tenstorrent/tt-metal issues and pull requests

#11133 - [Master issue] Fix all wheel installation issues

Issue - State: open - Opened by tt-rkim 3 months ago - 2 comments
Labels: infra-ci, P1_critical, pywheel

#11127 - #6232: Add embedding backwards op for training

Pull Request - State: closed - Opened by TT-BrianLiu 3 months ago - 1 comment

#11126 - Port over all old version GS Resnet demo tests to the new api version

Issue - State: open - Opened by mywoodstock 3 months ago
Labels: ResNet-50, P1_critical, CNNs, op_cat: conv2D, op_cat: maxpool2D

#11124 - Blackhole CNN bringup

Issue - State: open - Opened by abhullar-tt 3 months ago - 4 comments
Labels: P0_Showstopper, CNNs, blackhole

#11111 - [Cloud Support] Wormhole device freezes

Issue - State: closed - Opened by mcw-zwakeelTT 3 months ago - 11 comments

#11107 - #9932: Add support to configure static tlbs for dram and eth cores on BH

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

#11106 - #10703: Make ttnn::Buffer allocation thread/memory safe

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

#11104 - #0: target 32GB VM runners for post-commit

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

#11097 - Add better error handling over the static asserts being used

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

#11096 - Update ttnn to return the semantic version supported

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

#11095 - TTNN Version One

Issue - State: open - Opened by eyonland 3 months ago
Labels: master

#11089 - line_all_gather issue with device async enabled

Issue - State: closed - Opened by kpaigwar 3 months ago - 11 comments
Labels: bug, P0_Showstopper, LLM_bug, llama3, llm_tg

#11086 - Update documentation for Unary bw ops

Issue - State: open - Opened by VirdhatchaniKN 3 months ago
Labels: bug, op_cat: eltwise, backward, MCW

#11081 - #10881: update golden function

Pull Request - State: closed - Opened by bharane-ab 3 months ago

#11072 - MCW: Port Additional OPs to TTNN

Issue - State: open - Opened by eyonland 3 months ago - 1 comment
Labels: master, ttnn-merge

#11070 - [Llama3] Integrating dram sharded matmul for ff1.

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

#11069 - Sharded inputs support in line all-gather for TG llama

Issue - State: open - Opened by mikevin920 3 months ago - 7 comments
Labels: P1_critical, LLM_feature, LLMs on Metal, llama3, llm_tg

#11068 - Flash_Decode V2 not running fast enough on TG llama shapes

Issue - State: open - Opened by mikevin920 3 months ago
Labels: P1_critical, LLM_bug, LLMs on Metal, llama3, llm_tg

#11063 - TG llama3-70b sharded dram running around 120 GB/s

Issue - State: open - Opened by mikevin920 3 months ago - 15 comments
Labels: P1_critical, LLM_bug, LLMs on Metal, llama3, llm_tg

#11061 - Context Leakage - Mistral 7B

Issue - State: closed - Opened by mvanniasingheTT 3 months ago - 6 comments
Labels: bug, P0_Showstopper, mistral, LLM_bug

#11060 - Add Batching to Mistral 7B

Pull Request - State: closed - Opened by mvanniasingheTT 3 months ago - 2 comments
Labels: P0_Showstopper, mistral, LLM_bug

#11048 - Adding support for Width Sharded Convolutions

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

#11047 - #11042: Overload complex fw ops

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

#11046 - #8865: Fix dispatch time profiling infra

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

#11043 - Overload complex forward ops for Div and Mul (Set 2 ops)

Issue - State: open - Opened by VirdhatchaniKN 3 months ago - 2 comments
Labels: op_cat: eltwise, MCW, ttnn-merge

#11042 - Overload complex forward ops for Add and Sub (Set 1 ops)

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

#11041 - Overload complex forward ops

Issue - State: open - Opened by VirdhatchaniKN 3 months ago
Labels: op_cat: eltwise, master, MCW, ttnn-merge

#11025 - Aliu/rm wh arch env

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

#11017 - Slice Op for Multi-Device Tensors

Issue - State: open - Opened by kpaigwar 3 months ago - 1 comment
Labels: P2_should_have, LLM_feature, llm_tg

#11014 - Optimize mcast path reservation/linking for dispatch

Issue - State: closed - Opened by tt-aho 3 months ago
Labels: P1_critical

#11008 - Llama3-70b on TG FF2 unit test hangs

Issue - State: closed - Opened by johanna-rock-tt 3 months ago
Labels: bug, P1_critical, LLM_bug, llama3, llm_tg

#11005 - moreh: compile kernels from source in a string

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

#11003 - Add Llama3.1-8B tests to CI

Pull Request - State: open - Opened by mtairum 3 months ago
Labels: P1_critical, LLM_feature, llama3

#11002 - ResNet needs bcast removed.

Issue - State: closed - Opened by eyonland 3 months ago - 9 comments
Labels: P1_critical, CNNs

#11000 - Issues with uint8 and bfloat16 CBs in Compute Kernel

Issue - State: open - Opened by dongjin-na 3 months ago - 8 comments
Labels: bug, P0_Showstopper, moreh

#10976 - [Blackhole xfunc] Get faster BH machines

Issue - State: open - Opened by abhullar-tt 4 months ago - 3 comments
Labels: P0_Showstopper, blackhole

#10959 - Compilation issues with tt-metal whl (from assets)

Issue - State: closed - Opened by ksriv001 4 months ago - 18 comments
Labels: bug, infra-ci, P1_critical, pywheel

#10946 - ND PCC with TG Llama3-70b on CI

Issue - State: open - Opened by mikevin920 4 months ago - 14 comments
Labels: bug, LLM_bug, LLMs on Metal, llama2-70b, llama3

#10944 - Investigate why Weka mounting hangs often, especially in Nightly FD

Issue - State: open - Opened by tt-rkim 4 months ago - 3 comments
Labels: infra-ci, P2_should_have, ci-bug

#10942 - Test failing for unary_fmod and unary_remainder

Issue - State: open - Opened by mouliraj-mcw 4 months ago
Labels: bug, op_cat: eltwise, op_cat: element_wise, MCW

#10936 - 2D matmul with specific shapes gives bad PCC on a TG

Issue - State: open - Opened by johanna-rock-tt 4 months ago - 35 comments
Labels: bug, P1_critical, op_cat: mm, LLM_bug, LLMs on Metal, prefill, llama3, llm_tg

#10923 - SD Failures on N150 + N300

Issue - State: open - Opened by mywoodstock 4 months ago - 1 comment
Labels: P0_Showstopper, CNNs, CNN_bug, Stable Diffusion, ci-bug

#10896 - #10890: Add rounding mode support for rdiv op

Pull Request - State: closed - Opened by mouliraj-mcw 4 months ago - 4 comments

#10894 - #8865: Optimize bcast_h and bcast_w binary kernel override_runtime_ar…

Pull Request - State: open - Opened by nemanjagrujic 4 months ago

#10890 - rounding_mode support for forward rdiv op

Issue - State: closed - Opened by mouliraj-mcw 4 months ago
Labels: op_cat: eltwise, MCW

#10885 - #10874: Enable test cases for concurrent instances in CCL all gather

Pull Request - State: open - Opened by Aswinmcw 4 months ago - 6 comments

#10877 - #10876: Port llama31_8b to wh_n300

Pull Request - State: closed - Opened by keerthana-r-mcw 4 months ago - 2 comments

#10876 - Port Llama31_8b TTNN implementation to WH N300

Issue - State: closed - Opened by keerthana-r-mcw 4 months ago - 1 comment
Labels: bug

#10869 - Update CODEOWNERS

Issue - State: closed - Opened by eyonland 4 months ago

#10868 - ckernel::wait not working as expected at least on wormhole

Issue - State: open - Opened by amahmudTT 4 months ago
Labels: bug

#10860 - Refactor conv uops

Issue - State: open - Opened by mywoodstock 4 months ago
Labels: cleanup, P1_critical, CNNs, op_cat: conv2D

#10795 - [llama] llama 3 70B hangs after multiple batches of full context decode generation

Issue - State: open - Opened by tstescoTT 4 months ago - 16 comments
Labels: bug, P0_Showstopper, LLM_bug, didt_suspected, llama3

#10775 - #11187: yolov4 in WH

Pull Request - State: closed - Opened by punithsekar 4 months ago - 5 comments

#10754 - Bringup Shallow UNet on N300 via data-parallel

Issue - State: closed - Opened by dvartaniansTT 4 months ago - 2 comments
Labels: bug, P0_Showstopper, CNNs, feature, Customer_Bug, Unet-Shallow

#10732 - #10100: Add support for paged update_cache and fill_cache

Pull Request - State: open - Opened by cglagovichTT 4 months ago - 2 comments

#10718 - [Master issue] Harden data pipeline infrastructure with extra goodies and cleanup like retries, unit testing, gtest junit xml support, and documentation

Issue - State: open - Opened by tt-rkim 4 months ago - 2 comments
Labels: infra-ci, P2_should_have, data-collection

#10713 - update all gather tests to use t3k device

Pull Request - State: open - Opened by SeanNijjar 4 months ago

#10711 - Add handling for Galaxy reset in pytest xdist timeout infrastructure

Issue - State: open - Opened by tt-rkim 4 months ago - 2 comments
Labels: infra-ci, P1_critical, ci-bug

#10705 - #10612: Unit tests for Galaxy cluster

Pull Request - State: open - Opened by sagarwalTT 4 months ago - 2 comments

#10682 - Generate run/measurement CSVs for Mamba single-card demo test

Issue - State: closed - Opened by skhorasganiTT 4 months ago

#10677 - Generate benchmarking data for all model demo tests

Issue - State: open - Opened by skhorasganiTT 4 months ago
Labels: P1_critical, LLM_feature

#10673 - DRAM-sharded matmuls ND bad PCC when running on TG

Issue - State: open - Opened by mikevin920 4 months ago - 54 comments
Labels: bug, P1, op_cat: mm, LLM_bug, didt_suspected, llama3, llm_tg

#10668 - change dispatcher page free mechanism to track only the last block of pages

Issue - State: open - Opened by pgkeller 4 months ago
Labels: P1_critical

#10654 - Add ttnn.addalpha sweeps

Pull Request - State: closed - Opened by amalbasaTT 4 months ago

#10624 - Create stress tests for galaxy

Issue - State: open - Opened by tt-rkim 4 months ago - 7 comments
Labels: infra-ci, P2_should_have

#10617 - llama on TG codeowners change

Issue - State: closed - Opened by cglagovichTT 4 months ago - 1 comment

#10603 - TTNN backward log and neg - incomplete documentation

Issue - State: open - Opened by npetrovic-tenstorrent 4 months ago - 1 comment
Labels: bug, op_cat: eltwise, backward

#10601 - #10600: renamed execute_on_main_thread to operator()

Pull Request - State: closed - Opened by arakhmati 4 months ago

#10571 - Hang Issue in Compute Kernel

Issue - State: open - Opened by dongjin-na 4 months ago - 2 comments
Labels: bug, moreh, P1_critical

#10554 - #10550: Enable remote chip routing before profiler init

Pull Request - State: open - Opened by mo-tenstorrent 4 months ago

#10533 - Update event data to use correct original event for a `workflow_call` instance rather than `GITHUB_EVENT_NAME` for cicd data

Issue - State: open - Opened by tt-rkim 4 months ago
Labels: infra-ci, P1_critical, gh-workflow, data-collection

#10532 - Replace usage of tt_lib with ttnn

Issue - State: closed - Opened by ayerofieiev-tt 4 months ago
Labels: master, MCW, ttnn-merge

#10530 - #10395: Run single card fast dispatch nightly on WH X1 and X2

Pull Request - State: open - Opened by tt-rkim 4 months ago

#10526 - Issue with Packing Only 7 Bits in UINT8 Data Type CB

Issue - State: closed - Opened by dongjin-na 4 months ago - 4 comments
Labels: bug, P0_Showstopper, moreh

#10498 - [Blackhole Bringup] Test Unpack_tilize_A_B

Issue - State: closed - Opened by rtawfik01 4 months ago - 4 comments
Labels: P0_Showstopper, LLK, blackhole

#10485 - Add more explicit notes about isolated Python environments in INSTALLING.md so users aren't confused if they have a polluted system Python environment

Issue - State: open - Opened by tt-rkim 4 months ago - 1 comment
Labels: docs, infra-ci, P2_should_have, pywheel

GitHub / tenstorrent/tt-metal issues and pull requests