- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I have a client whose performance is drastically slow on an AMD 64 system. I've read online that there are manual tweaks to force the loading of processor-specific libraries at runtime to override the "auto detect" features of IPP. Can anyone recommend a course of action or research?
If we can't get IPP to run at least as quickly as on a PIII, then unfortunately, we'll have to switch to another image processing solution, which would be a bummer because I really like it.
Thanks much,
Matt
If we can't get IPP to run at least as quickly as on a PIII, then unfortunately, we'll have to switch to another image processing solution, which would be a bummer because I really like it.
Thanks much,
Matt
Link Copied
1 Reply
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Matt,
what version of IPP do you use? We got a lot of requests from customers to dispatch optimized code on base of detected processor features (for example SSE). In answering that IPP v5.1 will dispatch appropriate optimized code on AMD processors.
Regarding manually dispatching particular processor-specific code - it is simple: you need to link with IPP static libraries and call ippStaticInitCpu function with specifying desired processor-specific optimization. Note, you will get hardware exception if you will try to use absent features (for example, trying SSE code on processor which supports only MMX instruction set).
Regards,
Vladimir
Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page