Intel to Omit Release of Enthusiast-Class "Ivy Bridge" Processors This Year.

[Zebo]

So what? SB will last you guys years like Bloomfield has lasted me almost 3. It's really amazing how little we've gained (if any since Bloomfield) hence why I don't bother much with CPU forum anymore.

Get a 2500K, crank her and be happy. IB offers nothing or at least infinitesimal gains.

Don't get me wrong I get itches for the new shit being an enthusiast then I go look at these clock for clock benchmarks here and come to my senses that a twice as fast video card every year is probably more well spent than "upgrading" my 4.2 Ghz three years old bloomy.

 

[Rvenger]

I wish this happened... without the modules, oh and without the code name Zambezi.

 

[myocardia]

xx

 

[pelov]

I wish this happened... without the modules, oh and without the code name Zambezi.

8 Thuban cores would be ~1.2B transistors? It would have less L2 than the current BD chips and probably wouldn't clock all that high considering Stars on 32nm (Llano) doesn't get all that much high either. K7...errr, K10/10.5 was on its last legs. They needed a redesign, I just think we're not all too happy with their intended goals.

I remember reading an article a year before BD's release which said they'd never catch up to Intel with respect to IPC, thus they required a new direction: offer whatever Intel won't/can't.

Us desktop folk may not like it but AMD has quit caring about us

The rankings by form-factors in Q4 2011 were the following:

In the desktop PC processor segment, Intel's market share was 76.1%, a gain of 0.3%; while AMD had 23.8%, a loss of 0.3%.
Intel commanded 82.3% share in the mobile PC processor segment, no change compared to Q3 2011. AMD finished with 17.6% of the market, flat with Q3 2011.
In the server/workstation processor segment, Intel finished with 94.3% market share, a loss of 0.8%, and AMD earned 5.7%, a gain of 0.8%, which is rather remarkable and represents moderate success of Bulldozer-based AMD Opteron microprocessors.

http://www.xbitlabs.com/news/cpu/di...ightly_Gain_CPU_Market_Share_in_2011_IDC.html

It's the server numbers that need the most fixing with mobile right behind it.

Also want to reiterate the notion that Intel's screwup of 2011 SB-E is the likely reason for the omission of IB-E, with perhaps some 22nm tri-gate issues thrown in. The latter may become quite visible if IB isn't being made in enough volume or they decide to throw in some fused graphics chips in early.

 

[Edrick]

Edrick is not going to like the sound of this.

:''''''(

So there goes my plan right out the window (buy the 3820 and then go IB-E). The only good side to this is now I can save more money for a killer Haswell rig.

I would have no problems if they skipped IB-E and introduced Haswell-E in advance.

 

[Rvenger]

:''''''(

So there goes my plan right out the window (buy the 3820 and then go IB-E). The only good side to this is now I can save more money for a killer Haswell rig.

I would have no problems if they skipped IB-E and introduced Haswell-E in advance.

If I were you I would just strap down and just hold on to that SB-E until Haswell. I think Intel is just buying time and we can buy them time by minimizing their profits :sneaky:

 

[Edrick]

If I were you I would just strap down and just hold on to that SB-E until Haswell. I think Intel is just buying time and we can buy them time by minimizing their profits :sneaky:

Honestly, I like the 3820. And for the price, you really can't beat it. It will last me well into 2013 as it can handle anything I throw at it currently. Haswell is what I have been waiting for anyways, more so than IB-E.

 

[Genx87]

If I were building an IB computer it would be with either the 3570 or 3770K processor anyways. Just depends which one Microcenter sells for 179.99.

 

[Rvenger]

If I were building an IB computer it would be with either the 3570 or 3770K processor anyways. Just depends which one Microcenter sells for 179.99.

But I don't think MC will be launching them at $179??

 

[Genx87]

But I don't think MC will be launching them at $179??

They may not launch at that price. But I expect one of them to replace the 2500K at 179.99.

 

[greenhawk]

Just depends which one Microcenter sells for 179.99.

that would be the 3570. The other is a i7-2700K replacement looking at the slides about up coming cpus.

 

[Genx87]

that would be the 3570. The other is a i7-2700K replacement looking at the slides about up coming cpus.

Sounds great.

 

[Ajay]

You're missing the point. HWe will be released near-simultaneously or perhaps even a little before HW, ie right around the time people are predicting IBe now. So it's a whole generation leap without any extra waiting.

Maybe I missed it, but what's the info source of the simultaneous release of HW and HW-E?

 

[tynopik]

Maybe I missed it, but what's the info source of the simultaneous release of HW and HW-E?

no source, just my guess as to what makes the most sense

1. IB-E gets cancelled because it's so late it's irrelevant.

Besides IB-E never actually offered that much to the enthusiast platform. IB's main claim to fame is its improved graphics core, which won't be included in IB-E anyways.

2. With IB-E cancelled, they still need to refresh SB-E. They aren't going to let their enthusiast platform fall TWO generations behind.

Need meets opportunity and HW-E is born.

 

[Lyfer]

Could it be that both Intel/Nvidia just dont care about IB/Kepler and just want to shift more focus to tablets/mobile market?

 

[Ajay]

2. With IB-E cancelled, they still need to refresh SB-E. They aren't going to let their enthusiast platform fall TWO generations behind.

Need meets opportunity and HW-E is born.

Well, Intel will need a 22nm server CPU sooner, rather than later, in order to improve compute density for large server installations. There is no way to complete verification of HW-E in time for the Haswell launch. If Intel does skip IVB-E, then they'd have more resources to get HW-E out, say, six months after the launch of Haswell.

Bottom line: this is some heavy duty speculation. Hopefully Intel's roadmap will become clearer sometime this summer.

 

[tynopik]

There is no way to complete verification of HW-E in time for the Haswell launch.

Why do you say that? Nehalem-E (LGA-1366) launched 10 months before the consumer line (LGA-1156)

Intel really hit a double whammy with the memory controller and SAS controller on 2011. Now that they hopefully have a grip on them, the next generation can come much easier/sooner.

 

[aigomorla]

lulz... cuz the only competition intel has now is apple..

oh wait.. apple uses intel processors.. :X <sacasm>

id expect the enterprise line to be advancing if they do advance.
if the technology halts even on the enterprise line, then yeah, were screwed.

But im pretty sure i will be picking up a haswell server based system, and then overclocking that mofo after getting a overclocking server board..
But i havent touched a new processor since the 990X... more speed at this point doesnt justify the upgrade when your processor is already at 4.43ghz on 6c/12t.
I want more cores... and we have yet to see that 10C/20t monster anywhere outside a quad socket platform...

 

[Makaveli]

lulz... cuz the only competition intel has now is apple..

oh wait.. apple uses intel processors.. :X <sacasm>

id expect the enterprise line to be advancing if they do advance.
if the technology halts even on the enterprise line, then yeah, were screwed.

But im pretty sure i will be picking up a haswell server based system, and then overclocking that mofo after getting a overclocking server board..
But i havent touched a new processor since the 990X... more speed at this point doesnt justify the upgrade when your processor is already at 4.43ghz on 6c/12t.
I want more cores... and we have yet to see that 10C/20t monster anywhere outside a quad socket platform...

lmao

 

[Zebo]

lulz... cuz the only competition intel has now is apple..

oh wait.. apple uses intel processors.. :X <sacasm>

id expect the enterprise line to be advancing if they do advance.
if the technology halts even on the enterprise line, then yeah, were screwed.

But im pretty sure i will be picking up a haswell server based system, and then overclocking that mofo after getting a overclocking server board..
But i havent touched a new processor since the 990X... more speed at this point doesnt justify the upgrade when your processor is already at 4.43ghz on 6c/12t.
I want more cores... and we have yet to see that 10C/20t monster anywhere outside a quad socket platform...

LOL The only thing I've put on my mobo the last three years in a new video cards and a USB3 card/ 5.25" panel. Some ppl been kickin' almost 4 years with bloomys and I bet they will be good for another 4.

I don't see why I need more cores either. My benches in things I need the power for, games, are identical to the 6/12 cores tested in reviews.

 

[dac7nco]

lulz... cuz the only competition intel has now is apple..

oh wait.. apple uses intel processors.. :X <sacasm>

id expect the enterprise line to be advancing if they do advance.
if the technology halts even on the enterprise line, then yeah, were screwed.

But im pretty sure i will be picking up a haswell server based system, and then overclocking that mofo after getting a overclocking server board..
But i havent touched a new processor since the 990X... more speed at this point doesnt justify the upgrade when your processor is already at 4.43ghz on 6c/12t.
I want more cores... and we have yet to see that 10C/20t monster anywhere outside a quad socket platform...

Intel is locking down the Xeon CPUs now. ZERO overclocking via multiplier. Check out the EVGA forums... Intel isn't likely to release unlocked Xeons... ever again. That 16c/32t overclocked system isn't happening.

Daimon

 

[IntelUser2000]

Could it be that both Intel/Nvidia just dont care about IB/Kepler and just want to shift more focus to tablets/mobile market?

That would be a stupid business decision. Tablet/Mobile markets make fraction of the fraction of what PC business offers. No, they can't ignore it, but don't forget what's happening to HP that they've denounced the core business(PC) and went stray to a much smaller, mobile market.

1. IB-E gets cancelled because it's so late it's irrelevant.

Besides IB-E never actually offered that much to the enthusiast platform. IB's main claim to fame is its improved graphics core, which won't be included in IB-E anyways.

Why?

Besides the enthusiast SB-E and IB-E market which is sold in tiny volumes, Sandy Bridge EP(and its successors) will be a very successful product for servers and workstations. SB-E "spin-off", despite its low volume is justified because its development is more than paid off by its Xeon big brother.

Ivy Bridge EP for servers
-Going from 8 to 10 cores with necessary interconnect and router enhancements allow 20-25% performance gain.
-5-7% IPC improvement from enhanced CPU architecture, which might turn out better in servers
-Additional 5-10% clock speed increases for a total of 30-40% speed gain, or significant TDP reductions

I can't see why they need to skip Ivy Bridge EP. If things go well they can have it by very end of the year for another significant bump to the Xeon E5 platform. Sandy Bridge EP has been shipping in low volumes since Q4 2011. If things go well they can do a official mass launch in Q4 2012. We would then see Ivy Bridge E at the same time.

The only reason that a skip to Haswell EP(and therefore Haswell E) would make sense if Ivy Bridge EP is delayed another few quarters to consider skipping it instead.

 

[AtenRa]

99% of buyers dont need more than the Core i5 2400 right now. Intel will not prioritize the High-End socket 2011 for the 1% market segment (that is us ).

I believe the 1366 socket was the last time High-End Socket/CPUs where the first to be introduced in the market. Socket 1156 and especially 1155 was the tomb stone for the High-End market.

 

[tynopik]

Why?

It's cheaper.

Why bother investing all the resources need to create and validate IB-E and IB-EP when you can skip right to the next step and save yourself all that money?

The only reason to even consider investing in IB-E is if HW-E would take so long that the lost sales from continuing to sell the 'obsolete' SB-E vs introducing IB-E is sufficient to cover the expenses of creating IB-E.

And from my totally un-informed standpoint, I don't think it is. It appears that they could bring HW-E out not long after they could bring IB-E out if they so desired.

And in that case it makes imminent financial sense to just skip IB-E altogether.

If things go well they can do a official mass launch in Q4 2012. We would then see Ivy Bridge E at the same time.

Did you not see the article in the first post?

Those, who expected to upgrade their extreme-performance systems six-core Intel Core i7-3900-series "Sandy Bridge-E" LGA2011 chips inside with even faster "Ivy Bridge-E" chips later this year will probably have to wait until at least the second half of 2013, as Intel has no plans to launch new generation of enthusiast processors in 2012.

. . .

will not release its six-core or eight-core Ivy Bridge-E chips either this year or in the first half of 2013.

The only reason that a skip to Haswell EP(and therefore Haswell E) would make sense if Ivy Bridge EP is delayed another few quarters to consider skipping it instead.

Well there's your few quarters

So now do you agree that skipping IB-E makes sense?

 

[Ajay]

Why do you say that? Nehalem-E (LGA-1366) launched 10 months before the consumer line (LGA-1156)

Intel really hit a double whammy with the memory controller and SAS controller on 2011. Now that they hopefully have a grip on them, the next generation can come much easier/sooner.

Good point, I was really thinking about the server version (EP). So it is possible to get and enthusiast version (E) out faster, since it will require less rigorous verification.