Replies: 1 comment
-
Merged in 29ba1f6 to master. Thank you. This is going to save 20 TBW on my 500 re-plots. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Assuming that there is a 110G RamDisk for temp2, moving .p2.t2.sort_bucket_xxx.tmp to temp2 will speed up about 10 to 20 seconds.
The change would be really easy, just change the phase2.hpp line below near the bottom from:
out.sort[i] = std::make_shared(32, log_num_buckets, prefix + "t" + std::to_string(i + 1));
to
out.sort[i] = std::make_shared(32, log_num_buckets, (i == 1 ? prefix_2 : prefix) + "t" + std::to_string(i + 1));
This moves about 40G from temp1 and temp2 and will not exceed the 110G limit. I tried a bucket size of 512 and then 128 and both work fine.
Beta Was this translation helpful? Give feedback.
All reactions