dev-ucm-v1 merge main #452
Closed
Conversation
This file contains hidden or 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
* [Doc] update document link * [Doc] add the README.md of TraceReplay
add code owners
* Improve the quick_start.md * Add a Note
* add store intf with tensor addr ptr * fix interface doxy
v091_patch add commit
…r` with CUDA (#305) add batch interface for device ops and implement ScatterGather with CUDA
hotness management for gc Co-authored-by: lijiachen <[email protected]>
* fix mtp in ucm
…322) * linear buffer for device * check data consistency after embedding
* capacity check * recycle --------- Co-authored-by: lijiachen <[email protected]>
* refactor: reusable transport abstraction & optimized NSFStore pipeline * add memory pool * rewrite dramstore (1st version) * fix * fix * fix * fix * fix * add comment for later development * fix * fix * fix * fix * fix * complete dramstore_connector.py * fix * fix CMakeLists.txt * fix * fix * fix * fix * rewrite dram test:dump (1/2) * format * fix * rewrite dram test:fetch (2/2) * modify dram test script * fix * fix * naive try: add device intf into memory pool * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * update memPool intf * fix * fix * fix * try reconstruct threadPool in dramstore (naive version) * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * fix * remove redundant comments --------- Co-authored-by: Mag1c.H <[email protected]>
* [Feat] Toy proxy now supports PD-mixed round-robin scheduling * [Docs] modify the path of toy_proxy_server.py
#309) [Feat]Add import checking to trace replay and fix the issue of unclosed network resources Co-authored-by: t00939662 <[email protected]>
fix gc Co-authored-by: lijiachen <[email protected]>
Add bandwidth testing script
* delta kvcache block swap * clean code * add core bind method * clean code
reuse act Co-authored-by: lijiachen <[email protected]>
Co-authored-by: hek14 <[email protected]>
adapt to deepseek
* [Feat] Support launch from config file * [Docs] Update documents for launch with yaml * [Fix] Change load only on first rank into configuration * [Feat] Add support for hit ratio in yaml * [Fix] Fix load only first rank in non mla scene
refuse monkey patch
fix gqa bug
fix end == 0 bug
optimize generate_tensor
* adapt GQA & modify config.yaml * move process to UCMDirectConnector * fix comment * modify hash function * fix style * code style and modify hash * init parent_block_hash_value
* Adapt vllm_ascend_0110 and Add configurable options * avoid type conversion in init kvcache
seprate spase patch Co-authored-by: lijiachen19 <[email protected]>
Support tensor parallelism across servers
…#414) * [Feat] Build metrics frame * [Feat]add metrics(ucm_obser.py + metrics_configs.yaml) * [Feat] Implementation of metrics logger on the C++ side for storing and retrieving stats * [Fix] Provide simple grafana and fix bugs * [feat] change the log position of UCM metrics * [fix]modify grafana.json * [Feat] UCM supports metrics display online via Grafana and Promethues * [Fix] Remove configs to examples and add liscense --------- Co-authored-by: flesher0813 <[email protected]> Co-authored-by: hero<[email protected]>
* [fix] fix sparse attention (#397) fix ascend attention Co-authored-by: lijiachen19 <[email protected]> * [opt] Share Infra implementation and unify status codes (#399) share infra module Co-authored-by: Fang Run <[email protected]> * [bugfix] Fix ESA to be compatible with the latest NFSStore. (#401) fix esa to adapt latest NFSStore * release v0.1.0rc4 (#402) Co-authored-by: lijiachen19 <[email protected]> * [opt] Remove unused cc impl of dramstore (#406) remove unused cc impl of dramstore * [Fix]remove dram docs and modify quick-start doc (#411) * [Fix]remove dram docs and modify quick-start doc * modify index.md --------- Co-authored-by: t00939662 <[email protected]> * [Feature] Added performance testing tool based on the PyTest testing framework (#295) Performance testing tool based on the PyTest testing framework. * [Misc] Add cpp-linter.yml (#422) * [docs]add metrics doc (#416) * [docs]add metrics doc * modify metrics.md * modify metrics.md --------- Co-authored-by: t00939662 <[email protected]> * [perf] Modify CUDA SIMD and add Triton hash encoder (#408) * fix cpp code style --------- Co-authored-by: Lijiachen1018 <[email protected]> Co-authored-by: lijiachen19 <[email protected]> Co-authored-by: Mag1c.H <[email protected]> Co-authored-by: Fang Run <[email protected]> Co-authored-by: MaxWang <[email protected]> Co-authored-by: hero0307 <[email protected]> Co-authored-by: t00939662 <[email protected]> Co-authored-by: ML <[email protected]> Co-authored-by: ShiXiaolei <[email protected]>
Co-authored-by: lijiachen19 <[email protected]>
Fix(patch): fix patch for vllm-ascend volcengine/verl#2564 Co-authored-by: lijiachen19 <[email protected]>
cuda trans batch api bug fix (cherry picked from commit 77f5090)
* fix accuracy problem when chunked prefill
set default logger backend to spdlog
* fix num_schedule-tokens=1 * Simplify the code
Fix sparse patch Co-authored-by: lijiachen <[email protected]>
* Add an access bandwidth test script for 'ucm_connector'
adapt vllm0.9.1
#447) Set the multiprocessing start method of the test tool to 'spawn' and add NPU cleanup
* sparse to adapt new connector * Adapt the YAML configuration
renew docs for v1 Co-authored-by: lijiachen19 <[email protected]>
* adapt nfsstore
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
20 participants
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.
Purpose
What this PR does / why we need it?
Modifications
Does this PR introduce any user-facing change?
Test
How was this patch tested?