yorkfield late too?

[JumpingJack]

Intel Spec update for the QX9650:
ftp://download.intel.com/design/proc...t/31872701.pdf

There are, ironically, two TLB errata

[savantu]

Intel's Yorkfield has a problem at the FSB level ; it isn't a bug in the core itself , but at the I/O level.
As a result , the delay will be minimal since it doesn't involve messing with the core.

[JumpingJack]

Intel's Yorkfield has a problem at the FSB level ; it isn't a bug in the core itself , but at the I/O level.
As a result , the delay will be minimal since it doesn't involve messing with the core.

I suspect that to ... often times the press gets these type of confirmed announcements, then rush without getting enough detail. Even TR did not have the full story the first time.

This could be very serious or it could simply be Intel will push the QX9770 out and the french site lost that in translation and assumed all Yorkies. I don't know... the 9770 had weird (and very poor) thermals, and it was the first 1600 MHz processor -- leads me to think it is a packaging issue and the pushout there would make sense. This is purely speculation, it will only be a few weeks before we find out.

However, if this is not so simple as a packaging problem and the RELEASED QX9650 has this problem as well, then Intel deserves to get nailed to the wall as well, and suffer the same bad PR.

[saaya]

Who cares about Extreme Editions? Here we go.
http://www.newegg.com/Product/Produc...ion=harpertown
Starting from $252 (in stock).

who cares about extreme editions? well whats everybody benching with? xeons?

who cares about xeons is what it should say!
this is xtremesystems and not "pimp my 2u server rack.com" :P

Intel can't make enough XE cpus , demand is incredible due to big OEMs taking every chip available.

Secondly , how can a York field have a bug and Wolfdale not ? Yorkfield is after all , 2 Wolfdale dies in the same package.

IMO , Intel is using a new stepping for Yorkfields and that's why they are a month late if the rumour is true.

oems buy all the extreme editions? who? dell?

i doubt that yorkfields use a later revision than wolfdales... that wouldnt make much sense plus we have seen yorkfields for months now and wolfdales are only appearing now.

Hey, whats 4 pads between friends?

i think this bug is unlikely to affect us end users, otherwise all of the dozens of ocers pushing yorks to 5ghz and beyond would have found some problems. but the bug could be desastrous for companies using 45nm xeon processors pushing them to workloads you never see on the desktop and that 24/7... it seems intel sold most of its 45nm chips to so far so this would be quite serious for them. i wonder what intel will do now...

and i wonder if its possible to reproduce the bug on a desktop chip too... and if we can, will intel swap it or will people be stuck with buggy qx9650s? lol

there is a chance after all that this bug affects all 45nm cpus and even causes problems on desktop cpus. cause lets face it, who with a 45nm cpu has been running it under full load for days or even weeks? what if this isnt possible and the cpu crashes or freezes randomly after some days if youd give it a try? so far everybody has been running 45nm chips at so high speed that having them crash was nothing special and nobody would have wondered about it.

[hollo]

a front side bus bug *only* on the quads, and we know quad cores have more problems with high FSBusses than dual cores.

this could have implications for people who want to overclock the first wolfdales

[JumpingJack]

a front side bus bug *only* on the quads, and we know quad cores have more problems with high FSBusses than dual cores.

this could have implications for people who want to overclock the first wolfdales

I don't think you are thinking about this right... wolfdales are dual cores, the complexity of signalling within the package is simply not there.

If Intel is simply working out bugs on the packaging (it is not straight forward as gluing together), then this revelation makes more sense.

[JumpingJack]

Well, there appears to be confusion... some people think launch is Jan 6, others think later... Digitimes (rumor mill as well), has this:
http://www.digitimes.com/mobos/a20071206PD221.html

They are listing Quads on the map for launch....

I have seen Digitimes not get it right before and, for me anyways, is not enough to shoot down another HW site...

[Hornet331]

geizhalsnone in stock...
geizhals 9600 in stock @ everywhere.

stop telling BS and trash everything non-intel, they paperlaunched QX9650 just like amd did with the 9700, and this is a FACT


i doubt that intel has a problem with their FSB, they simply want to milk 65nm as long as possible just as amd did before C2D

hahaha, check the geizhals link again, first shop has it. an im sure next week we see more and more stores have it.

[KTE]

Yep they have an instability bug with high FSB which delayed Yorkfields. Kinda sad that I expected plus 2.5GHz Phenoms/Yorkies to be out and competing and neither will make it 3-4 months after initial release boasts.

Just like I've seen no retail 9500/9600 owner get any bug at 2.4GHz (9700 frequency) nor upto 2,86GHz I doubt you'll see it with Yokfield if ever, either. But they still have to stop the retail and get it sorted just like AMD has to. Server segment dreads these bugs and that's where it matters mainly.

[kl0012]

who cares about extreme editions? well whats everybody benching with? xeons?

Let me recall your first post:

they really stole a lot of the thunder of the phenom launch by sending out a lot of 45nm es chips and selling 10 or 20 retail chips here and there

if 45nm yields are so awesome like intel claims then where do all their millions and millions of cpus end up?
i think they havent started 45nm mass production yet

There were no words about extreme editions, just about all 45nm. products. This is pure BS because you can buy 45nm. Xeons almost everywhere.

[seamumc]

tiger has them and if you look to the right in this link they have a package deal with x-38 intel board and a game for a decent price.http://www.tigerdirect.com/applicati...&CMP=KNC-GOOGL

I got a retail QX9650 from newegg last night, already out of stock at moment. OEM QX9650's still available. One thing I like about newegg, they are very good with their stock status.

[Donnie27]

hallo

Hello, who would be drunk enough to spend that much money on one LOL! Howdy!

[Donnie27]

Without competition theres no need to migrate to 45nm yet - they could clear out as much 65nm stock now on the mainstream products,then migrate fully to 45nm when the inventory is cleared - rather than selling the 65 and 45nm products side by side.

Absol-Frackin'-lutely QFT!

A Bug becomes an excuse to hold back a certain targeted line of New Products. They then sell off the equivalent model/s from the previous generation. Then when they think they've sold enough, wa-la, problem fixed. IMHO, this is worse than if Intel had a real problem. Common sense guys. FSB problem on Quad Core processors at 1333MHz but Quad Core processors running at 1600MHz are OK? Get real I'm not about to believe something this lame. I'll say it again, Intel and AMD wouldn't know the truth if it bit them on the @$$!

See any low stock warnings of any of Intel's current Quad Cores including Xeons?

[saaya]

so your saying intel delaying the launch of yorkfields to q1 08 doesnt have to do with the bug they mentioned?
and the bug might just be completely made up?

if thats true then intel would be REALLY stupid, cause mentioning there is a bug in your current products is hurting your name quite a bit. and most people who are in charge of server farms have no idea how serious a bug is and if it affects them or not, so they just hear there is a bug and want to get all their cpus replaced.

so im pretty sure, yes, there is a bug, and it must be serious enough that intel thinks people will find out about it and they prefer to go ahead and anounce it before somebody else does it for them.

[Donnie27]

so your saying intel delaying the launch of yorkfields to q1 08 doesnt have to do with the bug they mentioned?
and the bug might just be completely made up?

if thats true then intel would be REALLY stupid, cause mentioning there is a bug in your current products is hurting your name quite a bit. and most people who are in charge of server farms have no idea how serious a bug is and if it affects them or not, so they just hear there is a bug and want to get all their cpus replaced.

so im pretty sure, yes, there is a bug, and it must be serious enough that intel thinks people will find out about it and they prefer to go ahead and anounce it before somebody else does it for them.

I'm saying it or there might be some obscure errata that might affect 1 in 300,000 or 889,000 models shipped. So they'd have their Assets covered.

No they will NOT not because only some ding-bat IT folks will have NOT ran thorough testing on any Server joining their Farm/Cluster or whatever=P Again, I'm saying I hope it is a bug and or something real and not Intel milking the market. I agree with the folks saying that if Phenom wasn't Phenomenally Blah, then we'd hear nothing about this. It might be just another line on the errata list, if it is real.

[Donnie27]

Well, there appears to be confusion... some people think launch is Jan 6, others think later... Digitimes (rumor mill as well), has this:
http://www.digitimes.com/mobos/a20071206PD221.html

They are listing Quads on the map for launch....

I have seen Digitimes not get it right before and, for me anyways, is not enough to shoot down another HW site...

Intel also plans to introduce a 45nm product for its Core 2 Due E4000 series featuring 1066MHz FSB, 3MB L2 cache and a 65W TDP, however the model number is not yet finalized.

The so called Legacy Core 2's I heard the rumor about. One guy from Germany no less called them Core 2.5 LOL! Some guy said there was talk about Intel not making the same 478 mistake and not wanting others feeling left out like the 939 users. 1066 FSB and 65W TDP might make a great overclocker? They should also have a very nice price, why wouldn't they?

[Shintai]

Without competition theres no need to migrate to 45nm yet - they could clear out as much 65nm stock now on the mainstream products,then migrate fully to 45nm when the inventory is cleared - rather than selling the 65 and 45nm products side by side.

Thats actually not right. Shrink is more or less indenpendent of competition.

Remember, at 45nm is cost Intel about half to make the same CPU as on 65nm. So even a 4Billion+ $ plant quickly gets cheap.

Intel got plenty of stuff to use "old" technology on. They can hardly today ship enough chipsets and such due to production limits. And soon they will ship GPUs aswell.

And competition will stay in some sort. Even if Intel is a sole player on x86. If they get too lazy we change CPU arch. However a time without competition might actually be very good andf extremely healthy. Because we are now in a somewhat deadwater situation with a duopoly and a 35year+ legacy bagage.

In the long run we could leap ahead basicly on a new arch, remove alot of constraints and get more players to the new playing field.

[hollo]

^

imagine playing raytrace rendered games on a 64 thread 8-core sparc t2 system running linux or similar
as in http://www.pcper.com/article.php?aid=455&type=expert
and when you're not using all your processing power for rendering it's 100% available for general purpose computing (unlike a gfx card, well nVidia's ones at least)

i hope i'm not too far ahead of myself here

[terrace215]

So was this claim of a Yorkfield delay even true?

I've seen no additional confirmation yet... somewhat surprising if Intel really "informed all partners" a couple days ago...

[Donnie27]

Yeah, meanwhile. I wonder how close these two really are?

http://techreport.com/discussions.x/13742

The basis for the kernel patch solution depends on the root cause of the L2 eviction problem. The only exposure for the problem is when the TLB needs to set an A or D bit in a page table entry. If the TLB never needs to set an A or D bit, the bug cannot occur. By emulating the A and D bits with the help of the Present and Writable bits, the patch will ensure the real A and D bits are always preset. It works by forcing a page fault when the first access is made to a page with the emulated A bit not set, and when the first write access is made to a writable page with the emulated D bit not set. Emulated A and D bits are stored in bits generally available to the OS in the page table entry.

[Jacky]

Shintai,
maybe it's about getting rid of old 65nm processors at good prices and building inventory for 45nm. But why should they release dual cores but no quad cores then?
Anyway that's to much of a coincidence. I don't believe it until confirmed by another source/Intel (I really want my 45nm system in early feb! lol)
What about shipped Harpertown QCs? Stop ship, recall or does no one care?

[KTE]

So was this claim of a Yorkfield delay even true?

I've seen no additional confirmation yet... somewhat surprising if Intel really "informed all partners" a couple days ago...

What do you mean by confirmation? Outline what makes you definitely believe in news or makes you disbelieve it? Since you do post a lot of news and most of the time with too much certainty and postulates while it's preliminary rumors or guesswork by someone. But you seem to be doing the opposite now.

This is basically the same "instability bug" as what AMD said about Phenom but AMD has 2 desktop SKU's out and Intel has one -> all three have the instability bugs they claim.
No QX9650 user I've heard of has reported this.
No Phenom user I've heard of has reported this.

Alot of childish cries though.

But it seems way too concidental. Intel 'aint daft enough to claim something this stupid and loose maximum possible gains at the best time especially for the bugged Xeons where it really counts but then again, some could be bugged while not others and it could be just the excuse Intel was looking for to delay until more Phenoms. Market tactics have to be playing at least a 20% take here.

[terrace215]

What do you mean by confirmation? Outline what makes you definitely believe in news or makes you disbelieve it? Since you do post a lot of news and most of the time with too much certainty and postulates while it's preliminary rumors or guesswork by someone. But you seem to be doing the opposite now.

This is basically the same "instability bug" as what AMD said about Phenom but AMD has 2 desktop SKU's out and Intel has one -> all three have the instability bugs they claim.
No QX9650 user I've heard of has reported this.
No Phenom user I've heard of has reported this.

Alot of childish cries though.

But it seems way too concidental. Intel 'aint daft enough to claim something this stupid and loose maximum possible gains at the best time especially for the bugged Xeons where it really counts but then again, some could be bugged while not others and it could be just the excuse Intel was looking for to delay until more Phenoms. Market tactics have to be playing at least a 20% take here.

Childish cries? No OEMs are shipping Barcelona servers! That tells you that the AMD erratum is a show-stopper. The original TR report on it contained quotes from AMD officials. It's now in the financial press.

The Yorkfield thing is a blurb from 2 days ago on a french website. They claim Intel confirmed, but no quotes. I'm surprised that nothing has emerged on TR, DT, Fudzilla, TheInq, or any other site 2 days later...

[nemrod]

So Intel gave a "response of normand" (french expression)

We asked Intel to comment and received this response from the company's PR manager, Dan Snyder:

45nm Core 2 Quad launch is planned for Q1'08, and we are still on track for that. We can't comment on web speculation.

From january 20 up to march is still Q1...
http://techreport.com/discussions.x/13756

May mention too that hardware.fr has removed the "intel has confimed"

So rumour or secret

[mascaras]

http://www.xtremesystems.org/forums/...&postcount=306

regards