131 CPUs Benched

[veri745]

so if I out run you, out swim you and out bike you 99% of the time, it would only be an opinion that I am the better overall triathlete? lol. must be nice to live in your world where facts are just opinions.

So I have a choice between two cars; both are completely adequate for my needs, but one is $3000 more than the other. Which is the better value?

 

[EnzoLT]

So I have a choice between two cars; both are completely adequate for my needs, but one is $3000 more than the other. Which is the better value?

wow you guys need to stop thread crapping. let the other person believe whatever they want to believe.

 

[toyota]

So I have a choice between two cars; both are completely adequate for my needs, but one is $3000 more than the other. Which is the better value?

what is wrong with some of you people? this isnt a car. this is a cpu and no one said anything about what cpu would be adequate as I am sure plenty of cpus could qualify for that. all I said was that compared to the P2 965, the i5 750 is the better overall performing cpu for the money. buy whatever you want to but it doesnt change that fact.

 

[Markfw]

what is wrong with some of you people? this isnt a car. this is a cpu and no one said anything about what cpu would be adequate as I am sure plenty of cpus could qualify for that. all I said was that compared to the P2 965, the i5 750 is the better overall performing cpu for the money. buy whatever you want to but it doesnt change that fact.

I agree. Truth is a valid defense. They are both fine, but the I7 is the better performer (but costs more)

 

[lopri]

wow you guys need to stop thread crapping.

For real. But I expected no less from the usual suspects.

 

[lopri]

On topic: Another thing I like about the bench is that it used recent versions of software. It looks like they tried to use 64-bit binaries where available, as well. The game choices may not look mainstream, but I suppose they wanted the most stressful ones for CPUs, so it makes sense.

 

[PsiStar]

I would like to see some Xeons added

 

[Jimbo]

So, if I'm interpreting (and then extrapolating) this right, my Q9650 overclocked to 3.6GHz, should have an overall performance index (page 14) of 156.6.

That puts it within 4.4 points or 3.8% of a stock 920.

I subtracted the difference in scores between a Q9650 and a QX9770 and divided by the difference in clock speeds. That gave me a difference per 100MHz in clock speed (2.4 points per 100MHz), then I multiplied that 2.4 by 6 (the 600 MHz difference between a stock Q9650 and my overclocked Q9650) and then added that to the base score of a Q9650. I didn't feel any adjustment for FSB was necessary because my Q9650 is now running at a 400 FSB just like the QX9770.

If any of this is wrong, feel free to correct me.

 

[Apocalypse23]

Awesome, bookmarked for reference.

^this

 

[lopri]

@Jimbo: I think your math is sound enough for rough extrapolating, but what is the difference per 100 MHz you've got? I looked at the chart and it gave QX9770 139.8 and Q9650 131.4. Since QX9770 is 3.2 GHz and Q9650 is 3.0 GHz, you could simply multiply the difference by x3 for 3.6 GHz.

 

[Jimbo]

I updated my post to show a difference of 2.4 points per 100 MHz of speed difference between the Q9650 and the QX9770.

 

[Makaveli]

Great article!

Thank you for the link.

 

[peonyu]

I wouldnt say AMD gets stomped by Intel since they are still price competitive, but its certainly not close like it was in the Athlon xp/P4 days. Hopefully AMD can get a chip out that is neck and neck with Intel for the sake of more options.

 

[lopri]

At first I didn't notice they also color-coded the graphs. For example;

Nehalem - Navy Blue
C2Q (Kentsfield and Yorkfield) - Cobalt Blue
C2D (Conroe and Wolfedale) - Blue
Pentium D (Smithfield and Presler) - Light Blue

It makes a lot of sense and helps understanding how the CPUs have been evolving for years. For example, we can see that AMD's 65nm was a total failure. It not only was slower than the competition but was also slower than its own 90nm predecessors. I always suspected that AM2 CPU's DDR2 controller was not up to par with the venerable Socket 939 CPU's DDR controller despite some hair-splitting reviews that attempted to show the new products in better light.

On Intel side, I can't help but pity my P-D 830 which sits in one of my drawers. That poor thing would throttle at stock speed, even with an XP-90, to eventual lock-ups.

 

[toyota]

I ran the GTA 4 built in benchmark at the settings they used and got 47 fps for my 3.8 E8500. then I realized they didnt use the standard benchmark scene. "We use the built-in benchmark but on a scene chosen by us for more weight than the default."

how do you use a different scene for the GTA 4 built in benchmark?

 

[PsiStar]

from toyota's comment ... isn't that a deal breaker for *any* comparison or dupication that others might do? Ergo, this is of no value other than for only what they tested ... old CPUs, to be harsh.

What about for future CPUs such as the next month or so? If the list is not promptly updated then it is of little value.

Time will tell.

 

[MrK6]

from toyota's comment ... isn't that a deal breaker for *any* comparison or dupication that others might do? Ergo, this is of no value other than for only what they tested ... old CPUs, to be harsh.

What about for future CPUs such as the next month or so? If the list is not promptly updated then it is of little value.

Time will tell.

There's always going to be new hardware around the corner; if enthusiast sites took this stance they'd never publish anything. This chart is invaluable for people looking to upgrade, especially on a price/performance ratio, as it gives them a "real world" comparison.

I especially enjoyed the gaming benchmarks, it definitely shows just how capable some current games are of bringing CPUs to their knees. Thanks for posting, lopri.

 

[ElFenix]

my opteron 165 @ 4800+ is looking a little short

i don't want to buy an 860 and then have next month come around with 32nm quads tho

 

[evolucion8]

AMD is looking very competitive compared to the Core 2 processors, it falls behind compared to the i7 CPU's but both are in different price points.

That chart reassured what I already knew, no need for a CPU upgrade yet.

 

[Oxaqata]

Edit

 

[ShawnD1]

Very nice, I especially liked the cumulative performance chart
http://www.behardware.com/articles/778/page14.html
Glad to see my Q6600 still holds its own pretty well.

That's crazy my super awesome E6600 is destroyed by about a 50% margin by the Phenom 9600. I think I paid like $300 when that Core 2 came out. I paid about $100 for the Phenom because Newegg was in a rush to get rid of them.

 

[toyota]

That's crazy my super awesome E6600 is destroyed by about a 50% margin by the Phenom 9600. I think I paid like $300 when that Core 2 came out. I paid about $100 for the Phenom because Newegg was in a rush to get rid of them.

you mean the 9650? the Phenom 9600 would probably be slower overall if it had the TLB patch. anyway the 9650 is only 36% better than the E6600 if my math is correct. the E6600 could easily be overclocked enough to remove that deficit.

 

[ShawnD1]

the E6600 could easily be overclocked enough to remove that deficit.

Not on stock cooling it can't. Stock settings run at about 60C, 3GHz at stock voltage runs at 70C. It doesn't go any higher than that without spending extra money on a special heatsink. Seems like the best settings are at 2800MHz, which is a 17% overclock. Phenom 9600 at stock voltage runs up to 2600MHz before hitting a wall, which would be a 13% overclock. Both processors consume a ridiculous amount of power, so I had to underclock and undervolt the Phenom just to prevent this computer room from getting so damn hot. The core 2 will remain overclocked because I play games on it.

 

[peonyu]

Im just nitpicking but you can regularly run a C2 at 75c and be perfectly fine. The BSOD from heat seems to showup at 80c with these chips [Core2duo, quads etc]

 

[toyota]

oops