You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I encountered this by accident when doing benchmarks and using glibc malloc by accident. Normally I use jemalloc, but it wasn't installed on the host on which I compiled db_bench. This is introduced by 8b74cea and the new allocation might be here.
I know we prefer jemalloc vs glibc malloc, but is it possible to reduce the amount of allocations?
Example output from the fwdrangewhilewriting benchmark step shows the impact. QPS drops from 512323 to 485274. The first line is from b82edff and the second is from 8b74cea. These diffs are adjacent in the repo (b82... precedes 8b7...).
From the throughput result and vmstat output (not shared here) I see that 8b74cea uses ~5% more CPU per query. I confirmed this does not reproduce when db_bench is linked with jemalloc.
A flamegraph for 8b74cea that shows the problem -- on the left side of the flamegraph the call stacks for __default_morecore, __libc_free and __libc_malloc are much wider:
The text was updated successfully, but these errors were encountered:
I encountered this by accident when doing benchmarks and using glibc malloc by accident. Normally I use jemalloc, but it wasn't installed on the host on which I compiled db_bench. This is introduced by 8b74cea and the new allocation might be here.
I know we prefer jemalloc vs glibc malloc, but is it possible to reduce the amount of allocations?
Example output from the fwdrangewhilewriting benchmark step shows the impact. QPS drops from 512323 to 485274. The first line is from b82edff and the second is from 8b74cea. These diffs are adjacent in the repo (b82... precedes 8b7...).
From the throughput result and vmstat output (not shared here) I see that 8b74cea uses ~5% more CPU per query. I confirmed this does not reproduce when db_bench is linked with jemalloc.
A reproduction script:
A flamegraph for b82edff (no regression here)
A flamegraph for 8b74cea that shows the problem -- on the left side of the flamegraph the call stacks for __default_morecore, __libc_free and __libc_malloc are much wider:
The text was updated successfully, but these errors were encountered: