| June 27, 2009 9:00 AM PDT | |
The Intel IPP v6.0 and later version supports the latest Intel® Core™ i7 processor (codenamed "Nehalem"). There are several Intel IPP string processing functions like ippsFind*Any() functions and tranformation functions that are specially optimized for Intel Core i7 processors for additional performance benefits . All Intel IPP functions will continue to use “p8” optimized libraries for IA-32 and “y8” optimized libraries for Intel® 64 when you target Intel Core i7 processors. The “p8” and “y8” optimized libraries in Intel IPP are generally optimized for Intel® Streaming SIMD Extensions 4 (Intel® SSE4).
For all complete Intel IPP supported cpu identifiers, please refer this article or check “Getting_Started.htm” or “userguide_*.pdf” from IPP \doc directory.
To find out more Intel IPP APIs performance results on Intel Core i7 processor, you can run Intel IPP Performance Test tool on this target system. The tool is available under IPP directory \tools\perfsys.
Please check "readme.htm" in this folder and also an article at Intel IPP Knowledge Base for more information.
A lot of Intel IPP samples like Audio/Video sample, new Unified Image codec (UIC) sample aslo provide performance results for decoding/encoding as part of output data. Please check Intel IPP Web site and click Sample link to download.
This article applies to: Intel® Integrated Performance Primitives Knowledge Base, Intel® Parallel Composer Knowledge Base
For more complete information about compiler optimizations, see our Optimization Notice.
Comments (16) 
November 9, 2008 1:41 AM PST
 Ray Moore Jr. |
Having perused other posts from this author, I believe my confusion was correct, in that there is either a kaput semi-colon key by Ying's keyboard or it is missed for the comma. Back to you. |
| November 19, 2008 9:54 AM PST
Ying Song (Intel)
|
Thanks for the feedback, we corrected the typo. |
| November 20, 2008 1:25 PM PST
Pink Anderson |
If I want to learn more about: 1) Xeon 7400, (Darlington) 2) Atom - Menlow 3) Intel I7 - Nehalem EP are their White Papers, Webcasts, newsletters, and especially Videos on Intel.com you can point me to? P Anderson |
| November 23, 2008 11:34 AM PST
eweber
| Both peryn and nehalem family of processors seems to use the same optimized library (p8 or y8). Is there actually some Nehalem specific optimization? |
| March 29, 2009 11:43 PM PDT
Ying Song (Intel)
|
Several string processing functions are specifically optimized for Nehalem based processor. In the latest IPP 6.1 beta via Intel Parallel Studio beta, there are few more signal processing functions additionally optimized on Nehalm ( Intel® Core™ i7 processor) |
| June 17, 2009 8:48 PM PDT
Vinz |
Hey guys, Nehalem works great for v6.0, but for existing users with IPP v4.0, does it work well for Nehalem? Any performance degrdation? |
| June 17, 2009 10:51 PM PDT
Ying Song (Intel)
| Your application building with Intel IPP v4.0 can still work on Nehalem and it would use the last optimzied code offered in v4.0, but it wont offer you the maximum performance on the latest Intel CPUs. That's why we continually provide the up-to-date optimizations for the latest Intel CPUs via each release. I would recommend you check out the latest version of Intel IPP for best performance. By the way, check the Intel IPP 6.1 via http://software.intel.com/en-us/intel-ipp which will be available shortly. |
| November 30, 2009 1:52 AM PST
matthieu.darbois
| eweber seemed to have a valid comment which lacks a clear response (I guess that Ying Song (Intel) is one). So here it is again : Both peryn and nehalem family of processors seems to use the same optimized library (p8 or y8). Is there actually some Nehalem specific optimization? If yes, why the same optimized library is used. Not enough difference to make another library ? This means that additional dispatching is used in the optimized library ? |
| February 11, 2010 2:56 AM PST
loosers | weak |
| October 10, 2010 10:49 PM PDT
Lamp
| The profermance is terrible when work with i7-870 in multiple-thread enviroment. |
| October 10, 2010 11:36 PM PDT
Ying Song (Intel)
|
Hi xiehuanxie, Do you have more specific performance data and Intel IPP version that we can try to reproduce the issue you reported here? I noticed your similiar report at http://software.intel.com/en-us/forums/showthread.php?t=77893&o=d &s=lr and we will exchange information via that forum thread. Thanks, Ying |
| October 11, 2010 6:08 PM PDT
Lamp
|
hello Ying, Sorry to trouble you here, we met a problem to use IPP work with i7, but seems it's really hard to get the root cause; I'll test it more, to replace some function call or use DMIP. By the way, if there is no 'p8' dlls, IPP will choose 'v8' for i7, the performace does not have any big difference. Regards, |
| October 12, 2010 1:59 AM PDT
Lamp
|
Hello Ying, finally I cached the method causes the delay: IPPAPI( IppStatus, ippiCrossCorrValid_NormLevel_8u32f_C1R, ( const Ipp8u* pSrc, int srcStep, IppiSize srcRoiSize, const Ipp8u* pTpl, int tplStep, IppiSize tplRoiSize, Ipp32f* pDst, int dstStep )) Params(pSrc, 96, {96,96}, pTpl, 26, {26,26}, pDst, 284) I already submitted an issue to the Premier Support, you can test it if you are interesting. |
| October 12, 2010 7:16 PM PDT
Lamp
|
Here is the solution: http://software.intel.com/en-us/articles/performance-of-cryp.....g-systems/ |
| October 12, 2010 8:34 PM PDT
Ying Song (Intel)
| Thanks for the notice, and I will let our support engineer know your finding. |
Trackbacks (0)
Leave a comment 
Ying Song (Intel)
|
Ray Moore Jr.
Performance benefits for the i7 in addition? I be lost.