计算 CUDA 数组中数字的出现次数 [英] Counting occurrences of numbers in a CUDA array
问题描述
我有一个使用 CUDA 存储在 GPU 上的无符号整数数组(通常是 1000000
元素).我想计算数组中每个数字的出现次数.只有几个不同的数字(大约 10
),但这些数字可以从 1 到 1000000
.大约9/10
的数字是0
,我不需要他们的计数.结果如下所示:
I have an array of unsigned integers stored on the GPU with CUDA (typically 1000000
elements). I would like to count the occurrence of every number in the array. There are only a few distinct numbers (about 10
), but these numbers can span from 1 to 1000000
. About 9/10
th of the numbers are 0
, I don't need the count of them. The result looks something like this:
58458 -> 1000 occurrences
15 -> 412 occurrences
我有一个使用 atomicAdd
s 的实现,但它太慢了(很多线程写入同一个地址).有人知道快速/有效的方法吗?
I have an implementation using atomicAdd
s, but it is too slow (a lot of threads write to the same address). Does someone know of a fast/efficient method?
推荐答案
您可以通过首先对数字进行排序,然后进行键控归约来实现直方图.
You can implement a histogram by first sorting the numbers, and then doing a keyed reduction.
最直接的方法是使用thrust::sort
,然后使用thrust::reduce_by_key
.它通常也比基于原子的临时分箱快得多.这是一个示例.
The most straightforward method would be to use thrust::sort
and then thrust::reduce_by_key
. It's also often much faster than ad hoc binning based on atomics. Here's an example.
这篇关于计算 CUDA 数组中数字的出现次数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!