• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

crossfire newbie, got a issue help!

IceShock

IceShock

IceShock

Soldato
Joined
16 Dec 2005
Posts
4,054
Location
Halifax W'Yorkshire
hey there iv just got my 2nd 7950 today,

and iv just tried firestrike extreme, in test 2 and combine test, my fps is jumping from 30 to 15 then to 10 then back to 30 then 15 then 10 then 30 then 15 then 10, any idea whats going on? because of this my end result is a score less than with a single card, tho, heaven 4.0 seems to work fine, not had chance to test games yet though. in test 1, gpu usage is 100, in other two they are both at 65% and they stay there. temps both are fine, ulps is turned off, neither are overclocked at moment. its definitely not a issue that cards aren't getting the power, iv just got a brand new corsair 1000w psu.

any one know whats going on?

Also, if you could, I have another slight issue you might be able to help me with here:

http://forums.overclockers.co.uk/showthread.php?p=26206196#post26206196

thanks!
 
IceShock said:
hey there iv just got my 2nd 7950 today,

and iv just tried firestrike extreme, in test 2 and combine test, my fps is jumping from 30 to 15 then to 10 then back to 30 then 15 then 10 then 30 then 15 then 10, any idea whats going on? because of this my end result is a score less than with a single card, tho, heaven 4.0 seems to work fine, not had chance to test games yet though. in test 1, gpu usage is 100, in other two they are both at 65% and they stay there. temps both are fine, ulps is turned off, neither are overclocked at moment. its definitely not a issue that cards aren't getting the power, iv just got a brand new corsair 1000w psu.

any one know whats going on?

Also, if you could, I have another slight issue you might be able to help me with here:

http://forums.overclockers.co.uk/showthread.php?p=26206196#post26206196

thanks!
Click to expand...

Look at the Useful Links thread pinned at the top of the forum. Look for my guide, 'How To Make Sure Crossfire Is Working'.
 
Last edited:
tried everything in that list, crossfire is showing to be working, also iv removed 1 crossfire bridge so I only have one but still, fps is jumping up and down drasticly and my score is less than with just 1 card
 
IceShock said:
tried everything in that list, crossfire is showing to be working, also iv removed 1 crossfire bridge so I only have one but still, fps is jumping up and down drasticly and my score is less than with just 1 card
Click to expand...

Sounds like your cards may be throttling. Are the core clocks changing when the fps goes up and down? If you used my guide as suggested you should be able to monitor the core clocks and gpu usage on screen while running firestrike.
 
nope, they are staying consistant at 850 core 1250 memory temps staying same, fans staying same speed, everything seems perfectly fine as if it should work, but in firestrike, its not for some reason
 
IceShock said:
nope, they are staying consistant at 850 core 1250 memory temps staying same, fans staying same speed, everything seems perfectly fine as if it should work, but in firestrike, its not for some reason
Click to expand...

Which cards do you have, got a link? From memory 850/1250 is the non boost state of Boost 7950's, with the actual boost state being 925/1250. Unless of course you have a non boost 7950 in which case that would be your normal clock.
 
tried at 950/1250, firestrike wont run at all now at that setting it just freezes and not responds the programme, and yep power is at its limit +20
 
IceShock said:
tried at 950/1250, firestrike wont run at all now at that setting it just freezes and not responds the programme, and yep power is at its limit +20
Click to expand...

Are the stock clocks the same on both cards? Are both cards boost cards? I'd leave both at stock clocks for now with the power limit set at +20%. No point overclocking until you're stable.

I think you might be getting the 7950 boost throttling problem. You have two choices. You can either use Tommys power limit hack. This raises the powerlimit to +50% so avoids the throttling issue by raising the power limit. Or you can take the better, but slightly more complicated choice and flash to a non boost bios that does not have the throttling problem. Flashing a gpu bios is easy if you follow the simple guide i made in the link below.

Tommys power limit hack
http://forums.overclockers.co.uk/showthread.php?t=18509458

The HIS Ice-Q Thread - Includes my bios flashing guide for 7950's in one of the spoiler tags -
http://forums.overclockers.co.uk/showthread.php?t=18539905

The bios download links in that thread are dead. So if you upload both of your cards bios files to sendspace i will alter them for you, remove the boost part of the bios and the throttling issues. Use GPU-Z to save each cards bios to your desktop, then upload to sendspace.

If you followed my crossfire guide from earlier you should also disabled ULPS using MSI Afterburner Beta 19 as well.
 
Last edited:
IceShock said:
here ya go pal,

http://www.sendspace.com/filegroup/XAcRvwrPhkQRrGzRjvJeCA

yep both are same clock speeds
Click to expand...

Ok well as far as i can see, neither card is a boost card. However one card, Tahiti 2.rom had a very low TDP, 167W vs 217W for the first card. No idea why that was. So i made both cards 220W reaching 267W at +20% power tune. Should be ample even for large overclocks. Increased max clock speeds in CCC to 1250/1750 and increased both cards fan profiles. 40C=25% - 60C=60% etc. You may want to set your own profile if you don't like these. Or i can further tweak them again if required. Don't forget to disable ULPS before using 7950 crossfire. Good luck.

http://www.sendspace.com/file/7papkl

Oh and make a backup of the bios incase you want to go back to the stock bios.
 
Last edited:
Okay, this has fixed that my 2nd card wouldn't overclock and was causing freezes and hangs etc, all 3d applications work perfectly now,

but still, every bench mark is working fine apart from 3dmark, im scoring lower than with a single card, fps jumping up and down drasticly in the 2nd graphics test and combined test :S

iv got the latest drivers and everything installed, might reinstall 3dmark now that iv got two cards, it might just be that somethings gone wrong with it lol
 
IceShock said:
Okay, this has fixed that my 2nd card wouldn't overclock and was causing freezes and hangs etc, all 3d applications work perfectly now,

but still, every bench mark is working fine apart from 3dmark, im scoring lower than with a single card, fps jumping up and down drasticly in the 2nd graphics test and combined test :S

iv got the latest drivers and everything installed, might reinstall 3dmark now that iv got two cards, it might just be that somethings gone wrong with it lol
Click to expand...

What's gpu usage like when the test is going on and fps is going up and down? I assume you're using 14.4 drivers? Don't forget if you update drivers you need to disable ULPS again.
 
14.3 drivers, didn't know 14.4 was out :O ill have a look now, yep I know I read about that ulps thing, ulps, that turns the second card off when its not in use correct?

and gpu useage is 64% exactly both cards, clocks on both cards running at 1100/1575 now, in test 1 on firestrike, iv tried both extreme and regular, in test 1 gpu usage is about 97-99% and in test two and combine test, gpu usage is 64 and it stays there, Everything else is fine, all games fine so far.

Also, may I ask is there anyway to turn cross fire on for windowed mode games? I regularly play windowed mode, especially mmos like warcraft etc so I can do other things at the same time, only play full screen in the likes of battlefield and racing games.

Thanks again matt for all your help :)
 
IceShock said:
14.3 drivers, didn't know 14.4 was out :O ill have a look now, yep I know I read about that ulps thing, ulps, that turns the second card off when its not in use correct?

and gpu useage is 64% exactly both cards, clocks on both cards running at 1100/1575 now, in test 1 on firestrike, iv tried both extreme and regular, in test 1 gpu usage is about 97-99% and in test two and combine test, gpu usage is 64 and it stays there, Everything else is fine, all games fine so far.

Also, may I ask is there anyway to turn cross fire on for windowed mode games? I regularly play windowed mode, especially mmos like warcraft etc so I can do other things at the same time, only play full screen in the likes of battlefield and racing games.

Thanks again matt for all your help :)
Click to expand...

That's correct.

What cpu have you got?

Crossfire only works in full screen mode im afraid. If you run Mantle when playing BF4 then you can use Xfire in Windowed mode but you get slightly less performance than you do proper full screen.

If games work ok with decent and equal gpu usage then i wouldn't worry too much about firestrike.
 
Uncle Petey said:
Even at 5Ghz, is it possible that his 2500K is slightly bottlenecking two cards in firestrike?
Click to expand...

Doubt it. Sounds like throttling to me, as my 7990 dropped to 64% when it throttled. Maybe only one card is set to 20% power tune? Not sure, it sounds like a throttling issue to me but its hard to say from behind a monitor.
 
Uncle Petey said:
Even at 5Ghz, is it possible that his 2500K is slightly bottlenecking two cards in firestrike?
Click to expand...

Let's find out.:D

2 stock 290Xs

4930k all 6 cores with HT @4.0

Graphics Score 20456

http://www.3dmark.com/fs/2057564



4930k on 4 cores no HT @4.7 (same as a 3570k)

Graphics Score 20620

http://www.3dmark.com/fs/2057476



4930k on 4 cores with HT @4.7 (same as a 3770k)

Graphics Score 20682

http://www.3dmark.com/fs/2057498



4930k on 6 cores with HT @4.7

Graphics Score 20469

http://www.3dmark.com/fs/2057516

So not a lot of difference with HT on or off and 4 or 6 cores, also very little difference @4.0 or 4.7ghz.

I am going to do one more test using PCI-E 1.0 which I suspect will lower the score by about 5%.
 
You must log in or register to reply here.
Back
Top Bottom