Maximum PC

Because clock speeds aren't a representation of performance? Look at AMD. Yes they have a 4.0GHz stock part. But it can't even run miles around a 3.4GHz part.

IBM's POWER6 went up to 5.0GHz. How come nobody was drooling over that?

I'm disappointed that we're stuck at 4/4 and 4/8.

I know, I know. I'm not the average user. What's the point of having fast SSDs, gobs of RAM, and all the other goodies if you only have four cores. VMs rule, especially when rocking a number at the same time.

Give me 8/16 and I'll be happy.

n0b0dykn0ws

Then get a server chip. Consumer chips weren't really made to run multiple VMs in mind.

Because they tend to start off in price where the Extreme chips end.

n0b0dykn0ws

There is a physical constraint, the equation P = C * V^2 * f * <# of transistors>. There's also the fact that temperature is a function of frequency (P = I^2 * R, and resistance usually converts electrical energy into thermal energy) and that the life span of a part is a function of its operating temperature.

So why can IBM or AMD afford to drive their chips at high speed? Probably because they don't really care about low TDP as much as Intel does. And Intel is trying to get into the tablet market as well as expand the AIO market. That or since POWER is a RISC architecture, it needs less transistors for similar performance. So yes, this suggests if you want your high frequency with current technology, you need to dump x86. Good luck with that!

If we get them on desktop chips from Intel, Ivy-Bridge E (Q3 2013) will be the first opportunity, but probably only as the Extreme Edition CPU...

Unfortunately, with the lack of serious competition from AMD, Intel can afford to take the stance of "anyone needing more than 6 cores / 12 threads needs to be running Xeon's anyway."

It doesn't help that AMD isn't making the idea that "more cores = better performance". However, while I do understand some of your guy's positions on running multi-threaded applications and/or multiple VMs, the point still stands: practically all of Intel's customer base in the general consumer field does not need more than 4 cores. At least, 4 cores on the performance level Intel's current offerings provide.

$990,000 does not equal 1 Million dollars, just as "practically all" does not equal "100% of users". There are consumers who can use that extra performance provided by 8-cores / 16-threads, and should be using Xeon based systems anyway, but cannot afford them in these trying times (or could otherwise make better use of the $$$ if we didn't have to spend it), so it would be nice if there were a more affordable alternative i.e. an Intel desktop chip with 8-cores & HT.

It's really not that different from the SLI question... who actually needs SLI??? Only about 1% of gamers. Who wants SLI? a lot more than 1%. Who knows that they don't need SLI but still wants SLI? at least 1%

Here's some other questions I have to ask...

• For those running VMs, why do you need that many? Do you really need more than three unique environments up and running all at once?
• For those who desire compute: how come we're still relying on CPUs for a majority of the workload when you have an iGPU that can accelerate the work? If your compute work is highly parallelized, you have hardware better suited to the task.

Yes, but you still represent a niche market. A market that may or may not be worth spending valuable engineering dollars on. And in these trying times, that's money better spent on a market guaranteed to generate a profit. I mean, I guess Intel can sell you such parts for cheap at the cost of QA and lack of warranty.


Except SLI and CrossFire are modular upgrades made possible because the infrastructure to support it is cheap and plentiful. And it's a multiprocessor solution, not a multicore one. Not to mention, asking to bring the cost down of an SLI/CrossFire means bringing the cost down of everything. Yeah, that's going to happen.

And you miss the point yet again. Multi-core is multi-core regardless if you have a single package; i.e. One GTX-690 GPU for two GPU cores (not counting "Cuda" cores) on one delivery package, or One Xeon E5-2650 for 8 CPU cores on 1 delivery package; or multiple packages i.e. two GTX-680 GPU's in SLI which still gives you the same two GPU cores but in two delivery packages, or two Xeon E5-2620 CPU's which still gives you 8 CPU cores but in two delivery packages.

However, other factors play a part, for two GTX-680's, you must have a motherboard with two PCI Express x16 slots available, for two Xeon E5-2620 CPU's, you must have a motherboard with two LGA-2011 sockets, two coolers, two sets of RAM, etc... Extra sockets & what-not add to the total system costs.

And none of that has anything to do with the point either That point being that, the market IS very limited for users that can utilize an 8-core desktop CPU, just as it is very limited for users that can utilize SLI / Crossfire. None-the-less, a market still exists for both.

However, there is a potential market that a lot of people tend not to think about.... Fortunately Intel is thinking about it & is virtually the only reason Intel is even looking at the possibility of 8, 10, & 12 core Ivy-Bridge E chips, "low cost" professional Workstations.

We can play this game back and forth of "I want this!" and "Intel's not going to give a shit about you" forever. I'll leave it at that.

And kleinkinstein's modus operandi is to throw out non sequiturs and ad hominens because he can't come up with a decent argument to save his life.

I went against the cool thing to do here and enjoyed myself with Windows 8, even wrote an post about it with rational reasons. What did you respond with? "You have a friend?" Really? That's the best you can do on your crusade with spreading the Windows 8 hate? And after that you brought up my supposed love of the OS to discredit me anywhere I go, because I'm assuming that hating Windows 8 makes you an expert and liking it makes you casual PC user who spends $2000 on a machine just to go on Facebook.

Have any of you that are complaining taken a step back at the CPU you have now? I'm building iTX's on i3's because they're everything I need for those builds. Are you all just focusing on your main builds and wondering why Intel is not looking directly at you?

Up front, the needle is still going in the right direction. You can't get something from nothing, the TDP steps back up a bit (please research that because most of you have it wrong still) and the performance moves another step UP too.

The CPU freq has topped out because it's stable while they work on shrinking. You have to pick one side or the other in order to go that route. Do you want more cores (no), do you want faster clocks (why) or do you keep shrinking like you're suppose to do.

Latios, I entirely disagree with your found function and reply, there is not a physical constraint. That looks like it would have worked 10 years ago, but was also proven false by some magician here and we're running on OTHER proofs of concept currently. I've been through it all and found we don't need CPU's that last 30 years. Mind you, I've been IN there. You're also being ignorant of what servers actually do today. Most don't need a 16-core CPU and can have a 4w Atom run their cloud service, in which AMD and ARM are aiming at as well. Never the less, it has NOTHING to do with the CPU's we're talking about. "some of our guy's position" includes your opinion posted here, does it not? I'm with ya LatiosXT, but not...

We are all still in a world where most of us with 107465 cores only use 2 threads daily. You can't blame Intel for that one bit and should move along. Thus the debate that we're having here: Why even bother?

SLI/Crossfire has always been pure fucking bullshit. Plop 2 6800's in there for $700, in a year there will be a $300 that does the same job. Go ahead. Plop a pair of 680's in your mega rig for a grand. You're not fixing a damned thing and detracting from the logical debate.

LatoisXT Just gave up (as I'm reading this thread). It's not back and forth but actual talk. Sorry you missed it or understand that you could be wrong.

Oops, Latios is back with fucking Latin. A dead language and forgot about English debates of fact or knowledge.

LatiosXT: you're right. In every way. I hope your dreams come true in your world. Thanks for the debate. Do you have a conclusion like I've made time and time again that's correct for decades, or just pissing in the wind and getting pissed off that others don't agree? You go up to this point, and then quit. When you run out, you treat as stupid. Really?

I assume a lot, but but blind. I'm ready for a debate any time, but don't shut members down because you have another ideal. It may just not fit. Hell, this is just as fun as Facebook right now. Just call each other names and force opinions. No one has pulled Latin before...ever. Why? It's a dead language unless your a biologist finding rare moths.

Wait until the next iteration of Intel's chips, Broadwell, when Intel switches to solder only chips. No future CPU upgrades.

I don't get it either. I guess 14 nM is getting too small to package.

Et tu Brutus? Hey, I am a biologist who finds rare moths. Conaris, esse CATACLYSMOS?

LatiosXT appears to have "won" the "debate" some time ago. Speaking of ponderous threads....