tenstorrent/tt-metal issues and pull requests

#11948 - #11691: Move tensor module to ttnn

Pull Request - State: open - Opened by KalaivaniMCW 3 months ago - 1 comment

#11943 - #11646: Replace ttnn.experimental.tensor.* in models/demos

Pull Request - State: open - Opened by mcw-anasuya 3 months ago

#11942 - #11809: migrate profiler into ttnn

Pull Request - State: open - Opened by bharane-ab 3 months ago - 2 comments

#11940 - Add fold to resnet

Pull Request - State: open - Opened by yugaoTT 3 months ago - 1 comment

#11938 - [Feature Request] Refactoring `moreh_bmm`

Issue - State: closed - Opened by dongjin-na 3 months ago
Labels: feature-request, moreh

#11937 - #11936: Make tests to show tranpose problems

Pull Request - State: open - Opened by nemanjagrujic 3 months ago

#11933 - #8868: Fixed conv for Stride>2

Pull Request - State: open - Opened by sankarmanoj-tt 3 months ago

#11930 - Change the limit of ttnn usage

Issue - State: closed - Opened by VirdhatchaniKN 3 months ago - 1 comment
Labels: ttnn, ttnn-merge

#11923 - build tt metal in docker in CI

Pull Request - State: open - Opened by ttmchiou 3 months ago - 2 comments

#11919 - Enable ttnn/tteager on BH CI

Issue - State: open - Opened by ttmchiou 3 months ago

#11917 - `test_resnet_with_module_cache` fails when writing to `/home/ubuntu/.cache/torch/hub`

Issue - State: closed - Opened by TT-billteng 3 months ago - 2 comments
Labels: bug, ci-bug

#11913 - Shard `ttnn.upsample` input in UNet so that we use more than 1 core

Issue - State: open - Opened by esmalTT 3 months ago
Labels: P1_critical, CNNs, CNN_bug, Unet-Shallow

#11912 - UNet Shallow device performance has regressed

Issue - State: open - Opened by esmalTT 3 months ago
Labels: P0_Showstopper, CNNs, CNN_bug, Unet-Shallow

#11910 - [Master issue] Blackhole Infrastructure List

Issue - State: closed - Opened by ttmchiou 3 months ago - 1 comment
Labels: P1_critical, feature, blackhole, machine-management, gh-workflow

#11889 - #8865: Optimize softmax dispatch time

Pull Request - State: closed - Opened by nemanjagrujic 3 months ago

#11885 - Tensor Module

Issue - State: closed - Opened by bharane-ab 3 months ago - 1 comment
Labels: MCW, ttnn-merge

#11884 - #11133: Improve various things about the wheel, including removal of `patchelf` and linking runtime assets to cwd

Pull Request - State: open - Opened by tt-rkim 3 months ago - 1 comment

#11880 - [MCW] Systematic Perf Optimisation/Utilisation analysis of every Conv/MM

Issue - State: open - Opened by punithsekar 3 months ago - 4 comments
Labels: segformer_b0, mcw_cst

#11879 - Add E75 Tests to Post Commit

Issue - State: open - Opened by tt-asaigal 3 months ago
Labels: P1_critical

#11878 - #11877: Make dispatch core order in the core descriptor match for E75 with 1 and 2 CQs

Pull Request - State: closed - Opened by tt-asaigal 3 months ago

#11877 - [Bug Report] failing to initialize a Grayskull e75

Issue - State: closed - Opened by marty1885 3 months ago - 5 comments
Labels: bug, community

#11876 - [Falcon7b] Re-enable demo perf-mode tests on galaxy, update targets, prevent multinomial errors (during perf-mode) using nan-to-num

Pull Request - State: open - Opened by skhorasganiTT 3 months ago

#11875 - [Blackhole Bringup] Add pack_untilize tests

Pull Request - State: open - Opened by rtawfik01 3 months ago

#11874 - #11873: Add shard/pad programming examples and basic programming exam…

Pull Request - State: open - Opened by jerryliangTT 3 months ago - 3 comments

#11873 - Add tensor manipulation programming examples and docs

Issue - State: open - Opened by jerryliangTT 3 months ago

#11872 - FD Optimizations/Cleanup

Pull Request - State: open - Opened by tt-aho 3 months ago

#11871 - [Falcon7b] Fix duplicate loading of rotary embeddings in prefill/decode

Pull Request - State: open - Opened by skhorasganiTT 3 months ago

#11870 - #11132 Run Post-Commit Python Tests against wheels

Pull Request - State: open - Opened by dimitri-tenstorrent 3 months ago - 2 comments

#11869 - #9630: assert out concat when concatenating along padded dimensions

Pull Request - State: open - Opened by sjameelTT 3 months ago

#11868 - [Falcon7b] Re-enable decode perplexity test with seq len 2048

Pull Request - State: open - Opened by skhorasganiTT 3 months ago

#11867 - Clean composite operations

Issue - State: open - Opened by ayerofieiev-tt 3 months ago

#11866 - #11865: Add docker-compose.yml

Pull Request - State: open - Opened by JushBJJ 3 months ago

#11865 - [Feature Request] Add docker-compose.yml

Issue - State: open - Opened by JushBJJ 3 months ago - 2 comments
Labels: feature-request, P2_should_have, community, docker

#11864 - #0: Remove llk_io.cc for WH and BH as well. GS was removed in 7b8e627

Pull Request - State: closed - Opened by tt-aho 3 months ago - 2 comments

#11863 - #0: fix cloud-virtual-machine label

Pull Request - State: closed - Opened by ttmchiou 3 months ago

#11862 - #11564: added test for generating sample data with many different use cases to the visualizer

Pull Request - State: closed - Opened by arakhmati 3 months ago

#11861 - #11858 , #11859: Fix Dockerfile 20.04 and 22.04 sequence and requirements

Pull Request - State: closed - Opened by JushBJJ 3 months ago - 3 comments

#11860 - #10524: Fix override_runtime_args for s2s_rm_concat

Pull Request - State: open - Opened by jaykru-tt 3 months ago

#11859 - [Bug Report] libxml2 and libxslt are not installed by default for 22.04 docker requirements

Issue - State: closed - Opened by JushBJJ 3 months ago - 8 comments
Labels: bug, infra-ci, P1_critical, community, docker, dependencies

#11858 - [Bug Report] Both dockerfiles for 22.04 and 20.04 are incorrectly sequenced

Issue - State: closed - Opened by JushBJJ 3 months ago - 1 comment
Labels: bug, community, docker

#11857 - Optimize Reduce Scatter Dispatch Latency

Issue - State: open - Opened by SeanNijjar 3 months ago
Labels: P1_critical, op_cat: ccl, perf

#11856 - Expose #workers and #buffers per channel attributes at op level (reduce scatter)

Issue - State: closed - Opened by SeanNijjar 3 months ago - 1 comment
Labels: P2_should_have, feature, op_cat: ccl, perf

#11855 - Expose #workers and #buffers per channel attributes at op level (all gather)

Issue - State: open - Opened by SeanNijjar 3 months ago
Labels: P2_should_have, feature, perf

#11854 - [Bug Report] Issues with TT-Metal Wheel Installed as Root, Running as Non-Root

Issue - State: open - Opened by asrinivasanTT 3 months ago - 8 comments
Labels: bug, infra-ci, P2_should_have, pywheel

#11853 - Enable double/triple buffered EDM channels for reduce-scatter

Issue - State: closed - Opened by SeanNijjar 3 months ago - 1 comment
Labels: P1_critical, feature, op_cat: ccl, perf

#11852 - modify keys within device_info

Pull Request - State: closed - Opened by FrancesssZ 3 months ago

#11851 - #11850: Remove Llama3.1-8B output matching to avoid blocking CI

Pull Request - State: closed - Opened by mtairum 3 months ago

#11850 - Llama3.1-8B and Mixtral non-deterministic output in demo_with_prefill.py

Issue - State: closed - Opened by mtairum 3 months ago - 8 comments
Labels: bug, models, P0_Showstopper, LLM_bug, LLMs on Metal, llama3

#11848 - #10855: Add device perf measurements to sweep infra

Pull Request - State: open - Opened by jdesousa-TT 3 months ago

#11847 - [Bug Report] ... Sweep runner failure with T3K device

Issue - State: open - Opened by Aswinmcw 3 months ago - 5 comments
Labels: bug

#11846 - #11845: fix worker ring direction assignment in reduce scatter

Pull Request - State: open - Opened by SeanNijjar 3 months ago

#11843 - Added ttnn.hypot_bw unit test

Pull Request - State: open - Opened by amalbasaTT 3 months ago

#11842 - Fix deterministic output issue on Llama 3.1 on single-card demos on both N150 + N300

Issue - State: closed - Opened by tt-rkim 3 months ago - 1 comment
Labels: P0_Showstopper, ci-bug

#11838 - Move `work_split`

Issue - State: closed - Opened by VirdhatchaniKN 3 months ago - 1 comment
Labels: MCW, ttnn-merge

#11834 - #11838: Move work_split

Pull Request - State: closed - Opened by VirdhatchaniKN 3 months ago - 8 comments

#11827 - #11814: Propagate test script status to CI job status

Pull Request - State: open - Opened by djordje-tt 3 months ago

#11823 - #11644: Migrate AutoFormat to TTNN Experimental

Pull Request - State: open - Opened by mouliraj-mcw 3 months ago

#11822 - #11559: Replace tt_lib in tests/ttnn files

Pull Request - State: open - Opened by VirdhatchaniKN 3 months ago

#11821 - #11646: Replace ttnn.experimental.tensor.* to ttnn.* in models/experimental, tests

Pull Request - State: closed - Opened by mcw-anasuya 3 months ago

#11820 - #11693: Move DeviceModule bindings and replace ttnn.experimental APIs

Pull Request - State: closed - Opened by KalaivaniMCW 3 months ago

#11819 - [MCW] Compare performance of segformer on n150 compared to e150

Issue - State: open - Opened by saichandax 3 months ago - 1 comment
Labels: mcw_cst

#11818 - [MCW] Port from n150 to e150 (with new Conv API)

Issue - State: open - Opened by saichandax 3 months ago
Labels: mcw_cst

#11816 - [Blackhole Bringup] Reshuffle Rows Support

Issue - State: open - Opened by ncvetkovicTT 3 months ago
Labels: LLK, blackhole

#11813 - [MCW] Performance improvements of Data Parallel Shallow Unet on n300

Issue - State: open - Opened by saichandax 3 months ago
Labels: mcw_cst

#11812 - [MCW] Enable Data parallelism for ShallowUnet on n300

Issue - State: closed - Opened by saichandax 3 months ago
Labels: mcw_cst

#11811 - [MCW] Port Shallow Unet with new Conv API on n150

Issue - State: closed - Opened by saichandax 3 months ago
Labels: mcw_cst

#11804 - [MCW] Tensor Parallel (TP) implementation of RMSNorm sub-module of Llama3.1_8b on n300

Issue - State: open - Opened by saichandax 3 months ago
Labels: mcw_cst

#11803 - [MCW] Tensor Parallel (TP) implementation of MLP sub-module of Llama3.1_8b on n300

Issue - State: closed - Opened by saichandax 3 months ago - 1 comment
Labels: mcw_cst

#11802 - [MCW] Tensor Parallel Implemenation of LLama3.1_8b on n300

Issue - State: open - Opened by saichandax 3 months ago - 1 comment
Labels: mcw_cst

#11801 - Move compute kernel config to TTNN

Pull Request - State: open - Opened by Aswinmcw 3 months ago

#11800 - #10524: Fix override_runtime_args for s2s_rm_concat

Pull Request - State: closed - Opened by jaykru-tt 3 months ago - 1 comment

#11798 - #0: Refactor Python dynamic modules creation

Pull Request - State: open - Opened by yan-zaretskiy 3 months ago

#11796 - fd low latency master todo list

Issue - State: open - Opened by pgkeller 3 months ago - 1 comment
Labels: P1_critical

#11793 - #0: Move Unary device operation to tmp

Pull Request - State: open - Opened by ayerofieiev-tt 3 months ago - 1 comment

#11788 - Add support for cur_pos tensor arg in sdpa decode

Pull Request - State: open - Opened by sraizada-tt 3 months ago

#11785 - #0: remove extra arch-wormhole labels for single-card workflows

Pull Request - State: closed - Opened by ttmchiou 3 months ago - 7 comments

#11784 - Move assets both needed during runtime and generated during runtime to a folder that users can actually modify outside of wherever the ttnn wheel is installed

Issue - State: open - Opened by tt-rkim 3 months ago
Labels: infra-ci, P1_critical, pywheel

#11783 - Update DPRINT core options to include workers/dispatch

Issue - State: closed - Opened by tt-dma 3 months ago - 1 comment
Labels: P1_critical, feature

#11782 - Investigate removing dependency on `patchelf` by instead packaging shared libs within wheel using `$ORIGIN`, which should be set at build time for the wheel

Issue - State: open - Opened by tt-rkim 3 months ago - 1 comment
Labels: infra-ci, P1_critical, pywheel

GitHub / tenstorrent/tt-metal issues and pull requests