Sign in with
Sign up | Sign in

Report: All Nvidia G84/G86 Chips Defective

By - Source: Tom's Hardware | B 60 comments

It appears that Nvidia’s notebook chips aren’t the only ones that could be defective, if an Inquirer report is to be believed.

According to sources, the entire line of G84 and G86 chips all suffer from the same defect, regardless of desktop or notebook application. The chip generation utilizes the same Application Specific Integrated Circuit (ASIC), which supposedly has an issue with an “unnamed substrate or bumping material, and it is heat related.”

As expected, Nvidia is staying mum on the topic, but analysts are spilling various details. One story is that HP was the recipient for the majority of the defective chips – only limited to a specific batch. Further explanation reveals that it was isolated to an end-of-life batch that used a different bonding/substrate process.

The story with HP didn’t fly with the Inq, as it believes that no changes in process were made throughout the chip’s cycle, leading to suspicions that the defect is widespread. Furthermore, Dell responded quickly to failure reports by quickly issuing a BIOS update that pumped up the cooling for the vulnerable chip parts, which clearly shows that the problem ventured beyond just HP. Now HP has also a new BIOS to run the fan at all times at the expense of battery life.

While evidence may point to only notebook chips being afflicted, one theory is that notebook chips go though more stress than their desktop counterparts. Notebooks are power cycled more often, and by extension the GPUs go through more heat cycles – quickly exposing the defect. GPUs in notebooks may also be harder to cool, with less space for elaborate heatsinks and fans.

Unfortunately for owners of Nvidia G84 and G86 hardware, this could just be the tip of a very troubled iceberg.

Discuss
Ask a Category Expert

Create a new thread in the News comments forum about this subject

Example: Notebook, Android, SSD hard drive

This thread is closed for comments
  • -1 Hide
    random1283 , August 12, 2008 11:42 PM
    I'm not surprised by this. It makes sense that if the mobile chips on that architecture fail so should logically the desktop chips. I have also experienced it first hand. My 8600gt (G84) Gigabyte Silent pipe, Started Making noises from the GPU itself after 1 month and after 2 months it broke completely. I was very disappionted
  • 0 Hide
    sphericaline , August 12, 2008 11:49 PM
    random1283I'm not surprised by this. It makes sense that if the mobile chips on that architecture fail so should logically the desktop chips. I have also experienced it first hand. My 8600gt (G84) Gigabyte Silent pipe, Started Making noises from the GPU itself after 1 month and after 2 months it broke completely. I was very disappionted


    Last time I checked silicon doesn't make noise. You "first hand experience" had to be a mechnical issue and thus not the same problem as the one mentioned here.
  • 0 Hide
    juvealert , August 12, 2008 11:52 PM
    So does this mean a product recall ?


    Well i own an XFX 8600GT @ 620Mhz (over clocked by manufacture)touch wood its still ok...

    But everest give me tempratures of between 50 - 55 degress celcius. Is this one of the issues?
  • Display all 60 comments.
  • 0 Hide
    dmacfour , August 13, 2008 12:11 AM
    50-55c doesn't sound bad. My 7900gt and 8600m gt both max at about 70c.
  • 1 Hide
    caamsa , August 13, 2008 12:14 AM
    "The sky is falling, the sky is falling!"


    http://www.wikinomics.com/blog/uploads/disney-chicken-little-sky-falling.jpg
  • -2 Hide
    I Hate Nvidia , August 13, 2008 12:18 AM
    Trueeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee.
    I had (had) 2 HP laptops ,DV 6599ee and DV6899ee ,both coasted me a fortune,both have the same nvidia disasterus 8400m GS,both now defective.
    The problem shows itself as a crash in the OS (regardless of OS type itself) whenever you try to run any Graphics card-related aplication,such as Videos and games,the OS crashes n a message notifys you that ( nvdllmkm has crashed and recovered).
    youcan actually bypass this problem by uninstalling the GC Driver,yet,you will notbe able to run any games or videos any more.
    Soooooooooooory,but now I really hate nvidia,I had to pay 3000 $ for my laptops,now I just gave them to my chuld n my youngest bro to learn using PCs,n also,I am learning them to hate nvidia!!!!
  • 0 Hide
    terror112 , August 13, 2008 12:19 AM
    hey, my 4850 runs 80C full load, you are fine...
  • 7 Hide
    ThreatDown , August 13, 2008 12:34 AM
    I would take anything the inquirer says with a grain of salt, or a whole ******* shaker.
  • 0 Hide
    jaywalker256 , August 13, 2008 12:38 AM
    8800gt's are G92 right? Certainly hope I dont have to deal with this crap...
  • 0 Hide
    anonymous x , August 13, 2008 1:00 AM
    yay, my 9800 GTX+ is G92!
  • 2 Hide
    fulle , August 13, 2008 1:14 AM
    ^Inquirer reported G92 and G94 chips as effected as well... but, as others have pointed out in the past, Charlie Demerjian seems to have a personal vendetta against Nvidia. Read some of his other work, its hilarious. To be honest, I'm not sure why the Inq site hasn't been sued for slander yet. Remember when he claimed EVGA was going to "defect" and stop selling NV boards? This guy will exaggerate any story he can to make Nvidia look a in worse light.

  • 1 Hide
    LAN_deRf_HA , August 13, 2008 1:36 AM
    sphericalineLast time I checked silicon doesn't make noise. You "first hand experience" had to be a mechnical issue and thus not the same problem as the one mentioned here.


    It's typically not the cpu itself, it's the motherboard. A lot of 680i boards suffer from the squeaky cpu problem. I had an old motherboard that had noisy ram slots, it was a commonly reported issue. I don't understand how but its true.
  • 0 Hide
    Titanius , August 13, 2008 2:35 AM
    Silicon will normally not make any noise, but pump enough voltage and heat and it will buzz or "I don't know how to describe it sound". Most people will never ever hear this happening, but some, mostly unlucky, people will get this behavior followed by the component failing.
  • 1 Hide
    TheDozer42 , August 13, 2008 3:01 AM
    The Inq is full of !(@*& as always. I stopped reading that trash site long ago.

    I've been hammering away at my 8800 for a year and a half, no problems whatsoever. I've never had a high-end graphics card stay on the leading edge of speed for so long (well over a year before they beat it by any real margin!)

    It has a lifetime warranty from eVga anyway, so who cares? I'll just get the latest greatest for free when this one bites the dust someday.

    I'd like ATI/AMD to survive just to give nVidia/Intel competition, but at the time the 8800 was the fastest thing on the planet by a large margin. 100% solid and reliable for me. =]

    Inq shouldn't even be lining the bottom of birdcages, it's that poor.
  • 2 Hide
    anhe64 , August 13, 2008 4:19 AM
    extremely disappointing to see tomshardware believe the nonsense of the Inquirer with Charlie Demerjian.

    As we have heart today from the CC, the Quarterly results and answers given on the questions asked by the analysts, this is absolutely not the case.
  • 0 Hide
    nukemaster , August 13, 2008 4:27 AM
    The buzzing is a vibration. The most common cause is the coils in the power regulation system on the card. This is normal as the coils are sold as low vibration. It will vary from card to card since no 2 coils are exactly the same.

    My X1900XT buzzes, replaced it with a 8800GTX, it buzzes even louder. so far the 4870 has not made a peep, its just luck. I have a motherboard that does it too.

    I never noticed the X1900XT making noise until i moved it out of my Sonata case. so many users with quiet or noise dampening cases will not hear it. Now the 8800GTX lives in a Sonata II and you would be hard pressed to hear it now.

    I do not think the buzzing had to do with the dead card, just bad luck.
  • 2 Hide
    sgmeridian , August 13, 2008 5:06 AM
    This report is completely false. Listening to the Q2 2009 Conference Call today, Jen-Hsun answered point blank that this issue was only with the notebook chips that they had earlier identified to be defective and thus took a $197 million write down for Q2.

    Unless he would like to be thrown in jail by the SEC for saying something completely blatantly wrong to Nvidia shareholders, the real press (unlike The Inquirer) and analysts and then hiding an issue that's much more severe (all desktop GPUs in the G84 and G86 family being defective), I'm pretty sure it's just Charlie's vendetta against Nvidia at this point.
  • 0 Hide
    Anonymous , August 13, 2008 6:06 AM
    @sgmeridia of course the "whoop ass"-man says the desktop chips all ok. If he would admit it the stocks would plummet and nvidia should immediately prepare for huge refunds.
  • 2 Hide
    klemkas , August 13, 2008 6:16 AM
    well maybe it's not the GPU problem, but the cooling. because my dell xps laptop started to act strange, downclocking and laging games after half year of usage. i did my research, and found out that the CPU unit was overheating, not the GPU. i dissasembled the laptop and the thermal paste on cpu and gpu was all dry, so i replaced it with new paste, cooling problems solved, and all went ok again.
  • 2 Hide
    apache_lives , August 13, 2008 8:26 AM
    have to agree with random - i had alot of high pitched electrical sounds coming from alot of 8600's (perhaps one of the resistors, but only ever seen on the 8600's and one 7600GT - DEFINITLY NOT A FAN OR VIBRATION NOISE, and was even louder during 3D load) and just yesterday yet another dud 8600GT (asus this time) and when they were first out we had a large batch of dead albatron 8600's, at first we blamed albatron and they were releasing new revision cards every few weeks (different coolers, pcb designs etc) it was odd but now it makes sence.

    ...... This cant be good for Nvidia
Display more comments