Yes, the 5970 is more power efficient than the fermi cards when considering performance/energy ratio....but when the system is under full load with 4 GPUs running at once, its kinda pushing the limits.
Each 5970 pulls about 400W of power. So both 5970s are pulling about 800W, which only leaves 300W for the system, which is not really much to play with.
I had the same problem, which is why I am only using 1 5970 right now. When I talked to an Alenware tech about this, and asked why the PC was shutting down at 100% load, he says the system shouldnt be pushed to 100% because its a powerful system.
My oversight of this whole thing is that, I dont think Dell tests these machines at 100% load to see if the system is even stable under the conditions. Now 2 5870s or other lower end cards on this system even with the 1KW PSU will be fine under 100% load.......but with the 480s and 5970s, this should be tested as they use a LOT more power.
I heard that there should be a 1.2KW PSU coming out soon for the ALX model, but I havent seen anything yet.
I bought Bad Company 2 from steam yesterday, and I was on that for like three hours on max settings and no problems. It's games like Age Of Empires 3 and Arma 2 that the system shuts down, so is it just old games won't run on my PC. I can play most, but some just shut the system down like the two above?
Yeah that is a power issue. When a computer just ups and shuts off like if you pulled the power plug out of the wall its a power issue but there my be hope for you.
Switch your connections and see if the problom goes away. If it doesnt after doing that then call Dell. They will trouble shoot at the end of the call they will replace the PSU if they cant fix it and I dont think they can. Its ether the connections or the PSU.
Well, if it can survive 3 hours of BC2, it sounds like your hardware (and likely software) is ok. That was with both cards still installed, right?
I would still check the above link on video cards power connections because those are 2 power hungry cards. Mine was built earily June 2010, and they were done correctly (but I only have the one 5870 ... so it's kinda hard to get that wrong).
Age of Empires 3 is kinda old, but I wouldn't call Arma II old. Basically, you have 4 GPUs so some games might have a problem with that ... most have only been tested/validated on 3 (if even that). You could try:
1. Update to the current version/patch. Check readmes or forums for possible .ini hacks required for extreme CrossFire setups.
2. Run the game in XP or Vista Compatibility Mode and/or with Admin Privilege (in Startup Icon). Sometimes you also have to set the main game .exe as well.
3. Lower your resolution (either desktop and/or game). Some games have problems with unknow (to them) resolutions.
4. Is there an option in CrossFire for a Compatibility Mode or to turn off some GPUs?
5. Main processor compatibility ... I've never had a problem running old DX9 games on Intel Dual/Quads, but I have seen a AMD x2 dual have problems. Had to set the Affinity ... there is a utility ... RunFirst worked for me on Serious Sam - The Second Encounter (old version) ... AMD 64-x2 Dual- WinXP32- nVidia 8600
had to add " -winxp " on game's Launch Options inside Steam. That seem to fix it so at least things have textures now. Game seems poorly optimized but this is probably an old version. I don't think I can get patches for a demo version. I only have one GPU but it works. Here are more ... there is a whole list over at their site.
-winxp use it in Vista/Windows 7 to enable multi-GPU support (Arma2) -noCB turns off multicore drawing, slow down rendering but may resolve visual glitches (Arma2) -cpuCount=X change X to a number less or equal than numbers of available cores. This will override auto detection (Arma2).
EDIT:
This works pretty good for starters I suppose:
-winxp -cpuCount=4 -maxmem=2047 -nosplash
... lots of Arma2 Tweak guides out there ... it's just a demo so I just want to play right now.
I have the exact same system as you, and by the looks of it the exact same problem also. Have had the system since the beginning of May, and so far the failed solutions include replacing gpu (at least 3 times), motherboard, cpu, psu (twice), ram, alien fx board, essentially every significant component in the system has been replaced to no avail. Eventually I was sent a full replacement system, but the same problem carried over.
For a while it was agreed the problem was the top 5970 overheating as the bottom one was directly below it leaving no more than a couple mm between them, essentially preventing it from cooling itself at all really. Since the service manual on Dell's own site states the bottom pic slot is 16x, I got hold of a longer Crossfire bridge and tried putting the 2nd 5970 in the lower pci slot to allow more space between the cards. Unfortunately it turns out the bottom slot is actually not a pci 16x slot, so that did not work.
For a short time it seemed like running the pci fan at almost full blast (using command centre) while gaming had sorted things, it certainly helped keeps temp below dangerous levels (although also pushed noise levels into ridiculously loud). But it turns out the problem isn't heat related after all, because last week, 2 minutes into a game with pci fans full blast (and temps at the time would have been no more than 75) the same shutdown occured.
So now I'm once again unsure what is causing it or how it can be solved, so I can't actually give an answer to this problem, but hopefully some of these failed solutions might save you going through the months of messing around I've had to go through. I'm going to have a look into this possible fix regarding the power cable configuration, but to be honest I will not be holding my breath :P
You guys really need to check your PCI power connections going to the video cards. This is a known issue so you gotta check them and make sure its connected right. If you are not sure if they are connected right post a picture and we can tell you. If its not the connections then the PSU is not powerfull enough or its bad. I dont think this is the case but that is the only other thing it can really be. So again check the connections! post pictures if you are not sure!
Just because you can play one game for hours on end doesnt mean anything. Some games stress the system more then others. While the cards have enough juice in one game another game may stress it more making it draw more power and that is when it will shut down the system.
Best way to test this is to use a stress test. Furmark is great for this but dont run it to long. If you can run Furmark for 2min then its not a power issue. Don't run it over 2min though to be safe. It can melt your GPU if you do.
The systems I have seen that are drawing to much off a rail can not run Furmark at all. They will crash soon as you start the test. Some will crash about 5secs into it. If you run it and this happens to you then its ether
1) Incorrect power connections
2)Inadequate PSU
3)Failed PSU
4)Gremlins
As far as the PSU having enough power goes...I ran 2 GTX 295's on the 1100watt PSU just fine. lots of people have come on the forums with your issue and they solved it by changed the power connections. So please go and make sure they are 100% connected like they should be. If they are not then the system WILL shutdown on you when stressing the cards.
You do not care how many connections are on each cable. You do care that only 1 cable is powering each 8-pin. You can ONLY run one 8-pin off of one cable/rail. You can ONLY run two 6-pins off one cable/rail. This is true for ANY system.
On my system my PCI power cables are Yellow,Blue and Black.
I run the 8-pin from the yellow cable to the one 8-pin on the top card.
I run the 8-pin from the blue cable to the 8-pin on the bottom card.
I run the two 6-pins from the black cable to the two 6-pins on each card.
Checked the connections, I guess I shouldn't really be surprised but sure enough the wrong cables have been used. So that's a few months and several thousands of pounds wasted because some dummy didn't build the system correctly (twice!). It's clearly enough of a known problem to warrant it's own space in the troubleshooting section, kinda makes you wonder why this solution was never suggested when I detailed this exact problem on the first or any of the subsequent 30+ phone calls to Dell technicians???
Haven't done enough testing yet to know for sure if it's sorted, but what is certain is the cables were previously not correctly setup, and now they are (according to the instructions on the troubleshooting page).
For anyone with this problem, you could be doing yourself a huge favour if you do as Morblore suggests and make the pci connections the very first thing you check, there's a decent chance you'll save yourself a whole lot of hassle.
yeah my cabels where wrong connected to but it dident change the powerup issue i have and also out of the blue my crossfirex stopped working. have bean in contact with dell support and they have concluded that the only way to know what is causing my problems is to "confiscate" my computer for abaut 1 week to find what my problem is. will wright when i know more.
Yeah Crossfire was disabled for me too and the system would crash if I tried to enable it. Uninstalling then reinstalling the video drivers sorted that out. I'm surprised they've agreed to collect your system, took me a long time to convince them to just take it away then send back when fixed, and when they finally did set it up it was then cancelled because apparently the parcel was too big.
Out of interest, do you use anything to control your gpu fans? If you have 2 5970's like my setup then having aggressive fan controls is an absolute must, otherwise the temps on the top card will reach dangerous levels very quickly. I use MSI afterburner which does the job nicely.
Yeah Crossfire was disabled for me too and the system would crash if I tried to enable it. Uninstalling then reinstalling the video drivers sorted that out. I'm surprised they've agreed to collect your system, took me a long time to convince them to just take it away then send back when fixed, and when they finally did set it up it was then cancelled because apparently the parcel was too big.
Out of interest, do you use anything to control your gpu fans? If you have 2 5970's like my setup then having aggressive fan controls is an absolute must, otherwise the temps on the top card will reach dangerous levels very quickly. I use MSI afterburner which does the job nicely.
i tried to reinstal the latest ati driver but it didn't work so i can not use the crossfire. no i dont use anyother program then the defult to control the temp and fan's of the gpu. you think that can be the problem?
Did you remove the drivers before trying the reinstall? A quick pass with driver sweeper should do the job if you're having trouble doing this - http://www.guru3d.com/category/driversweeper/ Then when you reinstall, Crossfire should be on by default.
I couldn't go as far as saying all your problems are down to not manually controlling the fans, but if you have 2 5970's setup like I do then it is certainly necessary. A pair of 5970's should either be watercooled or at the very least have a fair amount of space between them. The gap between mine is at best 3mm, but since the gpu fan is underneath the card, it essentially removes the ability of the top card to effectively cool itself at all really. Add to that the heat rising from the card below it and it doesn't take long to hit very high temps. Controlling the fans manually still won't stop the top card getting very hot (mine sits just above 52 degrees doing absolutely nothing), but it will keep it just below dangerous temps (at the expense of having a very loud system).
MSI Afterburner does the job well - http://event.msi.com/vga/afterburner/download.htm You can set a sliding scale so that as the temps increases, so do the fans (Not 100% sure but you may need to unlock the gpu in the overdrive section of catalyst control centre to allow this). That's half the job, the other half is to control the PCI fan in the command centre. I use 4 profiles, the default auto alien control for anything other than gaming, Light gaming (fans set to manual at 50%), Medium gaming (fans set to manual at 70%) and Heavy gaming (fans set to manual at 85%). It's then just a matter of choosing the appropriate fan profile depending on how intensive the game you are playing will be.
I've still not had the random reboot since switching the cables round so as long as that remains the same I'm at the point where I'm ok with how my system runs. Sure I'd rather I didn't have to be running the fans so loud, but to be fair it's only the most system intensive games that need the highest setting, and it only takes a couple minutes before I don't even notice them anyway. The alternative would be to remove one of the cards, and I don't fancy doing that after how much they cost anyway.
Hope you get yours sorted too, believe me I know how annoying all this can be :)
thanks for the info, did the sweep and reinstaled the 10.8 and the crossfire works will try 10.9 later. i was also able to play darksiders in 1024x720 for 25 minutes and that was something i could not do before changing the wires, then it was like 25 seconds, i was only able to play in 800x600 but atleast i could play and now it works as i should, no sending to DELL either cause i seam to have solved the problem myself with MY startup issue. all i did was disable the pre booth fan and vent system so it was nothing dangerus at all. thanks guys for all the help
m3th0d1c4l
265 Posts
0
July 18th, 2010 15:00
I can almost guarantee it is a power issue. Honestly, IMO, I dont think these Dell PSUs are sufficient enough to run 2 HD 5970s.
Try taking out 1 of your HD 5970s, and play your game and see if it shuts down again.
m3th0d1c4l
265 Posts
0
July 18th, 2010 16:00
Yes, the 5970 is more power efficient than the fermi cards when considering performance/energy ratio....but when the system is under full load with 4 GPUs running at once, its kinda pushing the limits.
Each 5970 pulls about 400W of power. So both 5970s are pulling about 800W, which only leaves 300W for the system, which is not really much to play with.
I had the same problem, which is why I am only using 1 5970 right now. When I talked to an Alenware tech about this, and asked why the PC was shutting down at 100% load, he says the system shouldnt be pushed to 100% because its a powerful system.
My oversight of this whole thing is that, I dont think Dell tests these machines at 100% load to see if the system is even stable under the conditions. Now 2 5870s or other lower end cards on this system even with the 1KW PSU will be fine under 100% load.......but with the 480s and 5970s, this should be tested as they use a LOT more power.
I heard that there should be a 1.2KW PSU coming out soon for the ALX model, but I havent seen anything yet.
Tesla1856
8 Wizard
•
17.3K Posts
0
July 18th, 2010 16:00
A Kill-A-Watt meter is pretty cheap and handy if you don't have one.
Some of the APC battery backup units will read/display current draw as well.
You both have nice (but power hungry) systems. But isn't the 5970 suppose to be more efficient than say ... something like a nVidia Fermi core?
Major_Cj
7 Posts
0
July 19th, 2010 10:00
Cheers for the posts guys.
I bought Bad Company 2 from steam yesterday, and I was on that for like three hours on max settings and no problems. It's games like Age Of Empires 3 and Arma 2 that the system shuts down, so is it just old games won't run on my PC. I can play most, but some just shut the system down like the two above?
I'll try and get a better setting on Catalyst.
morblore
2 Intern
•
2.4K Posts
0
July 21st, 2010 23:00
Yeah that is a power issue. When a computer just ups and shuts off like if you pulled the power plug out of the wall its a power issue but there my be hope for you.
Dell is not sending the systems with the power connections on the cards connected in the correct order. Go here: http://support.dell.com/support/topics/global.aspx/support/dsn/en/document?c=us&cs=19&dl=false&l=en&s=dhs&docid=84D6A6EDA59E9719E040AC0A64E97CC4&doclang=en
Switch your connections and see if the problom goes away. If it doesnt after doing that then call Dell. They will trouble shoot at the end of the call they will replace the PSU if they cant fix it and I dont think they can. Its ether the connections or the PSU.
Tesla1856
8 Wizard
•
17.3K Posts
0
July 22nd, 2010 13:00
Well, if it can survive 3 hours of BC2, it sounds like your hardware (and likely software) is ok. That was with both cards still installed, right?
I would still check the above link on video cards power connections because those are 2 power hungry cards. Mine was built earily June 2010, and they were done correctly (but I only have the one 5870 ... so it's kinda hard to get that wrong).
Age of Empires 3 is kinda old, but I wouldn't call Arma II old. Basically, you have 4 GPUs so some games might have a problem with that ... most have only been tested/validated on 3 (if even that). You could try:
1. Update to the current version/patch. Check readmes or forums for possible .ini hacks required for extreme CrossFire setups.
2. Run the game in XP or Vista Compatibility Mode and/or with Admin Privilege (in Startup Icon). Sometimes you also have to set the main game .exe as well.
3. Lower your resolution (either desktop and/or game). Some games have problems with unknow (to them) resolutions.
4. Is there an option in CrossFire for a Compatibility Mode or to turn off some GPUs?
5. Main processor compatibility ... I've never had a problem running old DX9 games on Intel Dual/Quads, but I have seen a AMD x2 dual have problems. Had to set the Affinity ... there is a utility ... RunFirst worked for me on Serious Sam - The Second Encounter (old version) ... AMD 64-x2 Dual- WinXP32- nVidia 8600
Tesla1856
8 Wizard
•
17.3K Posts
0
July 22nd, 2010 23:00
I got the Arma 2 demo on Steam a week or so ago, and have been meaning to try it.
Graphics looked like this on first run:
http://peecee.dk/upload/view/178780
had to add " -winxp " on game's Launch Options inside Steam. That seem to fix it so at least things have textures now. Game seems poorly optimized but this is probably an old version. I don't think I can get patches for a demo version. I only have one GPU but it works. Here are more ... there is a whole list over at their site.
-winxp use it in Vista/Windows 7 to enable multi-GPU support (Arma2)
-noCB turns off multicore drawing, slow down rendering but may resolve visual glitches (Arma2)
-cpuCount=X change X to a number less or equal than numbers of available cores. This will override auto detection (Arma2).
EDIT:
This works pretty good for starters I suppose:
-winxp -cpuCount=4 -maxmem=2047 -nosplash
... lots of Arma2 Tweak guides out there ... it's just a demo so I just want to play right now.
wedgie20
8 Posts
0
October 3rd, 2010 15:00
Hi,
I have the exact same system as you, and by the looks of it the exact same problem also. Have had the system since the beginning of May, and so far the failed solutions include replacing gpu (at least 3 times), motherboard, cpu, psu (twice), ram, alien fx board, essentially every significant component in the system has been replaced to no avail. Eventually I was sent a full replacement system, but the same problem carried over.
For a while it was agreed the problem was the top 5970 overheating as the bottom one was directly below it leaving no more than a couple mm between them, essentially preventing it from cooling itself at all really. Since the service manual on Dell's own site states the bottom pic slot is 16x, I got hold of a longer Crossfire bridge and tried putting the 2nd 5970 in the lower pci slot to allow more space between the cards. Unfortunately it turns out the bottom slot is actually not a pci 16x slot, so that did not work.
For a short time it seemed like running the pci fan at almost full blast (using command centre) while gaming had sorted things, it certainly helped keeps temp below dangerous levels (although also pushed noise levels into ridiculously loud). But it turns out the problem isn't heat related after all, because last week, 2 minutes into a game with pci fans full blast (and temps at the time would have been no more than 75) the same shutdown occured.
So now I'm once again unsure what is causing it or how it can be solved, so I can't actually give an answer to this problem, but hopefully some of these failed solutions might save you going through the months of messing around I've had to go through. I'm going to have a look into this possible fix regarding the power cable configuration, but to be honest I will not be holding my breath :P
morblore
2 Intern
•
2.4K Posts
0
October 3rd, 2010 21:00
You guys really need to check your PCI power connections going to the video cards. This is a known issue so you gotta check them and make sure its connected right. If you are not sure if they are connected right post a picture and we can tell you. If its not the connections then the PSU is not powerfull enough or its bad. I dont think this is the case but that is the only other thing it can really be. So again check the connections! post pictures if you are not sure!
Just because you can play one game for hours on end doesnt mean anything. Some games stress the system more then others. While the cards have enough juice in one game another game may stress it more making it draw more power and that is when it will shut down the system.
Best way to test this is to use a stress test. Furmark is great for this but dont run it to long. If you can run Furmark for 2min then its not a power issue. Don't run it over 2min though to be safe. It can melt your GPU if you do.
The systems I have seen that are drawing to much off a rail can not run Furmark at all. They will crash soon as you start the test. Some will crash about 5secs into it. If you run it and this happens to you then its ether
1) Incorrect power connections
2)Inadequate PSU
3)Failed PSU
4)Gremlins
As far as the PSU having enough power goes...I ran 2 GTX 295's on the 1100watt PSU just fine. lots of people have come on the forums with your issue and they solved it by changed the power connections. So please go and make sure they are 100% connected like they should be. If they are not then the system WILL shutdown on you when stressing the cards.
You do not care how many connections are on each cable. You do care that only 1 cable is powering each 8-pin. You can ONLY run one 8-pin off of one cable/rail. You can ONLY run two 6-pins off one cable/rail. This is true for ANY system.
On my system my PCI power cables are Yellow,Blue and Black.
I run the 8-pin from the yellow cable to the one 8-pin on the top card.
I run the 8-pin from the blue cable to the 8-pin on the bottom card.
I run the two 6-pins from the black cable to the two 6-pins on each card.
wedgie20
8 Posts
0
October 4th, 2010 00:00
Checked the connections, I guess I shouldn't really be surprised but sure enough the wrong cables have been used. So that's a few months and several thousands of pounds wasted because some dummy didn't build the system correctly (twice!). It's clearly enough of a known problem to warrant it's own space in the troubleshooting section, kinda makes you wonder why this solution was never suggested when I detailed this exact problem on the first or any of the subsequent 30+ phone calls to Dell technicians???
Haven't done enough testing yet to know for sure if it's sorted, but what is certain is the cables were previously not correctly setup, and now they are (according to the instructions on the troubleshooting page).
For anyone with this problem, you could be doing yourself a huge favour if you do as Morblore suggests and make the pci connections the very first thing you check, there's a decent chance you'll save yourself a whole lot of hassle.
Centclip
16 Posts
0
October 5th, 2010 03:00
yeah my cabels where wrong connected to but it dident change the powerup issue i have and also out of the blue my crossfirex stopped working. have bean in contact with dell support and they have concluded that the only way to know what is causing my problems is to "confiscate" my computer for abaut 1 week to find what my problem is. will wright when i know more.
wedgie20
8 Posts
0
October 5th, 2010 09:00
Yeah Crossfire was disabled for me too and the system would crash if I tried to enable it. Uninstalling then reinstalling the video drivers sorted that out. I'm surprised they've agreed to collect your system, took me a long time to convince them to just take it away then send back when fixed, and when they finally did set it up it was then cancelled because apparently the parcel was too big.
Out of interest, do you use anything to control your gpu fans? If you have 2 5970's like my setup then having aggressive fan controls is an absolute must, otherwise the temps on the top card will reach dangerous levels very quickly. I use MSI afterburner which does the job nicely.
Centclip
16 Posts
0
October 5th, 2010 23:00
i tried to reinstal the latest ati driver but it didn't work so i can not use the crossfire. no i dont use anyother program then the defult to control the temp and fan's of the gpu. you think that can be the problem?
wedgie20
8 Posts
0
October 6th, 2010 00:00
Did you remove the drivers before trying the reinstall? A quick pass with driver sweeper should do the job if you're having trouble doing this - http://www.guru3d.com/category/driversweeper/ Then when you reinstall, Crossfire should be on by default.
I couldn't go as far as saying all your problems are down to not manually controlling the fans, but if you have 2 5970's setup like I do then it is certainly necessary. A pair of 5970's should either be watercooled or at the very least have a fair amount of space between them. The gap between mine is at best 3mm, but since the gpu fan is underneath the card, it essentially removes the ability of the top card to effectively cool itself at all really. Add to that the heat rising from the card below it and it doesn't take long to hit very high temps. Controlling the fans manually still won't stop the top card getting very hot (mine sits just above 52 degrees doing absolutely nothing), but it will keep it just below dangerous temps (at the expense of having a very loud system).
MSI Afterburner does the job well - http://event.msi.com/vga/afterburner/download.htm You can set a sliding scale so that as the temps increases, so do the fans (Not 100% sure but you may need to unlock the gpu in the overdrive section of catalyst control centre to allow this). That's half the job, the other half is to control the PCI fan in the command centre. I use 4 profiles, the default auto alien control for anything other than gaming, Light gaming (fans set to manual at 50%), Medium gaming (fans set to manual at 70%) and Heavy gaming (fans set to manual at 85%). It's then just a matter of choosing the appropriate fan profile depending on how intensive the game you are playing will be.
I've still not had the random reboot since switching the cables round so as long as that remains the same I'm at the point where I'm ok with how my system runs. Sure I'd rather I didn't have to be running the fans so loud, but to be fair it's only the most system intensive games that need the highest setting, and it only takes a couple minutes before I don't even notice them anyway. The alternative would be to remove one of the cards, and I don't fancy doing that after how much they cost anyway.
Hope you get yours sorted too, believe me I know how annoying all this can be :)
Centclip
16 Posts
0
October 6th, 2010 02:00
thanks for the info, did the sweep and reinstaled the 10.8 and the crossfire works
will try 10.9 later. i was also able to play darksiders in 1024x720 for 25 minutes and that was something i could not do before changing the wires, then it was like 25 seconds, i was only able to play in 800x600
but atleast i could play and now it works as i should, no sending to DELL either cause i seam to have solved the problem myself with MY startup issue. all i did was disable the pre booth fan and vent system so it was nothing dangerus at all. thanks guys for all the help