Add initial support for Llama3-8B on BH #18697
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
4 tasks
mtairum
force-pushed
the
mtairum/llama3-blackhole
branch
from
e3c9847 to
e006a3c
Compare
mtairum
changed the title
mtairum
requested review from
cglagovichTT,
yieldthought,
uaydonat,
abhullar-tt,
pgkeller,
aliuTT,
tt-aho,
tt-dma,
tt-asaigal,
ubcheema,
cfjchu,
omilyutin-tt,
SeanNijjar and
a team
as code owners
cglagovichTT
approved these changes
Mar 6, 2025
| @@ -118,7 +122,11 @@ def count_devices(output): | |||
|
|
|||
| # Count total PCI devices (ignoring N/A) | |||
| total_pci_devices = len( | |||
| [line for line in available_boards.split("\n") if ("Wormhole" or "Grayskull" or "Blackhole") in line] | |||
There was a problem hiding this comment.
While you're here, want to remove Grayskull?
| @@ -81,9 +81,10 @@ def forward(self, x: ttnn.Tensor, mode) -> ttnn.Tensor: | |||
| pc_2 = self.model_config["DECODE_MLP_W2_PRG_CONFIG"] | |||
| pc_3 = self.model_config["DECODE_MLP_W1_W3_PRG_CONFIG"] | |||
| else: # Update the program configs based for prefill | |||
| if seq_len >= 1024: | |||
| prefill_len_cutoff = 512 if self.args.arch_name == "blackhole" else 1024 | |||
| if seq_len >= prefill_len_cutoff: | |||
| # Reshape input to to fit on device and parallelize computation | |||
There was a problem hiding this comment.
what do you think about making prefil_len_cutoff an attribute of model_config so this calculation isn't duplicated between MLP and model_config?
abhullar-tt
reviewed
Mar 6, 2025
| @@ -404,15 +404,6 @@ int main() { | |||
| uint8_t prev_noc_mode = DM_DEDICATED_NOC; | |||
| trigger_sync_register_init(); | |||
|
|
|||
|
|
|||
| #if defined(ARCH_BLACKHOLE) | |||
There was a problem hiding this comment.
The PR with the proper fix to handle this is still missing some cases so I think this revert can be included since we want demo changes in main.
mtairum
force-pushed
the
mtairum/llama3-blackhole
branch
from
d896b1f to
5e2355a
Compare
…hole" This reverts commit 57ba436.
mtairum
force-pushed
the
mtairum/llama3-blackhole
branch
from
eff47f0 to
abf91cc
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Issue: #18135
This PR adds initial support for Llama3-8B on BH.
Changes to
brisc.ccwill be dropped before I merge. That deals with an hang but I think it's going to be fixed in a separate commit.Main things that need review:
CI Tests