Remove MTE2-to-V event id reservation#677
Conversation
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request removes the reserved event ID for the PIPE_MTE2 to PIPE_V pipeline pair across the GraphSyncSolver and SyncEventIdAllocation components. It also introduces a new test case to verify that this synchronization pair now correctly utilizes EVENT_ID7. The reviewer suggested optimizing the reservedEventIdNum maps in Utility.cpp by declaring them as static const to avoid redundant allocations and recommended refactoring these maps into a single shared constant to reduce code duplication.
| @@ -267,7 +267,6 @@ int64_t getHWAvailableEventIdNum(SyncMode syncMode, pto::PIPE setPipe, | |||
| const llvm::DenseMap<std::tuple<PIPE, PIPE>, int64_t> reservedEventIdNum = { | |||
There was a problem hiding this comment.
The reservedEventIdNum map is constructed and initialized on every call to getHWAvailableEventIdNum. Since this map is constant and the function is likely called frequently during synchronization solving, it should be declared as static const to avoid redundant allocations and re-initialization. Additionally, consider refactoring this map into a single shared constant to avoid duplication with the identical map in getHWAvailableEventIds.
| const llvm::DenseMap<std::tuple<PIPE, PIPE>, int64_t> reservedEventIdNum = { | |
| static const llvm::DenseMap<std::tuple<PIPE, PIPE>, int64_t> reservedEventIdNum = { |
| @@ -299,7 +298,6 @@ SmallVector<int64_t> getHWAvailableEventIds(SyncMode syncMode, | |||
| const llvm::DenseMap<std::tuple<PIPE, PIPE>, int64_t> reservedEventIdNum = { | |||
There was a problem hiding this comment.
Similar to the instance in getHWAvailableEventIdNum, this map should be static const to improve performance by avoiding repeated construction and initialization on every function call.
| const llvm::DenseMap<std::tuple<PIPE, PIPE>, int64_t> reservedEventIdNum = { | |
| static const llvm::DenseMap<std::tuple<PIPE, PIPE>, int64_t> reservedEventIdNum = { |
Codex Review该评论由 review 机器人自动更新。
Summary未检查到 PR #677 存在问题 FindingsNo issues found. |
A5 板测失败
失败用例
|
A3 板测完成（有跳过）
|
|
/run a5 qwen3_decode_incore_0 qwen3_decode_incore_1 qwen3_decode_incore_2 qwen3_decode_incore_3 qwen3_decode_incore_4 qwen3_decode_incore_5 qwen3_decode_incore_6 qwen3_decode_incore_7 qwen3_decode_incore_8 qwen3_decode_incore_9 qwen3_decode_incore_10 qwen3_decode_incore_11 qwen3_decode_incore_12 qwen3_decode_incore_13 qwen3_decode_incore_14 qwen3_decode_incore_15 qwen3_decode_incore_16 --pto-level=level3 |
|
已接收
页面会自动刷新，可以直接看当前阶段、排队情况和最近结果。 |
A5 板测失败
失败用例
|
A5 板测失败详情：PR #677qwen3_decode_incore_6
|
|
/run a5 qwen3_decode_incore_6 qwen3_decode_incore_12 --pto-level=level3 |
|
已接收
页面会自动刷新，可以直接看当前阶段、排队情况和最近结果。 |
A5 板测成功
|
|
/run a5 qwen3_decode_incore_5 qwen3_decode_incore_6 qwen3_decode_incore_8 qwen3_decode_incore_12 --pto-level=level3 |
|
已接收
页面会自动刷新，可以直接看当前阶段、排队情况和最近结果。 |
A5 板测成功
|
3 participants
No description provided.