Skip to content
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

Add exerimental multicard mode #38

Merged
merged 9 commits into from
Nov 17, 2015
Merged

Add exerimental multicard mode #38

merged 9 commits into from
Nov 17, 2015

Conversation

fhaverkamp
Copy link
Collaborator

Adding our experimental multicard mode
Add code to set accelerator and card_no for gzip/gunzip
Enhance statistic printout

No real redundancy, first just to try out the performance when using
two cards. Set ZLIB_CARD=-1 to activate this.
Software like samtools always does deflateInit deflate deflateEnd.
If we select the initial card statically. E.g. 0, we will not be
able to distribute the load.
To judge if the multicard feature is working nicely, I added
some code to dump more statistics e.g. which card executed
how many requests (DDCB).

In this patch, I also set the worker thread tid to 0, if the
init was failing. This avoids problem with pthread_cancel if
the thread is already gone because it failed to open the afu
e.g. if it is busy by another process in dedicated AFU mode.
No real redundancy, first just to try out the performance when using
two cards. Set ZLIB_CARD=-1 to activate this.
Software like samtools always does deflateInit deflate deflateEnd.
If we select the initial card statically. E.g. 0, we will not be
able to distribute the load.
To judge if the multicard feature is working nicely, I added
some code to dump more statistics e.g. which card executed
how many requests (DDCB).

In this patch, I also set the worker thread tid to 0, if the
init was failing. This avoids problem with pthread_cancel if
the thread is already gone because it failed to open the afu
e.g. if it is busy by another process in dedicated AFU mode.
…tool

Instead of falling back to the environment variables to define the
accelerator type and the card number we add some proprietary interfaces
to define those for more convenience when using our gzip/gunzip tool.
@fhaverkamp
Copy link
Collaborator Author

I would like to get those patches in. They do not improve the problem with the 2nd card we are currently having and they do not make it worse. For dedicated mode the changes worked ok so far.

fhaverkamp added a commit that referenced this pull request Nov 17, 2015
Add exerimental multicard mode
@fhaverkamp fhaverkamp merged commit 90ccce5 into master Nov 17, 2015
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

1 participant