Uplift third_party/tt-metal to f8fe02d9f591f4c7f929131630ce5147cfe88f8e 2025-02-01

[vmilosevic]

This PR uplifts the third_party/tt-metal to the f8fe02d9f591f4c7f929131630ce5147cfe88f8e

• Updates in all mlir code for TTNN Shape related refactoring/updates
  • ttnn::SimpleShape renamed to ttnn::Shape in tt-metal 2df6e60cd9
  • Tensor get_shape() -> get_logical_shape() due to tt-metal 8063f0ab02
  • Removed some code in EmitC path to do with selecting between Shape and SimpleShape
    now just using Shape (previously called SimpleShape).

[kmabeeTT]

Pushed updates to match tt-metal renaming of ttnn::SimpleShape to ttnn::Shape in some code, and replacing get_shape() with get_logical_shape() on Tensor objects, can build default build locally now. I know there is a bit more updates missing (EmitC, gut the usage of both shape types, just use ttnn::Shape and also remove some LegacyShape usage) but wanted to get a pulse on CI run as baseline to make sure that code is covered. Edit: Passing after EmitC related updates. I didn't bother removing LegacyShape usage this time, since nothing was failing and CI was already passing here and in 3 FE's, so merging this while it's still hot.

[kmabeeTT]

Brings 27 tt-metal commits. Passing CI for all 3 FE projects:

tt-torch: link
tt-forge-fe: link
tt-xla: link

git l 9d69fb143bac50983dff914c5348539d0a7d2021..f8fe02d9f591f4c7f929131630ce5147cfe88f8e
2025-01-31 f8fe02d9f5 by GitHub (Author afuller@tenstorrent.com) : Extra cleanup.  With emphasis (#17466)
2025-01-31 4078e26b7f by GitHub (Author omilyutin@tenstorrent.com) : #17203: Resolve a deadlock resulting in graph tracker tracking allocation before it was completed  (#17436)
2025-01-31 f30bfcb0c6 by GitHub (Author dmakoviichuk@tenstorrent.com) : Optimized launch_op compilation (#17448) Tested locally only :(
2025-01-31 70206b9cf1 by GitHub (Author 159165450+skhorasganiTT@users.noreply.github.com) : [skip ci] Update vLLM commit for deepseek-distill-llama70b in README.md (#17459)
2025-01-31 2df6e60cd9 by GitHub (Author sminakov@tenstorrent.com) : Rename ttnn::SimpleShape to ttnn::Shape (#17454)
2025-01-31 b6d72ad545 by GitHub (Author afuller@tenstorrent.com) : Update the .rst to match code (#17455)
2025-01-31 8063f0ab02 by GitHub (Author sminakov@tenstorrent.com) : Remove ttnn::Shape and LegacyShape from the codebase (#17413)
2025-01-31 f9b52ff317 by Aditya Saigal (Author asaigal@tenstorrent.com) : #0: Modify DeviceRange semantics to match CoreRange
2025-01-31 c7825a1e62 by GitHub (Author 132708568+kpaigwar@users.noreply.github.com) : #0: fix bug in createqkv (#17452)
2025-01-31 e106618f7c by GitHub (Author williamly@tenstorrent.com) : [skip ci] #0: Add failure signature for runner shutdown (#17439)
2025-01-31 2b29d50289 by GitHub (Author 159165450+skhorasganiTT@users.noreply.github.com) : [Llama3.2-11b-vision] Add max_cross_attn_tokens property to vLLM generator class (#17401)
2025-01-31 12efa1f54f by GitHub (Author moconnor@tenstorrent.com) : [skip ci] Add DeepSeek perf and instructions (#17445)
2025-01-31 ed0d2e14ac by GitHub (Author afuller@tenstorrent.com) : Step 1 of normalizing Boost as a dependency (#17446)
2025-01-31 41d4b368f5 by GitHub (Author 159165450+skhorasganiTT@users.noreply.github.com) : Add Qwen2.5 vLLM generator (based on LlamaGenerator), fix batch 1 issue with generator's decode forward (#17422)
2025-01-31 47fe29f879 by GitHub (Author afuller@tenstorrent.com) : Address clang tidy error (#17437)
2025-01-31 c109516f95 by GitHub (Author 164946524+bbradelTT@users.noreply.github.com) : #16380: add nD support for generic reductions (#17399)
2025-01-31 6c87f4c91a by John Bauman (Author jbauman@tenstorrent.com) : #0: Update pgm_dispatch_golden.json
2025-01-31 d354e26cac by Austin Ho (Author aho@tenstorrent.com) : #0: Change noc_inline_dw_write to use write_at_cmd_buf to avoid overwriting cached register values of write_reg_cmd_buf
2025-01-31 e099072861 by GitHub (Author williamly@tenstorrent.com) : #15414: Read annotation data to determine job-level failure signature and reason (#17410)
2025-01-31 a53f8dd024 by GitHub (Author caixunshiren@gmail.com) : Xuncai/all gather fixes (#17155)
2025-01-31 9760f02266 by GitHub (Author 129077594+johanna-rock-tt@users.noreply.github.com) : Post all gather layernorm/rms norm with resharding (#17377)
2025-01-31 afe387ac62 by GitHub (Author 132708568+kpaigwar@users.noreply.github.com) : #0: Fuse batch slicing with create_qkv_heads (#16416)
2025-01-31 c9ddf8478b by GitHub (Author nardo@tenstorrent.com) : Page size assertion error (#16717)
2025-01-31 783d35aea7 by GitHub (Author nardo@tenstorrent.com) : Delete tilize cpp tests (#17329)
2025-01-31 25ee758217 by GitHub (Author skrstic@tenstorrent.com) : Removing skip_for_blackhole and fix some minor issues for blackhole conv2d (#17222)
2025-01-31 2d624db973 by GitHub (Author skrstic@tenstorrent.com) : Expanded im2col documentation (#16846)
2025-01-30 f708212d6c by Joseph Chu (Author jchu@tenstorrent.com) : #0: Add programming examples for TT-Metalium multi-device native APIs