USB 3.0 problems for Intel's Haswell

[Fjodor2001]

http://uk.hardware.info/news/33716/usb-30-problems-for-intels-haswell

"In the document, Intel purportedly informs manufacturers that a system with Haswell inside, when waking from S3 sleep mode, will experience issues with devices connected through USB 3.0. For example this would result in blank pages displaying in Acrobat Reader PDFs or video playback stopping, rather than continuing. A quick fix would be to restart the failing application.
[...]
The company indicates the issues will not impact the release of the Haswell processor series, but it is of the intention of resolving this problem in a future processor stepping."

If this is true, you might want to wait a few months after release until the next Haswell CPU stepping is on the market that fixes the problem...

 

[Fjodor2001]

BTW: Doesn't anyone find it strange that Intel considers it ok to release Haswell with such a hardware bug?

 

[TuxDave]

BTW: Doesn't anyone find it strange that Intel considers it ok to release Haswell with such a hardware bug?

Depends if this a bug with an pre-production stepping that they're sending out to partners for validation/review which will be fixed in a production stepping. Or if this bug be patched with an OS fix otherwise? If it's any of the two above, I'm not opposed to it.

Disclaimer: I'm trying my best to not be biased.

 

[Dark Shroud]

Yeah I have too many USB devices that I depend on. Looks like I'm going Ivy Bridge Extreme or Broadwell for my gaming rig.

 

[SomeoneSimple]

BTW: Doesn't anyone find it strange that Intel considers it ok to release Haswell with such a hardware bug?

No, not really. They have contracts and legal obligations to deliver a number of Haswell CPU's in a certain timeframe.

They can't just stop their CPU presses and throw away every Haswell they already have in stock. Letting their OEM partners wait until Intel completed preparing their fabs for a new stepping and starts pumping them out again. :hmm:

 

[Phynaz]

BTW: Doesn't anyone find it strange that Intel considers it ok to release Haswell with such a hardware bug?

Noticed your linked article doesn't name a source. Makes it rather suspect.

 

[blastingcap]

I wonder who paid to spread the anti-Intel FUD around... or maybe it's a slow news day and people want to make stuff up to get hits.

 

[videogames101]

Where are all those people telling us that IB-E is delayed because Intel has no competition? That Intel would never need that much time to validate server components?

Back to reality folks.

 

[RU482]

this kind of leak is funny to me

 

[Insert_Nickname]

Noticed your linked article doesn't name a source. Makes it rather suspect.

It has happened before. Remember the 6-series chipset debacle?. But I agree, no source makes it suspect.

 

[TuxDave]

this kind of leak is funny to me

I kind of wish they could describe a hint of about the source to set some level of credibility. Is this an engineer or some business major? Or is this just some undergraduate intern that wanted to show off to his friends that he has secret knowledge and overheard something and only caught on some key words but really has no idea what's going on?

This is what Hardware.Info learnt from a reliable source

I have read many leaks from "a reliable source" and some of them are so terribly wrong.

 

[Fjodor2001]

This is what Hardware.Info learnt from a reliable source...

I have read many leaks from "a reliable source" and some of them are so terribly wrong.

You skipped the rest of the part of that sentence from the article which said "...and saw confirmed in an internal document from one of the processor manufacturing giant's partners.". If Hardware.Info actually saw it confirmed in such a document I think that adds quite a lot of credibility.

Also, they are by that hinting at the source as being "one of the processor manufacturing giant's partners".

 

[ViRGE]

BTW: Doesn't anyone find it strange that Intel considers it ok to release Haswell with such a hardware bug?

Not at all. Processors ship with errata. Impressively large lists of errata. Intel will issue a microcode fix before Haswell even ships and this will be the only time anyone ever hears of it.

 

[TuxDave]

You skipped the rest of the part of that sentence from the article which said "...and saw confirmed in an internal document from one of the processor manufacturing giant's partners.". If Hardware.Info actually saw it confirmed in such a document I think that adds quite a lot of credibility.

Also, they are by that hinting at the source as being "one of the processor manufacturing giant's partners".

It's still missing the part that I'm asking. What qualification does the PERSON have that handed over the document? Is he technically capable to interpret it and confirm that production silicon will have uncorrectable USB 3.0 problems that the end user will see?

I'm not really questioning whether this leak is reliable or not. I'm just saying I wish they would at least say if this person actually knows what he's doing or if this is just some intern that got his hands on some slides.

 

[podspi]

The effects don't make sense to me. PDFs not displaying, videos not playing? If USB 3.0 devices don't reconnect after the sleep, that is very annoying and could cause data loss.

But: Are USB 2.0 devices affected? And is this leak even legit?

 

[Fjodor2001]

The effects don't make sense to me. PDFs not displaying, videos not playing?

Possibly that assumes they are stored on some USB memory stick, and the "connection" to the USB memory is broken on a SW level after resuming from sleep, so the PDF/video on the USB memory cannot be accessed without triggering a reconnect by restarting the application? Something like that perhaps?

But I agree more details would be nice of course...

 

[exar333]

Likely this will be fixed before release. Nothing to see here, move on. This would be news if it was 2 days before release.

 

[ShintaiDK]

Possibly that assumes they are stored on some USB memory stick, and the "connection" to the USB memory is broken on a SW level after resuming from sleep, so the PDF/video on the USB memory cannot be accessed without triggering a reconnect by restarting the application? Something like that perhaps?

But I agree more details would be nice of course...

Still doesnt make sense, such data(PDF) would never be on the USB since its either stored in memory(sleep) or the HD/SSD (hibernation) after the initial load.

 

[Abwx]

Likely this will be fixed before release. Nothing to see here, move on. This would be news if it was 2 days before release.

This will be fixed this way :

It is worth noting that Intel is requesting its partners to accept this issue, before it will supply them with processors.

Take or go see elsewhere..

 

[Grooveriding]

Nothing new. SB-E released to market with an errata they didn't fix until the C2 stepping chips.

 

[gevorg]

its hard out there for the king

 

[cytg111]

Dont use hibernate at all, so dont really care .. but haswell is aiming this chip at the mobile segment right? Pretty darn big bug if you cant put your laptop to sleep properly.
Which equals that this is not in final silicon. Cant be.

 

[Tsavo]

This could cost Intel 10's of millions of dollars. Maybe 100's of millions of dollars. 10's of 100's of millions of dollars. 100's of 10's of millions of dollars.

 

[NTMBK]

Wait and see. Could be something which requires you to throw out your motherboard, could be something which could just require a BIOS update.

 

[inf64]

While this is bad news since nobody(OEMs) wants to deal with issues like that ,I guess they can patch it out somehow or make some workaround. I wonder what went wrong in the design verification process when a bug like that can creep up unnoticed. Intel has the best tools in the industry,there is no doubt about it.