TGEMV_MX¶

指令示意图¶

TGEMV_MX tile operation

简介¶

带缩放 Tile 的 GEMV 变体，支持混合精度/量化矩阵向量计算。

数学语义¶

概念上（基础 GEMV 路径）：

\[ \mathrm{C}_{0,j} = \sum_{k=0}^{K-1} \mathrm{A}_{0,k} \cdot \mathrm{B}_{k,j} \]

对于 TGEMV_MX，缩放 tile 参与实现定义的混合精度重建/缩放。架构约定是输出对应于目标定义的 mx GEMV 语义。

汇编语法¶

PTO-AS 形式：参见 PTO-AS 规范。

示意形式：

%acc = tgemv.mx %a, %a_scale, %b, %b_scale : (!pto.tile<...>, !pto.tile<...>, !pto.tile<...>, !pto.tile<...>) -> !pto.tile<...>

AS Level 1（SSA）¶

%acc = pto.tgemv.mx %a, %a_scale, %b, %b_scale : (!pto.tile<...>, !pto.tile<...>, !pto.tile<...>, !pto.tile<...>) -> !pto.tile<...>

AS Level 2（DPS）¶

pto.tgemv.mx ins(%a, %a_scale, %b, %b_scale : (!pto.tile_buf<...>, !pto.tile_buf<...>, !pto.tile_buf<...>, !pto.tile_buf<...>)) outs(%acc : !pto.tile_buf<...>)

C++ 内建接口¶

声明于 include/pto/common/pto_instr.hpp：

template <typename TileRes, typename TileLeft, typename TileLeftScale, typename TileRight, typename TileRightScale,
          typename... WaitEvents>
PTO_INST RecordEvent TGEMV_MX(TileRes &cMatrix, TileLeft &aMatrix, TileLeftScale &aScaleMatrix, TileRight &bMatrix, TileRightScale &bScaleMatrix, WaitEvents &... events);

template <AccPhase Phase, typename TileRes, typename TileLeft, typename TileLeftScale, typename TileRight,
          typename TileRightScale, typename... WaitEvents>
PTO_INST RecordEvent TGEMV_MX(TileRes &cMatrix, TileLeft &aMatrix, TileLeftScale &aScaleMatrix, TileRight &bMatrix, TileRightScale &bScaleMatrix, WaitEvents &... events);

template <typename TileRes, typename TileLeft, typename TileLeftScale, typename TileRight, typename TileRightScale,
          typename... WaitEvents>
PTO_INST RecordEvent TGEMV_MX(TileRes &cOutMatrix, TileRes &cInMatrix, TileLeft &aMatrix, TileLeftScale &aScaleMatrix, TileRight &bMatrix, TileRightScale &bScaleMatrix, WaitEvents &... events);

template <AccPhase Phase, typename TileRes, typename TileLeft, typename TileLeftScale, typename TileRight,
          typename TileRightScale, typename... WaitEvents>
PTO_INST RecordEvent TGEMV_MX(TileRes &cOutMatrix, TileRes &cInMatrix, TileLeft &aMatrix, TileLeftScale &aScaleMatrix, TileRight &bMatrix, TileRightScale &bScaleMatrix, WaitEvents &... events);

template <typename TileRes, typename TileLeft, typename TileLeftScale, typename TileRight, typename TileRightScale,
          typename TileBias, typename... WaitEvents>
PTO_INST RecordEvent TGEMV_MX(TileRes &cMatrix, TileLeft &aMatrix, TileLeftScale &aScaleMatrix, TileRight &bMatrix, TileRightScale &bScaleMatrix, TileBias &biasData, WaitEvents &... events);

template <AccPhase Phase, typename TileRes, typename TileLeft, typename TileLeftScale, typename TileRight,
          typename TileRightScale, typename TileBias, typename... WaitEvents>
PTO_INST RecordEvent TGEMV_MX(TileRes &cMatrix, TileLeft &aMatrix, TileLeftScale &aScaleMatrix, TileRight &bMatrix, TileRightScale &bScaleMatrix, TileBias &biasData, WaitEvents &... events);

附加重载支持累加/偏置变体和 AccPhase 选择。

约束¶

使用后端特定的 mx 合法性检查，用于数据类型、tile 位置、分形/布局组合以及缩放格式。
缩放 tile 兼容性和累加器提升由目标后端的实现定义。
为了可移植性，请根据目标实现约束验证确切的 (A, B, scaleA, scaleB, C) 类型元组和 tile 布局。

示例¶

实际使用模式请参见：

docs/isa/TMATMUL_MX.md
docs/isa/TGEMV.md

汇编示例（ASM）¶

自动模式¶

# 自动模式：由编译器/运行时负责资源放置与调度。
%acc = pto.tgemv.mx %a, %a_scale, %b, %b_scale : (!pto.tile<...>, !pto.tile<...>, !pto.tile<...>, !pto.tile<...>) -> !pto.tile<...>

手动模式¶

# 手动模式：先显式绑定资源，再发射指令。
# 可选（当该指令包含 tile 操作数时）：
# pto.tassign %arg0, @tile(0x1000)
# pto.tassign %arg1, @tile(0x2000)
%acc = pto.tgemv.mx %a, %a_scale, %b, %b_scale : (!pto.tile<...>, !pto.tile<...>, !pto.tile<...>, !pto.tile<...>) -> !pto.tile<...>

PTO 汇编形式¶

%acc = pto.tgemv.mx %a, %a_scale, %b, %b_scale : (!pto.tile<...>, !pto.tile<...>, !pto.tile<...>, !pto.tile<...>) -> !pto.tile<...>
# AS Level 2 (DPS)
pto.tgemv.mx ins(%a, %a_scale, %b, %b_scale : (!pto.tile_buf<...>, !pto.tile_buf<...>, !pto.tile_buf<...>, !pto.tile_buf<...>)) outs(%acc : !pto.tile_buf<...>)