sort a matrix (1M rows x 100 columns) for each row in GPU# DataSciences - 数据科学
v*7
1 楼
遭中共通缉的外逃商人郭文贵反被连连爆料。最新消息显示，郭文贵曾和中共前国安高
官密会恐怖分子，和不同的恐怖组织、国家有关联：http://bit.ly/2fPV7s8
官密会恐怖分子，和不同的恐怖组织、国家有关联：http://bit.ly/2fPV7s8
c*l
2 楼
Do someone think it is feasible to sort a matrix (1M rows x 100 columns) for
each row in GPU? We keeping the repeating sorting every day and want to
know whether the performance could be improved to 10X or 20X faster (
Currently we just bought a server with 8 GPU K40).
each row in GPU? We keeping the repeating sorting every day and want to
know whether the performance could be improved to 10X or 20X faster (
Currently we just bought a server with 8 GPU K40).
l*m
3 楼
please refer https://solarianprogrammer.com/2013/02/04/sorting-data-in-
parallel-cpu-gpu/
In my opinion, cpu should be fast enough for the size if the sort alg and
implementation is correct. CPU-GPU data copy is a big overhead for such a
task
for
【在 c*****l 的大作中提到】
: Do someone think it is feasible to sort a matrix (1M rows x 100 columns) for
: each row in GPU? We keeping the repeating sorting every day and want to
: know whether the performance could be improved to 10X or 20X faster (
: Currently we just bought a server with 8 GPU K40).
parallel-cpu-gpu/
In my opinion, cpu should be fast enough for the size if the sort alg and
implementation is correct. CPU-GPU data copy is a big overhead for such a
task
for
【在 c*****l 的大作中提到】
: Do someone think it is feasible to sort a matrix (1M rows x 100 columns) for
: each row in GPU? We keeping the repeating sorting every day and want to
: know whether the performance could be improved to 10X or 20X faster (
: Currently we just bought a server with 8 GPU K40).
y*0
4 楼
没试过，大概想法是这样的。
因为你的columns比较少，所以log2 100才是7都不到，所以复杂度是7*matrix_size。
太小，不适合传到gpu上面。
cpu的cache用好了，直接每行走cache，直接在cpu上并行，应该是最佳选择。
for
【在 c*****l 的大作中提到】
: Do someone think it is feasible to sort a matrix (1M rows x 100 columns) for
: each row in GPU? We keeping the repeating sorting every day and want to
: know whether the performance could be improved to 10X or 20X faster (
: Currently we just bought a server with 8 GPU K40).
因为你的columns比较少，所以log2 100才是7都不到，所以复杂度是7*matrix_size。
太小，不适合传到gpu上面。
cpu的cache用好了，直接每行走cache，直接在cpu上并行，应该是最佳选择。
for
【在 c*****l 的大作中提到】
: Do someone think it is feasible to sort a matrix (1M rows x 100 columns) for
: each row in GPU? We keeping the repeating sorting every day and want to
: know whether the performance could be improved to 10X or 20X faster (
: Currently we just bought a server with 8 GPU K40).
相关阅读
data scientist对sql要求高吗新鲜面经San Diego 或者IRVINE 有·BIG DATA的MEETUP或者CONFERENCE吗？Hadoop Spark 学习小结[2014版] (转载)database startup の 广告~ (不是骗子)Hadoop Spark 学习小结[2014版] (转载)转：Coursera 推荐工作National Big Data Events这个cassandra paging的解决方案怎么样？海本统计 硕士读software engine还是ds要不要继续当ds？转： Twitter 各种职位内推！！！大量新职位最近的一些面经 （下）NYC area data scientist opportunities找工作屡败屡战，求提携，求建议Neural Network面试的时候会怎么问啊？Any 面经 for Workday Data Scientist?来美国两年半，工作-》绿卡-》创业，钱不重要，自己最重要Re: 攒人品，发Google Statistician/Data Scientist电面面经scala的主场在数学应用上 (转载)