MGATHER
指令示意图
简介
使用逐元素索引从全局内存收集加载元素到 Tile 中。
数学语义
对每个元素 (i, j) in the destination valid region:
dsti,j=mem[idxi,j]\mathrm{dst}_{i,j} = \mathrm{mem}[\mathrm{idx}_{i,j}]
汇编语法
PTO-AS 形式:参见 PTO-AS 规范。
同步形式:
%dst = mgather %mem, %idx : !pto.memref<...>, !pto.tile<...> -> !pto.tile<...>
AS Level 1(SSA)
%dst = pto.mgather %mem, %idx : (!pto.partition_tensor_view<MxNxdtype>, pto.tile<...>)
-> !pto.tile<loc, dtype, rows, cols, blayout, slayout, fractal, pad>
AS Level 2(DPS)
pto.mgather ins(%mem, %idx : !pto.partition_tensor_view<MxNxdtype>, !pto.tile_buf<...>) outs(%dst : !pto.tile_buf<...>)
C++ 内建接口
声明于 include/pto/common/pto_instr.hpp:
template <typename TileDst, typename GlobalData, typename TileInd, typename... WaitEvents>
PTO_INST RecordEvent MGATHER(TileDst& dst, GlobalData& src, TileInd& indexes, WaitEvents&... events);
约束
- Index interpretation is target-defined. The CPU simulator treats indices as linear element indices into
src.data(). - No bounds checks are enforced on
indexesby the CPU simulator.
示例
See related examples in docs/isa/ and docs/coding/tutorials/.
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。
%dst = pto.mgather %mem, %idx : (!pto.partition_tensor_view<MxNxdtype>, 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)
%dst = pto.mgather %mem, %idx : (!pto.partition_tensor_view<MxNxdtype>, pto.tile<...>)
PTO 汇编形式
%dst = mgather %mem, %idx : !pto.memref<...>, !pto.tile<...> -> !pto.tile<...>
# IR Level 2 (DPS)
pto.mgather ins(%mem, %idx : !pto.partition_tensor_view<MxNxdtype>, !pto.tile_buf<...>) outs(%dst : !pto.tile_buf<...>)