7707be8384
[Feature][KVCache] Implement Cache Manager V1 with GPU + CPU Cache Support (1/n) ( #7097 )
...
* [Feature][KVCache] Support cache manager v1 architecture
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com >
* Update cache manager and related modules
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com >
* chore: update cache_manager and related modules
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com >
* fix: add node to evictable set in complete_swap_to_device
When a node transitions from SWAP_TO_DEVICE to DEVICE via
complete_swap_to_device, it was not being added to the
_evictable_device set. This caused nodes with ref_count=0 to
become "orphaned" - not appearing in any evictable set despite
having cache_status=DEVICE.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com >
* feat: update cache manager v1 and related modules
- Add new cache_manager.py with cache management functionality
- Add radix_tree.py for prefix caching
- Update block_pool.py and metadata.py
- Update request.py and resource_manager_v1.py for scheduling
- Update gpu_model_runner.py for GPU model execution
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com >
* feat(cache): add cache controller v1 implementation
- Add CacheController class for cache management
- Update config.py with cache related configurations
- Refactor gpu_model_runner.py for improved cache handling
* feat(cache_manager): update cache manager v1
* fix(cache_manager): 修复 swap_cache H2D/D2H 方向的 block_ids 逻辑并清理 ForwardMeta
## Motivation
修复 swap_cache_optimized.cu 中 H2D 方向时 src/dst block_ids 使用错误的问题，
并清理 ForwardMeta 中已废弃的 cache_controller 字段。
## Modifications
- fix: swap_cache_optimized.cu 中根据 D2H 模板参数正确选取 src/dst block_ids，
修复 H2D 方向 src/dst 倒置 bug（同时修复 SwapCachePerLayerImpl 和 SwapCacheAllLayersBatchImpl）
- refactor: cache_manager/v1/__init__.py 将 LayerSwapTimeoutError 导入从
cache_controller 改为 cache_utils（正确来源）
- refactor: ForwardMeta 移除废弃的 cache_controller 字段
- refactor: gpu_model_runner.py 移除对应的 cache_controller 赋值语句
- test: 新增 tests/cache_manager/v1/test_swap_cache_ops.py 单元测试
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* feat(cache_manager): refactor cache manager v1 and optimize swap ops
## Motivation
对 cache manager v1 进行重构和优化，精简代码结构，提升可维护性。
## Modifications
- 重构 transfer_manager.py，大幅精简代码逻辑
- 优化 swap_cache_optimized.cu GPU 算子实现
- 调整 cache_manager.py、cache_controller.py 逻辑，修复 free_device_blocks 方法缺失问题
- 更新 block_pool.py、cache_utils.py、metadata.py、radix_tree.py
- 精简 gpu_model_runner.py、forward_meta.py、attention.py 中相关调用
- 更新对应单元测试（test_cache_controller、test_swap_cache_ops、test_transfer_manager）
- 调整 config.py 中相关配置项
* [KVCache][MTP] 支持 cache_manager_v1 下的 MTP KV Cache 初始化及多模态 hash
## Motivation
在 enable_cache_manager_v1 路径下，MTP（speculative decode）的 KV Cache 需要由
CacheController 统一管理，以复用 swap/transfer 能力，同时修复多模态场景下 block
hash 未携带 multimodal extra_keys 的问题。
## Modifications
- `cache_controller.py`
- 新增 `initialize_mtp_kv_cache`：通过 CacheController 初始化 MTP KV Cache，
并将其注册到 cache_kvs_map，使 transfer_manager 自动覆盖 MTP 层
- `initialize_host_cache` 中的 num_layers 改为包含 MTP 额外 cache 层数，保证
Host Cache 也为 MTP 分配足够空间
- `_free_gpu_cache` 改名为 `free_gpu_cache`（对外可调用）
- `cache_utils.py`
- 新增 `get_block_hash_extra_keys`：提取单个 block 内的多模态 hash 信息，
对齐 PrefixCacheManager 的 multimodal extra_keys 逻辑
- `get_request_block_hasher` 中在 hash_block_tokens 时携带 extra_keys，
修复多模态场景 prefix cache 命中率不准的问题
- `spec_decode/mtp.py`
- `update_mtp_block_num` 新增 `skip_cache_init` 参数，避免 v1 cache manager
路径下重复初始化 MTP KV Cache
- `gpu_model_runner.py`
- `initialize_kv_cache(v1)` 路径：在主模型 cache 初始化后，调用
`cache_controller.initialize_mtp_kv_cache` 完成 MTP cache 创建
- `clear_cache` / `wakeup` / `reset` 等路径：respect `enable_cache_manager_v1`
标志，跳过重复的 proposer.initialize_kv_cache 调用
## Usage or Command
```bash
# 启动支持 MTP + cache_manager_v1 的推理服务（示例）
bash run.sh
```
* fix(cache_manager): multi-GPU fix, mm hash boundary fix, and remove batch ops
1. Fix CuPy stream/event creation for multi-GPU: wrap all stream operations
with cp.cuda.Device(device_id) context to ensure streams/events are bound
to the correct device, preventing cross-device errors in multi-GPU setups.
2. Remove cudaSetDevice from SwapCacheAllLayers (handled by cupy context now).
3. Remove swap_cache_all_layers_batch op: simplified the implementation by
removing the batch upload variant; all-layer transfers now use the standard
swap_cache_all_layers with cupy device context.
4. Fix mm hash boundary comparison in get_block_hash_extra_keys: change
strict less-than (<) to less-than-or-equal (<=) so that multimodal items
ending exactly at block start are correctly excluded.
5. Extract config fields to KVCacheBase: model_config, cache_config,
quant_config, parallel_config are now set in the base class __init__ to
avoid duplication in CacheController and CacheManager subclasses.
6. Translate metadata.py docstrings from Chinese to English for broader
contributor accessibility.
7. Add test_cache_utils.py: comprehensive unit tests for
get_block_hash_extra_keys covering all boundary and overlap scenarios.
8. Expand test suite: test_request.py cache fields tests, test_radix_tree.py
backup candidate tests, test_transfer_manager.py and test_cache_manager.py
multi-GPU and concurrent operation tests.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [BugFix][KVCache] fix List import and move write_policy normalization to CacheManager
## Motivation
修复两处问题：
1. `fastdeploy/engine/request.py` 中 `List` 未导入导致 pre-commit F821 报错
2. `write_policy` 归一化逻辑（`write_through` → `write_through_selective`）不应放在 `FDConfig`，移至 `CacheManager.__init__` 中，使其只影响 Cache Manager V1 的内部逻辑
## Modifications
- `fastdeploy/engine/request.py`: 在 `typing` 导入中补充 `List`，删除重复的 `CacheSwapMetadata` TYPE_CHECKING 导入，修复 F821/F811
- `fastdeploy/config.py`: 删除 `write_policy` 归一化逻辑
- `fastdeploy/cache_manager/v1/cache_manager.py`: 将归一化逻辑移入 `CacheManager.__init__`
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [BugFix][KVCache] fix pre-commit code style issues
## Motivation
修复 CI pre-commit 代码风格检查失败问题。
## Modifications
- `fastdeploy/engine/common_engine.py`: black 格式化
- `fastdeploy/worker/worker_process.py`: black 格式化 + isort 修复
- `fastdeploy/cache_manager/v1/storage/__init__.py`: isort 修复
- `fastdeploy/worker/gpu_worker.py`: isort 修复
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [Feature][KVCache] update cache_manager_v1 modules
## Motivation
更新 Cache Manager V1 相关模块，完善版权信息、改进模块结构与可维护性。
## Modifications
- `fastdeploy/cache_manager/v1/` 系列模块：补充版权 header，优化代码结构
- `fastdeploy/config.py`：配置项更新
- `fastdeploy/engine/sched/resource_manager_v1.py`：调度相关更新
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [Feature][KVCache] add BatchRequest.from_tasks and refactor worker task parsing
## Motivation
将 worker_process 中重复的 task 解析逻辑收敛到 BatchRequest，减少代码冗余，提升可维护性。
## Modifications
- `fastdeploy/engine/request.py`：新增 `BatchRequest.from_tasks()` 类方法，统一将 task_queue 任务分类为推理请求和控制请求
- `fastdeploy/worker/worker_process.py`：使用 `BatchRequest.from_tasks()` 替代内联解析逻辑，并修复重复的 control_reqs 处理块
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [Feature][KVCache] add NUMA affinity for host cache and skip swap cache tests
## Motivation
优化 Host cache 内存分配的 NUMA 亲和性，减少跨 NUMA 访问延迟；
同时跳过 swap cache ops 测试（当前环境不支持）。
## Modifications
- `fastdeploy/cache_manager/v1/cache_controller.py`：
- 新增 `_get_numa_node_for_gpu()` 方法，通过 nvidia-smi 或 sysfs 获取 GPU 对应的 NUMA 节点
- 新增 `_bind_to_closest_numa_node()` 方法，绑定当前线程到 GPU 最近的 NUMA 节点
- 在 `initialize_host_cache()` 中调用 NUMA 绑定，优化 H2D 传输性能
- `tests/cache_manager/v1/test_swap_cache_ops.py`：跳过所有测试类（`TestSwapCacheAllLayersCorrectness`、`TestSwapCacheAllLayersPerformance`、`TestSwapCacheRandomBlockIndices`）
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [BugFix][KVCache] fix unittest failures for cache_manager_v1
三个单测因接口变更或 Mock 方式问题导致失败，需修复。
- tests/distributed/chunked_moe.py：`setup_model_runner` 使用 `__new__` 跳过 `__init__`，补加 `enable_cache_manager_v1 = False`，修复 `AttributeError`
- tests/engine/test_resource_manager.py：`PrefixCacheManager` 为局部导入，`patch` 路径改为定义位置 `fastdeploy.cache_manager.prefix_cache_manager.PrefixCacheManager`
- tests/v1/test_resource_manager_v1.py：`_trigger_preempt` 第四参数已由 `list` 改为 `BatchRequest`，更新测试传参和断言
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [BugFix][KVCache] remove debug logging code
## Modifications
- fastdeploy/engine/request.py：删除调试用 logger 及 prompt_hashes 中的 debug 日志
- fastdeploy/worker/worker_process.py：删除 __main__ 中的调试 import 和 print 语句
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [BugFix][KVCache] fix cupy device id caching and pickle for _match_result
## Motivation
修复两个 bug：
1. `transfer_manager.py` 中每次调用 `cp.cuda.runtime.getDevice()` 存在隐患，应在初始化时缓存为实例变量，保证后续操作使用一致的设备 ID。
2. `request.py` 的 `__getstate__` 未跳过 `_match_result`，该字段包含 BlockNode 树的父子循环引用，pickle 时会触发 `RecursionError`；同时补充 `__setstate__` 确保 unpickle 后字段恢复为安全默认值。
## Modifications
- `transfer_manager.py`：初始化时调用 `cp.cuda.runtime.getDevice()` 并缓存到 `self._cupy_device_id`，后续 `with cp.cuda.Device(...)` 和日志均使用该缓存值。
- `request.py`：
- `__getstate__` 中将 `_match_result` 加入跳过集合 `_SKIP_KEYS`，避免循环引用导致 pickle 失败。
- 新增 `__setstate__`，unpickle 后将 `_block_hasher` 和 `_match_result` 恢复为 `None`。
## Usage or Command
* fix(test): fix unit test errors for _trigger_preempt and wakeup with MTP
- Use BatchRequest instead of list in test_trigger_preempt_records_tasks
- Add missing enable_cache_manager_v1 attr in TestSleepWakeupBehavior._make_runner
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [BugFix][KVCache] fix gpu_free_block_list returning wrong block IDs
## Motivation
`gpu_free_block_list` 的兼容 property 中误用了 `list(range(N))`，
将 `available_blocks()` 的返回值当作整数传给 `range()`，
导致返回 `[0, 1, ..., N-1]` 的假列表，而非真实的空闲 block ID。
## Modifications
- `cache_manager/v1/cache_manager.py`：将 `list(range(self._device_pool.available_blocks()))` 改为 `list(self._device_pool.available_blocks())`
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [BugFix][KVCache] 修复 gpu_free_block_list 返回 int 导致 TypeError
## Motivation
gpu_free_block_list 属性中调用 BlockPool.available_blocks()，
该方法返回 int（空闲块数量），用 list() 包装 int 会触发
TypeError: 'int' object is not iterable。
## Modifications
将 list(self._device_pool.available_blocks()) 改为
list(self._device_pool._free_blocks)，直接返回空闲块索引列表。
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [KVCache][CacheManager] 适配 V1 CacheManager 的 pause/sleep/free_cache 操作
## Motivation
V1 CacheManager 引入了新的 reset_cache() 接口，pause 和 sleep 操作需要适配，
同时 free_cache 需要支持可选的 clear_storage 参数。
## Modifications
- cache_controller.py: free_cache 新增 clear_storage 参数（默认 False），
仅当 clear_storage=True 时才调用 _clear_storage()，避免不必要的 storage 清空
- common_engine.py: pause 和 sleep 操作中，当 ENABLE_V1_KVCACHE_MANAGER 时
使用 cache_manager.reset_cache() 替代旧的 reset() 和 pause_transfer 逻辑
- gpu_model_runner.py: sleep 时仅在非 V1 cache manager 下执行 MTP cache 清除
## Usage or Command
# 启动服务（V1 CacheManager）
python -m fastdeploy.entrypoints.openai.api_server \
--enable-v1-kvcache-manager \
...
* [BugFix][KVCache] fix missing enable_cache_manager_v1 in test mocks and remove unused select_blocks_for_backup
- Remove unused `select_blocks_for_backup` method from radix_tree.py
- Fix `match_prefix` default param `skip_storage=True` and log order in cache_manager.py
- Sync test_gpu_model_runner.py with upstream/develop (add TestInsertTasksV1SplitwiseSuffix)
- Add `enable_cache_manager_v1=False` to all mock runners to fix AttributeError in CI
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [BugFix][KVCache] simplify _free_blocks in ResourceManagerV1 for non-v1 path
Remove redundant prefix_caching branch in else path; always call
recycle_gpu_blocks with full block_tables for non-cache-manager-v1 case.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com >
* [KVCache][Optimization][BugFix] fix and optimize block_pool, cache_manager, transfer_manager, request
## Motivation
修复 cache_manager v1 中若干代码质量问题，提升性能并消除潜在的类型不一致 Bug。
## Modifications
1. **block_pool.py**：`BlockPool.allocate` 将逐个 pop 循环替换为切片 + 批量 set.update，消除 Python 循环开销，O(n) → O(k)（C 层批量操作）
2. **cache_manager.py**：`match_prefix` 在 prefix caching 关闭时提前 return 前写入空 `MatchResult()`，避免调用方解引用 `_match_result=None` 崩溃
3. **transfer_manager.py**：`_build_device_layer_indices` 在 `_cache_kvs_map` 为空时也重置四个层索引列表，防止残留旧 tensor 被 swap 算子使用
4. **request.py**：`BatchRequest.append_swap_metadata` / `append_evict_metadata` 构造 `CacheSwapMetadata` 时将 `src_type`/`dst_type` 从字符串改为 `CacheLevel` 枚举，与字段类型声明一致；补充 `CacheLevel` 导入；`match_result` 属性返回类型标注修正为 `Optional[MatchResult]`
5. **resource_manager_v1.py**：`_allocate_gpu_blocks` 日志从 `INFO` 降级为 `DEBUG`，消除高频调度路径的日志噪音
6. **tests/engine/test_request.py**：同步更新 `src_type`/`dst_type` 断言为 `CacheLevel` 枚举值，补充 `CacheLevel` 导入
## Usage or Command
单元测试：
```bash
source .venv/py310/bin/activate
cd baidu/FastDeploy
python -m pytest tests/cache_manager/v1/test_cache_manager.py -v
python -m pytest tests/cache_manager/v1/test_transfer_manager.py -v
python -m pytest tests/engine/test_request.py -v
```
* [BugFix][KVCache] Fix BlockPool.allocate returns all blocks when num_blocks=0
## Motivation
当 `allocate(num_blocks=0)` 被调用时，Python 负索引陷阱导致严重错误：
`-0 == 0`，所以 `self._free_blocks[-0:]` 等价于 `self._free_blocks[0:]`，
会返回并清空整个空闲块列表，而非返回空列表。
## Modifications
在 `BlockPool.allocate` 中增加对 `num_blocks == 0` 的提前判断，直接返回 `[]`，
避免触发 Python 负索引陷阱。
## Usage or Command
```bash
# 运行相关单元测试验证修复
python -m pytest tests/cache_manager/v1/test_cache_manager.py -vv -s
```
* [KVCache][Test] add unit tests for cache_manager v1 modules
## Motivation
补全 cache_manager/v1 各模块的单测覆盖，确保核心方法有完整的测试保障。
## Modifications
新增/补充以下测试文件，全部 326 个用例通过：
- tests/cache_manager/v1/test_block_pool.py（新建）
覆盖 BlockPool.get_metadata/set_metadata/resize、DeviceBlockPool/HostBlockPool
- tests/cache_manager/v1/test_metadata.py（新建）
覆盖 BlockNode、RadixTreeStats、MatchResult、CacheSwapMetadata、AsyncTaskHandler
- tests/cache_manager/v1/test_cache_utils.py（补充）
新增 hash_block_tokens、get_request_block_hasher、LayerDoneCounter 时间追踪及内部辅助方法
- tests/cache_manager/v1/test_radix_tree.py（补充）
新增 TestCompleteSwapToDevice 专项测试类（6 个用例）
- tests/cache_manager/v1/test_cache_manager.py（补充）
新增 offload_to_host、load_from_host、pending backup 系列、prepare_prefetch_metadata
- tests/cache_manager/v1/test_transfer_manager.py（补充）
新增 _swap_single_layer 校验路径、sync_input/output_stream、record_input_stream_event
## Usage or Command
```bash
# 运行所有新增单测
source .venv/py310/bin/activate
python -m pytest tests/cache_manager/v1/test_block_pool.py \
tests/cache_manager/v1/test_metadata.py \
tests/cache_manager/v1/test_cache_utils.py \
tests/cache_manager/v1/test_radix_tree.py \
tests/cache_manager/v1/test_cache_manager.py \
tests/cache_manager/v1/test_transfer_manager.py -v
# 期望结果：326 passed
```
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com >
Co-authored-by: Jiang-Jia-Jun <163579578+Jiang-Jia-Jun@users.noreply.github.com >
2026-04-21 14:39:00 +08:00
2d8338f9e4
[Optimization][DeepSeekV3.2]Reducing slot_mapping compute frequency from twice per layer to a single pre-processing step. ( #7367 )
2026-04-16 19:54:12 +08:00
29495b2cf1
[XPU] Unify Spec and non-spec branch.( #6947 ) ( #7180 )
...
* [XPU] cherry-pick PR-6947
* [XPU] use unified_update_model_status.
* refactor xpu_model_runner.
* refactor sampler.
* fix codestyle.
* Fix XPU speculative decoding: rename output tensors to cu_seqlens_q_output/batch_id_per_token_output, correct
WRAPPER_CHECK_PTR types, and fix dynamic gather shape in verify_draft_tokens path.
* fix codestyle.
* replace output_padding_offset with is_speculative flag in gather_next_token.
* rename hiddden_states.
* unify cu_seqlens_q_output and batch_id_per_token_output init.
---------
Co-authored-by: cmcamdy <1027740945@qq.com >
2026-04-16 14:58:38 +08:00
de0c5e68fb
[XPU] Split the block_attn operator into smaller operators ( #6798 )
...
* spliced block_attn
* adapt to latest vllm
* fix unit tests
* delete mtp+cudagraph 4 cards test
* fix vl model
* fix mtp
* fix slot mapping
2026-04-16 14:28:40 +08:00
1090f8b123
[Models]support GLM4.7 Flash && Ernie_MLA ( #7139 )
...
* support GLM4.7 Flash && Ernie_MLA
2026-04-03 17:41:33 +08:00
7a2e33098f
[XPU] Refactor pre process ( #6993 )
...
* [XPU] support speculate_pre_process
* merge develop
* fix codestype
* fix mtp, support cu_seqlens_q_output
* fix mtp, support cu_seqlens_q_output
* fix test
---------
Co-authored-by: lizan1999 <lizan03@baidu.com >
2026-04-01 20:29:55 +08:00
c29e86fc9d
[Feature] Support mtp overlap schedule ( #7001 )
2026-04-01 14:24:26 +08:00
bb925c605f
[Other] Adjust GPUModelRunner to enhance compatibility ( #6851 )
2026-03-16 14:49:19 +08:00
7591e0d6bc
fix eb5 mtp(mix) ( #6800 )
2026-03-13 17:36:57 +08:00
c3aceb6bdc
[Models][OP][Optimization] Support DeepSeek-v3.2 model, integrate DSA & Indexer architecture with FlashMLA/DeepGEMM ( #6689 )
...
* Support DeepSeek-v3.2 model, integrate DSA & Indexer architecture with FlashMLA/DeepGEMM
2026-03-10 15:05:14 +08:00
5e78c1ac87
[Graph Optimization] Support CUDAGraph for P/PD mixed Batch using SOT subgraph spliting mode ( #6196 )
...
* refine comment && refine variable name
* replace comment
2026-01-29 16:29:54 +08:00
1e3c35496c
[XPU][Graph Optimization] XPU Support CUDAGraph ( #6152 )
...
* support cuda graph
2026-01-22 14:41:56 +08:00
988e0bc338
[Feature] Add PaddleFormers fallback backend ( #5999 )
...
* feat(paddleformers): add dense text model fallback backend
* docs(paddleformers): add user guide and fix code review issues
* add fallback unit test
* precommit format
* fix pre-commit
* fix: address code review feedback
* docs: add PaddleFormers backend documentation (EN) and simplify installation
---------
Co-authored-by: Jiang-Jia-Jun <163579578+Jiang-Jia-Jun@users.noreply.github.com >
Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com >
2026-01-19 21:50:50 +08:00
1ee285c2d6
[Intel HPU] enable chunked prefill ( #5903 )
...
* [Intel HPU] enable chunked prefill
* fix bug by copilot comments
2026-01-06 21:01:50 +08:00
404cf0ece4
[Intel HPU] enable tensor_wise_fp8 ( #5324 )
...
* [Intel HPU] enable tensor_wise_fp8
* update code based on comments
* fix code style issue
* fix bug about RP 5138
* mv kv_cache modifications to HPU backend
* fix FP8 Precision Issues
* fix FP8 Precision Issues
* Add quantization UT
---------
Co-authored-by: yanfeich <yanfei.cheng@intel.com >
Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com >
2025-12-17 16:45:03 +08:00
888c4b992d
[XPU] refactor of block_attn param 'pos_emb_type' ( #5511 )
2025-12-12 14:30:09 +08:00
e58fed3665
[Graph Optimization][BugFix][CI] Fix 0size bug && add unitest ( #5495 )
2025-12-11 16:25:26 +08:00
b2908b8e82
[New][RL] Support Rollout Routing Replay ( #5405 )
...
* [RL] Support Rollout Routing Replay
* add routing indices cache
* fix config bug and moe forward bug
* R3 Support GLM
* support eb4.5
* fix merge bug
* Apply suggestion from @Copilot
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com >
* Apply suggestion from @Copilot
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com >
* Apply suggestion from @Copilot
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com >
* Apply suggestion from @Copilot
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com >
* add routing replay ci
* support glm topk
* support orther top_k
* fix ci bug
* pre-commit
* only support chatcmpl
* Revert "Revert "[RL] Support Rollout Routing Replay (#5321 )" (#5402 )"
This reverts commit c45e064f3d175728472+Copilot@users.noreply.github.com >
Co-authored-by: Yuanle Liu <yuanlehome@163.com >
2025-12-05 22:06:26 +08:00
c45e064f3d
Revert "[RL] Support Rollout Routing Replay ( #5321 )" ( #5402 )
...
This reverts commit 96d2d4877b
2025-12-05 20:19:39 +08:00
96d2d4877b
[RL] Support Rollout Routing Replay ( #5321 )
...
* [RL] Support Rollout Routing Replay
* add routing indices cache
* fix config bug and moe forward bug
* R3 Support GLM
* support eb4.5
* fix merge bug
* Apply suggestion from @Copilot
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com >
* Apply suggestion from @Copilot
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com >
* Apply suggestion from @Copilot
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com >
* Apply suggestion from @Copilot
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com >
* add routing replay ci
* support glm topk
* support orther top_k
* fix ci bug
* pre-commit
* only support chatcmpl
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com >
Co-authored-by: Yuanle Liu <yuanlehome@163.com >
2025-12-05 20:01:33 +08:00
5cd17fd662
[Models] Add forward_meta to moe models' forward function ( #5138 )
...
* [Models] Add forward_meta to moe models' forward function
* fix missing param
* fix
* fix
* fix forward_meta
* fix test and remove chunked MoE releated in config
* fix test
* fix
* fix
2025-12-04 13:26:58 +08:00
e70e2279ce
[PD Disaggregation][XPU] Add XPU support for PD disaggregation ( #5113 )
...
* [XPU] xpu support PD disaggregation
* [XPU] fix the issue of cache KV transfer process startup failure on non-zero XPU cards
* [XPU] xpu support PD disaggregation in v1 scheduler
---------
Co-authored-by: ddchenhao66 <dhaochen163.com>
2025-11-21 14:09:01 +08:00
43097a512a
[BugFix] [PD Disaggregation] fix v1 scheduler prefill node profile run & ipc transfer protocol ( #5132 )
...
CE Compile Job / ce_job_pre_check (push) Has been cancelled
CE Compile Job / print_ce_job_pre_check_outputs (push) Has been cancelled
CE Compile Job / FD-Clone-Linux (push) Has been cancelled
CE Compile Job / Show Code Archive Output (push) Has been cancelled
CE Compile Job / BUILD_SM8090 (push) Has been cancelled
CE Compile Job / BUILD_SM8689 (push) Has been cancelled
CE Compile Job / CE_UPLOAD (push) Has been cancelled
Deploy GitHub Pages / deploy (push) Has been cancelled
* [fix] fix v1 scheduler profile run for append attention in prefill node
* [fix] skip send_signal if kv signal not inited for gpu and xpu
* [fix] extend fix to flash_attn & mla_attn
* [fix] fix v1 pd run in ipc transfer protocol
* [ci] add test for v1 pd profile run using ipc transfer protocol
* [style] fix code style check
* [style] fix code style again
* [fix] fix profile run
* [update] remove --num-gpu-blocks-override in example script
* [chore] rename forward_meta is_profiling to is_dummy_or_profile_run
2025-11-20 21:39:22 +08:00
876e4a8935
remove input_ids from ForwardMeta ( #4793 )
CE Compile Job / ce_job_pre_check (push) Has been cancelled
CE Compile Job / print_ce_job_pre_check_outputs (push) Has been cancelled
CE Compile Job / FD-Clone-Linux (push) Has been cancelled
CE Compile Job / Show Code Archive Output (push) Has been cancelled
CE Compile Job / BUILD_SM8090 (push) Has been cancelled
CE Compile Job / BUILD_SM8689 (push) Has been cancelled
CE Compile Job / CE_UPLOAD (push) Has been cancelled
Deploy GitHub Pages / deploy (push) Has been cancelled
2025-11-05 11:55:51 +08:00
83d45af1f3
fix import image_ops error on some platforms ( #4559 )
2025-10-24 16:09:20 +08:00
a751d977bc
[Optimization] Fuse get_max_len and get_kv_max_len ( #4369 )
...
CE Compile Job / ce_job_pre_check (push) Has been cancelled
CE Compile Job / print_ce_job_pre_check_outputs (push) Has been cancelled
CE Compile Job / FD-Clone-Linux (push) Has been cancelled
CE Compile Job / Show Code Archive Output (push) Has been cancelled
CE Compile Job / BUILD_SM8090 (push) Has been cancelled
CE Compile Job / BUILD_SM8689 (push) Has been cancelled
CE Compile Job / CE_UPLOAD (push) Has been cancelled
Deploy GitHub Pages / deploy (push) Has been cancelled
* opt split_q_block
* fuse max_lens and max kv_len
2025-10-13 20:35:00 +08:00
20c7b741f4
[XPU] Support W4A8C8-TP4-300B Model ( #4068 )
...
* support w4a8
* delete ep block attn
* delete moe_topk_select
* update note
* update
* delte useless info
* update
* add some note
* fix some format
* update scale info
* add ans baseline
---------
Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com >
2025-10-10 15:41:32 +08:00
aa27b03bc0
[Executor]CUDAGraph support Speculate Decode ( #3769 )
...
CE Compile Job / ce_job_pre_check (push) Has been cancelled
CE Compile Job / print_ce_job_pre_check_outputs (push) Has been cancelled
CE Compile Job / FD-Clone-Linux (push) Has been cancelled
CE Compile Job / Show Code Archive Output (push) Has been cancelled
CE Compile Job / BUILD_SM8090 (push) Has been cancelled
CE Compile Job / BUILD_SM8689 (push) Has been cancelled
CE Compile Job / CE_UPLOAD (push) Has been cancelled
Deploy GitHub Pages / deploy (push) Has been cancelled
* success run ngram
* Revert "[Code Simplification] remove cum_offsets (#3410 )"
This reverts commit 32b39620bc8351e83993834639a7ff2694294196@qq.com >
Co-authored-by: littledgg <1658565283@qq.com >
Co-authored-by: zeroRains <linjunlu@zerorains.top >
Co-authored-by: gongshaotian <gstain5555@outlook.com >
2025-10-09 21:18:29 +08:00
87179cb744
[XPU] support XPU VL model inference ( #4030 )
...
* [XPU] support XPU VL model inference
* fix image op import and device check
* rebase develop
* fix perf
2025-09-25 14:34:15 +08:00
f1b5392e20
[Intel HPU] Support intel hpu platform ( #4161 )
...
* [Intel HPU] Support intel hpu platform
* fix some issues
* apply precommit and move AttentionBackend_HPU
* fix format issue
* correct ops import
* fix ci issue
* update code in layers
* fix code style issue
* remove dense tp moe ep mode
* fix enc_dec_block_num
* fix rebase issue
* rename hpu to gaudi in readme
* rename ForwardMeta_HPU to HPUForwardMeta
2025-09-24 12:27:50 +08:00
a7392a0ff9
【Inference Optimize】DeepSeek-V3-model MLA Optimize ( #3886 )
...
* support MLA chunk_size auto search & cuda_graph
2025-09-11 10:46:09 +08:00
3d0aaa5923
[Excutor] Experiment Feature-Support Prefill in cudagraph ( #3459 )
...
* Support prefill in Cudagraph
* Refactor GetBlockShapeAndSplitKVBlock Kernel V2
* Refactor GetBlockShapeAndSplitKVBlock Kernel V2.1
* Refactor GetBlockShapeAndSplitKVBlock Kernel V2.2
* Refactor GetBlockShapeAndSplitKVBlock Kernel V2.3
* Refactor GetBlockShapeAndSplitKVBlock Kernel V2.4
* Refactor GetBlockShapeAndSplitKVBlock Kernel V2.5
* Solve problem about encoder_num_blocks_x_cpu
* Add early-exit mechanism for attention kernel
* fix test case about append-attention
* Update testcode, Add annotations to related tensors
* move get_input_length_list
* solve test_code
* Add annotations about early-exit for attention kernel
* Add annotations about early-exit for attention kernel2
* solve comment
* solve mtp
---------
Co-authored-by: RAM <gstian5555@outlook.com >
2025-09-08 13:12:24 +08:00
72094d4d82
enable dcu ci ( #3402 )
2025-08-29 10:23:08 +08:00
4957908275
add input_processor plugin ( #3657 )
...
* add input_processor plugin
* update
* update
* update
* update
* update
* update
* update
* update
* update
* update
* update
2025-08-28 22:53:57 +08:00
ea4a3b479c
[Excutor] Increase buffer size to prevent address corruption; add forward metadata debug tool ( #3404 )
...
* 修复buffer申请不够大，增加打印forwardmetadata的工具
* fix mistake
* Make CPU tensor in CPUPlace
* Add test about forward_meta_str and Add unitest_requirement
---------
Co-authored-by: RAM <gstian5555@outlook.com >
2025-08-18 16:14:09 +08:00
b4fef2cf29
[MetaxGPU] Support FastDeploy on metax gpu ( #3241 )
...
* [MetaxGPU] Support FastDeploy on metax gpu
* Update metax_worker.py
1. change worker log;
2. remove custom allreduce, adapt it later;
3. remove cuda graph;
* Update __init__.py
1. remove metax's key work comment
* Update __init__.py
1. remove metax's key word comment;
2. add fused_moe_kernel_paddle import
---------
Co-authored-by: yongqiangma <xing.wo@163.com >
2025-08-13 11:11:54 +08:00
d850660872
[Executor] Refactor GetBlockShapeAndSplitKVBlock Kernel ( #2989 )
...
* reset decoder_block_shape_q buffer
* refactor GetBlockShapeAndSplitKVBlock Kernel and cudagraph padding batch
* update decode_max_tile_size
* fix pre-commit
* update block_multihead_attn_backend
* update flas attn backend
* update MLA Attention
* update XPU Attention
* update gcu,iluvatar model runner
* Update MTP
* fix MTP bug
2025-07-31 00:09:31 +08:00
2f74e93d7e
use dist.all_reduce(min) to sync num_blocks_local ( #2933 )
...
* pre-commit all files check
* reduce min num_blocks_local
* fix nranks=1
* pre-commit when commit-msg
2025-07-21 01:23:36 -07:00
8c5407d9e4
remove cum_offsets from ForwardMeta ( #2925 )
Deploy GitHub Pages / deploy (push) Has been cancelled
2025-07-19 23:57:27 +08:00
25698d56d1
polish code with new pre-commit rule ( #2923 )
2025-07-19 23:19:27 +08:00
ddb10ac509
[Inference, rename] remove padding_offsets from atten use batch_id_per_token ( #2880 )
...
* remove padding_offsets from atten
2025-07-17 18:41:31 +08:00
0fad10b35a
[Executor] CUDA Graph support padding batch ( #2844 )
...
* cuda graph support padding batch
* Integrate the startup parameters for the graph optimization backend and provide support for user - defined capture sizes.
* Do not insert max_num_seqs when the user specifies a capture list
* Support set graph optimization config from YAML file
* update cuda graph ci
* fix ci bug
* fix ci bug
2025-07-15 19:49:01 -07:00
59071268b6
[Executor] Move forward_meta.py to fastdeploy/model_executor ( #2774 )
...
* Use PEP 563 in attention.py and fix conflict
* merge commit
* Change what was left out last time
2025-07-10 20:36:51 +08:00
kevin