CQ traces
#3729 opened
Jun 10, 2025 -
Fix FileChecks in optimizer/output_layout_override.mlir
#3727 opened
Jun 10, 2025 -
Constraining the input tensors of the golden tests
#3723 opened
Jun 9, 2025 -
Fully implement dst register modeled compute region + blocking in optimize tensor layout
#3720 opened
Jun 9, 2025 -
[Optimizer] Error reporting when trying to run OpModel on machine without device
#3718 opened
Jun 9, 2025 -
[EmitC] Tests that requires OpModel in CI
#3717 opened
Jun 9, 2025 -
[EmitC] Run a consteval test in ttnn-standalone in CI
#3715 opened
Jun 9, 2025 -
D2M: Math fidelity override pass
#3711 opened
Jun 8, 2025 -
ttrt perf to skip const-eval ops
#3706 opened
Jun 7, 2025 -
Fusing pass on by default
#3705 opened
Jun 7, 2025 -
Add documentation for running benchmarks
#3704 opened
Jun 7, 2025 -
Add documentation for turning on optimizer
#3703 opened
Jun 7, 2025 -
Run FFE nightly with optimizer enabled by default
#3702 opened
Jun 7, 2025 -
Memory layout analysis on by default
#3701 opened
Jun 7, 2025 -
Restrict sharding to tested ops only
#3700 opened
Jun 7, 2025 -
Optimizer june release ☔
#3699 opened
Jun 7, 2025 -
Support for metal trace with 2 command queues
#3698 opened
Jun 7, 2025 -
Model trace host fallback in compiler
#3695 opened
Jun 6, 2025 -
TTIRBuilder ttir.reduce_and
#3685 opened
Jun 5, 2025 -
TTIRBuilder ttir.update_cache error
#3679 opened
Jun 4, 2025 -
Add trace support to emitC
#3672 opened
Jun 4, 2025
57 Unresolved conversations
Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.
-
Constant folding compile time (TTIR Reshape, Permute, Transpose, Quantize)
#3604 commented on
Jun 9, 2025 • 33 new comments -
Open Correctly Sized Device Depending on Flatbuffer
#3660 commented on
Jun 10, 2025 • 14 new comments -
Support TOSA negate
#3543 commented on
Jun 10, 2025 • 12 new comments -
#3369: Add support for all to all shardy conversion to stablehlo in shardy automatic parallelization pass.
#3600 commented on
Jun 11, 2025 • 12 new comments -
Shard join nodes
#3642 commented on
Jun 7, 2025 • 11 new comments -
insert-dst-register-access pass
#3613 commented on
Jun 10, 2025 • 8 new comments -
[OpModel] support concat op
#3651 commented on
Jun 10, 2025 • 4 new comments -
Metal Trace POC
#3606 commented on
Jun 11, 2025 • 3 new comments -
Dynamic includes generation based on function ops.
#3626 commented on
Jun 9, 2025 • 2 new comments -
1x1 convolution to matmul optimization first pass
#3412 commented on
Jun 9, 2025 • 1 new comment -
test_log1p golden pcc fix
#1799 commented on
Jun 11, 2025 • 0 new comments -
test_log golden pcc fix
#1798 commented on
Jun 11, 2025 • 0 new comments -
test_tan golden pcc fix
#1797 commented on
Jun 11, 2025 • 0 new comments -
Reorder TTIR Passes: Run TTIRFusing after EIO Pass
#3623 commented on
Jun 10, 2025 • 0 new comments -
Add Galaxy g15cs03 to tt-mlir CI
#2257 commented on
Jun 11, 2025 • 0 new comments -
Change default `TTMETAL_BUILD_TYPE` to `RelWithDebInfo`
#3119 commented on
Jun 4, 2025 • 0 new comments -
Support toLayout lowering of metal_layout change
#3037 commented on
Jun 10, 2025 • 0 new comments -
Add unique locs pass
#3360 commented on
Jun 7, 2025 • 0 new comments -
[Multi-device] Issues with ```lax.pmean()``` (parent issue)
#3645 commented on
Jun 4, 2025 • 0 new comments -
D2M: ttir.sum and ttir.max support
#3469 commented on
Jun 7, 2025 • 0 new comments -
Changing ttrt json handling to improve efficiency
#3535 commented on
Jun 10, 2025 • 0 new comments -
D2M: Remove port, address, pageSize, numBuffers from TTKernel CBType
#3557 commented on
Jun 10, 2025 • 0 new comments -
TTIR To TOSA path for CPU Fallback
#3573 commented on
Jun 4, 2025 • 0 new comments -
Erf and GELU fusion patterns
#3627 commented on
Jun 10, 2025 • 0 new comments -
Build `ttnncpp` lib instead of `ttnn` lib
#3633 commented on
Jun 6, 2025 • 0 new comments -
Use shared builders
#3648 commented on
Jun 9, 2025 • 0 new comments -
Metal Layout Rework
#3654 commented on
Jun 10, 2025 • 0 new comments -
D2M: Multi Core MM Fixes (nsmith/blocking0 base)
#3661 commented on
Jun 5, 2025 • 0 new comments -
[explorer] Feature: add separate run data
#3664 commented on
Jun 5, 2025 • 0 new comments -
`TTIRBuilder` `conv*` ops have weird stride mismatch w.r.t. Attributes
#2683 commented on
Jun 4, 2025 • 0 new comments -
Make `check-ttmlir` Run Golden Tests @ `test/python/golden/`
#3572 commented on
Jun 4, 2025 • 0 new comments -
Non-divisible shard shapes cause tensor spec mismatch error
#3662 commented on
Jun 4, 2025 • 0 new comments -
Support gather operation in automatic shardy pass
#3495 commented on
Jun 4, 2025 • 0 new comments -
HACKATHON: Create a D2M pass for single tile dst fusion
#3665 commented on
Jun 4, 2025 • 0 new comments -
[Bounty $500] Implement specific op includes generation
#772 commented on
Jun 5, 2025 • 0 new comments -
Support TOSA negate
#3502 commented on
Jun 5, 2025 • 0 new comments -
'ttir.scatter' op Dimension size to slice into must be 1
#3567 commented on
Jun 5, 2025 • 0 new comments -
implement a tactical set of static buffer allocation algorithms
#3034 commented on
Jun 5, 2025 • 0 new comments -
Add doxygen/sphinx documentation for ttir-builder
#2445 commented on
Jun 5, 2025 • 0 new comments -
Add doxygen/sphinx documentation for ttrt
#2446 commented on
Jun 5, 2025 • 0 new comments -
Add doxygen/sphinx documentation for tt-explorer
#2447 commented on
Jun 5, 2025 • 0 new comments -
PrepareConv2dWeightsOp should read from dram
#3390 commented on
Jun 7, 2025 • 0 new comments -
Add constraints for to_memory_config op
#2704 commented on
Jun 7, 2025 • 0 new comments -
Add caching for getOpConstraints calls
#2735 commented on
Jun 7, 2025 • 0 new comments -
Performance Benchmarking, TT_DNN -x CompileProgram,HWCommandQueue_write_buffer tracy_profile_log_host.tracy' returned non-zero exit status 1
#3570 commented on
Jun 7, 2025 • 0 new comments -
[Metal Direct Pass] Data Type selection
#1905 commented on
Jun 8, 2025 • 0 new comments -
[EmitC] Add `ttnn-standalone.cpp` to .gitignore and make it autogeneratable in `run` script
#3592 commented on
Jun 9, 2025 • 0 new comments -
ttrt perf adds ops without loc
#3316 commented on
Jun 9, 2025 • 0 new comments -
[EmitC] consteval broken in default emitc path
#3594 commented on
Jun 9, 2025 • 0 new comments -
Enable constraint calls with partial memory config
#3408 commented on
Jun 9, 2025 • 0 new comments -
Add support for sharding upsample pooling op
#3396 commented on
Jun 9, 2025 • 0 new comments -
ttrt perf same file error
#3491 commented on
Jun 9, 2025 • 0 new comments -
Adopt device layout for tensor types + metal_layout refactor
#3389 commented on
Jun 9, 2025 • 0 new comments -
EmitC dylib mangled names
#2478 commented on
Jun 10, 2025 • 0 new comments -
[Ops] Support needed for adaptive_max_pool2d op for non-square input shapes (H ≠ W)
#3630 commented on
Jun 10, 2025 • 0 new comments -
Enable ttrt perf to run with multiple programs
#3585 commented on
Jun 10, 2025 • 0 new comments -
Squeeze/Unsqueeze Don't Generate Perf Reports
#2371 commented on
Jun 10, 2025 • 0 new comments