USB 3.0 problems for Intel's Haswell

[MaxPayne63]

But I really don't see how people can say it's a non-issue!

It's a non-issue because haswell hasn't shipped yet. Until that point everything is either FUD or marketing. Take a few deep breaths.

 

[Ken g6]

I'd say an OS fix is more likely than a hardware fix. MS should be able to have Windows freeze all filesystem operations to USB drives and disconnect them before sleeping. It could then undo all that on wake.

Of course, people using Linux will be stuck with problems with new hardware. But what else is new?

 

[Phynaz]

So has the original rumor now changed to this being a chipset issue and not CPU bug? But wait, I thought they had seen the documents?

So both their source and their own eyes were wrong? Or have I misread something?

 

[Phynaz]

But I really don't see how people can say it's a non-issue!

Remember the thread you created all freaked about Haswell being delayed a month? Well maybe this is why. Bug was found, needed additional time to be fixed, so that it wouldn't be an issue in a retail CPU.

Even if this bug were there in some engineering stepping, it could be a complete non-issue because it doesn't exist anymore.

 

[Abwx]

A chipset bug that necessitate a CPU new stepping....

 

[Piroko]

If I understood the issue correctly I'd rate it pretty high alongside the Sata issues of P67 and the TLB bug of Phenom I. It probably won't affect many people but those who trip over it will get very annoyed.

Also, bad press for Haswell, no matter if it's a big deal or not, isn't exactly ideal for Intel. Haswell has to span a lot of time if Broadwell will be mobile only.

 

[Idontcare]

The irony here is that Intel likes to advertise their home-grown USB prowess, so borking their native USB3 implementation is kinda sad IMO.

But we have yet to see if this actually pans out as a practical concern for products that actually hit the market. AMD's TLB bug with Phenom wasn't a practical concern in the consumer space but AMD (over)reacted as if it were.

Intel still has time to fix this (if the rumor is even true), or they may elect to leave it as is just like they did with the 24fps "bug" in their integrated GPUs.

 

[Elixer]

Meh.
If they haven't solved the issue once OEMs get mass shipments, then is the time to worry if your Intel or the OEM.

 

[Charles Kozierok]

But we have yet to see if this actually pans out as a practical concern for products that actually hit the market.

Not that that ever stopped anyone from freaking out over something, of course.

 

[IntelUser2000]

Obviously this will affect different persons to a different degree depending on what use cases are affected by this. But I really don't see how people can say it's a non-issue!

How many people you know that really does this? There's no data loss, and for those that use sleep extensively probably still have to use things like log in or whatever, so I bet they won't be bothered by videos not playing.

So for the person to be annoyed by it you have to meet certain criterias:

1. Use sleep, which is important
2. Have a USB device connected
3. Have a file already open pre-sleep from that USB device
4. Then wanting to resume work soon as the device is woken up. The user never closed his applications or anything

That's different from the Cougar Point chipset recall where it had potential for data loss, or the Pentium FDIV bug where the results could be inaccurate, or even AMD's TLB bug where system lockups can happen. All three can happen in widely used scenarios and not having it fixed could indeed result in really serious issues.

Compared to this one, where you get... sleep can't allow video to resume assuming you have it already open, and assuming its playing from a USB device. Big deal. My desktop already has minor intermittent issues where keyboard enter might not work to log in, and have to use the mouse. There's no news sites crying about that, is there?

IMO they need to at least address that it exists and a fix is coming but something like a recall is pushing it. There would be LOT of products that would be recalled everwhere if that was the case.

 

[NTMBK]

How many people you know that really does this? There's no data loss, and for those that use sleep extensively probably still have to use things like log in or whatever, so I bet they won't be bothered by videos not playing.

So for the person to be annoyed by it you have to meet certain criterias:

1. Use sleep, which is important
2. Have a USB device connected
3. Have a file already open pre-sleep from that USB device
4. Then wanting to resume work soon as the device is woken up. The user never closed his applications or anything

That's different from the Cougar Point chipset recall where it had potential for data loss, or the Pentium FDIV bug where the results could be inaccurate, or even AMD's TLB bug where system lockups can happen. All three can happen in widely used scenarios and not having it fixed could indeed result in really serious issues.

Compared to this one, where you get... sleep can't allow video to resume assuming you have it already open, and assuming its playing from a USB device. Big deal. My desktop already has minor intermittent issues where keyboard enter might not work to log in, and have to use the mouse. There's no news sites crying about that, is there?

IMO they need to at least address that it exists and a fix is coming but something like a recall is pushing it. There would be LOT of products that would be recalled everwhere if that was the case.

Students doing work from USB sticks, for a start. This is very, very common, as its an easy way to carry work from home to the uni network machines.

 

[IntelUser2000]

Students doing work from USB sticks, for a start. This is very, very common, as its an easy way to carry work from home to the uni network machines.

Yea, like it says there, it would be relevant if you want to continue work on an pre-sleep application straight from wake up. Then a solution isn't even a reboot, but a mere restart of a program, with no other adverse effects. Wow, what a disaster.

Not to mention if the XS post is to be believed, it only affects certain setups and is intermittent.

 

[NTMBK]

Yea, like it says there, it would be relevant if you want to continue work on an pre-sleep application straight from wake up. Then a solution isn't even a reboot, but a mere restart of a program, with no other adverse effects. Wow, what a disaster.

Not to mention if the XS post is to be believed, it only affects certain setups and is intermittent.

Please point me to where I said it was a disaster? Stop snapping at people and assuming they're trying to undermine your precious Intel.

Be nice now...
-ViRGE

 

[IntelUser2000]

Please point me to where I said it was a disaster? Stop snapping at people and assuming they're trying to undermine your precious Intel.

I never snapped at you. But maybe you referencing my name for a personal insult tells something about you instead.

I was just talking in general as I usually do.

 

[Turbonium]

Honestly, this news (though it may be false) was a huge reason behind me deciding not to wait for Haswell and just going Ivy Bridge. I mean, assuming the bug is real, it probably will introduce a further delay of some sort until we can all get our hands on working Haswell systems.

Waiting until June to build a system is just

 

[NTMBK]

I never snapped at you. But maybe you referencing my name for a personal insult tells something about you instead.

I was just talking in general as I usually do.

Sorry, "snapping" was the wrong phrase. But the "wow, what a disaster" and sarcastic eyeroll was what got my hackles up. I've not made any attempt to turn this into some great calamity that will sink Intel. Look back at what I said on page 1, I said to wait and see. I just disliked your misrepresentation of my stance on the matter.

 

[sm625]

lol you just try putting any windows 7/8 machine to sleep while playing a video. When it wakes up, IF it wakes up, it may resume where it left off. Or it may not. It is by no means a guaranteed thing. How well would you trust it to work? This crap barely works, its always been that way. And apple aint no better in that regard.

 

[Abwx]

Just tested with an old Pentium 4 laptop and WXP...

When waking up the media player effectively start where it was halted...

 

[Fjodor2001]

How are problems like this normally handled by the hardware manufacturers?

Let's say I want to buy a computer with a Haswell CPU (and related Lynx Point chipset), how can I in advance know if it has this HW bug or not? How will they indicate when the computers that have this HW bug has been cleared out of inventory? Will they change the model number or similar?

I guess a lot of people don't want to gamble and buy a Haswell computer and just hope that it doesn't have this problem...

 

[Abwx]

How will they indicate when the computers that have this HW bug has been cleared out of inventory? Will they change the model number or similar?

The informed people will wait for a new stepping , the general
insuspecting public will be bound by implicit contracts stipulating
that by buying such items they accept the product as it is with its
(un)known issues.

 

[Fjodor2001]

The informed people will wait for a new stepping

But how do you know when the old stepping has been cleared out of inventory?

 

[Abwx]

What matters is that it is cleared down to the retailers inventory.
A new revision is surely already in the making , they will need
a few months to redesign very slightly a minor part of the CPU ,
create a new mask and then manufature the new chips
in a two month manufacturing cycle.

We can expect more than six months but less than a year
for the new revision to be the only sold iteration.

 

[Fjodor2001]

What matters is that it is cleared down to the retailers inventory.
A new revision is surely already in the making , they will need
a few months to redesign very slightly a minor part of the CPU ,
create a new mask and then manufature the new chips
in a two month manufacturing cycle.

We can expect more than six months but less than a year
for the new revision to be the only sold iteration.

So perhaps about 6 months before Broadwell is released?

That is if you're willing to bet on that the store you buy from does not keep some old Haswell stock of the Ultrabook you intend to buy...

 

[ShintaiDK]

What matters is that it is cleared down to the retailers inventory.
A new revision is surely already in the making , they will need
a few months to redesign very slightly a minor part of the CPU ,
create a new mask and then manufature the new chips
in a two month manufacturing cycle.

We can expect more than six months but less than a year
for the new revision to be the only sold iteration.

You do know its the chipset and not CPU...right? And it took 3 weeks to fix the SATA issue in the 6 series chipset. Also this bug might not ship to customers at all. or be fixed via software.

 

[Rifter]

You guys are blowing this WAY out of proportion.

1. This may or may not even be a real issue that may or may not make it to retail at all.

2. Even if it does a microcode fix might be all that is needed.

3. Have any of you tried to sleep and then resume in the middle of a video in win7/8? it is by no means a sure thing it will work correctly even with current hardware that ISNT effected by this bug.

You guys get so worked up over nothing its funny sometimes.