每个线程的CUDA最大寄存器：sm_12 vs sm_20 [英] CUDA maximum registers per thread: sm_12 vs sm_20
问题描述
我的内核广泛使用寄存器。
My kernel uses registers extensively.
当编译1.2设备 - ptxas-options = -v 时报告83个寄存器。当我试图编译为2.0时，只有63个寄存器在使用，其余的本地数据被放入本地内存。使用--maxrregcount的实验给出了每个线程为124个寄存器的限制，对于1.2个设备，只有63个寄存器。
When compiling for 1.2 devices --ptxas-options=-v reports 83 registers. When I am trying to compile for 2.0 there are only 63 registers in use, the rest of local data are put into local memory. Experiments with '--maxrregcount' give limit of 124 registers per thread for 1.2 devices and as few as 63 registers for 2.0.
可以将所有数据寄存器在2.0架构？
Is it possible to put all the data into registers on 2.0 architecture?
推荐答案
不幸的是，计算能力2.x卡的每线程寄存器限制为每个线程63个寄存器。如果你有一个非常复杂的内核消耗了很多寄存器，那么就没有办法停止本地内存溢出。
Unfortunately, the per-thread register limit for compute capability 2.x cards is 63 registers per thread. There isn't anyway to stop local memory spillage if you have a very complex kernel which consumes a lot of registers.
这篇关于每个线程的CUDA最大寄存器：sm_12 vs sm_20的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
