GPGPU
GPGPU
在三维的位置看矩阵乘法
一、GPU的基础知识点
一个简易版本的架构区别:
GPU索引包含blockIdx和threadIdx,blockIdx是在线程之间共享的
-Warps :The PC is shared; maintain thread mask for Writeback,同一warp内所有线程共用一个程序计数器(PC),因此它们在同一时刻执行相同的指令。由于条件分支或其他条件,GPU会维护一个线程掩码,在执行写回(Writeback)阶段时,只有那些处于活动状态的线程会将计算结果写回到寄存器或内存中。
there are thread groups that share control units,and those are streaming multiprocessors. the core components of our gpu
- Raster Engine
- ROP
- PolyMorph Engine
- RT core
- TEX unit
- Warp Scheduler
- Dispatch Unit
- SFU
This post is licensed under CC BY 4.0 by the author.