Article: Westmere Arrives
By: Vincent Diepeveen (diep.delete@this.xs4all.nl), March 17, 2010 4:14 pm
Room: Moderated Discussions
David Kanter (dkanter@realworldtech.com) on 3/17/10 wrote:
---------------------------
>I just finished the first of two articles on Westmere, the 32nm, 6-core shrink of
>Nehalem. This covers the improvements, including new instructions, minor microarchitectural
>tweaks and some basics on the products that are available:
>http://www.realworldtech.com/page.cfm?ArticleID=RWT031710140138
>
>The second piece will be a review that actually includes performance data. I'm
>still gathering the data, but this should be a nice short read.
>
>David
>
It is a very fast chip. For Diep if we look at performance numbers from 3.33Ghz clocked 980 part:
http://www.lostcircuits.com/mambo//index.php?option=com_content&task=view&id=79&Itemid=1&limit=1&limitstart=17
1.85M nps it gets.
If we take the Nehalem i7-965 number and multiply by 1.5 for moving up from 4 to 6 cores, that's 1247903 * 1.5 = 1.87M nps.
Achieved is 1.85M nps.
Now there is some loss in scaling as it has less memory controller per thread seen, to given an example. Note at 6 cores the software still scales near lineair, so that's not a problem at all.
So a very impressive show from intel there.
Note 130 watt per chip is quite a lot and $1663 is a tad much, but considering it's roughly 250 mm^2 i assume intel creams off market with this chip and can easily lower its price when needed.
---------------------------
>I just finished the first of two articles on Westmere, the 32nm, 6-core shrink of
>Nehalem. This covers the improvements, including new instructions, minor microarchitectural
>tweaks and some basics on the products that are available:
>http://www.realworldtech.com/page.cfm?ArticleID=RWT031710140138
>
>The second piece will be a review that actually includes performance data. I'm
>still gathering the data, but this should be a nice short read.
>
>David
>
It is a very fast chip. For Diep if we look at performance numbers from 3.33Ghz clocked 980 part:
http://www.lostcircuits.com/mambo//index.php?option=com_content&task=view&id=79&Itemid=1&limit=1&limitstart=17
1.85M nps it gets.
If we take the Nehalem i7-965 number and multiply by 1.5 for moving up from 4 to 6 cores, that's 1247903 * 1.5 = 1.87M nps.
Achieved is 1.85M nps.
Now there is some loss in scaling as it has less memory controller per thread seen, to given an example. Note at 6 cores the software still scales near lineair, so that's not a problem at all.
So a very impressive show from intel there.
Note 130 watt per chip is quite a lot and $1663 is a tad much, but considering it's roughly 250 mm^2 i assume intel creams off market with this chip and can easily lower its price when needed.
Topic | Posted By | Date |
---|---|---|
Westmere Launch article | David Kanter | 2010/03/17 01:27 PM |
Gulftown??? (NT) | MS | 2010/03/17 04:04 PM |
Gulftown??? | David Kanter | 2010/03/17 05:28 PM |
Gulftown??? | MS | 2010/03/17 06:24 PM |
Westmere Launch article | Vincent Diepeveen | 2010/03/17 04:14 PM |
Westmere Launch article | ? | 2010/03/20 03:18 AM |
Westmere Launch article (NT) | Matt Sayler | 2010/03/20 06:55 AM |
Westmere Launch article | MS | 2010/03/21 07:12 AM |
Westmere Launch article | ? | 2010/03/21 10:06 PM |
Westmere Launch article | MS | 2010/03/22 05:38 PM |
Westmere Launch article | anonymous | 2010/03/17 11:02 PM |
Westmere Launch article | David Kanter | 2010/03/18 12:21 AM |
Westmere Launch article | Rohit | 2010/03/18 01:40 AM |
Westmere Launch article | a reader | 2010/03/18 09:09 PM |
Westmere Launch article | David Kanter | 2010/03/18 09:30 PM |
Westmere Launch article | a reader | 2010/03/18 09:46 PM |
Westmere Launch article | David Kanter | 2010/03/19 09:39 AM |
Westmere Launch article | Rohit | 2010/03/19 11:16 AM |
Westmere Launch article | David Kanter | 2010/03/19 11:47 AM |
Westmere Launch article | a reader | 2010/03/19 07:55 PM |
Westmere Launch article | David Kanter | 2010/03/19 08:58 PM |
Westmere Launch article | a reader | 2010/03/20 08:23 AM |