Sign in with
Sign up | Sign in
Your question

Are these benchmarks accurate?

Last response: in CPUs
Share
February 5, 2010 11:07:18 AM

Anand just put up some new CPU benchmarks, such as for 7zip and a few new games. I was looking at the 965 and the i5, and some of these numbers just seem so off to me. In games, I've never seen a benchmark where these two CPus have been more than a few FPS apart at normal resolutions.

At 1680x1050, Anand has the following:

Dragon Age ... i5 winning by 30%
DoW 2 ... i5 winning by 20%
WoW ... i5 winning by 24%

The only difference in rig setups is that the 965 is on 32-bit Vista, while the i5 is only 64-bit. Could that possibly make a difference that large? Any other benchmark I've seen at 1680x1050 has the two CPUs very, very close in those games.

Plus in some benches, they have the PII x2 555 (3.2GHz) beating the PII x4 965 (3.4GHz). Now, I understand that some games don't utilize 4-core rendering, but the 965 still has a clock speed advantage. How would it even be possible for the 555 to win?

More about : benchmarks accurate

a c 131 à CPUs
February 5, 2010 12:11:31 PM

There is not enough of a variation between the 555 and 965 in the benchmarks where the 555 wins to warrant an actual "win" over just variation.

As for the i5 and phenom II that does seem a bit odd. However, I find the game benchmarks rather useless since they don't tell us the rest of the system setup they were tested on.
a b à CPUs
February 5, 2010 12:42:19 PM

the i5 is more for gaming out of the box, but if you OC the Ph II to high freq, it just nukes the i5, if you oced i5 vs Ph II, you will find that the i5 can be faster for games and only games, while the Ph II, being a true quad core would win at almost everything else, and it would make you be able to multi task better
Related resources
a b à CPUs
February 5, 2010 12:49:06 PM

theholylancer said:
while the Ph II, being a true quad core would win at almost everything else


:heink: 

Im pretty sure that the core i5 750 is a "true quad core" as well.
a b à CPUs
February 5, 2010 1:13:13 PM

Anand is an Intel shill and whenever possible he will bend and twist benchmarks to make them look better.

A few years ago Anand's site was first rate.

Since he sold out to Intel I wouldn't trust anything posted on the site in terms of benchmarks.

In saying that, the i5 benchies for a number of games are stellar, and they make a great gaming rig ... just not the best, or the best value for money ... as they are still overpriced in many areas.

Once the volume increases ... the prices will drop ... they will become better value for money.

Thats just my view.

Note any benchies or reviews at THG from our German editors here are usually Intel biased as well.

The US / UK reviewers here are very balanced.

a b à CPUs
February 5, 2010 1:31:58 PM

*** are you thinking?? comparing highly oced proc to a stock speedish 965 when they can hit 3.8 under air easy?!

btw toms you ate my post again with the censor

here is the post of them comparing a i5 500 at 4 GHz vs a 965 at 3.4 when they should be at least 3.8 if not 4

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=372...

the full post actually pointed out the differences between the chip and the insane comparison between a highly oced chip vs a lightly oced chip... i hate the censor thing on here
a b à CPUs
February 5, 2010 1:37:24 PM

What MOBO and what RAM did they have the AMD on ? - they tend to put them on a crippled AM2+ MOBO with less and slower DDR2 RAM instead of using an AM3 MOBO with the same DDR3 RAM as the i5.

On the x2 955 - I believe they unlocked the cores on that chip successfully so it may be that the bench result you are looking at is the unlocked x2 955 making it perform like a P2 X4 !
a b à CPUs
February 5, 2010 1:42:09 PM

JDFan said:
What MOBO and what RAM did they have the AMD on ? - they tend to put them on a crippled AM2+ MOBO with less and slower DDR2 RAM instead of using an AM3 MOBO with the same DDR3 RAM as the i5.

On the x2 955 - I believe they unlocked the cores on that chip successfully so it may be that the bench result you are looking at is the unlocked x2 955 making it perform like a P2 X4 !



whaaa amd don't make Ph II X2 955, there is only a X4 955 and a X2 550 and another few things

a b à CPUs
February 5, 2010 2:12:48 PM

theholylancer said:
whaaa amd don't make Ph II X2 955, there is only a X4 955 and a X2 550 and another few things


was a typo meant X2 555 (which unlocked shows up as a X4 B55!! From the first article I saw on Benches at the site since there wasn't a link in th OP

Quote:
The chip also performs just like a 3.2GHz quad-core Phenom II, because it is one at this point:
Processor 1st Pass 2nd Pass
AMD Phenom II X4 965 72.1 fps 22.2 fps
AMD Phenom II X4 B55 70.6 fps 21.1 fps
AMD Phenom II X2 555 45.2 fps 10.9 fps

February 5, 2010 2:57:01 PM

Quote:
What happens when you overclock the I5 to 4+Ghz?

Thats right. The X4 gets it's ass kicked because it can't compete clock for clock

http://www.anandtech.com/bench/default.aspx?p=109&p2=85


In those benchmarks, the i5 was running at 3.2GHz due to turbo boost, while the Phenom was running at 2.66GHz. If you want to accurately compare clock speeds, turbo needs to be disabled. It's a great feature, but it definitely needs to be off to accurately compare the i5 to other CPUs in terms of clock for clock performance.

As for the benchmarks the OP posted, they definitely do appear to be off. The i5 may be a bit ahead at stock speeds, but definitley not by that much. The rig setups were similar (you can see them by viewing any specific benchmarks individually). The only difference is that the Phenom was on Vista 32, while the i5 was on Vista 64. That almost certainly wouldn't make a huge difference in gaming.
a b à CPUs
February 5, 2010 3:00:02 PM

Psycho know that full well Atranox, he's just trolling with his usual garbage 'clock per clock' comparison, which is actually a large clock speed boost for the i5 due to turbo.
a b à CPUs
February 5, 2010 5:20:12 PM

i think the i5s and what nots may actually have a ipc advantage (from some time ago when they all did a same 3.6 ghz comp with no turbo and etc), but the issue is that we are talking duals vs quads, no matter how much ipc advantage you have, you will NOT win, if you crank a i5-600/500 to 4.5 Ghz (the max I have heard on air with standard chips) or a Ph II X4 to 4 Ghz (max i have heard on air with standard chips) the Ph II X4 will kick butt

compare a i5-600 to a 555 BE locked in at 2 cores, and i'd say your are not comparing the right price point and should be an i3 vs 555/550 BE, when you want to say well if i3 is oced it gets top, i say B55 + oc and end it there.

the low / mid end for intel is just miserable atm, if you are going to get a i5-750 and oc the hell out of it, then great, its a very good platform and it's going to be close to a i7-920, if you went with a i7-800, i'd slap you and say spend the tiny extra and get a i7-920 with a cheap board and oc to 4 Ghz, both solutions works and can hand amd it's own butt, but not at anywhere near the price point of a Ph II X4 955.
a b à CPUs
February 5, 2010 11:42:59 PM

+1 to the holy guy.

His passionate expose pretty much sums up my view.

I wouldn't recmmend a dual core anymore for gaming ... now the newer titles are really making better use of the extra cores.

Unless it is a cheapy.

Then making silicon scream on an E5/ E8 series is far easier than those i3 / i5's and much cheaper ... albeit EOL.

a b à CPUs
February 5, 2010 11:59:37 PM

It is most likely that the games on the hardware and resolution tested(which is low by the way) it is more cpu bound. All three of those titles you mentioned have the FPS continuing to go up with the higher end Intel's. The Intel's are faster when GPU is not the bottleneck.
a b à CPUs
February 6, 2010 1:06:33 PM

Quote:
Here we go with the "Turbo is wrong" bullshit from the amd lovers. Don't get mad because amd's too technologically retarded to add it to there processors.


Don't think anyone was saying TURBO is wrong they are just saying to downclock an AMD chip to run at a slower clock to supposedly get a clock for clock comparison while keeping the Intel Chips Turbo mode which OCs the CPU to a different clock rate makes the comparison no longer clock for clock - so either rerun the tests with the Turbo mode disabled so they both run at the lower clock rate or up the AMD clock back to the actual clock rate the Intel chip is running and you'll get different results !!

As it is it is like your ISP implementing Turbo Boost to your download so that when you run a speedtest it shows 20Mb\s because the first few seconds of your download are boosted only to have it slow to around 1Mb\s speed on your 5Mb\s connection after that so your small download takes forever but you can't complain because the line tested to speed according to our testing methods !! (but then since your an Intel fanboy you probably believe the speedtest and ignore real world results anyway !!)
March 18, 2011 1:04:00 PM

B55 means an unlocked 555 Black edition.

The owner of the CPU unlocked it to a Quad Core. The 5xx Series black edition CPUs can be unlocked to a triple or quad core, a little risk with a large reward.
a b à CPUs
March 18, 2011 3:14:56 PM

Aecursis said:
Anand just put up some new CPU benchmarks, such as for 7zip and a few new games. I was looking at the 965 and the i5, and some of these numbers just seem so off to me. In games, I've never seen a benchmark where these two CPus have been more than a few FPS apart at normal resolutions.

At 1680x1050, Anand has the following:

Dragon Age ... i5 winning by 30%
DoW 2 ... i5 winning by 20%
WoW ... i5 winning by 24%

The only difference in rig setups is that the 965 is on 32-bit Vista, while the i5 is only 64-bit. Could that possibly make a difference that large? Any other benchmark I've seen at 1680x1050 has the two CPUs very, very close in those games.

Plus in some benches, they have the PII x2 555 (3.2GHz) beating the PII x4 965 (3.4GHz). Now, I understand that some games don't utilize 4-core rendering, but the 965 still has a clock speed advantage. How would it even be possible for the 555 to win?


yes they are but it will be different result by an extreme user, you can beat that benchmark depending on your own build system.
benchmarks are just guides, not a bible.
!