Closed Solved

HD6950 crossfire issues.

My specs: AMD Phenom 2 940BE @ 3.4GHz | Gigabyte GA-MA790GP-DS4H | 4GB G. Skill DDR2 1066 | 2x HiS IceQ 6950 2GB | 3x Asus 24" screens for Eyefinity awesomeness! | PC P&C 750 PSU

So I received my second HiS 6950 IceQ in the mail last night and hooked it up. Crossfire link cable installed, both 6 pin power plugs hooked up. CCC detected the 2nd card just fine and enabled it, hardware info shows both cards and says they are running at x8 pci-e speeds which I knew would happen.

Here's the issue: BFBC2 and Dragon Age both get higher frame rates with a single card then with crossfire. Dirt 3 and 3DMark 11 do much better with crossfire. So far that's all I've been able to test. As BFBC2 is my primary game at the moment I'm a bit po'ed that crossfire isn't working correctly.

Here's what I've done so far: Uninstalled cat's, cleaned with Driver Sweeper from safe mode, re-installed 11.9's and CAP4. That didn't do it so I formated C:\ and did a fresh install of Windows 7 Ultimate 64 bit, installed all updates then cat 11.9/CAP4, still didn't help. I have tried using one crossfire link cable as well as both at the same time. All power cables are hooked up, etc.

While running BFBC2 I noticed GPU-Z was reporting that the primary card was getting ~60% GPU load and the secondary card is at about ~30%. Dragon Age is worse, the secondary card is hardly getting used and the primary is still low. Once I turn off crossfire the single card get's ~80-100% usage.

Any ideas here guys? I really want this to work correctly. I game at 5760x1080 with high settings and was not happy with the performance of a single card. Thanks for any input!
21 answers Last reply Best Answer
More about hd6950 crossfire issues
  1. Best answer
    Check this out :

    Apparently the CF 6950's do not do so well with BFBC2.

    Have you checked your processor usage in the task manager while playing these games?

    I'm guessing your CPU is bottlenecking your CF setup.
  2. Interesting... Guess it should be noted that I was using 1xAA and 1xAF while testing, though I did try higher levels which didn't seem to effect frame rates at all up or down.

    I'll check CPU usage tonight, I know in the past with a single card CPU usage was no where near 100% but I haven't checked with both cards running.

    Thanks for the help, Chainzsaw.
  3. so you are doing Eyefinity, i always recommend a 3-way CrossFireX for this resolution...
    to be accurate in your info, tell us how many FPS do you get when playing BC2 ? and how many do you get when playing other games ?
    and best way to know if your CF working properly or not, just game on a single monitor and check benchmarks, there aren't a lot of benchmarks for Eyefinity out there, and i have seen yesterday a 580s 4-way SLI getting 47 FPS wehn playing metro 2033 with 3 monitors Eyefintiy (can you believe that)

    the CPU should be okay and capable of handling that Crossfire
  4. I'll get some exact-ish frame rate numbers tonight and post them. BFBC2, Grid 3, and Dragon Age (which is about all I play atm) get very playable frame rates with a single 6950. I run BFBC2 with medium settings across the board and 2xAA/AF no HSAF(sp?) and I rarely feel any kind of slow down with the single card. Now I do run DX10 with the single card as DX11 doesn't run as well. Crossfire is a different story... Once I enable crossfire frames are all over the place and drop down below 30 at times. The Oasis map really hates crossfire for some reason.

    I have tried all low settings, all high settings, DX10, DX11, 1xAA/AF to max AA/AF and nothing really seems to have much effect on FPS while in crossfire mode.

    To add (edit): Unfortunately I can't go 3-way CFX as I only have two PCI-E slots. Don't think I wouldn't consider it! :D
  5. Ok, so I tried turning off Eyefinity to see what would happen at 1920x1080. With crossfire, all settings on high, AA and AF maxed on Oasis I was getting 30-70FPS, AVG about 40. Without crossfire, same settings, I got 50-110 (indoors), AVG about 60.

    CPU load with a single card was anywhere from 60-95%, it rarely hit 100% if ever. GPU load on a single card was near 100% most of the time.

    CPU load with crossfire was about the same, however GPU load was constant 40% on each card almost the whole time.


    I get similar results at 5760x1080, not quite as high max FPS but the same issue where frames are higher with the single card then with crossfire. CPU and GPU usage is about the same as well

    Oh, at this point I'm running the 11.10 preview drivers with the 11.8 CAP4's. I installed the BF3 beta thus the reason for the preview drivers. I must say, so far it looks like crossfire is going to work for BF3, but it's to early to tell just how well being in beta. I will say this. BF3 sure heats up the hardware, holy crap!
  6. can't you push the CPU to 4.0 GHz and give it a try ? just ensure proper cooling and proper voltages.
    some games benefit from CF and some others not, and all the FPS problems you're stating is related to drivers issue, can'tsay the CPU is bottlenecking the GPUs but it's really strange to have a 95% CPU usage while running a game....

    BF3 seems to be heating both nVidia and AMD cards, i have seen a thread reporting such issue, the official release of CCC is 11.9 try it with 11.8 CAP4 for CF
  7. I was not able to get 4GHz to post, 3.8 would post but would probably require volts that I don't want to push. I can run 3.6GHz pretty well but usually just stick with 3.4 and save the heat.

    So far I've tried CCC 11.8, 11.9, and 11.10 preview all while using the 11.8CAP4's (uninstall and reinstall each time). Only difference is the better performance with the 11.10's in BF3.

    I'm seriously considering picking up a Phenom 2 x4 975 and O/Cing from there just for the hell of it.
  8. so if you're going to pick up one, get a six core 1090T as long as you're CPU is being used that much, the additional 2 cores will be helping much in gaming.
    the per-core Deneb performance will beat the "Thuban" only in multi-threaded apllications.
  9. Thought I read somewhere that due to the lack of enough L3 cache the hexa-cores really wern't that impressive. Was thinking about pushing that x4 to 4GHz or a little more, seems pretty easy to do from what I've read. Would a slightly lower clocked x6 help you think?
  10. 1090T is a BE CPU, you just raise the multiplier and you hit 4.0 GHz stable with eyes closed. i overclocked my 1055T just easily too.. and was stable 24/7
    i told you above, if you take the "Thuban" and disable 2 cores, the "Deneb" will beat it, but the benefit from the additional 2 cores gives more room for background applications to consume a little from the CPU.
    generally, the extra lets you run more processes without making a full usage for the CPU.
  11. I may just hold off for a bit honestly. Was thinking about picking up a i5-2500K, new mobo, and 8Gigs of DDR3 closer to the end of the year. Unless a faster Phenom x4/x6 will give me a good boost, but I don't see it helping a whole bunch.n
  12. switching to Intel will be a big mistake while the Bulldozer is on the way... and can be on your MOBO with simple Bios update......
    all the issues you're experiencing is related to a CPU bottleneck or CF drivers that's why i'm always considering a Dual GPU to get over CF problems.
  13. I'm pretty sure this mobo is AM2+ only and does not support DDR3. Wouldn't that mean that Bulldozer is a no go? Or does it support AM2+ and DDR2?
  14. what is the Model of your Mobo ? i thought it's AM3 socket. and No AM3+ Zambezi based CPUs won't be compatible with AM2+ sockets.
    so you'd better wait and see Bulldozer benchmarks and then decide whether to go AMD or Intel.
  15. Gigabyte GA-MA790GP-DS4H
  16. According to AMD, AM3+ CPUs will be backward compatible with AM3 sockets and fully incompatible with AM2+
  17. Yep, so that's not gonna work, at least not on this mobo. I'm not pulling my wallet out just yet anyway so we'll see what happens.
  18. Fixed it, I think!

    So I decided to swap my cards around, bottom card to the top, etc. While doing so I took a look at the under side of the crossfire link cable and noticed that one connector block was labeled "J1A" and the other "J2A." The way I originally had them was with the "J2A" connected to the top card. After swapping the cards I installed the link cables with "J1A" on the top and "J2A" on the bottm...

    Fired up BFBC2 and what do ya know, higher frame rates with crossfire then with a single card. Granted it's not double the FPS but after some testing I found that I can fully max the game out, all high settings, full AA/AF and rarely drop below 50FPS where as with a single card it drops much lower.

    I also kept an eye on CPU/GPU usage. CPU usage dropped across the board and GPU usage on BOTH cards was much higher then before often staying in the upper 80's to 90's.

    I then tested Dragon Age which still suffers from the same low FPS while using crossfire but I'm willing to blame the software rather then the hardware on this one. I hear DA: O just doesn't play nice with crossfire.

    TL: DR: WOOT!
  19. glad that you fixed it, but it sounds really strange.....
    not all the games benefit from crossfire, especially the older games but i guess all the modern games benefit from crossfire, so all games FPS should be raised up.
  20. Best answer selected by C0rehound.
  21. This topic has been closed by Mousemonkey
Ask a new question

Read More

Radeon Crossfire Graphics