f436ccff98
feat: consolidate cache, worker_process, and paddle logs into unified files ( #7432 )
...
Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com >
2026-04-22 21:47:35 +08:00
a6a740f759
support deepgemm without bias input ( #7559 )
2026-04-22 19:49:48 +08:00
e580cf0fef
[OP] latent moe deepgemm support ( #7537 )
...
* commit
* commit
* commit
* commit
* commit
* commit
* commit
* commit
2026-04-22 13:47:06 +08:00
36d47aa23e
[XPU] get_infer_param use inplace copy, remove block_tables abundant d2h copy ( #7431 )
...
* inplace_copy: encoder_batch_idx/decoder_batch_idx bs == 9 ok
* inplace_copy: encoder_seq_lod/decoder_seq_lod bs == 9 ok
* inplace_copy: all bs == 9 ok
* inplace_copy: all cpu bs == 9 ok
* inplace_copy: len_info_cpu bs == 9 ok
* finished and rm unused code
* prefix_block_tables reuse
* refine
* improve performance
* remove block_table copy to cpu
* fix unit test
* fix
* resolve conflict
* refine code
* fix
* fix
* fix
* fix
* fix
* try fix unit tests
* fix
* tmp save
* fix unit test
* get_infer_param try less return values
* add yinwei fix
---------
Co-authored-by: yinwei <yinwei_hust@163.com >
2026-04-22 11:01:32 +08:00
534c43c888
[BugFix] Fix skip_x_record_stream incompatibility across deep_ep versions ( #7542 )
...
* fix skip_x_record_stream
* fix
* optim
2026-04-21 06:31:00 -07:00
c618a39562
add m_grouped_bf16_gemm_nn_contiguous, del paddle.batch_gemm ( #7527 )
2026-04-21 20:19:46 +08:00
d8cdda86cb
[RL][Cherry-Pick] Fix the out-of-bounds issue caused by int32 in the R3 kernel ( #7155 )
...
* [RL]Perf: Optimize batch delete prefix and fused put in R3 (#6604 )
* Optimizate delete batch and fused put
* refine code
* refine code
* refine code
* Support suspend r3
* [RL] Fix R3 Empty bug with TP=1 (#6777 )
* Fix int32 overflow
* refine code
---------
Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com >
2026-04-21 01:51:09 -07:00
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
e4a4573080
[Iluvatar] Fix cannot import name mtp_save_first_token ( #7495 )
...
Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com >
2026-04-21 14:09:08 +08:00
609f649dd7
[OP] Add flashmla baseline implementation and precision test ( #7477 )
2026-04-21 13:37:52 +08:00
a0c39cc9af
[Typo] Fix parameter name typo in slice_fn: paramter -> parameter ( #7462 )
...
Fix the typo in the internal parameter name of slice_fn():
weight_or_paramter -> weight_or_parameter.
No functional changes.
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com >
2026-04-20 10:06:02 +08:00
df3b4e12f4
[Speculative Decoding] Add MTP logprob support for PD disaggregation ( #7442 )
...
* support mtp logprob in pd
* fix
* fix
* fix
* fix xpu bugs
2026-04-17 21:37:38 +08:00
3b9d6c60d3
[Iiluvatar] fix ci error and update readme ( #7453 )
2026-04-17 20:42:56 +08:00
6ce4854714
[Feature] Support MOE Cutlass backend for latent MOE ( #7428 )
...
* support moe cutlass backend latent moe
2026-04-16 22:11:49 +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
420a8c1af5
fix deep gemm import ( #7425 )
2026-04-16 17:56:56 +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
6b891da02b
[Optimization] enable trtllm_all_reduce fusion kernel in glm model ( #6660 )
...
* enable trtllm_all_reduce fusion kernel in glm model
* fix conflict
* format update
* fix a bug
* modify test
* modify test
* support empty tensor and modify test
* fix test_linear config issues
* modify test name
* add edge test case
* modify format
* fix conflict
* modify default max token num in trtllm_allreduce_fusion
* add max token num branch for trtllm_allreduce_fusion
* fix format
* fix rmsnorm config issue
* modify 2025 to 2026
* using compat grard
* Lazily import flashinfer.comm and fix test config issue
* fix test issues
* add flashinfer cache dir clean machine
* fix some issues
2026-04-16 14:10:19 +08:00
a498720a75
[RL] Add clear_graph_opt_backend for glm4_mtp ( #7378 )
...
* add clear_grpah func
* fix spell
2026-04-15 19:44:15 +08:00
5e54770b2e
[Feature] 添加 MoE 层 latent mode 支持 ( #7382 )
2026-04-15 13:57:07 +08:00
f7a2418ce2
[Speculate Decoding] Fix reasoning_phase_token_constraint call args in SpeculativeSampler ( #7402 )
2026-04-15 12:45:23 +08:00
8995a38fa4
fix dsa indexer norm to layernorm ( #7398 )
2026-04-15 11:42:45 +08:00
bb30f88f1a
[Models] support MLA gate attention ( #7404 )
...
* support mla gate attn
* support mla gate attn
2026-04-15 11:42:34 +08:00
616b29ce08
check init_flash_attn_version log ( #7399 )
2026-04-15 11:05:10 +08:00
8819a039c9
[Others] Fix typo ( #7280 )
...
* typo
* typo
* typo
* typo
2026-04-14 17:28:22 +08:00
abba29b348
[BugFix] fix mm rope ( #7274 )
2026-04-14 11:36:08 +08:00
27b00cf385
[XPU] glm-4.5-air ( #7071 )
2026-04-14 11:31:49 +08:00
73bd4ab318
[Feature] 为 FusedMoE 添加 hidden_size 显式参数支持 ( #7361 )
...
[Feature] 为 FusedMoE 添加 hidden_size 显式参数支持
2026-04-13 20:24:58 +08:00
31e2a8bbad
[Speculative Decoding] Support mtp super ultra overlap in pd-split mode with insert_task overlap ( #7323 )
...
* support mtp overlap in pd-split mode with insert_task overlap
2026-04-13 19:41:17 +08:00
1fb8194191
[OP][Models][Optimization] 优化 RoPE CUDA kernel 并更新 DeepSeek V3 配置 ( #7359 )
...
* dsk del prefill mask
* dsk support 1M+ seq_len rope
* update rope tests
* Replace max_position_embeddings with max_model_len
* 1D grid: gridDim.x has a maximum size of 2^31-1, far exceeding the actual number of tokens.
2026-04-13 19:12:36 +08:00
b34708604c
[TI-consistent] support quant use pow2scale ( #7308 )
...
* support quant use pow2scale
* fix
* fix
2026-04-13 00:01:53 -07:00
6213ad5340
[Docs][BugFix] fix mla log ( #7243 )
...
* [Docs] Fix Chinese punctuation issues
2026-04-13 12:15:43 +08:00
d659099415
[Cleanup] Replace torch proxy alias with public compat API ( #7348 )
2026-04-13 11:43:26 +08:00
cb03958b52
[XPU] Refactor get_padding_offset to single kernel. ( #7029 )
...
* [XPU] Refactor get_padding_offset to single kernel.
* add unittest.
* fix codestyle.
* remove cum_offsets_now.
* remove max_len.
2026-04-13 11:04:50 +08:00
225fc8d222
use self.hidden_size not use self.fd_config.model_config.hidden_size ( #7340 )
2026-04-11 22:39:43 +08:00
4982aa000e
[RL]moe bf16 ep support paddle batch_gemm ( #7337 )
...
* moe bf16 ep support paddle batch_gemm
2026-04-11 21:51:12 +08:00
ba01d7a823
[Optimization] [OP] [Models] dsk del prefill mask ( #7313 )
...
* dsk del prefill mask
* dsk support 1M+ seq_len rope
* update rope tests
2026-04-11 19:32:27 +08:00
076ab07528
[RL] change glm rope_emb calculation ( #7316 )
...
* change glm rope_emb calculation
* glm without EnforceFmulRN
* fix ci
2026-04-11 18:36:28 +08:00
627f0d9cc8
[RL] change rms norm for glm ( #7269 )
...
* change rms norm for glm
* refine code
* refine code
* refine code
2026-04-10 01:02:37 -07:00
870dbac370
Use triton qk_norm both in Prefill and Decode ( #7213 )
...
Co-authored-by: “liuruian” <liuruian@baidu.com >
2026-04-10 15:44:01 +08:00
14d46181b8
[Loader] add multi-thread model loading ( #6877 )
...
* multi-thread-loader
* fix ut
2026-04-09 23:40:15 -07:00
c1fb3112f8
[FDConfig] Support CLI args for quantization params and add cudagraph validation ( #7281 )
...
* refactor quant cli param
2026-04-10 14:13:42 +08:00
613f92ee8f
[Feature] support nvfp4 tbo ( #7259 )
2026-04-09 17:29:39 +08:00
39ff38aba1
[OP]Unify MoE op with moe_permute path for bf16 GLM ( #7164 )
2026-04-09 16:17:56 +08:00
51efe27d76
[BugFix] Fix batch_size derivation and relax shape checks in SM90 flash_mask_attn ( #7210 )
...
* [BugFix] fix_flash_mask_attn_sm90
* [BugFix] fix_flash_mask_attn_sm90
* [BugFix] Fix batch_size derivation and relax shape checks in SM90 flash_mask_attn
* [BugFix] Fix batch_size derivation and relax shape checks in SM90 flash_mask_attn
2026-04-09 11:05:10 +08:00
43ace7af25
[RL] support moe-topk use topk_reduce_func ( #7218 )
...
* support moe-topk use topk_reduce_func
* fix ep error
* fix ut
* fix ut
2026-04-09 11:01:03 +08:00
7005404ce3
[DeepSeekV3.2][Graph Optimization]Remove synchronous operation to avoid capture fail and unnecessary contiguous in DSA Backend ( #7253 )
...
* Delete contiguous ops.
* fix scale
* Delete unnecessary comments
* fix style
2026-04-09 11:00:13 +08:00
48d2bbeb74
fix dsa ( #7252 )
2026-04-08 20:21:38 +08:00
bb48bcbaa2
Split enable_mm ( #7183 )
...
Co-authored-by: liuruian <liuruian@MacBook-Pro.local >
2026-04-08 11:25:41 +08:00
zhouchong