Coolaler shows us just how pathetic the i7-4960X will be...

[rk7p5]

Keep on dreaming. There is no way it will be 20% on average, it will be way lower. Maybe in some selected benchmarks.

I said around 20%, hey i'm just trying to stay positive. There is very little reason for me to upgrade as my 2500k is still very fast for my games and apps, i just want a better platform as a whole and something new to OC.

My dream of haswell 20% over sandy is a lot closer then steamroller having 15% more than Sb lol, even though i would love for Amd to surprise with SR.

If Steamroller came out the same time as haswell, i would build a Amd SR rig in a heartbeat.

[TRANCEFORMER]

http://vr-zone.com/articles/intel-iv...013/19911.html

[lutjens]

Continuing with their yearly trend, Intel will release a new generation of ultra high end microproessors as part of their Ivy Bridge-E platform.

The one interesting thing I read though:

Meanwhile, a fully unlocked 8 core/20 MB CPU will be sold under the Xeon branding, albeit for a much higher price tag (than $999).

Very interesting and encouraging. Not the unlocked dually 12 core that I'd like to see, but it's a start. I wonder if they will they be leftover Sandy Bridge EP dies or 1/3 disabled Ivy Bridge EP dies. I also don't see any source in the article or the linked article proving this to be true, but I'm hoping it is. They also don't mention if they will be dual processor enabled or be limited to one socket.

[BenchZowner]

Thread title is cranking me up, good job lutjens!

[tiro_uspsss]

From what I've seen, the 39xx cpu's have less mem bandwidth then the 3770k.
The 49xx cpu's aught to have more bandwidth then the 3770k.

But paying $1000 for a new cpu, no...
I'll try to go with the 4770k, unless they put out a $400-500 4930k or something.

And besides, wasn't this cpu actually supposed to have 12 cores?, and they apparently disabled half of it .

eh? quad channel less mem bandwidth than dual channel??

[twin_savage]

The Xeon side of the new Ivy Bridges is looking relatively interesting judging from this slide. The L3 caches get a nice bump as does clock speed.

[zanzabar]

Will these unlocked top parts finally get VT-d? It's so stupid that the 3770k doesn't support VT-d, while the 3770 non-k version does

all socket 2011 support vt-d, they want to force you up the line if you want to overclock and have vt-d. that said, amd and NV will both have on card vitalization, and nothing uses vt-d now so there is no real point in it as you only need it for graphics.

[Andi64]

all socket 2011 support vt-d, they want to force you up the line if you want to overclock and have vt-d. that said, amd and NV will both have on card vitalization, and nothing uses vt-d now so there is no real point in it as you only need it for graphics.

C1 stepping didn't support VT-d, at least on consumer parts (3930, 3960).
I went from 2600K to 3820 because of the VT-d. But I guess I'm on the .01% of the market who overclocks and uses VT-d. But I also don't get why Intel removes VT-d on socket 1155. Nobody will use an overclocked 3770K for an actual server, it doesn't make any sense.

[zanzabar]

C1 stepping didn't support VT-d, at least on consumer parts (3930, 3960).
I went from 2600K to 3820 because of the VT-d. But I guess I'm on the .01% of the market who overclocks and uses VT-d. But I also don't get why Intel removes VT-d on socket 1155. Nobody will use an overclocked 3770K for an actual server, it doesn't make any sense.

it did not have that warning before, i wounder if they lied like they did with the SSDs and encryption.

[Splave]

No Cold bug and strong IMC, these will be fun on ln2. Dont be a hater

[NEOAethyr]

eh? quad channel less mem bandwidth than dual channel??

Yep, that's what I said.
Though I should of said single threaded mem bandwith just to be clear...
This time around though it's hard saying.
Haswell does look nice, hard saying if ivy-e will beable to keep up in mem bandwith or not.

[lutjens]

No Cold bug and strong IMC, these will be fun on ln2. Dont be a hater

So we should be grateful to Intel for permitting us the honor of being able to bestow our patronage upon them and happily consume their rejects? If we do that, Intel will think we are content and happy with their scraps and said scraps will likely become all they're ever going to be inclined to provide in the future. Personally, for the premium price they are demanding, I expect better from a company that could very easily do so. The fact that they refuse to offer a fully functional, fully enabled CPU at any price shows a great deal of disrespect for the many segments of the market that could make use of such a chip. These include graphics designers, people in the stock market, and even some people looking for a bit extra out of a server. They are clearly and deliberately ignoring what could potentially be a very lucrative market due to their own short-sightedness and arrogance.

[xVeinx]

They are clearly and deliberately ignoring what could potentially be a very lucrative market due to their own short-sightedness and arrogance.

I'm not so sure. They might sell a reasonable number of them, but would it be worth the cost of producing that specific sku? Would it cannibalize other sales heavily? It's a business decision at the end of the day, whether we like it or not. If you argue that there aren't that many people who will buy them (no cannibalization), then you lose the cost of production argument, If you argue that there are plenty of people who will, then you lose the cannibalization argument. Seems like a no win situation .

[Kain665]

So we should be grateful to Intel for permitting us the honor of being able to bestow our patronage upon them and happily consume their rejects? If we do that, Intel will think we are content and happy with their scraps and said scraps will likely become all they're ever going to be inclined to provide in the future. Personally, for the premium price they are demanding, I expect better from a company that could very easily do so. The fact that they refuse to offer a fully functional, fully enabled CPU at any price shows a great deal of disrespect for the many segments of the market that could make use of such a chip. These include graphics designers, people in the stock market, and even some people looking for a bit extra out of a server. They are clearly and deliberately ignoring what could potentially be a very lucrative market due to their own short-sightedness and arrogance.

The extreme consumer market?

Lucrative indeed.

[lutjens]

The extreme consumer market?

Lucrative indeed.

In my post, I specified other parties that would likely be very interested in such a CPU. I admit extreme hardware enthusiasts are a very small portion of the overall market, but stock brokers, graphics/video folks, people running non-mission critical servers that need maximum performance could all benefit from such a chip.

The extreme consumer market...small. All the other segments I mentioned AND the extreme consumer market? Less so...

The argument that such a ship would cannibalize sales is largely invalid...Intel could produce such a chip at the same price as their other top offerings...and let the market decide which they prefer. The cost for Intel to produce that SKU would be the same as the cost to produce the existing chips. The ultra-simple solution would be if they simply unlocked the existing top bin chips, then the total cost to Intel would be that of unlocking the multiplier...which is zero. It may even save them time, as it would save them the hassle and time of locking the multiplier in the first place.

Net cost of unlocking the top chips=zero
Potential for cannibalization of sales of existing chips=zero, as there would be no new SKUs...only the top chips would be unlocked.
Potential to better serve many people who could benefit from increased performance=very high.

[Kain665]

In my post, I specified other parties that would likely be very interested in such a CPU. I admit extreme hardware enthusiasts are a very small portion of the overall market, but stock brokers, graphics/video folks, people running non-mission critical servers that need maximum performance could all benefit from such a chip.

The extreme consumer market...small. All the other segments I mentioned AND the extreme consumer market? Less so...

The argument that such a ship would cannibalize sales is largely invalid...Intel could produce such a chip at the same price as their other top offerings...and let the market decide which they prefer. The cost for Intel to produce that SKU would be the same as the cost to produce the existing chips. The ultra-simple solution would be if they simply unlocked the existing top bin chips, then the total cost to Intel would be that of unlocking the multiplier...which is zero. It may even save them time, as it would save them the hassle and time of locking the multiplier in the first place.

Net cost of unlocking the top chips=zero
Potential for cannibalization of sales of existing chips=zero, as there would be no new SKUs...only the top chips would be unlocked.
Potential to better serve many people who could benefit from increased performance=very high.

Lutjens, I don't even know what you're arguing any more but I'm going to assume you've dropped your arguements of offering an unlocked 10c chip under the "Extreme" brand for $999. There's one simple reason for locking their high end xeons - who overclocks a server? Who overclocks a workstation? Why create more liability for yourself when the majority of the intended market does not require it?

Also, could you give me a hint as to how you're running the dual GTX 690s in that ML370? As far as I know, it doesn't have any PCI-E power connectors (After all, no enterprise EDGE cards need PCI-E power, it's strictly a consumer/workstation standard), and being designed as a rackmount server, HP doesn't offer any replacement PDBs that offer any PCI-E 6 Pin or 8 Pin connectors.

[Kallenator]

Lutjens, I don't even know what you're arguing any more but I'm going to assume you've dropped your arguements of offering an unlocked 10c chip under the "Extreme" brand for $999. There's one simple reason for locking their high end xeons - who overclocks a server? Who overclocks a workstation? Why create more liability for yourself when the majority of the intended market does not require it?

Because we can.

[Kain665]

That's great and all, but CPUs have become extremely commoditized. This market is significantly different than even 5 years ago. Any CPU, especially starting from the midrange offerings and up from either AMD and Intel will work well for 99% of the applications out there.

In this market, there's really no good that comes with doing things "because we can" anymore. It's completely irrelevant that you could produce a 3.5Ghz 10 core, 20 thread chip for $999. Nobody cares if your $4k enterprise offering comes with an unlocked multiplier. The simple fact is that we don't NEED those features. There's no advantages to offering it.

It's not like Intel has any competition on having the fastest CPU.

If you're so annoyed that we don't have Intel releasing crazy products "because they can", stop whining about Intel's margins and go bug developers to creates something that legitimately NEEDS that performance.

[lutjens]

Lutjens, I don't even know what you're arguing any more but I'm going to assume you've dropped your arguements of offering an unlocked 10c chip under the "Extreme" brand for $999. There's one simple reason for locking their high end xeons - who overclocks a server? Who overclocks a workstation? Why create more liability for yourself when the majority of the intended market does not require it?

Also, could you give me a hint as to how you're running the dual GTX 690s in that ML370? As far as I know, it doesn't have any PCI-E power connectors (After all, no enterprise EDGE cards need PCI-E power, it's strictly a consumer/workstation standard), and being designed as a rackmount server, HP doesn't offer any replacement PDBs that offer any PCI-E 6 Pin or 8 Pin connectors.

I'm not arguing for a 10 core Extreme chip for $999. Far from it. I'm arguing that the top chips should be unlocked.

Lots of people overclock servers...the non-mission critical servers that benefit from more performance. Buckeye builds them. Who overclocks a workstation? No one at the moment, because nobody can. How is it a liability to unlock the top chips? If people don't want to use such a feature, they're not bound to and can run the chip at default.

You're making assumptions about the ML370 G6, much like assumptions made elsewhere in your argument. Assumptions about how others use hardware are often surprisingly inaccurate. I'm running GTX690s in my ML370 G6 using PCI-E power adapters that are an option from HP. The ML370 G6 has two additional power connectors for use with video cards, HP being the only server manufacturer to my knowledge that provides such connections. A third is avilable as well, but only if I forgo the use of the third drive enclosure. SLi is possible with the use of an SLi hack and even though this is an "enterprise" server, it plays games surprisingly well...

And provides memory and I/O expansion abilities that consumer machines can only dream about...

[Kallenator]

That's great and all, but CPUs have become extremely commoditized. This market is significantly different than even 5 years ago. Any CPU, especially starting from the midrange offerings and up from either AMD and Intel will work well for 99% of the applications out there.

In this market, there's really no good that comes with doing things "because we can" anymore. It's completely irrelevant that you could produce a 3.5Ghz 10 core, 20 thread chip for $999. Nobody cares if your $4k enterprise offering comes with an unlocked multiplier. The simple fact is that we don't NEED those features. There's no advantages to offering it.

It's not like Intel has any competition on having the fastest CPU.

If you're so annoyed that we don't have Intel releasing crazy products "because they can", stop whining about Intel's margins and go bug developers to creates something that legitimately NEEDS that performance.

Xtremesystems and enthusiast rigs in general has never been about sensible choices that necessarily has any application outside the realm of synthetic benchmarks or folding.

And I do not need to invent an application to use more than 4-cores, my workstation would gladly accept more at a more affordable price, because developing under Vmware takes it's toll on your CPU.
Also ever tried to live-stream what you are gaming? Not many games that enables you to do that unless you have an absolute beast of a CPU. (Planetside 2 is another application where CPU power is really needed under worst circumstances.)
So there is absolutely a market for this, but more than 4-cores has indeed been held back by Intel, and AMD CPU's just don't cut it for this type of application even though they are great at H264 encoding.

Might not be large enough to give a net profit though, I will give you that at least.

A theory, would the 2011 class CPU's be more popular if they represented their value in a more realistic way? The old 3960k with it's six unlocked cores dwarf the price of the 3770k by almost 100%, and it is nowhere near twice as fast as it. I do realize Intel's silicon is expensive in such die sizes, but there just isn't enough power to justify it for the user's who actually can put it to work.
So be it like this, the 4930k should cost no more than 50% on top of the 4770k. Representing it's premium over it's underdog in a way that makes more sense to any given consumer.

[Andi64]

It's nothing new .They price their processors as they like when there is no competition.

Anyway, 50% over a 3770K gives you around ~U$S500. It's not that overpriced.

I really hope they release an 8c/16t monster at U$S999 to drive down the 6c to lower prices. But it won't happen

[FlanK3r]

I think, we will see 8C/16t in desktop segment with Haswell-E

[lutjens]

I think, we will see 8C/16t in desktop segment with Haswell-E

Possibly, but such an event would seem equally pathetic when Haswell-EP comes out with 16 cores on the die. It'll simply be another half-a-chip reject being pawned off as something impressive.

[BenchZowner]

Possibly, but such an event would seem equally pathetic when Haswell-EP comes out with 16 cores on the die. It'll simply be another half-a-chip reject being pawned off as something impressive.

They'll make a less-profitable 16c non-Xeon CPU just for you, only $199.

Business 101 is a good read and will more than likely bring you to terms with the reality.

[lutjens]

They'll make a less-profitable 16c non-Xeon CPU just for you, only $199.

Business 101 is a good read and will more than likely bring you to terms with the reality.

That's the thing...both myself and others would be willing to buy said 16 core CPU, and pay handsomely for it, but Intel refuses to produce one, for any price. All they would need to do to satisfy many people would be to unlock the top Xeons...those who want to use the feature can and those who don't can run the chip at default.