Cyberpower’s Gamer Dragon: Can AMD Bring The Game?

[lolwuut]

It appears the only bias involved can be found in the zealous AMD defenders.
I love and have loved AMD, and I'm all for the underdog. I still love the memories of my old K6/2 400mhz with 96mb ram and an ATI rage 128 + Voodoo2. Since then I've always been looking first at AMD as that company holds a special place in my heart.
The unfortunate truth is that right now AMD is in a bad way, and while they have a few chips that are best bang for the buck in their respective price points, if I'm going to be spending between $250-$300 for a CPU, it's going to be the i7 920 hands down. Don't get off topic with motherboards and ram and a bunch of stuff that negates a chip vs chip debate.
Some will say the i7 architecture is a rip off of AMD's, maybe so, but do I care? They did it right, that's what matters in the computer world. Did Microsoft invent the mouse? No, they bought it from Xerox. Did Apple sell the first MP3 player or build the first smartphone? It's the nature of the industry, but kudos to AMD for a fantastic design, even if the mark was missed.
The fact of the matter is that the i7 920 even at stock speeds just destroys both the stock and the overclocked PII setup, even when the PII is equipped with bigger "guns" per say. The gaming tests would have been a complete joke had the i7 system been equipped with two 4890s, but as it stands it won even with inferior graphics hardware. There is simply no arguing this point, as much as we love AMD.
Are Tomshardware.com editors biased? No. They completely acknowledged that the PII platform does severely dominate several CPU price points, just don't try to say that this is one of them, please.

 

[philosofool]

This is coming from a guy who's currently looking at a Phenom II 720 triple core from AMD for his next buy: I will be the first one to say that a controlled experiment to test these two processors should use the same graphics cards, but I think the evidence here pretty clearly shows that there's some bottle necking at the processor because there's no significant change with a change of resolution with the Phenom machine. That indicates that the graphics sub-system is not the bottle neck; if it were, reducing the resolution would increase performance, which we do not observe. Yes, a perfectly controlled experiment would have grabbed the pair of 260 GTXs and slapped them in the other machine, and it's especially sloppy to have the graphics systems different when they're SLI/CrossFire because of the known issues with SLI/Crossfire scaling. However, I think the evidence here points to the processor, not the graphics, as the source of weaker performance.

 

[megamanx00]

Seems to me like overall the games benched were indeed biased towards NVIDIA cards which kind of portrays the system more negatively. Oh well, still an interesting article. Yikes on that power draw. I'm hoping that the mid range 40nm DX11 cards will offer at least 4890 performance with much less power draw ad I don't want to be replacing my 500W blue storm II. Not only that, think of the power bill, ouch .

 

[ssguy24]

I've read several reviews in the past over different websites and I've NEVER seen an AMD 955 system perform so poorly as the one today. To say I was flabbergasted would be an understatement. I can't say why this system did poorly, but I can say that it's the first 955 system I've seen to so consistently get sub 100fps over such a variety of titles.

 

[meatwad53186]

[citation][nom]superhoss[/nom]I don't concur. Let's not forget that the supposed "cheaper" option is the AMD build. Right? Yet, Tom's i7 build here is actually significantly cheaper.I think the best conclusion you can draw from all this, after applying a healthy dose of common sense, is that you can spend a little bit more on the i7, less on your graphics subsystem, and STILL get superior performance. In my estimation that's nothing but a sterling recommendation of Intel's offering.Brand loyalties have nothing to do with this decision.[/citation]

And why is the i7 build cheaper? Because they built it themselves. As everyone else said, its just not fair comparing a crappy boutique built PC to a hand built rig.

 

[dirtmountain]

Missing the point? No, it was a sloppy article based on one of the best SBM builds compared to an inferior Cyberpower system. Since when is a 3.2GHz stock clocked CPU overclocked to 3.6GHz count as "impressive"... except in this slop article.

"With the next SBM a couple months away, is there a better way to find out if the Phenom II’s price advantage over the Core i7 will allow it to excel, thanks to that stronger graphics system?"

Yes, build a decent $1,300 AMD system and actually do a fair comparison.

 

[jassao]

the phenom II is no competition for the i7, the i7 is in a different level (and price), the 2 4890's were bottlenecked by the phenom because of the architecture and whatnot the system is running.

if you put those ati card in xfire in the i7 920 system then the difference in performance would have been even greater.

so if you want to get a good price/performance ratio, go with a phenom, oc it, put a high end card and DDR2 ram.
But if you have the money, get the i7 920, oc it, put 2 ati cards in xfire and let it fly.

if you get more than that set up (like crazy quad-sli/xfire configurations, $999 i7, server-like ram amounts, etc) then you're paying a lot more, for just a marginal increase in performance (gaming wise). Just my opinion.

 

[falchard]

Whats with the game selection? All nVidia optimized.

 

[witcherx]

This article is fudged and biased.

http://www.tomshardware.com/reviews/phenom-x4-955,2278-9.html
This article is fudged and biased.

 

[superhoss]

[citation][nom]meatwad53186[/nom]And why is the i7 build cheaper? Because they built it themselves. As everyone else said, its just not fair comparing a crappy boutique built PC to a hand built rig.[/citation]

I do agree with that on principle, but in all reality, both machines were hand built. In fact, this Cyberpower rig was overclocked past "factory" settings, which suggests that they gave it a little extra TLC.

You can definitely find weaknesses in this build if you compare it component by component to something you yourself would hand build. For example, it was suggested by one commenter that a better mobo would help, and possibly more RAM. However, I don't believe that these deficiencies would account for the substandard performance. I believe it would be even more noticeable if they had put a couple 260's in the Cyberpower machine.

However, I do agree that the test could have been more thorough. I'm disappointed that it wasn't, because it allows these types of arguments to flourish.

After pricing components for the build I'm currently planning, I can't justify going with an AMD system, when the i7 is comparably priced and superior in a large cross-section of testing that I've seen. Unless, of course, it's a vast conspiracy and all of that testing was compromised and invalid.

 

[Guest]

After comparing the two articles (http://www.tomshardware.com/reviews/phenom-x4-955,2278-10.html, and this one), it seems like the numbers make sense, with the exception of the WIC tests, where the single gtx 260 outperforms the two 4890's. Seems like the crossfire drivers are to blame for that one.

 

[Schip]

It is a well known fact that the i7 outperforms PII in many (most) scenarios. It is also well known that in certain scenarios the PII matches (and occasionally outperforms) i7. So performance is mostly cut and dry. The tricky thing is value, being subjective by definition. Value is usually thought of as price vs performance which is a very difficult thing to quantify due to the wide variety of software and individual computer usage. The thing for the individual to do is identify their needs, then identify the best value. For extreme/high-end gaming (at least 1900x1200, very high details, etc) and media encoding (and other multi-thread optimized apps) The best way to meet great performance is with an i7 platform. But maybe your not interested in the absolute highest details or heavy encoding, then maybe the PII holds a higher value, because it can cost less. How much less OBVIOUSLY depends on the setup of each system, which again would depend on your needs.

 

[Guest]

2 gigs of ram at cas 9 vs 6 gig at cas8
16 meg cache hd vs 32 meg cache hd
8x x 8x crossfire vs 16x x 16x sli
HMMMM??????

 

[Guest]

Objective: Do a full-fledged article about gaming benchmarks, but only benchmark a measily 4 games, because those games will all favor Nvidia... Throw in video encoding(because it favors the i7), synthetic benchmarks(because those have always favored Intel, even in the P4 days), that Excel MonteCarlo benchmark(uber-rigged)....

Conclusion: AMD Sucks

 

[playerone]

I dont trust these numbers at all.

Something is not right here, Compare this TH review (as well as most other sites)of AM2 955 vs I7:

http://www.tomshardware.com/reviews/phenom-x4-955,2278-9.html

Whats up with those Farcry numbers Tom? You should pull this review until it can be explained.

Is I7 a stronger CPU? Of course, but these numbers Do Not look as if settings where on a par with one another. Not everyone understands display driver interaction, hey who knows what it is, but these numbers do not add up when stacked against so may other reviews, even right here on Tom's

 

[chaohsiangchen]

[citation][nom]ssguy24[/nom]I've read several reviews in the past over different websites and I've NEVER seen an AMD 955 system perform so poorly as the one today. To say I was flabbergasted would be an understatement. I can't say why this system did poorly, but I can say that it's the first 955 system I've seen to so consistently get sub 100fps over such a variety of titles.[/citation]

It is known fairly well that Core i7 doesn't do much better in single card, single GPU set up in games. In some cases, they are actually worse. Where Core i7 shines is when you put SLI/CF configuration which multiple cards/multiple GPUs stretch CPU power to the limit. Tom's staff have done that test months ago.

AMD is also put into disadvantage by compilers.

 

[Proximon]

I spent some time tracking down some reviews, and it seems the Cyberpower system is gimped somehow. I'm seeing some better numbers for CF 4870 with an X4 940@ 3.8, on a 790FX.
I think someone needs to have a look at these 790X boards a bit closer.

 

[Guest]

Only 4 games to test a gaming intended system ,only 4!!!
Where are the tests with COD 4 ,COD W@W ,Call of Juarez,UT 3,Prince of Persia,GTA 4 and so on ?

 

[oldscotch]

[citation][nom]lolwuut[/nom]It appears the only bias involved can be found in the zealous AMD defenders.I love and have loved AMD, and I'm all for the underdog. I still love the memories of my old K6/2 400mhz with 96mb ram and an ATI rage 128 + Voodoo2. Since then I've always been looking first at AMD as that company holds a special place in my heart.The unfortunate truth is that right now AMD is in a bad way, and while they have a few chips that are best bang for the buck in their respective price points, if I'm going to be spending between $250-$300 for a CPU, it's going to be the i7 920 hands down. Don't get off topic with motherboards and ram and a bunch of stuff that negates a chip vs chip debate.Some will say the i7 architecture is a rip off of AMD's, maybe so, but do I care? They did it right, that's what matters in the computer world. Did Microsoft invent the mouse? No, they bought it from Xerox. Did Apple sell the first MP3 player or build the first smartphone? It's the nature of the industry, but kudos to AMD for a fantastic design, even if the mark was missed.The fact of the matter is that the i7 920 even at stock speeds just destroys both the stock and the overclocked PII setup, even when the PII is equipped with bigger "guns" per say. The gaming tests would have been a complete joke had the i7 system been equipped with two 4890s, but as it stands it won even with inferior graphics hardware. There is simply no arguing this point, as much as we love AMD.Are Tomshardware.com editors biased? No. They completely acknowledged that the PII platform does severely dominate several CPU price points, just don't try to say that this is one of them, please.[/citation]

I like AMD.

As someone who likes AMD, I am in no way whatsoever disputing the i7 920 outperforms any desktop processor with an AMD label on it. And with the price of the i7 920 vs. the not-much-cheaper phII 955, not to mention the affordability of the more modestly priced and featured LGA1336 motherboards - it's hard to argue that AMD is offering more for the money. At least at the ~$300 CPU price point.

Coming down from that though, Intel doesn't have anything that competes against the value and futureproof(ness?) of a DDR3/x3 720 setup. For less than the i7 920 alone, you can get an ASUS AM3 790Fx board with a x3 720. Core i5 may well change all this, but for now the x3 remains a fantastic deal as well as simply being an interesting processor.

Anyhow, the point from the perspective of a self-labelled objective AMD user is that there's something wrong with this setup. The testing is likely sound, in as much as it's well established that one machine is definitely outperformning the other. Forget about intel/amd, boutique/homemade, ati/nvidia, $1700/$1300 - one of these machines simply works better and that is firmly established with these tests.

The reason for this performance disparity that some, well many, have suggested is what I disagree with. Tom's own tests have shown that when the GPU is the (likely) bottleneck, there's very little difference between the AMD and Intel platform - with the edge usually going to the i7.

http://www.hardocp.com/article.html?art=MTY0NCwxLCxoZW50aHVzaWFzdA - now, this test, which I unsuccesfully tried to link earlier, demonstrates that when the GPU bottlneck is lessened by a fair bit (~$1000 for the cards and power supply to handle all that) - the difference between the i7 and phenom II platforms becomes more pronounced, this time with the i7 more consistently edging out phII and by larger margins. However, this difference is much less than what we've seen in this article. Not only that, but here we're also looking at a tighter GPU (dual-SLI/CF vs. quad CF)

This leads me to believe that something else has caused the larger than expected performance gap, whether the CF wasn't implemented properly, whether there was a bad card, whether the power supply wasn't enough, whether there was a driver problem, chipset or lord knows what - heck, it could have been a bad processor. Regardless, I don't agree that the rather large disparity here is representative of what one can expect from the AMD platform.

 

[oldscotch]

Oh, and for those that say the point of this article was missed, it was titled "...Can AMD Bring The Game?" not "...Can Cyberpower Bring The Game?"

 

[kevin1212]

I have nothing against the article really, but clearly in the conclusion, they are addressing the ones backing up AMD, saying why they prefer i7... but they still seem biased.

 

[Guest]

AMD should just pack it in and go home. Loser CPUs for losers!

 

[Guest]

Why oh why aren't the two Radeon HD 4890 OVERCLOCKED?? 850 mhz gpu... when 1 ghz is (almost) easily attainable... That would have made one hell of a difference in game benchmarks.

Yes. I understand you might want to do an apples to apples comparison... but it isn't: you're comparing a system with OC'd CPU to a system with OC'd CPU AND GPU, and another without any OC...

Just a thought

 

[balancedthinking]

I

 

[balancedthinking]

What have you done to the AMD system to cripple it this way? I just can not believe that you post those numbers before you even think about that something must have gone terribly wrong.

Far Cry 2 THG 955BE @ 3.2ghz OLD Test with ONE GTX 260:

1920X1200 high No AA NO AF 74.5 fps

GPU limited setting

http://www.tomshardware.com/reviews/phenom-x4-955,2278-9.html

Far Cry 2 this Test 955BE @ 3.6ghz with TWO 4890s

1920X1200 very high NO AA NO AF 62 fps

Do you realy want us to believe that switching from high to very high makes that much a difference? Even though the high setting was already GPU limited and now we have TWO cards that on their own are already more powerfull than that one in the older Test?

Intel must have paid shitloads of cash to make this review happen.