Saturday, April 6th 2013

Intel Fixes 8-series Chipset USB 3.0 Erratum

Intel issued a product-change notification (PCN) to partners in the motherboard industry, informing them of a new C2 stepping of the 8-series "Lynx Point" chipset. The stepping introduces a metal layers change that fixes the USB 3.0 remuneration erratum, which causes devices plugged in to USB 3.0 ports to fail to reinitialize after waking up from sleep states such as S3, requiring uses to unplug and replug them, which could get particularly irritating for people with external RAID devices that rely on USB 3.0 for host connectivity.

According to the PCN, the first socket LGA1150 motherboards in the market may not feature C2-stepping chipset. It predicts samples to be available to motherboard manufacturers by April 19, 2013; availability of qualification data (when Intel has finalized design after taking feedback from partners), by July 1 2013; customers should be ready to receive C2-stepping chips by July 31. These dates indicate that the very first batches of socket LGA1150 motherboards will still feature C1-stepping chipset that are affected by the erratum, and if you can't live with it, you should ideally wait until late-August or mid-September for the first motherboards with C2-stepping chipset to make it to the markets. Intel expects to launch its 4th generation Core "Haswell" processors and compatible LGA1150 motherboards by mid-June.
Add your own comment

40 Comments on Intel Fixes 8-series Chipset USB 3.0 Erratum

#1
buildzoid
The highest end parts argument is BS and i7 3770k is only 100mhz faster than a i5 3570k in single thread processes also the highest end part currently offered by intel is the 3970x and that thing is equal to a 3770k in single threaded so. Also what if you wanna game and also render/edit movies. Well for rendering a Xeon 2687w is great but it's clocking in at 3.2Ghz at which point it will get beat in gaming by an i5 3570k so if you want both then you find an i7 3930k/3960x/3970x and take it to 4.5Ghz at which point they are great for gaming but they also can render at a speed equal to that 8 core Xeon. Because currently as CPU price goes up the core count goes up and the frequency either doesn't change or drops.
Posted on Reply
#2
cadaveca
My name is Dave
by: RejZoR
Eh? You want a guaranteed overclock? It's called "buying a higher end CPU model". For low and mid end, no one guarantees you anything but in most cases you can get massive free gains. So what you're saying is "i don't want overclocking, i want only high end CPU's". Sorry, but i can't agree with you.
No, I did not say I wanted "Guaranteed" overclock...I want WARRANTIED OC.


Which Intel has.


By the way, I ONLY have high-end CPUs....FX-8350, i5 3750K, i7 3770k, i7 3960X. One of those has no warranty for OC, the other three do.

by: Ferrum Master
And btw you remember that on ancient times we oc'ed by changing clock gen crystals? So where is the problem? OC is a privilegy for those who can and know, not for those who buy a K series CPU and simply toggle the multiplier up... it feels like cheating...
Yes, I would like real OC to take some skill, and not cost money only. Because that's what it is right now, and has been for many years. I STILL solder shit up...because I like to push further than the average folks.




Errata isn't that big of a deal, but with P67, it was drive controller that might die, now it's USB that doesn't work with some drives. This is far more minor than it was with P67, so no, I'm not really that concerned about it. And you shouldn't be either.

by: buildzoid
The highest end parts argument is BS and i7 3770k is only 100mhz faster than a i5 3570k in single thread processes also the highest end part currently offered by intel is the 3970x and that thing is equal to a 3770k in single threaded so. Also what if you wanna game and also render/edit movies. Well for rendering a Xeon 2687w is great but it's clocking in at 3.2Ghz at which point it will get beat in gaming by an i5 3570k so if you want both then you find an i7 3930k/3960x/3970x and take it to 4.5Ghz at which point they are great for gaming but they also can render at a speed equal to that 8 core Xeon. Because currently as CPU price goes up the core count goes up and the frequency either doesn't change or drops.
3770k and 3960x/3970x are NOT equal in single-threaded workloads. IVB has faster cache, which makes for most of the performance difference between SNB and IVB.
Posted on Reply
#3
KithKhan
Consumer Ignorance is Bliss

What portion of the populace is going to be aware of whether or not they're buying a C1 or C2 chip? Probably close to 0%?

Also, what portion of the populace is going to connect the dots when (if...?) they have USB 3.0 stuff plugged in after it sleeps? Again, probably close to 0%? They're going to blame Dell or Windows for a while, and eventually they're just going to disable sleep mode or leave their USB 3.0 stuff unplugged much of the time.

These chips are going to sell fine. Intel has already invested money in producing the first batch, and they're going to make a bundle selling them. Good call Intel! $$

PS. That's some clean looking soldering! Props. What exactly does it do in the case of the picture above?
Posted on Reply
#4
cadaveca
My name is Dave
by: KithKhan
PS. That's some clean looking soldering! Props. What exactly does it do in the case of the picture above?
Thanks! ;)

That's an ASUS HD 7970 Direct CUII. It comes with spots to attach the wires to already, for vCORE, vPLL, and vMEM. One wire for each(colored), plus one for measure. I also added the VRM control wires and used an alternate point for the vCORE mod.

I love soldering, not sure why. Ended up removing all of that, then soldering this on:

Posted on Reply
#5
Jurassic1024
by: The Von Matrices
Is it unusual to be on the 3rd major version of the chipset (i.e. "C")? I can only remember Intel having A and B chipset (not processor) steppings before. I would think that having to do two major revisions of the chipset would have been responsible for a large delay. Am I reading too much into this?
Yes you are. Revisions if anything are minor, not major. It's rare for AMD or nVIDIA to release A1 revision silicon first. If it happens, it's a miracle. Revisions are tweaks, to help with keeping production costs down, and lower power consumption etc.

Look at console revisions for example. None of those revisions made your console any better than anyone elses, just yours may run a bit cooler due to a CPU or GPU shrink, or an added heatsink, etc. Performance-wise there was no difference, so what's major about that?

Also, I got a B3 revision Intel P67 chipset motherboard... does that mean mine runs any better than a B2, or is mine just less likely to fail?

Nothing "major" about any of the above scenarios.
Posted on Reply
#6
cadaveca
My name is Dave
by: Jurassic1024
or is mine just less likely to fail?
Less likely to fail.


However, word was with P67 that EVERY chipset would fail within warranty periods and make the system unusable, hence the recall.

But you are right, usually errata are minor, every part has some, but P67 B2's issue was hardly minor. They do not offer world-wide recalls from every brand for minor issues.
Posted on Reply
#7
Jurassic1024
by: cadaveca
Less likely to fail.


However, word was with P67 that EVERY chipset would fail within warranty periods and make the system unusable, hence the recall.

But you are right, usually errata are minor, every part has some, but P67 B2's issue was hardly minor. They do not offer world-wide recalls from every brand for minor issues.
Um, that was a rhetorical question. :|

LMAO@word was. How about you do some actual research. It's not hard. Maybe if you stop fiddling with posting graphics card mod pics under an article about motherboard chipsets, you'd be able to actually contribute with more than, "word was" in your replies on the subject.

B2 was minor. It would take a few YEARS and a lot of disk activity for those Sata III controllers to degrade.
Posted on Reply
#8
LAN_deRf_HA
by: cadaveca
I mean...you're asking a bit much.
In this case I don't think so. I mean if you're buying a storage device now it's going to be USB 3, and the average person makes use of sleep these days. Even if it were obscure it would bug me to know I was buying something flawed, but in this case a great many people will notice. It's a little ridiculous they're trying to sweep this under the carpet. I can picture it now all the people complaining on WD forums about their external storage always going MIA, then getting directed to Intel, and then thinking well this stinks. Even more so when they find out it was fixed before it was ever released but delayed so they could ship some defective inventory at your expense.

Even an always on enthusiast is going to be annoyed when they have to resell their stuff for 70-85% the price of a C2 version.
Posted on Reply
#9
cadaveca
My name is Dave
by: Jurassic1024
How about you do some actual research.
I don't need to do that sort of research. As the motherboard reviewer here, this info was given to me by OEMs. It was also matched with info given out by Intel. What's the big deal?

Do I know of people that had B2 boards die? Do I have a dead one here on my shelf?


YUUP!


:roll:


Research. :p :laugh: I leave that to the pros.:roll:
Posted on Reply
#10
Am*
by: TRWOV
LOL it's the B3 thing again
While I do not in any way approve them still launching this broken chipset, it's nowhere near that bad. B2 rendered SATA ports of your motherboard unusable. This is just a minor inconvenience (having to replug USB 3 devices on wake). Most of the USB 3.0 controllers currently in use on motherboards are garbage anyway. The biggest problem I see is for laptop users who may only get USB 3.0 or the vast majority of ports as USB 3.0. They will be pretty pissed and I don't blame them.

Speaking of USB 3.0, I avoid the shitty Etron USB 3.0 controller on my mobo whenever I can -- half the time, it does not recognise the most common devices, which aren't even USB 3.0. Even had a BSOD thanks to this turd once. The ones I've seen on AM3 motherboards are just as bad, if not worse. I really can't see the point in USB 3.0 anyway -- if you have a device that needs that sort of bandwidth, go with eSATA or Thunderbolt/Lightning instead of wasting your CPU cycles on shitty USB 3.0 controllers.
Posted on Reply
#11
TheHunter
Wait, so even though they fixed it they will still ship "broken" chipset first?
Posted on Reply
#12
cadaveca
My name is Dave
by: TheHunter
Wait, so even though they fixed it they will still ship "broken" chipset first?
Yes, because the problem only affects a single USB 3.0 controller, that is used in a limited number of consumer drives. Simply put, there is a an incompatibility that required a hardware change. They found the problem, have fixed it, and are will be shipping revised chips to OEMs in two weeks, apparently, for testing. So Intel must be making them right now.

Of course, the platform hasn't launched yet.


But, OEMs need to build boards and test products (qualification testing), so the revised edition of the boards will take some time before they hit the market. You can find out how long qualification testing takes pretty easily.
Posted on Reply
#13
RCoon
Forum Gypsy
by: Protagonist
If I were Intel i would delay the launch till issue is ironed out
Problem is motherboard vendors likely have stacks of MoBo's by this point that they've made up in preparation. If intel denied the option of selling defective boards for this issue, then those vendors will have lost out on all the money it cost them to make the boards.
I wouldnt be surprised if we saw half of these C1 boards get sold without USB 3.0 headers/ports on them, as the basic range of motherboards. If there's no connectivity options, there's no problem. And plenty of domestic users arent gonna care about a blue USB port on their computer. Half of the household computers i see today are barely past the Core2Duo stage.
They probably wont even mention the issue on most shopping sites anyway, they will "expect" buyers to check the C1 or C2 revision for themselves and be wise enough to choose accordingly. Otherwise its their own fault for buying a motherboard that has a fault.
Posted on Reply
#14
jihadjoe
by: Steevo
Most of us are waiting on software to catch up with SMP.
So if you could get that taken care of that would be great.......
/holds coffee cup
yeah.....
Isn't TSX one of the things Intel is doing with Haswell to help with that? By implementing fine grained locking control at the hardware level, developers don't have to do it in their own code. This helps improve multithread/multi-core performance even with code that isn't very well optimized for either.
Posted on Reply
#15
NeoXF
...'K... now show me some non-prototype black/red gamer/OC/enthusiast grade boards based on these. ASUS and ASRock are my first choices, but MSI didn't look half bad last time I checked either.
Posted on Reply
Add your own comment