Report: New chipset revision to fix Haswell USB 3.0 bug

Earlier this month, rumors started floating around about a USB 3.0 bug in next-gen Haswell systems. Hardware.Info, which broke the story, said the bug causes USB 3.0 devices to act up when Haswell machines awake from an S3 sleep state. The site went on to say Intel planned a new CPU stepping to fix the problem.

Well, that last part may not be accurate. Word is now that a new chipset stepping will take care of the issue. Both Hardware.Info and Fudzilla are saying so, with the former citing the latter. The citation suggests the original rumor was indeed incorrect, at least about the CPU needing a new stepping.

A chipset re-spin does sound like a more plausible fix for a USB 3.0 problem. Of course, from an end user’s standpoint, the effect should be largely the same. Since Intel is reportedly proceeding with its early June launch plan, folks may have to put up with wonky USB 3.0 in the first wave of Haswell PCs.

Comments closed
    • briskly
    • 9 years ago

    Intel plans to have a fix out and available by early August. You’ll have [i<]that[/i<] issue sorted out. Source: [url<]http://qdms.intel.com/dm/d.aspx/C79FC2E6-6B75-4063-8687-660F4668FFC8/PCN112101-00.pdf[/url<]

    • JosiahBradley
    • 10 years ago

    Why not just disable the USB3 portion from the chipset and ship boards with a 3rd party module thrown in somewhere; or, if past board development time, give early customers a free PCI-E 1x USB3 controller for the trouble, on Intel’s dollar of course.

    • BaronMatrix
    • 10 years ago

    If it had been AMD it would be because they can’t do better… And people would scream “don’t buy…”

    It makes me sick…

    • NeelyCam
    • 10 years ago

    No.

    • MadManOriginal
    • 10 years ago

    You’ve never seen [url=http://en.wikipedia.org/wiki/History_of_the_World,_Part_I<]History of the World, Part 1[/url<]?

    • Deanjo
    • 10 years ago

    There was no hw firewall with the Nforce 5 series.

    • Airmantharp
    • 10 years ago

    Intel’s controllers have been the rare exception- so this really is sad. At least they own up to it, most of the controllers just plain suck, but they use them anyway.

    • Mr. Eco
    • 10 years ago

    [url<]http://www.theinquirer.net/inquirer/news/1009010/nforce-chipset-corruption[/url<] Without looking any further I remember there were issues in NVidia chipsets with the embeded firewall and the sound controller. But the worse of all was the data corruption issue. I also had Athlon XP with NForce2 chipset, and had to use specific drivers to avoid the crashes on disk access.

    • Krogoth
    • 10 years ago

    Nvidia chipsets work for the most part, but several of them had their own stupid little issues.

    Nforce 1: Just bland and dated at the time, but it is expect for the first-generation.

    Nforce 2: There are some driver issues with PATA controller if you use Nvidia drivers under a RAID configuration.

    Nforce 3/4: The hardware firewall for southbridge was broken if you use it (corrupted packets). The standard NIC driver work fine though. The northbridge was toasty which meant that vendor threw in el cheapo HSF on them that often failed and you need a massive passive heatsnik to keep it cool. They was massive outrage over the loss of “Soundforce” APU.

    Nforce 5: I don’t recall anything bad for this chipset, correct me if I’m wrong. I do believe it was inferior in terms of performance and features on the Intel font. It still had the same firewall problem. The only gimmick it offer was SLI support.

    Nforce 6: Issues with memory speed compatibility and Core 2 Quad chips on the Intel front.

    Nforce 7: Nvidia finally got it right, but it was already too late. Nvidia gave up on chipset market.

    • NeelyCam
    • 10 years ago

    Damn that was awful. Like Benny Hill awful

    • BoilerGamer
    • 10 years ago

    And you won’t have to, the Chipset respin should only take the Intel Fabs a quarter max so as this is already getting fixed right now the revision Chipset will probably start showing up in [b<]July[/b<] bearing the designation Z87 Rev2 or R2.

    • BoilerGamer
    • 10 years ago

    Kaveri is end of the year, this is coming in June so the initial batch of buggy Haswell Chipsets would have been cleared out of the market way before then.

    • Prion
    • 10 years ago

    [quote<]folks may have to put up with wonky USB 3.0[/quote<] Business as usual, then.

    • Sahrin
    • 10 years ago

    Especially when they [s<]intentionally delayed[/s<] wrote the standard. Gotta sell Thunderbolt somehow.

    • smilingcrow
    • 10 years ago

    Hardly likely as have you seen the cost of the TB chips? OEM’s didn’t like paying the peanuts to license Firewire and the chips themselves weren’t that expensive. TB is overkill for mainstream computing right now although Apple love it but they like stuff that means they can have less ports and cables.
    I’m happy with USB 3 and think TB is good tech but not mainstream right now.
    If Intel release a sub 10nm SoC for smartphones with built in TB for docking then it will be mainstream. We will see.

    • DeadOfKnight
    • 10 years ago

    Interesting theory, you may be right.

    • yenic
    • 10 years ago

    Kaveri here.

    • Duck
    • 10 years ago

    I had some nforce2 Athlon XP setup. This is the first I am hearing of any chipset issues.

    • kroker
    • 10 years ago

    [quote<]Intel fixes a bug before the CPU is even released, and you're angry at them?[/quote<] I don't think he's angry about that, I think he's angry about the part where Intel still wants to sell the chipsets for which the bug [i<]isn't[/i<] fixed...

    • Deanjo
    • 10 years ago

    Yup, same thing here. Zero issues with Nforce 2, 3 Pro, 570, 780a, 8200 chipsets here. None, zero, nada, zilch.

    • Prototyped
    • 10 years ago

    Speak for yourself — my PC happily uses an nForce 720a (aka GeForce 8200 mGPU) with a Thuban. Solid.

    (Yes, I’m still using five year old technology. What’s it to you? :p)

    • HisDivineOrder
    • 10 years ago

    This is why I’m really glad nVidia was forced out of the chipset business. They couldn’t make a reliable chipset for AMD or Intel to save their life. At least, not one I could find.

    • HisDivineOrder
    • 10 years ago

    You may change your mind about hibernation as SSD’s become more prevalent and faster. Still, I agree that for now Sleep has always been far more reliable and useful to me than Hibernation.

    Keep in mind any bug in Hibernation would affect Windows 8 users pretty horribly since Windows 8 does not “Shut down” when you “Shut Down,” it hibernates a lot of the system and logs out the user. A lot of users won’t know they’re actually using HIbernation all the time.

    This bug’ll sneak up and bite ’em on the butt.

  1. They had better have the new chipset revision out by the holiday shopping season. I plan to do an upgrade then but I am not going to buy something with a buggy chipset. I have had to deal with that way too much already (cough nvidia cough). No one should ever have to deal with any chipset bugs or any bugs that could possibly corrupt data.

    • HisDivineOrder
    • 10 years ago

    Because they really, really don’t want to. They delayed moving to USB3 for years. It wasn’t because they couldn’t get it working. It was because they’d prefer everyone use Thunderpants…er… I mean, Thunderbolt.

    • HisDivineOrder
    • 10 years ago

    Building a rig and selling a laptop, they seem relatively unconnected. If you sold your laptop and built yourself a desktop, you can still buy a Haswell laptop later.

    • HisDivineOrder
    • 10 years ago

    Oh, Phenom. You will always be remembered “fondly.”

    • Farting Bob
    • 10 years ago

    Intel fixes a bug before the CPU is even released, and you’re angry at them?

    Also, AMD has a far from perfect record of actual shipped products that contain defects.

    • AlumarX
    • 10 years ago

    I’m not willing to wait 3 months. I’m selling my laptop in 2 days and I’m building a Z77 rig that same day.

    • Wildchild
    • 10 years ago

    I don’t think I would ever run into this problem, but the idea of giving Intel money for something that they knew was broken just drives me crazy.

    Here’s to hoping Steamroller is competitive.

    • MadManOriginal
    • 10 years ago

    [url<]http://25.media.tumblr.com/tumblr_lz3gd4LugV1r6i3cdo1_500.jpg[/url<]

    • Deanjo
    • 10 years ago

    Why would they officially announce a revision of an unreleased product? Stuff like this happens all the time pre-release. It is rare that it actually admit anything had to be fixed unless it is released to the wild like what happened with the SATA issue of a few years ago that could be found in released product.

    • MadManOriginal
    • 10 years ago

    NVidia droolz, Qualcomm roolz!

    Or something like that.

    • MadManOriginal
    • 10 years ago

    Considering Intel hasn’t officially stated anything regarding this, we will just have to wait and see, won’t we Mr Negative?!

    • Deanjo
    • 10 years ago

    [quote<]Like I've said, a software solution might not completely take care of every circumstance, but I think a software solution could take care of the vast majority of problems.[/quote<] As you said, we do not have enough information to say 100% for sure but I would think that if it was a possibility, intel would have taken care of the situation that way. I very much doubt intel engineering would all of a sudden forget that some issues can be worked around via software. I don't think that intel would forget a cheaper alternative solution.

    • MadManOriginal
    • 10 years ago

    tl;dr version:
    [url<]https://www.youtube.com/watch?v=KuMQjKiaDTg[/url<] extended cut: [url<]https://www.youtube.com/watch?v=StJS51d1Fzg[/url<]

    • Farting Bob
    • 10 years ago

    What you going to do? Use AMD??? HAHAHAHAHA!!

    • MadManOriginal
    • 10 years ago

    That requires more information than I, or you, have at this point. Is the device actually disconnected during sleep? Does the access problem crop up during the sleep routine, or during the wake routine, or (for Connected Standby) *during* standby/sleep itself? If a device is still ‘connected’ during sleep for USB Wake, but just needs to be reinitialized for access when awake, a software solution would work fine. Or maybe USB wake would just not be possible with a software fix – not good, but that doesn’t mean a software solution shouldn’t be pursued for situations where that doesn’t apply.

    Like I’ve said, a software solution might not completely take care of every circumstance, but I think a software solution could take care of the vast majority of problems.

    My idea isn’t incredible, I’m 100% sure someone smart at Intel can think the same thing. They just haven’t replied in any meaningful way yet, maybe when they do before the launch we’ll know whether a software solution is possible. And even if it is ‘possible’, whether it happens would depend upon how conservative or risky – based on the risks of important things like data corruption – Intel is willing to be.

    • Deanjo
    • 10 years ago

    And if your settings are to wake from a USB event what do you propose? Disconnection and redetection of the controller?

    • MadManOriginal
    • 10 years ago

    It’s good to be the king.

    • MadManOriginal
    • 10 years ago

    Read my other reply – it would be part of the wakeup routine and utilize the hot-plug nature of USB. The problem doesn’t occur when going to sleep [er se or ‘during the sleep state’ (how could a problem occur ‘during the sleep state’*?) but when waking from sleep.

    *perhaps there’s more of a problem with Connected Standby, but again, a software driver workaround should be able to handle that.

    • MadManOriginal
    • 10 years ago

    No doubt fixing the hardware is the right thing to do. I just wonder why a software fix couldn’t take care of most of the problem.

    I’m thinking of USB devices and I can basically come up with a few broad categories: 1) storage (thumb drives, hard drives, etc) which are accessed in a discrete nature and 2) ‘real-time’ devices (networking, audio devices, audio/video streaming) which are accessed in a continuous, on-demand way. For 1) reconnecting the device would take care of the next discrete access before it happens, for 2) there could be a problem but well-behaved programs will pause or stop streams before sleep. If the reconnect is part of the wake routine then they’d be ready by the time the user does anything with them.

    To put it another way, a software solution would just use the hot-plug nature of USB to bypass the hardware problem.

    • Deanjo
    • 10 years ago

    You are absolutely right, one of the issues however is that the layman user does not see ugliness that has to occur to get things working in Windows with it’s rather poor logging hardware issues. How many people knew for example that many hard drives in the past advertised an ATA spec only to have it dropped to a lower level because of drive blacklisting in the ATA drivers to avoid compatibility and reliability issues.

    • chuckula
    • 10 years ago

    [quote<] I don't see AMD flubbing their chipsets[/quote<] AMD would like to publish an official document that disagrees with your assertion: [url<]http://support.amd.com/us/ChipsetMotherboard_TechDocs/48671.pdf[/url<] Since you think USB 3.0 is so trivially simple, you obviously know that XHCI refers to the host controller for USB 3. Go read the errata that AMD knows exists in its chipsets [b<][i<]and that AMD has officially and publicly stated will not be fixed[/i<][/b<]. Anyone who runs around claiming that chipsets from company A are perfect while chipsets from company B are somehow fatally flawed is a delusional fanboy who only wants to argue with strawmen instead of being grounded in reality.

    • Deanjo
    • 10 years ago

    [quote<]Couldn't there be a simple software or driver patch for this?[/quote<] Since the issue lies while in S3 sleep state when no OS is actually active the answer would be no.

    • someuid
    • 10 years ago

    Aside from your opinion that I don’t know what simple means, and your perception to how many times I keep using the word simple, you still haven’t answered my question.

    Why can’t Intel, the desiger of the most powerful x86 CPUs and industry leader in semiconducter design and smaller nm manufacturing processes, get something as run of the mill as USB 3.0 to work properly? This isn’t a problem with anyone else. I don’t see AMD flubbing their chipsets and they have 2x-3x more SATA 6, USB 3.0 and PCI lanes as Intel. I don’t see anyone else releasing their flagship products and the only suggestion is “live with it until we respin or step the chipset. Thanks for cash.”

    USB 3.0 and sleep states are not undocumented technologies that Intel had to reverse engineer on their own.

    • Voldenuit
    • 10 years ago

    From the hardware.info article:

    [quote<]It is worth noting that Intel is requesting its partners to accept this issue, before it will supply them with processors.[/quote<] So, accept and ship our buggy chipset, or we'll starve you of the new CPUs and put you on the backlog for the rev2 chipsets?

    • nanoflower
    • 10 years ago

    I feel the same but I don’t think I would ever run into this problem. I tried hibernation on my desktop once and that was it. I’ll never use it again so even a wonky Haswell chip would work well for me so long as the only issue is waking from sleep mode.

    • nanoflower
    • 10 years ago

    That might work well for some devices but I imagine there are some USB devices where the disconnect/reconnect might cause issues. Better to do a real fix and avoid the complaints down the road.

    • indeego
    • 10 years ago

    I am planning for an October build. I won’t put up with wonky [i<]anything[/i<].

    • MadManOriginal
    • 10 years ago

    Couldn’t there be a simple software or driver patch for this? I may be overreaching my technical knowledge here but I think the following is at least logical…if plugging and unplugging the USB 3.0 device after resuming from sleep reconnects the device properly, why couldn’t there just be a software patch that does reconnects the device?

    It might work funny with certain more complex software, but I just tried opening a simple text file stored on a USB drive and unplugged the drive. Naturally I couldn’t save the file (‘path not found’). Plugged the drive back in the same port and it could save fine.

    • bthylafh
    • 10 years ago

    Indeed. The 6-series chipset’s SATA problem was traced to one stinking transistor out of however many hundreds of millions.

    • DeadOfKnight
    • 10 years ago

    It’s not USB 3.0 that’s not working, it’s how the new sleep states are affecting it. Waking from sleep has always been broken for PCs.

    • chuckula
    • 10 years ago

    [quote<]simple[/quote<] You keep using that word in relation to modern chipsets. I do not think it means what you think it means.

    • DeadOfKnight
    • 10 years ago

    Or wait..

    • someuid
    • 10 years ago

    I don’t understand how Intel, the Behemoth of x86 Land, who can run circles around AMD and ARM all day long, can’t get an open standard like USB 3.0 to work [i<]on a simple chipset.[/i<]

    • HTarlek
    • 10 years ago

    [quote<]folks may have to put up with wonky USB 3.0 in the first wave of Haswell PCs[/quote<] Or, you know, buy something else.

Pin It on Pinterest

Share This

Share this post with your friends!