Reviews and pricing for 4x4 are out comparing it to Kentsfield. Here are some of them:
http://www.firingsquad.com/hardware/amd_athlon_64_fx_74_4x4/default.asp
http://enthusiast.hardocp.com/article.html?art=MTIzMyw1LCxoZW50aHVzaWFzdA==
http://www.tomshardware.com/2006/11/30/brute_force_quad_cores/page13.html
http://www.hothardware.com/viewarticle.aspx?articleid=911&cid=1
I'm not the right guy to dissect the technical details but there are two things I was expecting:
1. This will be a power hungry, large and loud beast of a system. Which is correct.
2. That this would be a hard kick in the performance cojones for Kentsfield. Which to my astonishment is not correct.
From a pricing standpoint...this thing is priced extremely agressively (in pairs):
2 x FX70 = $599
2 x FX72 = $799
2 x FX74 = $999
Considering the FX62 is at 713$, this is quite a steep price drop. However, the issues with the system price overall is the cost of the mobo and the fact that there is only one supplier - Asus. My hypothesis is that AMD knows at this point Quad FX is only a marketing ploy. It is not a market winning solution until they bring out Barcelona. Hence, they do not want to enable all their partners at this stage until they have a better sense of whether they want to market Barcelona by itself or as part of Quad FX.
Having seen that the performance is not spanking Kentsfield, I think Intel will now accelerate their quad core ramp. They will bring pricing further down in Q107. Their intent will be the same as it was on dual core. To make quad core mainstream faster than expected and hence place supply and capacity pressure on AMD. Their hope will be AMD cannot ramp Barcelona fast enough and Quad FX is hardly a mainstream desktop solution. Hence, customers (starting with channel I guess) will once again feel AMD is unable to meet their needs.
So do expect Kentsfield to drop below 999$ in Q1. I'd assume this will be accompanied with a new SKU with a higher clock speed just to land a 1-2 punch on AMD.
----------------------------------------------------------------------------
UPDATE
Anandtech in his review shows a Quad COre at 2.4GHz in Q1 priced at $851. So Intel will bring Quad Core mainstreamish fast.
http://anandtech.com/cpuchipsets/showdoc.aspx?i=2879&p=3
Intel Core 2 Extreme QX6700 2.66GHz 4MB per 2 cores $999
Intel Core 2 Quad Q6600** 2.40GHz 4MB per 2 cores $851
Intel Core 2 Extreme X6800 2.93GHz 4MB $999
I suspect the scenarios AMD is playing with right now are:
1. If Barcelona kicks Conroe's performance and perhaps more importantly performance/watt ass then 4x4 will die an unnoticed death.
2. However, if Barcelona does not completely wipe the floor with Conroe. Even if it's better but only marginally better then AMD will increase the focus on 4x4 and try and bring platform price and thermals down. They may even introduce dual core flavours of Barcelona (not sure if this is already on their public plans) to drop into 4x4 just as Intel is potentially introducing single core versions of Conroe for the lower end of the market.
Of course, everyone (including me) are assuming Barcelona will wipe Conroe off the map. But I was also expecting 4x4 to do the same to Kentsfield and that is not the case so let's assume nothing till we see it for ourselves.
Subscribe to:
Post Comments (Atom)
36 comments:
You may be right they still do have the option of releasing a quad core extreme edition. But there are several other factors to consider before doing so and AMD’s competitiveness is only one aspect of it. I’m inclined to think that Intel at this point is more interested in optimizing profit rather than humiliating AMD any further. They’ve already made AMD eat their own words about power efficiency, forced to show a “simulated” quad core demo and most of all, sell their top binning parts at a BUY1 TAKE1 free price giveaway.
I think the other reason to bring in another Kentsfield SKU is to ensure they don't vacate the 999$ price point even as they introduce something lower. So the 1-2 punch makes them feel good...but there's another more market competitive motive to do that too.
cheers...
Well, no doubt AMD should be embarassed by this dog. I'm surprised they released it. But maybe there's no such thing as BAD press?
I this this was more of a technical oddity to prove AMD is still around and working on multicore processors. Certainly no one is going to buy this processor (unless they feel they have something to prove).
In the summer of 07 when multicores are more mainstream this should be a fun tech race if AMD can get its act together and get back in the game.
There's a lot of talk going around that the apps where AMD has a severe performance deficit to the Intel system were the better threaded applications, which meant that with AMD's NUMA system, they could, and constantly did, switch cores, thus saturating the HT bus, and increasing latency. Vista would fix this, but the only benchmarks I've seen in vista sucked.
Even then this is hypothetical, but when you look at the apps where there isn't a strangely huge deficit, the performance is actually pretty much on par or slightly better. Ya, it's an inferior system, but did Intel just not introduce dual core when the resulting product was really hot, and much slower than AMD's solution? Of course not, this is just as much a PR stunt as Intel's quad core, and isn't really a worthwhile solution yet. HardOCP notes that the entire platform AMD presents, is pretty much on par with Intel's due to the insane options the 680a chipset provides. Plus, this solution more looks forward to a time of quad core processors that could be plugged into it quite easily.
Both AMD and Intel are trying to give themselves headroom for improvement on their current tech, so that they don't waste all the architectural improvements they've gained immediately. That's why AMD's 65nm parts and Intel's core parts aren't at higher clockspeeds, even though I'm sure they both easily overclock to them.
Greg you are comparing a single socket pentium D drop in solution to a complete system revamp which is the 4x4. Yea pentium D was hot but compared to 4x4 it is sitting on ice. ....
4x4 is a PR stunt and not a viable quad core solution dont put kentfield in the same category since it gives you quad core performance which is better than 4x4, lower power consumption and drop in upgradability.
And please link to me the site that you got the 65nm overclocking benchmarks cuz i have yet to find a single one. Could it be another paper launch by AMD???
After the PR murder, it seems they are going to be priced even more competitively..
http://www.theinquirer.net/default.aspx?article=36194
Still don't see them on Newegg.. Paper launch8)
Since when does anyone buy an inferior platform in the hope that it'll be better in almost a year's time? :D At best it'll be, twice as fast as four cores but still hot and noisy.
Well, being that they're SHIPPING on the fifth, it's pretty retarded to think they'd be available for purchase that day, or immediately afterwards. I'm assuming they'll be up by monday, and they just didn't bother putting up review samples.
Yes, compared to the pentium D, the 4x4 is extremely hot, but I'm not comparing 4x4 to the pentium d (you really need to reread my post please).
Being that kentsfield will only be 3% of Intel's production "soon" I'm guessing they aren't planning to make a whole lot of money on that yet. Thus, it's a PR stunt. If it gets good PR directly from customers, even better.
What makes you think I have benchmarking sites? I don't. I'm just pointing out, that with the higher transistor performance offered by 65 nm to both AMD and Intel, they could utilize it for higher clocks, but don't, so that they have plenty of head room for newer products in the future. You can already assume AMD will overclock much better with their 65 nm process because it's a smaller process, and because of the white papers they release to the public concerning the individual transistor performance. This is what yields or impedes overclocking (not only heat, as so many idiotically claim). This is why AMD lags behind Intel in terms of overclocking, even though when overclocked to its limit, an AMD processor is still cooler than some stock Intel processors, because Intel has faster transistor technology.
I'm not saying people will necessarily want to buy a product just because of what it can be upgraded to (though there will be those that inevitably will). I'm saying that AMD is establishing a product line, and getting the support for it bit by bit, so that when quad core and other such items are out, the 4x4 system is already established enough for serious consumer recognition.
Also, considering that hardly anyone except for enthusiasts (the small majority) hear about this, and that AMD is actually receiving positive PR from another company (Dell, and now HP), which will actually be noticeable by the masses, I'd say that AMD is having a net gain in PR right now.
The 130 to 90 nm die shrink for AMD yielded how much in terms of faster parts? 400 MHZ about 2 years AFTER the switch to 90 nm. I predict, based on what I've read, the VERY FASTEST K8 on 65 nm tech will be clocked at 3.2 GHZ. This part will not be available initially - it will be the "last of the breed" on AM2 before a socket and architecture change.
It will still be slower than the X6800 released last August.
The 90 to 65 nm die shrink yielded 400 Mhz (PDP840 @ 3.2 versus PDP960 @ 3.6, I think) and about 5% overall less power consumption in terms of maximum envelope.
AMD has proven that their current K8 on 90 or 65 nm tech just does not have a lot of headroom to move the clocks by more than about 400 Mhz. There are many folks that are gettin 1+ Ghz overclocks ON AIR with current Intel product.
And the Intel scales pretty darn linearly, unlike old Netburst.
TWO Words: They paniced..
Hey guys, did you notice Sharikou retracted his latest waco story!
Hahaha nice to see you post this 180sharikou..... notice how scientia after throwing a big fit about me being anonymous and not requesting it on his site hasnt said anything about the 4x4 on the site. Well i dont blame him.... hard to defend garbage even if you are a fanboy.
Well amd fans are used to being duped by their beloved company. Remember how amd was touting the benefits of 64 bit and released the athlon64 saying that vista was around the corner almost 4 years back.....yep all those computers are futureproof THANK YOU AMD.
First anonymous. The 90nm to 130 nm process switch was really hard on both AMD and Intel. AMD took a really long time to get it out the door, while Intel rushed it and ended up with voltage leaks. Both were calculated, as Intel was able to move onto 65nm faster because it had its 90nm architecture out the door, and AMD had a cooler process.
Your second argument (if you can call it that) actually need evidence please.
Architectures have absolutely nothing to do with clockspeeds (except for what performance increases clockspeed increases yield). As I pointed out, the only thing that effects this is transistor performance, which yes, Intel has better, but AMD is improving their's very quickly as well.
??? What part of what scales linearly. Please explain, as I have no idea what you're saying.
Second anonymous, he who is a fool thinks he is wise, while he who is wise knows he is a fool. So quit thinking the people that run these companies are idiots. They know what they're doing better than we'll be able to for quite some time and they've known about what performance improvements would be yielded by Intel's processor plans. AMD is run by engineers you know (and they kinda know what would yield what) and Intel has quite a few engineers up at the top ranks, so it's not like they can't see where either will be in the next year or so.
Anonymous 4, no computer is future proof, especially brand new cutting edge computers. Seeing as we have no idea exactly how well this board will support quad core processors, and how well the platform can handle it (in terms of not having ridiculous power consumption as the number of cores increase) we have no idea how future proof it is. Same with kentsfield. If Intel comes out with IMC and CSI when they said they would initially, then ketnsfield is just about as future proof as 4x4 (since both will almost certainly need a new socket, not to mention new chipsets, and voltage lanes). Also, seeing how their current voltage control system doesn't work on kentsfield correctly (at least as considering how they said it would work), chances are their next set of processors will need a new chipset or a new socket.
Ya, AMD used their customers, Intel used their customers, every company used their customers, your friends use you, and your family uses you, so get over it because that's the way the world works.
By the way, the whole anonymous # thing is getting really annoying, so please get names if you're going to post on here alot or use your google account, because it makes replying to you much easier, much less confusing if a comment gets deleted, and only takes a couple seconds (sorry if that's too much work for you red).
notice how scientia after throwing a big fit about me being anonymous and not requesting it on his site hasnt said anything about the 4x4 on the site. Well i dont blame him.... hard to defend garbage even if you are a fanboy.
The problem with Scientia is he is bounded by truth and logic when he tries to make an argument. Unlike Sharikou who seems to be free from these basic rule.
Here's a good summary of what AMD is in rigth now.
Greg, I don't see the point of them saying "Hey everyone we're 'shipping' these parts!" Need I say more?:) At the very worst, this 'launch' would've at least let them say that they got 65 nm out in 2006.. But where are they?
We should know soon enough when AMD can find enough cherry picked overclockers:D.. I wouldn't bother too much with those technical papers and numbers about transistor performance..
http://www.theregister.co.uk/2005/12/07/amd_ibm_stressed_silicon/
Really, that 40% improvement is just fluff to me.
Quad FX being vapordud..
Barcelona running task manager with 40%/70% numbers[Oh Scientia, what happened to that 95% hehe]
'OEM launching' 65nm with no benches..
Barcelona to top out at 2.5GHz according to HKEPC, as opposed to previously rumored 2.9GHz..
Great press:D
Greg: Yes, compared to the pentium D, the 4x4 is extremely hot, but I'm not comparing 4x4 to the pentium d (you really need to reread my post please).
You are wrong in your analogy firstly intel was the first to bring dual core to the market by three months before x2s came out, infact they were in the retail chains with the pentium Ds a good 5 or 6 months before the first x2s even appeared, it was not released after and you are comparing a pentium D processor to a whole platform like the 4x4. Whatever point you were trying to make was wrong.
Greg:Being that kentsfield will only be 3% of Intel's production "soon" I'm guessing they aren't planning to make a whole lot of money on that yet. Thus, it's a PR stunt. If it gets good PR directly from customers, even better.
Yep kentsfield is a PR stunt. I guess the upcoming Q6600 is also a PR stunt. Infact every processor intel is coming out with is one big marketing PR for them. ARE YOU NUTS!!! Kentfield is drop in 775 upgradable which is a very realistic solution for people looking to do some hardcore multitasking THIS WAS NO STOP GAP REBRAND OPTERON SOLUTION LIKE THE 4x4. Their announcement for the 4x4 rite after conroe whooping the k8 should be more than evidence for you.
Greg: What makes you think I have benchmarking sites? I don't.
Dat is exactly my point you are throwing assumptions on your part around as if they are facts. Yes 65nm will be a better overclocker but it wont be anything like intel. If their die shrink lets you overclock on stock air from 2.8ghz to 3.3ghz that is not a very big deal and if you dont know what kinda headroom you will be getting dont talk about it soo authoritatively since there is NO PROOF, NO ES, NO BENCHMARKS, NOTHING WHATSOEVER YET!!!. Infact looking at AMD's roadmap they are not planing on shifting their high bin chips to 65nm till very late in the year so dont get too excited yet about that die shrink equaling to higher stock clocked amd chips.
Greg: the 4x4 system is already established enough for serious consumer recognition.
Yes it will have established itself as AMD's dumbest move. THe k8L better have awesome power saving features because even if it beats kentfield the power consumption on this beast is ridiculous not to mention the lack of an eight core processor need in the mainstream market.
Face it 4x4 is the only one that is a PR move to downplay the conroe's release. No matter how you cut it AMD released a dud and you cant spray cologne on sh*t and expect it to smell good. Infact AMD has yet to do anything significant after releasing the X2s and i hope for their sake that the k8L is all that they say it is otherwise AMD is in for some roughtimes.
Of course, everyone (including me) are assuming Barcelona will wipe Conroe off the map. But I was also expecting 4x4 to do the same to Kentsfield and that is not the case so let's assume nothing till we see it for ourselves.
actually by the time, the Barcelona is not to compared with Conroe but its 45nm derivative which rumor to top at 3.73GHz and with full SSE4. though luck :)
Anonymous, good job ignoring the content of my previous post concerning transistor numbers (which I said favored intel). I guess we should just ignore numbers and only go based on a qualitative experience like Intel tells us to (core2, the "best" processor?).
AMD doesn't get to decide what overclocking numbers get published and hasn't ever really seemed to favor any reviewer over others, so I'm not getting where you're finding this "cherry picked" stuff from.
"Vapordud" (I'm assuming you mean a failed vaporware) is something meant to be sold to consumers that never becomes widely available. Right now it's way too early to assume quad fx is vaporware, much less 65 nm.
I don't know, maybe getting people's attention so they start to look for information on 65 nm is a good enough reason to announce shipment? Seems like sony, nintendo, and microsoft all picked up on that when they announce their shipment dates (or don't in the case of the ps3).
Yes, good job HKEPC. Yet again, they show their ability to be even less reliable and more speculative than even charlie. Oh, let's look at his latest post on Dell's AMD orders. Woops. If anything, starting the chips at a lower speed shows greater confidence in the efficient of the architecture. Like I said before (again, thanks for reading my comment carefully) they release lower clockspeed chips than they might necessarily be able to achieve so that they don't max out the potential of their current process and transistors.
Also, to comment on the update, assuming they'd drop 4x4 just because of its performance right now, seems really shortsighted 180, and I expected better of you. Seeing as the r690 has been announced to be 4x4 capable, and that ATI's chipsets are by far more power efficient than Nvidia's (not to mention the fact that there probably will only be one of them) this should lower power consumption quite a bit. I could also see AMD moving its FX line into a lower power budget with 65nm so that it doesn't seem so ridiculously wasteful to get a 4x4 system.
My question is, how likely are we to see Intel parts as cheap as they are supposed to be? The 6700 is selling for $1200 right now on newegg and retailed for $1500 on release. If AMD has as low of demand as you guys think it will, chances are it wont have that big of a royalty placed on it. But that remains to be seen.
hey sharikou 180, any particular reason why you think that K8L will outgun Conroe?
Before I go on to point out an obvious falsehood, I wanna say that I agree with a lot of what this person said. Especially the part of us being amatuers in comparison to those entrusted to run the multi-billion dollar Co.'s in question. And the proof is the fact no one has corrected this statement yet.
"Architectures have absolutely nothing to do with clockspeeds (except for what performance increases clockspeed increases yield). As I pointed out, the only thing that effects this is transistor performance, which yes, Intel has better, but AMD is improving their's very quickly as well."
Architecture has everything to do with clockspeed, and that's not to say the quality of transistors have nothing to do with it. But it was architecture (short pipeline) that caused the PIII to struggle to get beyond 1Ghz until in later iterations, better transistors made up for it. It was architecture that allowed the PIV to do 3Ghz+ while at the same time it had worse transistor performance than AMD. And its the architecture of Conroe that caused clockspeeds to drop despite having better transistors and less leakage than they've had in a long time.
A couple of things:
Greg - the reason I think AMD will de-focus from 4x4 if Barcelona is a Conroe killer is linked to a statement I've made before. With the ATI acquisition, AMD has to manage their cash & debt carefully which means they have to make some hard choices on what to resource and what not. At that point, I said projects like PIC might die. And it has. Similarly - what AMD needs is to re-take the performance crown and be able to take that product mainstream while making good margins. It is far easier to do this with a simple CPU (Barcelona) vs a complete platform (4x4). However, it will consumer a lot of their resources if they need to enable and market both Barcelona and 4x4. Hence, depending on Barcelona performance I think they will make a choice. Someone else appropriately pointed out they will need to compete with Penryn which will come after the Barcelona launch. So I would expect 4x4 to continue to be hyped by AMD till well into 2007 but for them not to put their entire weight behind it till we see performance results from Barcelona at which point they decide what to do. Remember - they have to keep watching Intel's performance because it's obvious that Conroe has a fair amount of clock speed that can be squeezed out.
One other thing - I'm expecting low demand for 4x4 in it's current version. However, at the inflection point above, if AMD decides to put their resources behind 4x4 in the marketplace then I expect them to make significant improvements to thermals, etc.
Anonymous - why do I think Barcelona will beat Conroe. Two reasons - It is a newer architecture. Secondly - integrated memory controller. But I'm not an expert so I could be wrong.
Ashenman - I don't think we will see 65nm versions of Quad FX chips very quickly. As AMD ramps their 65nm, I think they will first commit this capacity to server and mobile where they can increase their volumes and margins in order to maximise profit. my guess is it will be at least late Q1 or perhaps Q2 before they would even announce a 65nm version of Quad FX and availability would probably be spotty to start with.
The issue of efficiency is only a factor as long as AMD is still on 90nm. AMD will be mostly 65nm in another two quarters.
Intel's Kentsfield easily beats 4x4 in terms of SSE but not in any other category. For memory intensive code 4X4 is faster than Kentsfield. Neither Anandtech nor THG actually bother doing competent testing so their numbers don't really support the notion of Kentsfield superiority.
Also, your assumption that 4x4 will quietly disappear is way off. AMD will provide a quad core K8L for 4X4 in Q3 07. This will in fact put pressure on Intel to also provide a dual socket system. And, an Intel dual socket quad core would have similar power draw to 4x4.
I have to say that I am puzzled about calling 4x4 a rebranded Opteron. All FX's and Athlon 64's are the same die as Opteron as they have been since 2003. And, just as all C2D chips are basically the same die.
I have to laugh when people call me an AMD fanboy. Even sharikou180 said he thinks Barcelona will beat Conroe while I have only estimated Barcelona to close 3/4 of the current integer IPC gap. I think Barcelona could end up faster in SSE but not by a large amount. And, not at all if THG keeps using the Intel Compiler in their testing.
Intel's Kentsfield is comparatively low volume however it will still be a much higher ratio of production than 4X4 so these two aren't exactly equal. Kentsfield should have more market penetration.
I do have to say that the talk about humiliation and one two punches is a bit odd. In spite of C2D Intel has still lost share for the last six months.
Scientia you amd fan boy where are the quadfx reviews that you were telling me to ask you for???? THAT IS EXACTLY WHY I DONT REGISTER TO POST ON YOUR SITE!! Because this is the first real thing AMD has released and you choose not to even make a passing reference to it instead you go around talking about how every professional site has a bias towards AMD and how the intel compiler was used for the review of Intel's clowertown.
Yes, Kentsfield is a PR tool right now, seeing as they're not making much money from it or making it so that it can easily proliferate the entire market.
Since when did a product targetted at a niche high end gaming PC intended to making tons of profit?
Since when did an extreme system proliferated to the entire market? You need to look all around you and get a litle understanding of market segmentation and how a company can maximize profit by doing so. Airlines and automotive companies are just a few examples that practices the same principles.
Bottomline is, Kenstfield is available to supply the market it is targetted to while at the same time its forcing the competition to make misteps and price cutbacks on similar premium products.
Ofcourse its a PR tool. So what.
Sorry to go off topic, but does anyone know what happened to Sharikou? He hasn't updated his blog in a while.
Maybe all the bad press AMD has been getting lately is finally getting to him. ;)
Roborat, sorry I wasn't more clear, but by entire market, I meant the entire gaming market. By proliferate, I meant to make a substantial portion of. So I really don't get what the purpose of your first paragraph is if in the end you agree with me completely.
180, while I can see that they might hold back on releasing 65nm fx parts, I don't think they will, as they need to start making quad fx a viable option for somebody if they want it to get off the ground. They'd need to do that before q2 if they want to support barcelona with this platform in any respectable way. Seeing as all their 65nm chips are being binned at 65 watts (at least almost all of them), it would make a perfect 4x4 option. Plus, since the fx parts are so similar to server parts, I could easily see them shifting a little bit of extra processors over to 4x4, and not have to devote too much of their infrastructure to doing so. Plus, most of their stuff seems to be going out to Dell and the like before any of the channel can get there hands on it, so they wouldn't have to worry about marketing it anyway.
There are dozens of reviews of Quad FX. Surely AMD could've tried to spin some of them in their favor? I'm still wondering what these memory intensive apps are. You claimed encoding and we all know how that turned out.
http://www.xbitlabs.com/images/cpu/amd-quad-fx/charts/multitasking.png
Here's one unreal world situation that favors Quad FX:)
http://www.xbitlabs.com/images/cpu/amd-quad-fx/charts/photoshop+3dsmax+xvid.png
And another that doesn't;)
Once again, Quad FX is still going to be extreme wattage, are they going to sell 8 cores for $1000 max, where are the apps and where is the need?:)
Well, a real world bench that nobody does that would theoretically do much better on AMD processors is a folding benchmark (since folding is FP intensive). And a lot of people are very obsessed with folding for some reason.
Also, I don't see how the first benchmark you point out is not real world. If I had to encode a bunch of videos, compress a bunch of files, or other things like that and I wanted to game at the same time, I wouldn't bother with seeing how fast those processes happen, and I would definitely be interested in how well the game would play. And finding a bunch of those processes to do at the same time would be pretty easy.
http://techreport.com/reviews/2006q4/quad-fx/folding-total.gif
http://techreport.com/reviews/2006q4/quad-fx/cine-power-qx6700.gif
http://techreport.com/reviews/2006q4/quad-fx/cine-power-fx74.gif
Yeah it edges out in the only official Folding bench I could find, but it comes down to wattage again. You'd expect more with that much more heat.
I can't imagine loading up 2, let alone 8 instances of WinRAR, especially while gaming..
Perhaps 4 instances of WinRAR is memory intensive enough to choke the FSB, but in the 2nd bench (with a diversified set of 3 apps), the QX6700 is so ahead of the others in that I think it is highly unlikely that QX6700 falls behind unless people just throw apps at it for the sake of it.
Well, sadly, I can. I work a lot with open source software, CVS packages, SVN packages, Inventor and Solidworks work, and then the resulting video rendering of said assemblies. As such, if it's a bad day, I can end up rendering a couple of assemblies at once while decompressing a lot of files. If I could do all of that and some gaming on the side to pass the ridiculous amount of time all of that takes.
So for the megatasker who needs a workstation gaming rig combo with a bit more emphasis on gaming, I'm sure 4x4 fits the bill. Kentsfield probably fits most bills better, but seeing as people like me are likely a small minority (I'm an engineering student right now) I don't see a market that makes sense for either of these.
But people are stupid, and like to spend money, so if they're dumb enough to think they "need" either, then it really doesn't matter which one they buy, because they're the type of people who basically don't care, and probably don't pay attention to anything people like us say.
If you use your machine that extremely then I suppose Quad FX could be in your order:)
With the stuff that you work with though, I would think that a PC with server RAM would be more fitting? And there are Opteron boards with SLI also, then again, that server RAM degrades gaming performance, and Quad FX is a lot cheaper.
Hey Greg:
When you get that engineering degree, then you can join all of us who already have one.
Gee, just saw some AMD 65 nm reviews today on a couple different sites. Guess what - my predictions were pretty much spot-on. ANY 65 will not overclock 200-400 MHz beyond it's stock frequencies. No more headroom for K8.
What was referred to as my 'second' argument above now has the evidence in some enthusiast web sites. It was speculation when I wrote it; now it has a lot of written support in the review sites.
Scaling: a % increase in clocks has a roughly equal % in computations. Netburst did not. Core does. Intel keeps fine-tuning is steppings, they'll keep dropping power and upping clocks. It's not possible, it's inevitable.
Xeon "clovertown." Xeon is made for business - they already have need for SMP in the corporate world, quad core is not novelty.
Who cares about who beat who? All I know is that for every 5 PDP's I saw, I saw 1 X2.
Anonymous, please make sure your first paragraph is actually making a point instead of trying to attack me personally, as I certainly wasn't trying to use being an engineer in training as some way to give myself more clout (as you could probably figure out already that I'm an ME, and thus don't have the background that would matter to this argument).
Also, if you have an electrical or computer engineering degree, then please go into specific details within your next set of arguments to explain your statements, as I feel your breadth of knowledge would greatly clarify any and all arguments we're making.
I've seen fairly divergent accounts of how well the processors overclock, and on many the motherboards being used were not updated to use .5 multipliers (which may or may not adversely effect overclocking potential). If anyone finds one with a board that specifically has an updated bios that supports the procs, then please link here, as it would be greatly appreciated.
While it's somewhat unexpected that their processors would not overclock somewhat better at this specific process, it's still pretty obvious that they wont immediately be able to milk the process change for all it's worth, as they have a lot of transistor technology invested in 90nm that could not immediately be transfered over to 65nm.
Also, could you specificy which argument, as I'm having a really hard time re-reading through these comments and figuring out exactly who said what (again, please register or use your google account if in any way possible).
I realize now that you are right about architectures and how they effect clockspeeds (due to stepping, though I have yet to have it explained to me). However, it's not like AMD isn't working on refining its stepping as well, or working on other methods of architectural enhancement that will preemptively attack penryn.
To the other anonymous, I would gladly invest in a server machine if I had the money, as such platforms are even more expensive in that class.
I've only got a minute, so my apologies if someone's already yelled at you about this-as I'm going to do it right now.
WHY on EARTH would you expect 4x4 to cream Kentsfield? It's 2x3.0GHZ(x2) (FX-74) vs 1x 2.93GHZ(x4), K8L vs C2D (let's not get into semantics here, a QX6700 is still a C2D in my book). It's (roughly) the same thing as a 2.4GHZ E6600 vs a 2.4GHZ 4600/4800 X2...and we all know what happens there. True, the AMD chips have a slight clockspeed advantage in this matchup, but we all also know that given their somewhat large clock-for-clock deficit, ~100MHZ isn't going to make up for it.
In all honesty, I'm amazed it did as well as it did-and you should be, too.
That said, the power consumption is through the roof, and the motherboard's a joke. I don't believe anyone in their right mind would buy it.
Sometimes I laugh at the things you say-sometimes I laugh at the things he says.
Post a Comment