Bugs item #3181175, was opened at 2011-02-14 19:59
Message generated for change (Tracker Item Submitted) made by megurikuru
You can respond by visiting: 

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: megurikuru (megurikuru)
Assigned to: Nobody/Anonymous (nobody)
Summary: clock_gettime nanosecond resolution

Initial Comment:
In the patch in attachments I've implemented the clock_gettime function.
This patch is related to my discussion on the forum and it's not ready for 
release yet.
It contains some bug for sure, due to the storage of time references in 
nanoseconds instead of microseconds.
Maybe it's not convenient to change the storage to save some multiplications, 
especially on <64bit systems.
A bug exist for sure, it's the "delms" parameter of [packet_enqueue(int 
dir,const unsigned char *buf,int size,int delms)].
It's of type int, while long or long long is needed. I've not changed this 
because I'm not sure about the type to use and I don't want to exaggerate.
Using wirefilter "flat" (without limitations from command line and/or from mgmt 
console) should work and you can test that for large bandwidth
the "counter" isn't used anymore (use the new "showqstaus" from the console if 
you don't belive me :) ).
Some other performance improvements are on the way.
Notice the patch contains previous posted patches.
To compile with clock_gettime, you will need a patch for configure.ac file. 
I'll attach this patch in another bug I'll post soon.


You can respond by visiting: 

The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE:
Pinpoint memory and threading errors before they happen.
Find and fix more than 250 security defects in the development cycle.
Locate bottlenecks in serial and parallel code that limit performance.
vde-users mailing list

Reply via email to