-
Notifications
You must be signed in to change notification settings - Fork 59
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Multi-thread read performance #99
Comments
yes you should be seeing higher concurrency if there are more than 3 threads. |
OK, thanks. I'll do some more digging, including putting the data back on the SSD volume. Do you think it's possible the performance characteristics of the ramdisk may be at fault? |
I'm seeing a speedup of about 3.7 on 6 threads. The serial version of the code is seeing about 60% system CPU; I wonder if I'm just hitting disk overheads. Data records are small; 8 byte keys and 10 byte records. 600M entries. (Prior version of this code loaded up an in-memory hashtable from a memory map; trying to get the app working with as little startup overhead as possible without the current solution of requiring either vmtouch-locked data or persistent server) |
I don't know |
I've been experimenting with NuDB and ran into the following issue - I would like to have multiple worker threads access the database concurrently, entirely read-only. There seems to be a bottleneck where I only get 2-3x speedups; intel profiler showed what looks to be locking overhead. Should I be able to see higher concurrency for this? Or perhaps am I seeing I/O read rate issues (the data file is in /dev/shm ramdisk for this testing)
Thanks,
Bob
The text was updated successfully, but these errors were encountered: