Does Cache Size Really Boost Performance?

Sidney · 25th October 2007, 02:37

The next Core 2 generation based on the 45-nm Penryn core will even carry up to 6 MB of L2 cache. Is this just marketing bluff or does the swelling of L2 cache capacities indeed lead to better performance? Let's find out.

http://www.tomshardware.com/2007/10/...ter/index.html

jmke · 25th October 2007, 08:37

interesting would be to find out at what point extra L2 cache no longer makes a difference, 6Mb will still give a nice boost

Rutar · 25th October 2007, 12:05

I think, considering cache costs a lot of money, Intel and AMD know exactly why they give the chips a certain amount of cache.

jmke · 25th October 2007, 12:29

ignoring price, what would be the saturation point?

Kougar · 25th October 2007, 13:11

Saturation point couldn't be that much higher than 12MB? There is only so much available bandwidth / time available for prefetching use before anything extra would be moot. Intel's 65nm Tulsa had a 16mb L3 cache... anyone ever do any cache analysis with that one?

Everything will change with CSI though, I feel more cache will continue to be added with the 32nm Nehalem, aka "Westmere" since CSI will allow faster prefetching ability.

25th October 2007, 02:37	#1
Sidney [M] Reviewer Join Date: Mar 2004 Posts: 16,730	Does Cache Size Really Boost Performance? The next Core 2 generation based on the 45-nm Penryn core will even carry up to 6 MB of L2 cache. Is this just marketing bluff or does the swelling of L2 cache capacities indeed lead to better performance? Let's find out. http://www.tomshardware.com/2007/10/...ter/index.html __________________ lazyman Opteron 165 (2) @2.85 1.42 vcore AMD Stock HSF + Chill Vent II

25th October 2007, 08:37	#2
jmke Madshrimp Join Date: May 2002 Location: 7090/Belgium Posts: 82,497	interesting would be to find out at what point extra L2 cache no longer makes a difference, 6Mb will still give a nice boost __________________

25th October 2007, 12:29	#4
jmke Madshrimp Join Date: May 2002 Location: 7090/Belgium Posts: 82,497	ignoring price, what would be the saturation point? __________________

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
EVGA Gives Away ''Free Performance Boost'' to its GTX 460 Users	jmke	WebNews	0	6th August 2010 16:58
ATI's latest Catalyst driver reduces IQ for performance boost?	jmke	WebNews	1	7th April 2010 16:10
Expect 22.8% performance boost from next week's Firefox 3.6 beta	jmke	WebNews	0	15th October 2009 15:05
Samsung to boost SSD performance 30 times with PRAM	jmke	WebNews	0	8th May 2009 15:32
ATI Catalyst 9.2 Delivers Nice Performance Boost	jmke	WebNews	0	23rd February 2009 16:29
NVIDIA Forceware 177.92 Beta Driver Promises Performance Boost	jmke	WebNews	0	26th August 2008 17:14
Intel Core 2 Duo Performance - L2 cache Sizes Compared	jmke	WebNews	2	10th September 2007 07:14
Intel Set to Boost Performance of Celeron D Chips	jmke	WebNews	0	15th May 2006 23:24
Performance Impact on Changing Memory Size from 1GByte to 2 GByte	Sidney	WebNews	4	30th August 2005 20:45
ASUSTeK to Make Add-In-Cards that Boost Physics Performance in Games	jmke	WebNews	0	18th May 2005 16:34

25th October 2007, 12:05	#3
Rutar Member Join Date: Apr 2005 Posts: 2,267	I think, considering cache costs a lot of money, Intel and AMD know exactly why they give the chips a certain amount of cache.

25th October 2007, 13:11	#5
Kougar Member Join Date: Jun 2006 Posts: 1,640	Saturation point couldn't be that much higher than 12MB? There is only so much available bandwidth / time available for prefetching use before anything extra would be moot. Intel's 65nm Tulsa had a 16mb L3 cache... anyone ever do any cache analysis with that one? Everything will change with CSI though, I feel more cache will continue to be added with the 32nm Nehalem, aka "Westmere" since CSI will allow faster prefetching ability.