Port flat PA from habana_next to habana_main

[dolszewska]

FILL IN THE PR DESCRIPTION HERE

FIX #xxxx (link existing issues this PR will resolve)

BEFORE SUBMITTING, PLEASE READ THE CHECKLIST BELOW AND FILL IN THE DESCRIPTION ABOVE

---

PR Checklist (Click to Expand)

Thank you for your contribution to vLLM! Before submitting the pull request, please ensure the PR meets the following criteria. This helps vLLM maintain the code quality and improve the efficiency of the review process.

PR Title and Classification

Only specific types of PRs will be reviewed. The PR title is prefixed appropriately to indicate the type of change. Please use one of the following:

• [Bugfix] for bug fixes.
• [CI/Build] for build or continuous integration improvements.
• [Doc] for documentation fixes and improvements.
• [Model] for adding a new model or improving an existing model. Model name should appear in the title.
• [Frontend] For changes on the vLLM frontend (e.g., OpenAI API server, LLM class, etc.)
• [Kernel] for changes affecting CUDA kernels or other compute kernels.
• [Core] for changes in the core vLLM logic (e.g., LLMEngine, AsyncLLMEngine, Scheduler, etc.)
• [Hardware][Vendor] for hardware-specific changes. Vendor name should appear in the prefix (e.g., [Hardware][AMD]).
• [Misc] for PRs that do not fit the above categories. Please use this sparingly.

Note: If the PR spans more than one category, please include all relevant prefixes.

Code Quality

The PR need to meet the following code quality standards:

• We adhere to Google Python style guide and Google C++ style guide.
• Pass all linter checks. Please use format.sh to format your code.
• The code need to be well-documented to ensure future contributors can easily understand the code.
• Include sufficient tests to ensure the project to stay correct and robust. This includes both unit tests and integration tests.
• Please add documentation to docs/source/ if the PR modifies the user-facing behaviors of vLLM. It helps vLLM user understand and utilize the new features or changes.

Notes for Large Changes

Please keep the changes as concise as possible. For major architectural changes (>500 LOC excluding kernel/data/config/test), we would expect a GitHub issue (RFC) discussing the technical design and justification. Otherwise, we will tag it with rfc-required and might not go through the PR.

What to Expect for the Reviews

The goal of the vLLM team is to be a transparent reviewing machine. We would like to make the review process transparent and efficient and make sure no contributor feel confused or frustrated. However, the vLLM team is small, so we need to prioritize some PRs over others. Here is what you can expect from the review process:

• After the PR is submitted, the PR will be assigned to a reviewer. Every reviewer will pick up the PRs based on their expertise and availability.
• After the PR is assigned, the reviewer will provide status update every 2-3 days. If the PR is not reviewed within 7 days, please feel free to ping the reviewer or the vLLM team.
• After the review, the reviewer will put an action-required label on the PR if there are changes required. The contributor should address the comments and ping the reviewer to re-review the PR.
• Please respond to all comments within a reasonable time frame. If a comment isn't clear or you disagree with a suggestion, feel free to ask for clarification or discuss the suggestion.

Thank You

Finally, thank you for taking the time to read these guidelines and for your interest in contributing to vLLM. Your contributions make vLLM a great tool for everyone!

[kzawora-intel]

Token budget was introduced in #206 and these functions should take the budget into account (filter out "too expensive" buckets)

[kzawora-intel]

let's not have these hardcoded values as defaults, and use max_model_len instead

[kzawora-intel]

Suggested change

	return list(sorted(buckets, key=lambda b: (b[0] * b[1], b[1], b[0])))
	filtered_buckets = filter(
	lambda bucket: bucket[0] * bucket[1] * block_size <= max_num_batched_tokens,
	buckets)
	return list(sorted(filtered_buckets, key=lambda b: (b[0] * b[1], b[1], b[0])))

and add block_size and max_num_batched_tokens as fn args

[dolszewska]

max_num_batched_tokens was removed for decode buckets

[kzawora-intel]

Suggested change

	return list(sorted(buckets, key=lambda b: (b[0] * b[1], b[1], b[0])))
	filtered_buckets = filter(
	lambda bucket: bucket[0] * bucket[1] <= max_num_batched_tokens,
	buckets)
	return list(
	sorted(filtered_buckets, key=lambda b: (b[0] * b[1], b[1], b[0])))

and add max_num_batched_tokens as fn arg

[kzawora-intel]

caution: prepare_input_tensors is affected by #158; you need to modify the following code not to duplicate first sequence's block tables, as that will give you incorrect block sums:

        seq_group_metadata_list = seq_group_metadata_list.copy()
        seq_group_metadata_list.extend(seq_group_metadata_list[0]
                                       for _ in range(batch_size_padding))

please sync with @iboiko-habana who prepares a fix for that

[dolszewska]

I saw that Iryna's fix is ready for habana_next and I think it can be cherry-pick after flat-PA is merged

[kzawora-intel]

the bucket printout is now extremely confusing, can we mention here that this is (bs, total_blocks) and (bs, seq) in prompt?

[jikunshang]

please note there is fix for this magic number. feel free to cherry pick my commit
#244

[dolszewska]

The fix was tested and it does not help for accuracy in Llama

[madamczykhabana]

LGTM