KitGuru KitGuru.net – Tech News | Hardware News | Hardware Reviews | IOS | Mobile | Gaming | Graphics Cards
When Dirt 3 was released the developers pointed out that their advanced code took advantage of extra physical processing cores. Putting this to the test, we decided to investigate this further, by comparing an AMD quad-core and hex-core processor, on the basis of an upgrade.
DiRT 3 was only released about a month or so ago but has received a lot of praise from gamers and reviewers across the globe. It is the latest iteration of the Colin McRae Rally series, despite Codemasters dropping the Colin McRae branding. It supports DirectX 11 which enhances detail and brings a number of other visual enhancements to the gaming experience.
We will be comparing the AMD Phenom II X6 1090T processor which is clocked at 3.2 GHz to the AMD Athlon II X4 635 processor which is clocked at 2.9 GHz. To make the test as fair as possible we overclocked the Athlon II X4 635 processor to 3.2 GHz, keeping the clock speeds identical. We wanted to focus on real world CPU's today to show differences in upgrading to six cores.
As we weren't able to adjust the multiplier on the Athlon II 635 processor, we boosted the baseclock from 200 MHz to 221 MHz, giving us a final clock speed of 3215 MHz using the standard multiplier of 14.5x. Please bear in mind that the Athlon X4 system has a slight advantage as the memory is overclocked a little when the baseclock is increased to 1774 MHz from the 1600Mhz of the Phenom X6 system. Additionally, there are some architectural differences between the processors so its not a purely scientific test, but as a topic of interest we thought it would be worth a look.
Moving onto the testing itself, we used exactly the same test system for the two CPUs. We used the latest Crosshair V Formula motherboard from Asus combined with a reference AMD Radeon 6950 graphics card and 4GB (2x 2GB) of Kingston HyperX Genesis memory @ 1600 MHz.
The test system was built inside a Corsair Obsidian 650D case and was powered with a Thermaltake ToughPower XT 750W power supply. Microsoft Windows 7 Home Premium (64-bit) was configured onto a Intel 320 Series 160 GB SSD for testing with all the latest drivers installed.
We used the graphics settings displayed in the screenshot below for testing.
When using the Phenom 1090T processor in our system there was a 14% performance increase over the similarly clocked Athlon 635 CPU, highlighting that the 2 additional physical cores are making a difference, with the level 3 cache playing a part also. Some people still question whether there is really a need for six cores at the moment but this test shows that they do give you a performance bump. This is currently looking to be a good upgrade move and an indication of things to come.
KitGuru says: Having six cores may boost your gaming performance but we look forward to seeing how eight cores perform when Bulldozer is released!
Very interesting. codemasters have clearly got some nifty coders to utilise all the threads. nice.
Good game too, best in the series yet I think. shame they dropped the mcrae name from it. I understand the reasons, but it would have been nice to keep his name alive over the years with qualtiy games like this.
nice idea for someone with an older CPU to see the differences. my buddy has an older 4 core chip. Wonder if this will make much difference over the coming year.
interesting.
I suppose cores could have been disabled, but as a test it shows the benefits of a real world processor and upgrading to a new system. my brother runs a slower X4 overclocked, so basically even overclocked its still slower than a reference X6…. if speeds are the same.
A pity it wasn’t a Phenom to Phenom comparison. The Phenoms have an L3 cache that the Athlons don’t, and this could account for some of the difference. I searched a little for benchmarks, and for example at one Anandtech test the Athlon X4 635 @ 2.95 got 40.2FPS on Far Cry 2, while a Phenom X3 720 @ 2.8 got 45FPS.
I think its interesting. I know its just a ‘clock for clock’ and there are technical differences with architecture (it says that in the article and that its not meant to be a purely scientifc analysis).its more a ‘what if I upgrade to an X6 and keep the clocks the same’.
Sorry, that was meant to be Athlon X4 635 @ 2.9 (GHz), not 2.95. Another thing is that if only few cores are used, then the Phenom II X6 would use Turbo Core to up the speed. Therefore for example a game that uses only two cores would end up much faster on the Phenom because it would run at 3.6GHz with an L3 cache when compared to the 3.2GHz Athlon. This might not be the case here, but this point needs to be addressed.
In short, not a great comparison.
Here’s a comparison of how the L3 cache affects gaming performance: http://www.tomshardware.com/reviews/athlon-l3-cache,2416-6.html
Left 4 Dead gets 20% better performance with the L3 cache, so 14% in DiRT 3 could easily be the result of only the cache.
Am i missing something, isnt this an upgrade article showing the benefits of moving from an older X4 to a new X6 at the same clocks. I have read on anandtech that the cores do make a difference. I doubt level 3 cache is giving an extra 10 fps at the clocks.
Anyway its been a fairly interesting read, perhaps codemasters could teach other companies how to actually code a fricking game.
I think xbit did an article once on level 3 cache in gaming. it doesnt make a huge difference (whereas it does with rendering and some other floating point processes), depending on the status of the DirectX code. I think it was more prevalent with DX9 games.
I dont think this is meant to be scientific going on the text in the article, which is fine. looks like an interesting upgrade article, showing that a user with an older OC’d cpu at 3.2ghz gets a pretty huge increase, even running stock with a new X6 CPU.
this review is utterly pointless and false. the review used an athlon II based CPU against a phenom II? did AMD pay him for this to help push the myth of six core cpus as gaming CPUs? (proved false by every web site out there!)
the review should have been done using a phenom II X955 vs 1090T
techspot (a more competent web site) did a Dirt 3 performance review showing the phenom II 980 vs 1100T having equal FPS while both outdistancing themselves from the athlon II x4 645
@MIke: All reviews on KitGuru are independent. To be honest, I think we’re victims of our fixed-length headlines for stories. The full headline could have been “Upgrade: Does DiRT 3 play better if you upgrade a mainstream 4 core part to a modern 6 core processor?”.
We’d ask you to focus on the conclusion. The result was really straight forward…
“When using the Phenom 1090T processor in our system there was a 14% performance increase over the similarly clocked Athlon 635 CPU, highlighting that the 2 additional physical cores are making a difference, with the level 3 cache playing a part also”.
The 1090 is a valid upgrade for gaming with products like Dirt 3.
Do you disagree?
Well it was useful to read for me, ive an older 4 core athlon which i overclock to around 3ghz. This has shown me if i buy a newer tech 6 core at 3ghz it will be faster with dirt 3.
Which was the point of the article. I dont think people read the text right and just make their own assumptions. I dont see any claims of an architectural clock for clock comparison, its for upgraders…. Its even in the title!
Thanks for the clarification, Faith. I’m still a little disappointed with this review because I didn’t get enough real info about the subject, but I guess that a better title would have made things easier to swallow.
In this particular case I don’t much see the benefit of such an upgrade, since GRiD 3 is getting a minimum of 60 FPS on the overclocked Athlon.
Well its not meant to be an indepth analysis or ‘review’ ET. its a single page with one test showing possible improvements for a user with a last generation AMD CPU who might be thinking of changing to a newer design. Dirt 3 was at hand and it was used, due to Codemasters commentary on their code in the past.
Obviously with Bulldozer around the corner, we expect the 6 core prices to drop soon so people could end up with a bargain, and tangible gains. The pricing is already good.
If you think its a waste of time, then we take the negative feedback on board and can look at these ‘quick, one page’ articles which we thought were interesting reading for people who might have a 5 minute break in work or at home, for some ‘quick’ reading. We always like to try new things.
Sadly we don’t have enough space front page to put ‘If you currently own an older four core AMD processor and overclock, is it worth trading up to a new 6 core design? Today we have a look at Dirt 3 to see how they match up at the same clock speed. *caveat – this is not a direct head to head due to architectural differences between the designs.
Obviously if people read the text properly it might make more sense, but next time we will try and spell it out very carefully.
Perhaps you could take the phenom II X6 and disable cores and show a simple graph of processor scaling for Dirt 3? Windows can set processor affinity can’t it?
Zardon, I’m sorry, I’m a certain type of reader, it’s possible that this article is fine for others. I’m always looking for articles which will provide me with new information, and based on the title this looked promising, which was why I was disappointed.
The problem isn’t with the title or the length of the article, rather the content. It should have been possible to write an article of the same length which really tests what’s described in the title. That would have told how well DiRT 3 scales with the number of cores.
As is, the article might say that upgrading from an Athlon X4 to a Phenom X6 will provide a benefit, but that’s something I’d have considered a given. I would have been surprised if there wasn’t some difference, even for a game that doesn’t take advantage of 6 cores. So while I certainly agree with your thoughts about the Phenom X6, and I’ve already bought an X6 to replace my X3 (and should get it any day now), I don’t see this as an article that would affect such a buying decision.
So my take is that the idea of this article is good, and the length is fine, it’s just that a better test could have been made to test the particular subject of DiRT’s scaling. The subject of upgrading to a Phenom X6 could use a different article, perhaps a longer one.