1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Do I need to RMA my XFX XXX HD 5870? Please take a look--lots of info provided. Long.

Discussion in 'AMD / ATI' started by KantVersusHume, May 3, 2010.

  1. KantVersusHume New Member

    Joined:
    May 3, 2010
    Messages:
    11 (0.01/day)
    Thanks Received:
    1
    Do I need to RMA this card? And if I do, could you explain what has possibly happened to it? Up front, sorry about the length, but, I am trying to provide you
    as much information up front as I can. I see that often, in cases like this, you folks who help out have to drag additional information out of people like pulling teeth, so I am trying to avoid that. That said, I am ignorant about these things, so if I am providing a lot of info that is useless, please forgive me, and lead me the right way.

    Nothing on my system is overclocked, except my GPU, which was bought OC'd from XFX. I did try the Auto-tune feature in the ATI Overdrive screen in the CCC, but it only offered less than a 2% upgrade on the GPU clock/none on the memory clock.
    Tried playing Trine with it set up as such, and after 20 to 30 minutes, it locked up with a black screen accompanied by a loud oscillating, buzzing sound. I had to kill the power at the switch and reboot. I immediately reset my GPU clock back to its original settings. Since then, as before, I have had no problems, no artifacts, no lock-ups, but, paying more attention to GPU-Z, I noticed readings that I doubt are possible with a healthy card.

    All tests below were done with Trine. Looking at the logs, I did not notice any throttling in these tests. That said, when I played for a couple of hours yesterday, I did notice the card throttling down to 600/900 and back when the temps
    got too high—-higher than those in the tests. Then, I saw the GPU die temps break 100 C. The throttling was not noticeable in game but was recorded in the sensor log. Somehow I do not think Trine should be able to push this card in such a way, and considering that the card was throttling yet the game play did not suffer, it seems confirmed that my card should not have been working so hard as it obviously did not need to do so. Anyway, I think those temps are unacceptable, but, I come to you for verification. While I wait for help on this, I am going to play Trine for a couple of hours and see if I can replicate the results I got yesterday.



    Without further ado, here are my specs.

    CPU--------------AMD Phenom IIx4 965BE 125W, RB-C3
    CPU Cooler-------Zalman CNPS10X Extreme
    GPU--------------XFX XXX Radeon HD 5870, GPU Clock: 875MHz/GPU Memory: 1300MHz
    Memory-----------G.SKILL Ripjaws Series 4GB (2 x 2GB) DDR3 1333, 1.5v, 9-9-9-24
    PSU--------------Antec Truepower New 750W, (+3.3V@25A,+5V@25A,+12V1@25A,+12V2@25A,+12V3@25A,+12V4@25A,-12V@0.8A,+5VSB@3.0A)
    Motherboard------MSI 790FX-GD70
    Tower Case-------Cooler Master HAF 932, stock fans
    Hard-drive--------Western Digital WD5001AALS Caviar Black Hard Drive
    Optic Drive------Lite-On IHES208-08, BD-ROM 8X, DVD-ROM 16X, CD-ROM 48X
    Monitor----------HP 2509m
    Surge Protector--BELKIN BE112230-08 8 ft. 12 Outlets 3940 Joules Surge Protector with Telephone and Coaxial Protection
    OS---------------Microsoft Windows 7 x64 Home Premium


    FPS in all tests basically remained between 62-67.

    Test 1: With GPU fan set at auto, case fans and CPU fan at max, box filtered, adaptive multi-sample AA, AA and AF using application settings: Texture Detail-very high, AFx8, AAx4, 1920x1080.
    Run: 30 minutes



    Test 2: With GPU fan set at 80%, case fans and CPU fan at max, box filtered, adaptive multi-sample AA, AA and AF using application settings: Texture Detail-very high, AFx8, AAx4, 1920x1080.
    Run: 30 minutes



    Test 3: With GPU fan set at 80%, case fans and CPU fan at max, edge-detect filtered, super-sample AA, AA and AF using application settings: Texture Detail-very high, AFx8, AAx4, 1920x1080.
    Run: 30 minutes






    (1.)
    Test 1
    Max Core Temp: 52.0C
    Max NB Temp: 67.0C
    Max CPU Temp: 55.0C
    Max Ambient Temp: 39.0C

    Test 2
    Max Core Temp: 52.0C
    Max NB Temp: 64.0C
    Max CPU Temp: 55.0C
    Max Ambient Temp: 37.0C

    Test 3
    Max Core Temp: 52.0C
    Max NB Temp: 64.0C
    Max CPU Temp: 54.0C
    Max Ambient Temp: 37.0C

    (2.)
    Test 1
    Max GPU Temp: 88.0C
    Avg GPU Temp: 69.8C

    Test 2
    Max GPU Temp: 65.0C
    Avg GPU Temp: 51.8C

    Test 3
    Max GPU Temp: 59.0C
    Avg GPU Temp: 45.0C

    (3.)
    Test 1
    Max Fan Speed: 40%
    Avg Fan Speed: 30%

    Test 2
    Max Fan Speed: 80%

    Test 3
    Max Fan Speed: 80%

    (4.)
    Test 1
    Max GPU Load: 100%
    Avg GPU Load: 40%

    Test 2
    Max GPU Load: 96.0%
    Avg GPU Load: 57.0%

    Test 3
    Max GPU Load: 99%
    Avg GPU Load: 42.0%

    (5.)
    Test 1
    Max Fan Speed: 2357RPM
    Avg Fan Speed: 1700RPM

    Test 2
    Max Fan Speed: 4334RPM

    Test 3
    Max Fan Speed: 4342RPM

    (6.)
    Test 1
    Max GPU Temp#1: 87.0C
    Avg GPU Temp#1: 69.9C

    Test 2
    Max GPU Temp#1: 59.5C
    Avg GPU Temp#1: 52.0C

    Test 3
    Max GPU Temp#1: 60.0C
    Avg GPU Temp#1: 45.2C

    (7.)
    Test 1
    Max GPU Temp #2: 94.0C
    Avg GPU Temp #2: 74.8C

    Test 2
    Max GPU Temp #2: 66.0C
    Avg GPU Temp #2: 57.2C

    Test 3
    Max GPU Temp #2: 65.5C
    Avg GPU Temp #2: 49.4C

    (8.)
    Test 1
    Max GPU Temp #3: 91.0C
    Avg GPU Temp #3: 72.8C

    Test 2
    Max GPU Temp #3: 64.5C
    Avg GPU Temp #3: 56.7C

    Test 3
    Max GPU Temp #3: 64.5C
    Avg GPU Temp #3: 49.5C

    (9.)
    Test 1
    Min VDDC Current: 0.0A
    Max VDDC Current: 61.5A
    Avg VDDC Current: 32.8A

    Test 2
    Min VDDC Current: 0.0A
    Max VDDC Current: 58.1A
    Avg VDDC Current: 37.3A

    Test 3
    Min VDDC Current: -1.2A
    Max VDDC Current: 56.9A
    Avg VDDC Current: 25.7A

    (10.)
    Test 1
    Min VDDCI Current: -0.3A
    Max VDDCI Current: 15.7A
    Avg VDDCI Current: 10.7A

    Test 2
    Min VDDCI Current: 0.0C
    Max VDDCI Current: 15.4A
    Avg VDDCI Current: 13.1A

    Test 3
    Min VDDCI Current: -10.7A
    Max VDDCI Current: 15.4A
    Avg VDDCI Current: 9.4A

    (11.)
    Test 1
    Min VDDC Phase #1 Temp: 0.0C
    Max VDDC Phase #1 Temp: 72.0C
    Avg VDDC Phase #1 Temp: 58.6C

    Test 2
    Min VDDC Phase #1 Temp: 0.0C
    Max VDDC Phase #1 Temp: 54.0C
    Avg VDDC Phase #1 Temp: 47.3C

    Test 3
    Min VDDC Phase #1 Temp: 0.0C
    Max VDDC Phase #1 Temp: 53.0C
    Avg VDDC Phase #1 Temp: 42.1C

    (12.)
    Test 1
    Min VDDC Phase #2 Temp: 0.0C
    Max VDDC Phase #2 Temp: 255.0C
    Avg VDDC Phase #2 Temp: 60.3C

    Test 2
    Min VDDC Phase #2 Temp: 0.0C
    Max VDDC Phase #2 Temp: 56.0C
    Avg VDDC Phase #2 Temp: 49.3C

    Test 3
    Min VDDC Phase #2 Temp: 0.0C
    Max VDDC Phase #2 Temp: 56.0C
    Avg VDDC Phase #2 Temp: 43.7C

    (13.)
    Test 1
    Min VDDC Phase #3 Temp: 0.0C
    Max VDDC Phase #3 Temp: 255.0C
    Avg VDDC Phase #3 Temp: 60.7C

    Test 2
    Min VDDC Phase #3 Temp: 0.0C
    Max VDDC Phase #3 Temp: 255.0C
    Avg VDDC Phase #3 Temp: 50.1C

    Test 3
    Min VDDC Phase #3 Temp: 0.0 C
    Max VDDC Phase #3 Temp: 255.0C
    Avg VDDC Phase #3 Temp: 44.1C

    (14.)
    Test 1
    Min VDDC Phase #4 Temp: 0.0C
    Max VDDC Phase #4 Temp: 255.0C
    Avg VDDC Phase #4 Temp: 69.6C

    Test 2
    Min VDDC Phase #4 Temp: 0.0C
    Max VDDC Phase #4 Temp: 56.0C
    Avg VDDC Phase #4 Temp: 48.2C

    Test 3
    Min VDDC Phase #4 Temp: 0.0C
    Max VDDC Phase #4 Temp: 255.0C
    Avg VDDC Phase #4 Temp: 42.8C

    (15.)
    Test 1
    Min VDDCI Phase #1 Temp: 0.0C
    Max VDDCI Phase #1 Temp: 86.0C
    Avg VDDCI Phase #1 Temp: 68.6C

    Test 2
    Min VDDCI Phase #1 Temp: 0.0C
    Max VDDCI Phase #1 Temp: 62.0C
    Avg VDDCI Phase #1 Temp: 55.4C

    Test 3
    Min VDDCI Phase #1 Temp: 0.0C
    Max VDDCI Phase #1 Temp: 62.0C
    Avg VDDCI Phase #1 Temp: 48.0C

    (16.)
    Test 1
    Min VDDC: 0.9500 V
    Max VDDC: 2.0375 V
    Avg VDDC: 1.#QNB V

    Test 2
    Min VDDC: 0.9500V
    Max VDDC: 2.0375 V
    Avg VDDC: 1.#QNB V

    Test 3
    Min VDDC: 0.9500V
    Max VDDC: 2.0375V
    Avg VDDC: 1.#QNB V

    (17.)
    Test 1
    Min VDDCI: 0.9500 V
    Max VDDCI: 2.0375 V
    Avg VDDCI: 1.#QNB V

    Test 2
    Min VDDCI: 0.9500V
    Max VDDCI: 1.1500 V
    Avg VDDCI: 1.#QNB V

    Test 3
    Min VDDCI: 0.9500V
    Max VDDCI: 2.0375V
    Avg VDDCI: 1.#QNB V




    Observations and Questions

    1. CPU, Core, NB and Ambient Temps went about as I expected. Not that I am happy with the CPU temps, but, evidently, I need to switch out the large stock intake fans and the large exhaust fan with groupings of 120mm fans,
    which should double to quadruple the CFM airflow throughout the tower.



    2. Are the GPU temps roughly within a range of normal operation, and why
    is there a lower max GPU temp and lower average GPU die temps on Test 3
    than Test 2?

    I would have thought that using the edge-detect filter and
    super-sampling would have contributed more heat rather than be more efficient--if it did anything at all, as I was not certain if the filter
    and sampling method would be applied while letting the application
    determine AA. For the record if I saw any real difference, it was
    minimal, as I was trying to make certain that I played roughly the same
    area of the game in the same amount of time, in basically the same way, and so I
    wasn't focusing on aesthetics.

    The only reason I even changed those settings is because I know in some
    games, like NWN2, doing so can significantly change overall FPS, even
    when AA and AF are turned off in the game settings and the GPU is given
    charge of the AA and AF globally. Still, seeing the temps drop is a
    mystery I hope somebody might solve. Could Trine possible be coded to
    make best use of those settings?



    3. Average GPU Load seems to confirm the results above. Again, why?



    4a. What should I expect in amperage output for VDDC and VDDCI current?
    More importantly, what does the O.OA and negative amperage readings
    indicate?

    Now, I know next to absolutely nothing about electricity, but
    doesn't zero or negative amperage indicate that some type of voltage regulator is not supplying the current needed at load? That is--at least
    with the negative reading--doesn't this indicate that current is being
    demanded greater than current being supplied, like a negative amp reading
    on a automobile's alternator?

    I should say that, using HWinfo32, I found that the VDDC current
    reading never displayed lower than in the 1+ amp range--no zero, no
    negatives. HWinfo32 did not have a corresponding reading for VDDCI to
    compare. The average and max reading on HWinfo32 did correspond to the
    comparable VDDC current readings in GPU-Z.

    If somebody could offer up a good explanation about what VDDC and VDDCI
    are, what they do, and how to analyze them for proper behavior, I would
    greatly appreciate the tutelage.



    4b. Looking at the VDDC and VDDCI phase temps, I see minimum temps that I
    just cannot believe to be true. Not to say they aren't true--I may be
    ignorant as to how they could be achieved on stock cooling, and, if so, p
    please educate me--but I doubt anything on my card is literally freezing, even for even for one polling cycle, whatever the polling frequency is.

    VDDC Current, VDDCI Current, VDDC Phase #1 Temp, VDDC Phase #2 Temp, VDDC Phase #3 Temp, VDDC Phase #4 Temp, VDDCI Phase #1 Temp, and VDDC all show occasional, very brief registrations at 0--though VDDC will not list it as a minimum reading, all of the others do. VDDC, like all of the above, show a complete break in the red continuous bar graph, but the O reading does not register in the reading field for VDDC when it is set to show the minimum current reading. When these 0 readings appear they seem to occur in groups of 4 to 6 occurrences over roughly a 30 to 60 second interval. This is just what I have observed while watching real-time when the memory and GPU clocks are on the lowest frequency.



    4c. Getting max readings on VDDC Phase #2, #3, and #4 of 255.0C?!? Surely
    if that had happened I wouldn't be using this computer right now to post
    this to the forum without--at least--needing to insert another GPU into
    my mobo, if not insert everything not fried to a crisp into a new mobo.

    Right?

    Looking through the logs, these 255.0 C temps, 0.0 C temps and zero or
    negative amp readings are very rare. If they are real, I suspect they
    last less than a second certainly, as they are not showing up very often
    in the polling cycles. I would definitely say that the 0.0A and 0.0C
    are not showing up in the logs as often as they have been visually
    represented in the red bar graphs, so that would suggest to me that
    they, if real, are far shorter than 1 second in duration.

    Still, I shouldn't have temps ever spiking to 255.0 C anywhere in the
    card at any time correct? Could that momentarily happen and yet the
    card still function without locking, displaying artifacts, or frying
    my cat sleeping three feet away into feline bacon bits?



    5. Finally, all of my average VDDC and VDDCI readings are showing up as
    "1.#QNB V". Obviously that is not correct. If it is correct, and all
    of you reading this understand it, then maybe its time for me to begin
    laboriously attempting to grow and groom a unibrow as well as working on
    communicating in monosyllabic grunts. Is that some type of error
    code?




    Just from what I have read trying to understand what is cropping up in
    GPU-Z, this is not behavior indicative of proper operation, at least in the case of 4 and 5 above. Is it as simple as my card is shot, or soon will be, RMA it? Or could it be a corrupt Catalyst driver, a problem on the GPU-Z side, a mobo BIOS that needs to be reset with CMOS cleared, etc? Is it possibly a bad sensor on the card which is leading to false readings?

    So, what am I looking at?

    Thank you so much for your time and patience in reading through this long post and for whatever suggestions you provide. I figured if I were going to seek your assistance, it was my responsibility to provide as much information up front as possible. If there are any things you wish me to do next in analyzing this problem, tell me what to do, and I will do it and get back to you.
  2. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    13,770 (4.54/day)
    Thanks Received:
    6,837
    Location:
    Edmonton, Alberta
    Contact XFX and create a support ticket. They will provide you with a file you can burn to a CD that will automatically flash the bios on the card.

    Seems that the XXX cards have issues running 1300mhz memory. I have 2 cards here that do the same, and the very first response from XFX about me having issues was the bios file.


    I was told that the XXX cards shipping today have this new bios, with 900mhz GPU, and 1250 mhz memory. For me, this still has a negative impact on performance, and I think they are confirming my results before doing anything about it(Even though the tech admitted there was a difference in some apps, but it was "unnoticible".:rolleyes:).

    The momentary readings you have that seem weird are probably just reading errors. The bios should fix your crashes, and you'll then ber able to adjust clocks in CCC without having any issues, although I doubt you'll get 1300mhz memory fully stable. 1300mhz is fine for benches, and short gaming, but most apps that use the gpu's memory @ high load for extended periods of time seem to crash.

    I no longer have the crashes, but I'm still missing some performance, so I might end up getting new cards...I have yet to hear back from XFX. They should have recalled these cards, it seems.
    KantVersusHume says thanks.
  3. KantVersusHume New Member

    Joined:
    May 3, 2010
    Messages:
    11 (0.01/day)
    Thanks Received:
    1
    Well, first off, thank you very much for the quick and useful reply. To tell you the truth, I imagine that I can take the slight boost in GPU and the slight drop in memory and live with that, if it will stop the throttling issue. Like I said, I didn't notice it in Trine, but, I imagine that I would notice it in other games. Luckily for me, I don't really need the card for anything but gaming, in regards to apps that might stress it, so, a possible slight drop in performance will not cause me any professional or artistic concerns. I just want to know my card is stable and will perform at a high rate consistently.

    It doesn't seem the best business strategy for XFX to sell a card that they must then slightly under-clock on the whole with a new BIOS, however. This is the kind of thing I will remember come time to buy a new card. If it runs smoothly after the BIOS update and I don't suffer for the loss of memory, I probably would give them equal shot at my money next time, but now, the psychological effect, unfortunately and maybe irrationally, maybe not, will be to hold XFX responsible for performance issues that may or may not result from the slight under-performance of their product. If it hadn't cost me so much, that would mitigate this as well.

    Regardless, thank you very much. I will start a support ticket forthwith. You were one of the people I hoped would reply. :toast:
  4. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    13,770 (4.54/day)
    Thanks Received:
    6,837
    Location:
    Edmonton, Alberta
    If you are comfortable in reseating the cooler, I'd suggest you do so as well. Those temps are high, but it's more than possible that unstable ram and error correction are causing your high temps and throttle.

    Many people are having issues with these cards...but then there are also probably jsut as many, if not more, that have no issues at all. There a few apps that I could run @ 875/1300 without issues, so I kinda question XFX's binning process at this point...

    I was one of those that got under-performing 4890's too. The cards I was sent on RMA performed WORSE than my 4870's. They did replace them with reference-design cards though, and I haven't had an issue with those cards since.

    I buy XFX due to the warranty, but this is too many cards from them that I have had now that I ahve had to return. Each time cost me shipping charges, although XFX did pick up the shipping tab for sending the bunk 4890's back.

    I don't know what to make of them at this point. This is almost class-action lawsuit sort of stuff, given the number of people that I've seen with issues and XXX 5870 cards...
  5. KantVersusHume New Member

    Joined:
    May 3, 2010
    Messages:
    11 (0.01/day)
    Thanks Received:
    1
    I'm not averse resetting the cooler, so long as doing so doesn't invalidate the warranty. I have an anti-static mat, with a wristband that plugs into ground on a grounded outlet that I can work with. I have the support order in and now I just have to wait to hear back from them. After that is settled, assuming that the technician is not going to suggest to RMA the card, then, once seeing how it acts after the new BIOS, I will do as you suggest. I may come back and ask for pointers if I find I am uncomfortable about some aspect of the task.

    Any general advice you might have about resetting the cooler? I still have some AS5 from when I built this rig around December. I just don't what the shelf life is for it, or if I would need to recondition it before use. I also don't know if it is the right thermal interface material, or if the GPU and memory need something else specific.

    By the way, what exactly does invalidate the XFX warranty? One of the reasons I went with the XXX 5870 was the apparent strength of the warranty. Do you have to actually physically modify the card, like say water-cooling it or say OC it using non ATI software or OC it via some type of hardware mod?
  6. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    13,770 (4.54/day)
    Thanks Received:
    6,837
    Location:
    Edmonton, Alberta
    Overclocking, Changing coolers, and flashing bioses(provided you can flash to orginal bios before RMA) are all covered under XFX warranty, provided in doing so, you do no physical damage to the card.

    Add that to the life-time warranty(which is stipulated as lifetime of the owner, not the product line) in certain regions, and that warranty being transferrable to one other person, makes it hard for me to NOT buy XFX.

    I mean, they got custom coolers and watercoolers covered in their warranty...that's pretty good.


    AS5 will be fine...just be very careful removing the cooler, as the power-area of teh card also has a white paste that covers a fairly large gap that other paste will not be sufficient for, and is easily ruined if you are not careful.

    Anti-static mat and wristband should have you completely covered. It's not hard to do, if you take your time.
    Last edited: May 4, 2010
  7. mastrdrver

    mastrdrver

    Joined:
    Feb 24, 2009
    Messages:
    3,115 (1.57/day)
    Thanks Received:
    565
    Its possible that the memory chips that they use may not be able to run at the specs that they were told they could run.

    What are you playing that you're seeing a performance hit by lowering the memory clocks?
  8. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    13,770 (4.54/day)
    Thanks Received:
    6,837
    Location:
    Edmonton, Alberta
    Almost ANY newer title. Eyefinity, and 2560x1600 4xaa and 8xaa take a hit. It's small, but it's there.

    I know, not exactly the standard config for gaming, but I have multiple cards for just that reason. Noone needs more than one card, really.

    Which also makes any and all loses double, of course.


    Now, of course, some of that can be alleviated by increasing core clocks, but of course, that requires a bios flash yet again, nevermind a custom bios, so XFX are kinda stuck between a rock and a hard place with that. The card I bought, stock, also had room for overclocking(minor, of course, but still possible), and now, has absolutely none, as CCC gpu clocks are maxed, and memory won't hit the 1300 max stably. Nevermind AA and high res aren't really gpu bound so much as memory, so my chosen config is just one that gets the raw end of this deal.

    Kinda irks me that they had this bios since February(02/23 bios date, so they knew about the problem before that), but didn't pull back cards already in stores, but unsold, that they knew would potentially have issues. If they took steps to make a bios, obviously it's more than just a small problem.

    I mean, sure, could have just been a bad batch of mem, but that doesn't explain them changing the clocks of new cards to 900/1250....obviously no memory they can get is really capable of those clocks. Maybe not thier fault, but it's them that sold the cards without proper testing, so they've got to deal with it.

    I've had issues since day one of getting the cards, so thier binning process is flawed. I can bench all I want, looped, even, as data in memory for benches doesn't stay there long, but in games, of course, it can be a long time before you load into another map, or die, or whatever. They didn't test real-world, or I'd not have two cards with serial numbers just 5 apart, and from the same XFX batch, with the same problem.

    Let me repeat that...XFX batch. Both cards have OEM serial numbers 50k apart, so either XFX binned them, or whoever made the cards for them did...and seeing as no other card maker has such a memory bin, i think it's most likely XFX binned the cards themselves.
  9. KantVersusHume New Member

    Joined:
    May 3, 2010
    Messages:
    11 (0.01/day)
    Thanks Received:
    1
    That is very good to know. I had thought that the warranty was that extensive, but I could not find the information that I had previously read stipulating the specifics of the warranty. In fact, their warranty page specifies parameters that it would seem they could refer to if they wished to claim a warranty invalidated by custom coolers, water blocks, etc. However, I am fairly certain that the warranty page on their site is not the specific warranty contract that applies to the cards in question.

    Could you guide me to your source for warranty information, if you wouldn't mind?

    I would have responded sooner, but I was waiting for an original reply to my support ticket. Any estimations on how long I might need to wait before they make contact with me?
  10. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    13,770 (4.54/day)
    Thanks Received:
    6,837
    Location:
    Edmonton, Alberta
    My info comes from thier RMA form. It very clearly says that any aftermarket coolers should be removed and the stock heatsink put back, and any custom firmware must be removed. I also asked a tech specifically about changing coolers, and he said it was fine, as long as there was no physical damage.

    Normally they are pretty quick to answer..within 24 hours or so. I've been waiting since last week for them to respond to me though...
  11. KantVersusHume New Member

    Joined:
    May 3, 2010
    Messages:
    11 (0.01/day)
    Thanks Received:
    1
    Well then, I must have read that information in a post on some forum or another, as I don't recall or even know if I can look their RMA form prior to beginning the RMA process--though I imagine I could...too lazy to search for it at the moment. Anyway, no need to--your word is gold to me on this.

    I just got an answer back from a tech. Basically told me the same thing they told you in regards to the clock changes, that, for most apps, I should see a gain in performance. He did say that he didn't think the BIOS update would fix my issues, but, he gave me the BIOS, told me to give it a shot, and that, if the problems persisted, we would go on to the next step, which, I assume would be an RMA.

    Going to flash the BIOS now and hopefully his speculation will be wrong.
  12. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    13,770 (4.54/day)
    Thanks Received:
    6,837
    Location:
    Edmonton, Alberta
    He might be right...you may just need to reseat to cooler. You can always clock the ram back up to 1300mhz in CCC if that wasn't the issue, whereas the old bios had issues with adjusting clocks via CCC. No matter the end cause of your problems, the new bios is far better than the stock one, and heck, it's even an official bios, and they've told you to go ahead, so you know it's perfectly safe, too.
    KantVersusHume says thanks.
  13. KantVersusHume New Member

    Joined:
    May 3, 2010
    Messages:
    11 (0.01/day)
    Thanks Received:
    1
    He might have been right, but, he wasn't. You were. :toast:

    At least, I haven't been able to produce any of the previous issues since flashing the new BIOS. Many, many thanks to you cadaveca. I signed up at this forum seeking help partially because this is the home of GPU-Z, but also because, while looking for information on my issue, I noticed that you and a few other people on here seemed like just the type of people that would be helpful. This was the first forum I went to for assistance, and it turned out, in this case, to be the only one I needed.

    I wish I had something to offer back, but I really don't. Maybe you could find use with this program I found while trying to solve a micro-stuttering issue I had with Geometry Wars Retro Evolved. It's called SMP Seesaw Pro .

    Basically, it allows you to isolate one or more programs to any one or more cores of your processor, up to 32 cores, exceedingly quick. You just execute it, highlight the program(s) you want to isolate or simply move, right click, choose which option you want to use, and then choose the cores. Stopped the processor interrupts that I suspect was leading to the micro-stuttering. My theory could be complete BS, but it worked, and I am a practical type of person, so I'm satisfied. Anyway, the guy who wrote it actually has four or five other tiny yet useful programs on his site. Crosshair, StartupMonitor, and Clipomatic are all very useful to me. Maybe you won't find use in any of them or maybe you already have better alternatives--if so, let me know about them--but, I just wanted to offer something back as a show of gratitude for your help.

    Thanks again.
    cadaveca says thanks.
  14. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    13,770 (4.54/day)
    Thanks Received:
    6,837
    Location:
    Edmonton, Alberta
    Let us know if you have any further issues, and if you don't let us know too...I do have specific interest on how all these XXX XFX cards pan out.
  15. LifeOnMars

    LifeOnMars New Member

    Joined:
    Jun 20, 2008
    Messages:
    2,566 (1.15/day)
    Thanks Received:
    396
    Location:
    Kettering,Northants. UK
    Nice one Cad :toast: Glad to hear it's all sorted Kant :)
  16. KantVersusHume New Member

    Joined:
    May 3, 2010
    Messages:
    11 (0.01/day)
    Thanks Received:
    1
    I'm glad to LifeOnMars. Takes a lot of anxiety off of my shoulders. This is the first rig I have built, and the first time I have had a desktop in ages, so, I don't have any spare parts on which to fall back. If I ever have to RMA this card before I intend to buy a new one, I will have to buy a cheap card to get by while I wait for my card to come back. So, needless to say, I'm thrilled with cadaveca's guidance. If it holds up and the BIOS was the only issue, cadaveca probably saved me a needless RMA, and some coin, because the tech I have a support ticket open with did not think the BIOS would make any difference. Ergo, he probably wouldn't have offered it had I not asked for it.

    I'll keep you up to date on how things go with the card cadaveca. Next, I am going to switch out the 10.4 Catalyst with the OpenGl4.0 preview drivers they released in late march to see if it covers the Nvidia extensions that were used in NWN2. I should be able to see what is covered in Everest. It might if those extensions were accepted as standards under OpenGL3.3 or before or OpenGl4.0. If so, well, maybe the FPS loss ATI cards suffer due to how lighting was rendered in NWN2, via proprietary Nvidia extensions, might be a thing of the past. Worth a look anyway.
  17. alexsubri

    alexsubri New Member

    Joined:
    Feb 7, 2010
    Messages:
    1,391 (0.85/day)
    Thanks Received:
    199
    There was a newer XFX XXX 5870 Bios that LoneWolf had here a couple of weeks ago...

    If you want it, I will look it up...

    I have two 5850's (one which is XFX XXX BE) and I bios flashed them with a newerXFX XXX 5870's BIOS and I am running everything maxed out with wicked FPS's @ 900/1250...I've tried 1300, but it would make my card freeze. I know I would have to up my voltage to 1.3-1.4v's to reach 1300, but since I'm running it all at stock voltages with fan speeds of 45-50% (not too noisy), I feel more comfortable with what I got.

    FPS:

    BFBC2: 80-130 FPS vs 60-90 FPS
    Crysis Max Settings: 40-45 FPS vs 30-35
    L4D2: 250-300 FPS
    MW2: 120-250 FPS

    BTW, my average temp is around 35-39c and 65-70c on load with stock fans

    The list goes on and on, but I hope all is well with you card :toast:
  18. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    13,770 (4.54/day)
    Thanks Received:
    6,837
    Location:
    Edmonton, Alberta
    Same bios. XFX gave it to me as well.
  19. KantVersusHume New Member

    Joined:
    May 3, 2010
    Messages:
    11 (0.01/day)
    Thanks Received:
    1
    Alexsubri, thank you for the offer. As cadaveca said though this must be the same BIOS I would think.

    BIOS Version 012.019.000.006.036339
    BIOS Date 02/23/10 03:08
  20. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    13,770 (4.54/day)
    Thanks Received:
    6,837
    Location:
    Edmonton, Alberta
    You know, Kant, if in the end this really does fix your issue, you should point that out(and this thread) to XFX. They will continue to address problems like this incorrectly until shown how to fix it. I mean, you might have had to send in your card...cost both you, and them. Both sides of the story have won out here, if my solution worked for you.
  21. KantVersusHume New Member

    Joined:
    May 3, 2010
    Messages:
    11 (0.01/day)
    Thanks Received:
    1
    Oh, I intend to. I am testing the card out on a few different applications to see if the results stay consistent and am waiting for another reply from the tech that has my support ticket. I intend to point him to this thread and email XFX directly about it as well. I just want to make certain that it is fixed, though at the moment mysteriously, and not some type of momentary ghost-in-the-machine anomaly. Anyway, I doubt I will hear back from the tech today; he is either in Africa or the UK by the time-stamp on the support ticket, and if he is in the UK, which I assume he is, then there is a good chance he is participating or focused on what may very well be the most consequential election of his lifetime. Heck, I'm in Texas and I am fixated upon it.

    I will update when I have talked to the tech or if the card reverts back to previous behavior before I talk to the tech.
  22. KantVersusHume New Member

    Joined:
    May 3, 2010
    Messages:
    11 (0.01/day)
    Thanks Received:
    1
    Errr, things just took a left turn, but, I now definitively know what was causing the bizarre readings.

    Well, the crazy high and low temps, zero and negative amp readings and whatnot made a reappearance. It wasn't the BIOS. It was running Everest Ultimate sensor checks in conjunction with running GPU-Z. I assume that both Everest and GPU-Z occasionally polled the sensors on the 5870 simultaneously. When that occurs, and Everest has control first, GPU-Z is seeking a reading that it cannot access. Then it returns the strange data.

    Now, I'm not an expert--far, far, far from it--on how either GPU-Z or Everest accesses the card, and what each specifically accesses in order to get the data each polls for, but, I have isolated the anomalies to specifically, and only, occurring when both Everest Ultimate is polling sensors and GPU-Z is active. If anyone might competently explain the possible reason instead of people who might look at this thread with the same issue having to depend upon my ignorant musings, please do so.

    HWInfo32 can poll for the various GPU temps, VDDC temps, voltage and amperage without causing the conflict--at least I haven't been able to reproduce it with both running. I don't know about Sandra as I just tried to use it, and, at the moment, I don't have enviroment sensors available.

    Nevertheless, I know that, at least for me, running Everest Ultimate sensor polling and GPU-Z simultaneously with an XFX XXX HD Radeon 5870 is a no go.

    If anyone else would try using the sensor monitoring with Everest Ultimate along with GPU-Z and see if they could replicate my results, that would be great.

    It appears that the new BIOS fixed my temp problems and the throttling, so, I would say my issues are resolved. I just need to get back with the XFX tech.
  23. cadaveca

    cadaveca My name is Dave

    Joined:
    Apr 10, 2006
    Messages:
    13,770 (4.54/day)
    Thanks Received:
    6,837
    Location:
    Edmonton, Alberta
    Yeah, i mentioned before that the weird readings were probably just software errors. No surprise there.

    GPU-Z and Afterburner(when using the right versions) will also set gpu volts to 1.5v!!!

    :laugh:

    I own Everest, and can confirm your issues. Hence my recommendations to flash the bios, based on the OTHER problems you had.
  24. KantVersusHume New Member

    Joined:
    May 3, 2010
    Messages:
    11 (0.01/day)
    Thanks Received:
    1
    I assume that is for when you might want to use your card to produce metallic incense.
    :D

    Yeah, I don't know why I didn't immediately look at the various other sensor polling programs I had open along with GPU-Z for the false readings after you basically said, hey, you're getting false readings from software incompatibility. I have a lot on my mind right now and so I think I panicked a bit and my brain throttled down to idle.:ohwell:

    That said, everything is running smooth with the new BIOS you directed me to, and over the next few days I will see how much of that lost memory I can get back stable.

    Now, to play a few games for enjoyment for the next few days and then I will look at changing out the TIM when I do my regular dust maintenance.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Share This Page