will extra 512K cache improve gaming performance?
- Thread starter holycpu
- Start date
Not really. If there is an improvement it would be insignificant, about a 1-2 fps boost in games. The extra 512k cache is better for things such as video editing and encoding.
Check this link out:THG CPU Chart
Compare the 3800+ (2.4ghz 512k) and 4000+ (2.4ghz 1024k). In Doom3, the 4000+ beats the 3800+ by 2.1 fps.
..so yes, an extra 512k will improve gaming peformance. But 2.1 fps isn't much.
That's a horrible chart, they're bottlenecking somewhere, I'm thinking the GPU, there's no way in hell a stock 3000+ will bench 87 points lower than a 4000+ in 3dmark 2k5. If I OC my NC from 2.0ghz to 2.35ghz I get about a 2000 point difference.
yes, the extra cache will improve performance but the reason why it's discouraged is that the cost doesn't justify the improvement
More cache will always give you performance increases, but the amount that it increases by, isn't always that great.
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2249&p=9
Yes, but is it really worth the money? For me, it's $200 CAD for usually 5-10 FPS. However, in all likelyhood, I will opt for the 4000+ for that reason exactly.
Yes, but is it really worth the money? For me, it's $200 CAD for usually 5-10 FPS. However, in all likelyhood, I will opt for the 4000+ for that reason exactly.
Apparently, AMD thinks that the extra 512k cache translates to a 200MHz difference when it comes to processors. They are pricing the San Diego (which has an extra 512k cache) at the same level of the next higher level Venice, which runs 200Mhz faster.
I am sure you have all heard me rant the AMD pr rating are bogus and that a same speed chip is ranked 200+ points higher because it has 512 more of cache....The fact is the L2 cache does not deliver in MOST things....I did look at far better reviews then Toms...I looked at Xbits and AT and the results were about:
Doom 3 = 2.2-3.5%
UT2004 = 4.4-6.2%
HL2 = 8.2% (wow seen it in 3 diff reviews....)
Halo 1.05 = 1.8%
Far Cry = 2.3-2.6%
Quake 3 = 2-3% on one test and 7.1-7.5% on the other
Wolfenstein = 1.5-1.7%
The Sims 2 = 2.0%
Star Wars Battlefront = 0.7%
Battlefield Vietnam = 0.0%
Warcraft 3 = 0.0%
Hope that helps.....
IN other apps it rarely goes above 2% with the average within 0-2%...
Doom 3 = 2.2-3.5%
UT2004 = 4.4-6.2%
HL2 = 8.2% (wow seen it in 3 diff reviews....)
Halo 1.05 = 1.8%
Far Cry = 2.3-2.6%
Quake 3 = 2-3% on one test and 7.1-7.5% on the other
Wolfenstein = 1.5-1.7%
The Sims 2 = 2.0%
Star Wars Battlefront = 0.7%
Battlefield Vietnam = 0.0%
Warcraft 3 = 0.0%
Hope that helps.....
IN other apps it rarely goes above 2% with the average within 0-2%...
Originally posted by: Jeff7181
Originally posted by: g33k
How do you get a 2000 point difference in 3dmark05 with a 350mhz clock increase on your cpu?
You forget that 3DMark 2001 and 3DMark 2005 are different...
Yes I know they are different. 05 is much more gpu dependent and 01 is more cpu dependent. Icopoli stated that he scored 2000 more marks in 3dmark05 by overclocking his cpu by 350mhz.
Jeff7181
Lifer
- Aug 21, 2002
- 18,368
- 11
- 81
Originally posted by: g33k
Originally posted by: Jeff7181
Originally posted by: g33k
How do you get a 2000 point difference in 3dmark05 with a 350mhz clock increase on your cpu?
You forget that 3DMark 2001 and 3DMark 2005 are different...
Yes I know they are different. 05 is much more gpu dependent and 01 is more cpu dependent. Icopoli stated that he scored 2000 more marks in 3dmark05 by overclocking his cpu by 350mhz.
I was answering your question.
Well 200+ more points on average in the range of winchesters and clawhammer is usually 5%....As you can see it gets 5% on some games but average is sligtly below that...Outside of games it doesn't deliver anywhere near the 5% so ppl need to watch before they get suckered in on higher prices to pay for added cache that may not even help you......IN most instances that price diff is like 50+ bucks...So you yourself need to figure is 50 bucks worth a 5% more PR rating or apply it towards actually getting 200mhz more of cpu clock....I know which one I would want....
To be fair clawhammers seem to OC better as they are direct decendents of the FX models...
In games where you will actually need the performance difference (read: recent FPS), you'll usually see 2-5% performance in an A64 going to 1meg. Up to you if it's worth it. If you don't already have a 6800GT, x800 XT or better, I'd buy a cheaper 512k chip and upgrade my video card.
Originally posted by: Jeff7181
Originally posted by: g33k
Originally posted by: Jeff7181
Originally posted by: g33k
How do you get a 2000 point difference in 3dmark05 with a 350mhz clock increase on your cpu?
You forget that 3DMark 2001 and 3DMark 2005 are different...
Yes I know they are different. 05 is much more gpu dependent and 01 is more cpu dependent. Icopoli stated that he scored 2000 more marks in 3dmark05 by overclocking his cpu by 350mhz.
I was answering your question.
Haha. The techie equivalent of a yo momma joke, making someone put their foot in their mouth.
Originally posted by: Jeff7181
I was answering your question.
Well I don't know how you answered it, 3dmark01 is not even in that Toms' Hardware Article. It's 3dmark05 and 3dmark03. And how can you assume that he means 3dmark01 when he stated 3dmark05?
Originally posted by: Lithan
Originally posted by: Jeff7181
Originally posted by: g33k
Originally posted by: Jeff7181
Originally posted by: g33k
How do you get a 2000 point difference in 3dmark05 with a 350mhz clock increase on your cpu?
You forget that 3DMark 2001 and 3DMark 2005 are different...
Yes I know they are different. 05 is much more gpu dependent and 01 is more cpu dependent. Icopoli stated that he scored 2000 more marks in 3dmark05 by overclocking his cpu by 350mhz.
I was answering your question.
Haha. The techie equivalent of a yo momma joke, making someone put their foot in their mouth.
wth are you talking about?
Jeff7181
Lifer
- Aug 21, 2002
- 18,368
- 11
- 81
Q: How do you get a 2000 point difference in 3dmark05 with a 350mhz clock increase on your cpu?
A: You forget that 3DMark 2001 and 3DMark 2005 are different...
Because only 2001 would see a 2000 point increase from a 350 MHz increase in CPU clock speed alone.
A: You forget that 3DMark 2001 and 3DMark 2005 are different...
Because only 2001 would see a 2000 point increase from a 350 MHz increase in CPU clock speed alone.
Hmm well assuming he did mean 3dmark01...
3dmark01 and 3dmark05 are two different benchmarks. If Icopoli wanted to look at a more cpu dependent benchmark, he sould look at 3dmark03 on that chart. 87 in 3dmark05 does not mean the same thing as 2000 in 3dmark01. But I'll just let Icopoli answer that. Thanks Jeff7187
3dmark01 and 3dmark05 are two different benchmarks. If Icopoli wanted to look at a more cpu dependent benchmark, he sould look at 3dmark03 on that chart. 87 in 3dmark05 does not mean the same thing as 2000 in 3dmark01. But I'll just let Icopoli answer that. Thanks Jeff7187
Gotta agree with g33k here, the only reference Icopoli makes in his post is to '3dmark 2k5,' and the post seems to imply a 2000 point difference in that benchmark. He doesn't even have a 2k1 score listed in his sig. So, what do you mean Icopoli?
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 24K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter