[H] Core 2 Gaming Benchmarks

[Entropy]

It is also worth noting that during their last Pre-Conroe CPU article, AM2 review, HardOCP made a heavy use of low-resolution testing in the gaming section of the article. As they say, ladies and gentlemen, draw your own conclusions….

And as Dave Glue and others have pointed out, using only GPU-limited conditions to test a CPU is useless.

BTW, latest Source survey data show that less than 5% game at 1400x900 or above, and more than 60 % game at 1024x768 or below.

The Conroes do not just target enthusiast gamers because they are tested by gaming oriented sites, they target everybody. Good performance, lower power draw than AMDs offerings as well as Intels own will benefit everyone, and a reasonable price structure. The Conroes offer compelling benefits in all niches.

(Difficult to see what AMD can do for the next year. Just about anyone buying a new desktop and can afford an E6300 will gravitate to Conroe, and they have locked out their installed base of enthusiast socket 939 and 754 users from their immediately upcoming CPUs, forcing these customers to change their platform. A less than brilliant strategy under the circumstances.)

Entropy

 

[KimB]

(Difficult to see what AMD can do for the next year. Just about anyone buying a new desktop and can afford an E6300 will gravitate to Conroe, and they have locked out their installed base of enthusiast socket 939 and 754 users from their immediately upcoming CPUs, forcing these customers to change their platform. A less than brilliant strategy under the circumstances.)

Well, AMD really did need to change the socket type to move to DDR2, and now was the right time to move to DDR2 given AMD's hardware. If they didn't, they'd be even harder-pressed to compete with Intel's offerings. So I'd say that moving to AM2 now is likely to be the lesser of two evils.

Edit: And I don't think the Source survey really says anything about this situation. Most of those people also have slower video cards that would be more CPU-limited. It is highly unlikely that anybody upgrading from any Athlon 64 X2 would see any tangible benefit in today's games at the resolution they play at. But if you're upgrading from something old/slow enough, a Conroe would obviously be the choice to make (given current prices, the only hope for AMD to have a winner in price/performance would be to undercut Intel's lowest price). Benchmarks at lower resolutions, even if they don't accurately reflect what users will experience right now, are essential in showing this fact.

 

[Tahir2]

AMD have a response and it is 4x4 - two dual core CPU's on one motherboard with Quad SLI.

That is the plan, to win the performance benchmarks and give the impression that for absolute speed AMD is still the fastest. However if Intel execute well with Core 2 Duo, have cheap motherboards and plentiful supply then AMD is in trouble until it answers back with its own Core 2 beating CPU which could be over 12 months away.

 

[IgnorancePersonified]

I hope there's some thing else in that 2nd socket.

 

[Entropy]

Well, AMD really did need to change the socket type to move to DDR2, and now was the right time to move to DDR2 given AMD's hardware. If they didn't, they'd be even harder-pressed to compete with Intel's offerings. So I'd say that moving to AM2 now is likely to be the lesser of two evils.

But introducing AM2 is unrelated to not bringing out new CPUs for socket 939 (and 754). That is solely a marketing decision - and a marketing decision that looks remarkably ill adviced for AMD, since anyone moving from these sockets, whatever their motivation really, will find better offerings from Intel.

Edit: And I don't think the Source survey really says anything about this situation. Most of those people also have slower video cards that would be more CPU-limited. It is highly unlikely that anybody upgrading from any Athlon 64 X2 would see any tangible benefit in today's games at the resolution they play at. But if you're upgrading from something old/slow enough, a Conroe would obviously be the choice to make (given current prices, the only hope for AMD to have a winner in price/performance would be to undercut Intel's lowest price). Benchmarks at lower resolutions, even if they don't accurately reflect what users will experience right now, are essential in showing this fact.

I completely agree with your final conclusion.

I wanted to include the Source data to show that when we speak of gamers today, the "1600x1200 + candy" that hardware reviewers like to use routinely to show differences between video cards is pretty far removed from what people actually use, completely apart from the fact that it's idiotic for CPU reviewing. It is NOT the goal of everyone who buys a new system to game at those resolutions (or even at all). People who play games may after all use their computers for other things as well. Money is always an issue and a cost/benefit analysis may show that adding 250 bucks for another notch up in resolution just isn't worth it for most gamers. Source data overwhelmingly show that people tend to select low midrange to midrange graphics cards.

There is a large and seemingly growing gulf between hardware review site equipment and interests, and the actual market. The Source surveys is the best statistics available of what people who game online actually use, and most probably overestimates the level of hardware, given the genres that are represented.
And it shows what AMDs 4x4 is actually for - marketing. How many will ever buy it as opposed to having a review/testing sample? Hell, looking at the statistics, even Crossfire and SLI barely exists at all, their value being more as a tactical weapon for nVidia and ATI in the fight for chipset market share.

Just because the Conroes outperform the current high-end CPUs, that doesn't mean they will go into high-end systems. The price/performance/power landscape simply got redefined.

Entropy

 

[John Reynolds]

So what do you guys think of AMD's 4x4? Personally, I still think most hardware enthusiasts willing to spend that money would go instead with the Core 2 Extreme part, particularly if the system is for gaming.

 

[BRiT]

So what do you guys think of AMD's 4x4? Personally, I still think most hardware enthusiasts willing to spend that money would go instead with the Core 2 Extreme part, particularly if the system is for gaming.

Completely. Perhaps the best indicator of what hardware enthusiasts will do is obvious by looking at the XtremeSystems forums. Everyone and their mother have been selling off their AMD Socket 754 and Socket 939 equipment over the past month. Some already have the Conroe-compatible motherboards. They're just waiting on the actual CPUs to be available. There are plenty of posts talking about which model of Conroe to buy. There's plenty of talk about what extreme cooling equipment (H2O, Phase) is best to use too. Intel has indeed won over the enthusiasts market.

 

[Tahir2]

The way Conroe is performing even a 4x4 architecture may end up slower than Conroe..

My own experience is that the "average joe" customer that AMD has won over will take a lot longer to switch back to Intel - and they will only switch when their favourite game or application requires it.
I remember once, in the shop I work for, Doom3 was released and someone came in for an X800XT (IIRC) and a PSU - I said to him gleefully, you know that came didn't cost you £39.99 - it cost you £499.99.

My boss who was behind me gave me the most evil look... but the truth had been spoken

 

[Bouncing Zabaglione Bros.]

All things being equal, it will be pretty difficult for anyone with AMD brand loyalty to turn down the performance and price that Conroe offers. It might have been different a few years back when performance was more equal and AMD gave you a much better motherboard in the form of Nvidia's Nforce chipset, but now it seems that Intel will be able to offer a significantly faster, cheaper, cooler CPU with ATI and Nvidia chipsets giving us great motherboards.

AMD may be able to keep bragging rights with an expensive and exotic high end 4x4 config, but a person who has a set amount of money to spend on their new PC will look at Conroe and decide they are getting more for their money by going with Intel. It's looking unarguable at the moment.

[H] may try and twist that by looking at GPU bottlenecked scenarios, but most people use their PCs for more than just gaming, and in non-GPU bottlenecked scenarios, or future use (including with newer, faster GPUs), Conroe's performance is difficult to dismiss, especially given the very aggressive pricing.

Really, AMD have no one but themselves to blame for the spanking they are about to get. AMD took advantage with a great architechture while Intel went down a dead end, and then didn't follow up on their success. AMD eeked out as much money from their products as possible and failed to innovate while Intel gathered themselves back together.

 

[KimB]

The way Conroe is performing even a 4x4 architecture may end up slower than Conroe..

For games, yes. For applications that make full use of multithreading? No.

 

[digitalwanderer]

So what do you guys think of AMD's 4x4? Personally, I still think most hardware enthusiasts willing to spend that money would go instead with the Core 2 Extreme part, particularly if the system is for gaming.

Not sure yet, but I find it extremely weird in light of the fact that my daughter Maddy stuck a purple & yellow "4x4" sticker on Bubbles case a few months back without my permission that I just can't get off clean...

..this could fix that.

 

[Skinner]

[H] has Core 2 gaming benchmarks up, and........ Core 2 not all what its cracked up to be for gaming.

Why am I not surprised?

What did you expect in GPU-limited envoirements?
It's silly to bench Oblivion in 1600x1200 with all the bells with only one 7900.

Run those tests with X1900 CF and look again It will get even better with G80 or R600 nearing the low res differences.

 

[Acert93]

AMD have a response and it is 4x4 - two dual core CPU's on one motherboard with Quad SLI.

That is the plan, to win the performance benchmarks and give the impression that for absolute speed AMD is still the fastest.

Very few games have a noticable benefit from multi-core so 4x4, besides being a $2000 CPU solution, is not going to give AMD the performance wins in games they will be seeking. I think 4x4 is in for a very touch introduction in that a $300 Conroe will outperform it in most games. Ouch. Obviously future looking I like the dual socket idea and with multi-core catching on it seems to be a realistic option, but I don't think 4x4 is going to give AMD the immediate results they want.

 

[Baraclese]

Conroe is looking good... I like better performance/watt ratio and the great overclocking possibility. [H] benchmark is a joke. It's been a long time since I've last been hyped about a tech product. 4x4 'solution' is ugly, while Conroe is just slick.
Of course 4x4 solution is a nice welcome for enthusiasts that like to hook up as many monitors as possible to get a good view playing their flight simulators, but that's just not for me . I hope that AMD now doubles their efforts to regain the performance crown.

 

[KimB]

Er, 4x4 has no effect at all on the number of monitors you want to use....

 

[Entropy]

So what do you guys think of AMD's 4x4? Personally, I still think most hardware enthusiasts willing to spend that money would go instead with the Core 2 Extreme part, particularly if the system is for gaming.

I think the number of gaming enthusiasts willing to spend that amount of money amounts to less than a millionth of the total number of PCs sold in a year.

Shamelessly nicked:
http://www.amd.com/us-en/Weblets/0,,7832_8366_7595~110132,00.html - "AMD’s upcoming four-core, multi-socket platform is based on AMD’s Direct Connect Architecture, which is the next iteration of the enthusiast platform for AMD Athlon™ 64 FX dual-core processors."

So you're looking at roughly $4000 for CPUs + gfx-cards + MB + memory. Add the surroundings for a total of some $6000. That's a lot of money, power, and noise to pay for some more frames per second in Harry Potter. Now, I used to use graphics systems costing ten to twenty times as much, so the cost is peanuts in some applications, but - gaming is not one of them. Particularly as Conroe is probably going to make it look very silly in gaming benchmarks. Of course, something like this isn't bought for any rational benefit, but for e-penis reasons. Those who are into that kind of thing, and have access to the money to indulge themselves, don't seem to be the kind of folks who would appreciate buying something extreme and then winding up looking ridiculous even to their peers.

Besides, you have been able to build similar Opteron based systems sans Quad SLI for a couple of years now. And some have been sold. But not to gamers. Very nice workstations though, particularly for code that parallellizes well.

AMD 4x4 is marketing damage control. It's not even physical reality, and once it is - will anyone be interested in buying? I doubt it. Thus, I even doubt it will be produced beyond possibly a few samples spread to review sites. The idea is DOA.

Ironically, when A64 was known as "Hammer", I was very enthusiastic and positive about how their architecture would finally usher in the era of affordable and efficient multiprocessing on the desktop given the combination of chip interconnect and bandwidth scaling. Of course, AMD neutered that by segmenting off the multiprocessing capabilities to the Opteron line. Today, multichip multiprocessing on the desktop is pointless. We have sufficient silicon real estate to realize duals, and soon quads in a single package. Given the state of x86 code, seeing much benefit beyond dual cores is unlikely except for very specific codes. Beyond four - and we will be there in months through Kentsfield - no way. And since production cost will more than double with every doubling of the number of cores, I suspect legacy codes and Amdahls law together will ensure that the pie in the sky projections for number of cores per die in the future will face an unsympathetic market.

By the way, do the folks who run benchies with liquid nitrogen and cascaded phase change setups qualify as a "market" at all? (They sure as hell don't qualify as "users".)
Their influence on the computer market should be similar to the influence top fuel dragracing has on cars.

Edit: Actually, if you are interested in this kind of thing, the Opteron route is a better option, and a more cost effective one at that, since you don't have to go for the most expensive and highest power draw CPUs. I fail to see it making any sense for gaming, but as I said, they made nice workstations. Whether they are preferable to the new Woodcrest solutions is probably situational.

 

[swaaye]

Yeah and it's too bad too. I am fascinated by triple monitor gaming. I wonder why more people aren't, especially with the SLI hardware out there now.

 

[ANova]

On the subject of AMD's 4x4, if someone is really willing to spend that kind of money he could just as well get a dual Woodcrest Xeon based system. The Woodcrests are identical to the Conroe only they run at a 1333 MHz fsb.

 

[JCLW]

As soon as AMD brings out 4x4, intel will push out Kentsfield (if not before).

The guys over at Xtremesys already have Kentsfield chips running (two conroe cores on one package, a la Smithfield): http://www.xtremesystems.org/forums/showthread.php?t=103982.

Many people are claiming that the FSB will hurt intel at 4+ cores but so far multithreaded benchmarks are showing that Kentsfield actually scales considerably better then K8, at least up to four cores.

The other thing about the 4x4 idea is that you'll have to use a "Pro" version of a given MS OS, instead of the "Home" version ("Home" only supports one physical CPU, while "Pro" supports two). Not that it really changes anything, but it does bump the base price up.

 

[poopypoo]

Well, AMD really did need to change the socket type to move to DDR2, and now was the right time to move to DDR2 given AMD's hardware. If they didn't, they'd be even harder-pressed to compete with Intel's offerings. So I'd say that moving to AM2 now is likely to be the lesser of two evils.

i completely disagree, but it may be due to my customerbase. these are home ans small office users who want a quick upgrade on the cheap -- sure, they're willing to spend $150 on an upgrade to dualcore, but not willing to also buy a mobo and RAM -- for 2% performance difference!? ridiculous. to say nothing of reinstalling drivers/the possibility of things not working 100% right for a little bit. Now, the bulk of the market may be buying new Dells, I dunno, so maybe it makes sense to offer BOTH -- not just one luckily, for now, you can still buy leftover 939 stock...