|
关注:1
2013-05-23 12:21
求翻译:在CUDA 平台上,一个线程块最多可知其512个线程,而GPU上的内存控制器从某个规定倍数地址开始读取,才会有最高的效率是什么意思? 待解决
悬赏分:1
- 离问题结束还有
在CUDA 平台上,一个线程块最多可知其512个线程,而GPU上的内存控制器从某个规定倍数地址开始读取,才会有最高的效率
问题补充: |
|
2013-05-23 12:21:38
In the CUDA platform, we can see a thread block up to its 512 threads, and memory controller on the GPU multiple addresses starting from a requirement to read, will have the highest efficiency
|
|
2013-05-23 12:23:18
In CUDA platform, did you know that at most one thread blocks, and its 512 threads on GPU the memory controller from a certain specified multiple start address read, can the highest level of efficiency
|
|
2013-05-23 12:24:58
In the CUDA platform, a thread block are most may know its 512 threads, but on the GPU memory controller starts from some stipulation multiple address to read, only then can have the maximum efficiency
|
|
2013-05-23 12:26:38
On the CUDA platform, one thread blocks up to 512 threads, and GPU memory controller from a provision on multiple address to begin the read will only have the highest efficiency
|
|
2013-05-23 12:28:18
In CUDA terrace, first thread piece can know his 512 pieces of thread at most, and the memory controller at GPU reads from a certain fixed multiple address, there will be supreme efficiency
|
湖北省互联网违法和不良信息举报平台 | 网上有害信息举报专区 | 电信诈骗举报专区 | 涉历史虚无主义有害信息举报专区 | 涉企侵权举报专区