Radeon RX 480 crossfire performance problem

Hi all,

I have problem with the performance of these card. With single card enable I achieve a good amount of fps in games but with cf enable I get lower fps than a single(in most cases) + freezes. I have latest amd driver's (Crimson 16.12.1) and updated windows, latest motherboard bios version so I don't rly know where the problem is ;/


I suspect it's a problem with power draw when im running it in crossfire but I have got 1350w platinum psu so it shouldn't be the problem. Here are screen of gpu power draw diagram when only single card is enable:

http://screenshot.sh/n7gAAFyBEx0fb

Diagram when cf is enabled:
http://screenshot.sh/oemfX8WKM9OJ9

Here are my specs:

MotherBoard: ASUS M5A99FX PRO R2.0
CPU: AMD FX-9590 4.70GHz
PSU: Enermax 1350W Platimax
HDD: Plextor 128GB 2,5" SATA SSD M6S Series
SEGATE st4000dx001-1ce168 4 TB
GPU: 2x ASUS Radeon RX 480 8GB STRIX OC
RAM: 4x DDR3 GOOD RAM 8GB 1600MHz Play CL 10

Rly I don't know where the problem lies ;/ I have searching for solution like 3 days but none of them works.
25 answers Last reply
More about radeon 480 crossfire performance problem
  1. I would think its more the CPU is bottlenecking the system. The AMD FX-9590 is just to slow per core for most games. I suggest you upgrade to Ryzen if its better than Intel next month.
  2. Well with single card enable I get more fps than in most benchamarks so I don't think it's the problem.
  3. CF and SLI performance is rather dismal ... while SLI scaling was great w/ the 9xx series (scaling @ 70% average, 96+% in demanding games) , making the twin 970 option a virtual no brainer with 40% more performance than the same price single 980. I can't justify it at this point with the 10xx / 4xx series

    Only 39% at 1080p

  4. Sure bottleneck plus the more overhead of the second card. You can try optimizing your OS for games. Try ccleaner and advanced systemcare. Use them clean up the system and shut down services. You could try reseating the cards by pulling them out and putting them back in.
    https://www.piriform.com/ccleaner/download
    http://www.iobit.com/en/advancedsystemcarefree.php
  5. Well, that CPU is just bad and probably not overclockable, since it is an overclocked FX-8350.
    You can try to overclock it with AMD Overdrive to get better performance, but chances are slim.
    Best wait for Ryzen and then decide whether you go AMD or Intel and get a CPU.
  6. That is not my problem. When I tried to compare my fps when using only single cards and same settings to this AC benchmark:
    https://www.youtube.com/watch?v=Q0uwwqFa5gU&t=76s

    I have got more fps than him.

    But when I compared crossfire version with https://www.youtube.com/watch?v=Qpye4YujA6s&t=15s I have got like 20 - 30 fps less on average :/

    When cf is enable in most cases i get less fps than with it disabled ;/ I don't think that cpu bottleneck.
  7. Some games do not support dual gpu at all.

    More likely, your games are cpu limited and not graphics limited.
    They do worse in cf because it takes more cpu power to manage two cards than one.
  8. Well the cpu is max on 60% use so like i said before I don't think it bottleneck these cards. Also the AC support crossfire:

    https://www.youtube.com/watch?v=Qpye4YujA6s&t=15s

    When I enable crossfire in AC I get on average 30 fps lower than him.
  9. Monitor seperate cores, there's no way a game will use all 8 cores to 100%.
    If the first few cores come to 100%, while the GPU's are under 100%, it's a CPU bottleneck.
  10. Pawemol12 said:
    Well the cpu is max on 60% use so like i said before I don't think it bottleneck these cards.


    Many games depend primarily on a single fast master thread for performance.
    Windows will spread such activity around among all the available threads.
    A single task that needs 100% of a single thread will show as an average of 13% busy on an 8 core cpu.

    To test this assertion, do an experiment with YOUR games.

    Try removing one core. You can do this in the windows msconfig boot advanced options option.
    You will need to reboot for the change to take effect. Set the number of processors to less than you have.
    This will tell you how sensitive your games are to the benefits of many threads.
    I am guessing that if you operate on 7 threads you will see no difference in performance.
    You probably will do about as well with 4 threads before you lose significant performance.
  11. Pawemol12 said:
    That is not my problem. When I tried to compare my fps when using only single cards and same settings to this AC benchmark:
    https://www.youtube.com/watch?v=Q0uwwqFa5gU&t=76s

    I have got more fps than him.

    But when I compared crossfire version with https://www.youtube.com/watch?v=Qpye4YujA6s&t=15s I have got like 20 - 30 fps less on average :/

    When cf is enable in most cases i get less fps than with it disabled ;/ I don't think that cpu bottleneck.

    This is how long it takes CPU's to create the fps per frame in crossfire. As you can see it takes the AMD way longer then even the old 3770k. This is due to fact each core is slower on the AMD.

  12. I have installed msi afterburner to monitor my gpu usage, cpu usage in game. I have noticed that when only 1 gpu is running I have nearly always 100% gpu use but when cf is enable then it is on 20% or even lower (on both gpus) ;/ Cpu clocks on all cores are 60 - 80% so I think I will not have to replace my cpu becouse im not able right now.

    I will post YT videos later to show you the comparasion of using only one gpu and using both in cf.

    Edit:

    In heaven benchmark on max settings both cards work on 100%. Idk why in game they won't :/
  13. Pawemol12 said:
    That is not my problem. When I tried to compare my fps when using only single cards and same settings to this AC benchmark:
    https://www.youtube.com/watch?v=Q0uwwqFa5gU&t=76s

    I have got more fps than him.

    But when I compared crossfire version with https://www.youtube.com/watch?v=Qpye4YujA6s&t=15s I have got like 20 - 30 fps less on average :/

    When cf is enable in most cases i get less fps than with it disabled ;/ I don't think that cpu bottleneck.


    1. I would avoid youtube reviews / compares. Anybody w/ a web cam can post a video, that doesnt men they have the knowledge and experience to conduct a reproducible test result. Unless associated w/a well respected site, I'd just ignore them.

    Search youtube for "970 3.5 GB problem" and you will see dozens of yahoos demonstrating the problem ...a problem which a dozen reputable web sites tried to reproduce but were unable to do so under normal usage without doing really really strange things.

    2. Assasins Creed is a freak ... AC:Unity was an extremely poor console port sucked RAM like crazy. As for AC:Syndicate, you should expect a 82% boost in fps.



    3. Is your card throttling ? Dunno what case you have or number / specs of fans but the 480 runs very hot... the intake cooling air for the top card is essentially the hot exhaust from the bottom card. In CF, the system will regulate the speed of both cards to the lower of the 2 cards. So if the top card throttles, the bottom card will downclock match it.

    4. Cores rarely provide any significant impact to gaming. Windows will not automatically spread this out to use all cores. You can look at affinity setting in Task Manager and most games will default to a single core and it will stay like that unless you change it. In my experience, setting it from 1 -2 has a noticeable impact, going from 2 to 3 not so much but the number changes minutely, going from 4, to 5, 6, 7, 8 has never changed anything in my experience.
  14. 2. I have maybe 10% boost in fps example. Normal I have on average 75 fps in one scene and when I repeat it in crossfire I get 85 fps lol.
    3. I have very good cooling, clocks are always on max (1310 MHZ). I have run 20 min max settings heaven benchmark and always 100% gpu useage on both gpus.
    4. I have noticed cores 1 - 6 in AC Syndicate are most used but with cf enabled (60% - 70%) average use on one core.

    I think my problem is software related as in heaven benchmarks gpu usage stay the same (both 100%) but in game they are jumping like one is 100% and another is 0%. Or one is 99% and other is 10% etc.

    So from crossfire in games I get no benefits, in some cases lags and worse performance than single gpu.

    Edit:
    Well when heaven is performing these cards actually are making really loud noise but when gaming they are silent :/
  15. Pawemol12 said:
    Hi all,

    I have problem with the performance of these card. With single card enable I achieve a good amount of fps in games but with cf enable I get lower fps than a single(in most cases) + freezes. I have latest amd driver's (Crimson 16.12.1) and updated windows, latest motherboard bios version so I don't rly know where the problem is ;/


    I suspect it's a problem with power draw when im running it in crossfire but I have got 1350w platinum psu so it shouldn't be the problem. Here are screen of gpu power draw diagram when only single card is enable:

    http://screenshot.sh/n7gAAFyBEx0fb

    Diagram when cf is enabled:
    http://screenshot.sh/oemfX8WKM9OJ9

    Here are my specs:

    MotherBoard: ASUS M5A99FX PRO R2.0
    CPU: AMD FX-9590 4.70GHz
    PSU: Enermax 1350W Platimax
    HDD: Plextor 128GB 2,5" SATA SSD M6S Series
    SEGATE st4000dx001-1ce168 4 TB
    GPU: 2x ASUS Radeon RX 480 8GB STRIX OC
    RAM: 4x DDR3 GOOD RAM 8GB 1600MHz Play CL 10

    Rly I don't know where the problem lies ;/ I have searching for solution like 3 days but none of them works.




    Did you try setting the profile for the game you want to play in the Crimson settings menu? I had similar issue with BF1untill I setup the profile.
  16. Yes I did. What is more intresting - for some games: Shadow Of Mordor, Far Cry 4 I get both gpu to 100% utilization and very high fpses(more than on benchmarks) but yeah, that are only 2 games that are taking advantage of my cf setup ;/

    Also I posted about my problem here:
    http://www.tomshardware.co.uk/answers/id-3273164/radeon-480-crossfire-performance-problem.html

    I have cleaned my OS with Advanced System Care 10.1 free and ccleaner but it didn't help.
  17. My guess it that since your CPU would already bottlenecking the performance of a single 480 in quite a lot of things, adding a further one is making the CPU struggle more and therefore performs worse. I may be wrong but it makes some slight sense to me.
  18. It's not cpu bottleneck - my cpu is not even in 50% use in most games ;/
  19. CPU usage means nothing if the things you are using only utilize 1/2/4 cores, as most games do. Other things come into consideration such how much memory the CPU can handle and send to the GPU and such. There is a bottleneck, there was one when comparing my OC'd fx 8350 @ 4.8 GHz and a single GTX 970, compared to my Ryzen 1700x and the same GTX 970.
  20. But when I run for example. AC Syndicate on max possible settings I get no stuterring and better fps than on medium or low settings ;/ That is the problem here not cpu ;/
  21. That would make even more sense that it is CPU related, more graphics = more GPU load, lesser graphics can, in some circumstances make more CPU load.
  22. Actually the usage of cpu is higer.
  23. Here's the problem, you came here for help and everyone's trying to help you.
    You are just denying what they're saying because you don't like it.
  24. lakimens said:
    Here's the problem, you came here for help and everyone's trying to help you.
    You are just denying what they're saying because you don't like it.


    I have to say this is pretty rude and where is your proof or knowledge that it is a cpu bottleneck I run cfx 480s with a fx 8370 and have no issues with cpu bottlenecking.
  25. It's not rude, when three people are giving the same answer.
Ask a new question

Read More

Performance Crossfire