XFX R9 280X crashing issue
Tags:
- R9 280x
- XFX
-
Graphics
- crashing
Last response: in Graphics & Displays
crabbypatties
September 6, 2014 3:18:25 PM
This card has gave me nothing but trouble for days now. It is brand new, less then a month old when i acquired it. Initially i have been having black screen system crashes requiring hardboot. I have tried many options, guru drive sweepers, uninstalling and reinstalling numerous driver packages, reinstalling a fresh factory windows 8.1, reflashed my MB BIOS to several different versions and reinstalling latest currently. So far the one and only way i have been able to get this card stable was this morning when i decreased the card performance in msi afterburner as shown here:
voltage: stock
power limit: stock
core clock: decreased to 850
memory clock: decreased to 1050
fan speed: min of 50 with sharp 2 o clock increase
Temps dont reach more then 47 celsius with this setting. It can also for the first time complete a 3DMark test. Before it would crash within a minute.
System specs
AMD FX-4100 OC
Corsair H60 water cooler
XFX R9 280X BIOS 015.044.010.000000....
Sabertooth 990fx latest BIOS 1604
1000w coolmax psu
120 gb SSD
1Tb HDD
I am not happy with this. Why have such a beast card only to tame it? It runs stable underclocked, so i think the card still has hope and is not broke right?? There has got to be something i am doing wrong i would think. So i wont give up on it just yet.
voltage: stock
power limit: stock
core clock: decreased to 850
memory clock: decreased to 1050
fan speed: min of 50 with sharp 2 o clock increase
Temps dont reach more then 47 celsius with this setting. It can also for the first time complete a 3DMark test. Before it would crash within a minute.
System specs
AMD FX-4100 OC
Corsair H60 water cooler
XFX R9 280X BIOS 015.044.010.000000....
Sabertooth 990fx latest BIOS 1604
1000w coolmax psu
120 gb SSD
1Tb HDD
I am not happy with this. Why have such a beast card only to tame it? It runs stable underclocked, so i think the card still has hope and is not broke right?? There has got to be something i am doing wrong i would think. So i wont give up on it just yet.
More about : xfx 280x crashing issue
-
Reply to crabbypatties
jimthenagual
September 6, 2014 3:22:07 PM
There is a good chance that your PSU isn't producing stable enough power to the GPU.
Your psu is listed as a tier5 on this list
Tier5: Replace immediately. These units are NOT recommended for any system, no matter the purpose. Reference to higher tiered models for a better and potentially money-saving unit
Your psu is listed as a tier5 on this list
Tier5: Replace immediately. These units are NOT recommended for any system, no matter the purpose. Reference to higher tiered models for a better and potentially money-saving unit
-
Reply to jimthenagual
m
0
l
fkr
September 6, 2014 3:23:33 PM
Related resources
- XFX R9 280X 3GB - Strange No Signal/Detection Issue - Please Help! - Forum
- Overheating issue with XFX R9 280x cards - Forum
- XFX R9 280x Voltage Issues - Forum
- Problems with a brand new XFX R9 280x Black Edition, crashes and FPS spikes - Forum
- XFX R9 280x - Instant Crash - Forum
crabbypatties
September 6, 2014 3:49:15 PM
crabbypatties
September 6, 2014 3:58:31 PM
Alright, crashed on 3rd 3Dmark test when i up the core clock to 950 and the +20 power setting. Will try 900 core clock.
Stayed stable at 900 core clock with +20 power. Might could push 925 but the temp raised up to 52 celsius. Normal for 3DMark testing i am assuming. So i will hang with that area for now. Not real sure what to do about the memory clock or if that is relevant. I read somewhere here that the reference clock numbers AMD put out for this card was lower then what comes on the card. Just all manufacturers push the envelop which is fine of course. Not sure what the reference was for the memory. Seems like my card wont do much better then this though for some weird reason.
No hardware failures that i can find and driver is fine when i am underclocked like this so dont think it is the driver now. Would a bottleneck somewhere cause a black screen crash requiring a hardboot? The build never did this with my nvidia 550Ti but to be fair here the 550Ti dont compare to the 280x.
I have a backup antec 500w 80 plus cert. but not sure that will have enough to juice this build and test the possible faulty 1000w coolmax. Think its possible?
Stayed stable at 900 core clock with +20 power. Might could push 925 but the temp raised up to 52 celsius. Normal for 3DMark testing i am assuming. So i will hang with that area for now. Not real sure what to do about the memory clock or if that is relevant. I read somewhere here that the reference clock numbers AMD put out for this card was lower then what comes on the card. Just all manufacturers push the envelop which is fine of course. Not sure what the reference was for the memory. Seems like my card wont do much better then this though for some weird reason.
No hardware failures that i can find and driver is fine when i am underclocked like this so dont think it is the driver now. Would a bottleneck somewhere cause a black screen crash requiring a hardboot? The build never did this with my nvidia 550Ti but to be fair here the 550Ti dont compare to the 280x.
I have a backup antec 500w 80 plus cert. but not sure that will have enough to juice this build and test the possible faulty 1000w coolmax. Think its possible?
-
Reply to crabbypatties
m
0
l
fkr
September 6, 2014 4:57:04 PM
jimthenagual said:
There is a good chance that your PSU isn't producing stable enough power to the GPU.Your psu is listed as a tier5 on this list
Tier5: Replace immediately. These units are NOT recommended for any system, no matter the purpose. Reference to higher tiered models for a better and potentially money-saving unit
while i agree that you are asking for problems using that power supply it should have plenty of amps on the 12 v rail and should still power that GPU if it is the one i see on new egg it has like 80 amps.
http://www.newegg.com/Product/Product.aspx?Item=N82E168...
i would get a xfx or seasonic power supply. there are many other good ones out there but you are guaranteed a reliable one with these manufacturers
-
Reply to fkr
m
0
l
fkr
September 6, 2014 4:58:34 PM
crabbypatties said:
Alright, crashed on 3rd 3Dmark test when i up the core clock to 950 and the +20 power setting. Will try 900 core clock.Stayed stable at 900 core clock with +20 power. Might could push 925 but the temp raised up to 52 celsius. Normal for 3DMark testing i am assuming. So i will hang with that area for now. Not real sure what to do about the memory clock or if that is relevant. I read somewhere here that the reference clock numbers AMD put out for this card was lower then what comes on the card. Just all manufacturers push the envelop which is fine of course. Not sure what the reference was for the memory. Seems like my card wont do much better then this though for some weird reason.
No hardware failures that i can find and driver is fine when i am underclocked like this so dont think it is the driver now. Would a bottleneck somewhere cause a black screen crash requiring a hardboot? The build never did this with my nvidia 550Ti but to be fair here the 550Ti dont compare to the 280x.
I have a backup antec 500w 80 plus cert. but not sure that will have enough to juice this build and test the possible faulty 1000w coolmax. Think its possible?
at this point if you want to keep overclocking you are going to have to bump the voltage up
i do wonder if your power supply is just not working since you should have no issue with a core clock of 1000 and a memory clock of 1500
i do want to add that your 500w antec should be just enough to power your system. i would need to see the exact model to check the amps on the 12v rail
-
Reply to fkr
m
0
l
crabbypatties
September 6, 2014 6:57:33 PM
exact antec model is http://www.newegg.com/Product/Product.aspx?Item=N82E168...
The coolmax model is a older one non modular version listed here, have had it for 3 years or so now http://www.newegg.com/Product/Product.aspx?Item=N82E168...
The coolmax model is a older one non modular version listed here, have had it for 3 years or so now http://www.newegg.com/Product/Product.aspx?Item=N82E168...
-
Reply to crabbypatties
m
0
l
Best solution
fkr
September 6, 2014 9:57:03 PM
so the antec only puts out 18 amps on the 12v rail so that equates to 216 watts and that is not enough
thecoolmax looks like this
+3.3V@30A, +5V@30A, +12V1@20A, +12V2@20A, +12V3@25A, +12V4@25A, -12V@0.8A, +5VSB@6.0A
it says your 3 and 4 12v rail put out more power. you may want to try and use those rails or power leads to connect to the GPU or a combonation of one of each
at 25 amps that is 300 watts per rail and that should be plenty, actually 2 20 amp rails should be plenty also
btw amps * vollts = watts
25 amps * 12v rail = 300watts
thecoolmax looks like this
+3.3V@30A, +5V@30A, +12V1@20A, +12V2@20A, +12V3@25A, +12V4@25A, -12V@0.8A, +5VSB@6.0A
it says your 3 and 4 12v rail put out more power. you may want to try and use those rails or power leads to connect to the GPU or a combonation of one of each
at 25 amps that is 300 watts per rail and that should be plenty, actually 2 20 amp rails should be plenty also
btw amps * vollts = watts
25 amps * 12v rail = 300watts
-
Reply to fkr
Share
crabbypatties
September 7, 2014 7:40:27 AM
fkr said:
so the antec only puts out 18 amps on the 12v rail so that equates to 216 watts and that is not enoughthecoolmax looks like this
+3.3V@30A, +5V@30A, +12V1@20A, +12V2@20A, +12V3@25A, +12V4@25A, -12V@0.8A, +5VSB@6.0A
it says your 3 and 4 12v rail put out more power. you may want to try and use those rails or power leads to connect to the GPU or a combonation of one of each
at 25 amps that is 300 watts per rail and that should be plenty, actually 2 20 amp rails should be plenty also
btw amps * vollts = watts
25 amps * 12v rail = 300watts
How would i tell which rail is what?
I originally tried using the two 6 pin rails, both which have a 6+2 connector option. So i did a 6+2 in the 8 pin connector and the other 6 pin rail for the 6 pin connector.
Now i have both of the connections setup with splitters that came with the card. The 8 pin splitter using both 6 pin rails. The 6 pin splitter is using 2 different rails of those regular 4 pin molex connectors. I read in the owners manual illustration not to use any two connectors of the same rails because it would not provide suffice power for i didnt. Not sure if i used the right rails though as i haven't noticed any numbering anywhere. Will take a look at them now.
-
Reply to crabbypatties
m
0
l
crabbypatties
September 7, 2014 5:19:01 PM
Well this is odd. Went back to stock clock settings after i switch the 8 pin splitter out and just plugged in a single 6+2 into the cards 8 pin and kept the cards 6 pin splitter on the 2 molex. I just made it through a full 3DMark firestrike test with no crashes with a score of 5075. WTF!! I am even more confused now...
-
Reply to crabbypatties
m
0
l
fkr
September 7, 2014 9:11:39 PM
crabbypatties
September 7, 2014 9:32:58 PM
Seems to be stable for now, also for reference in case anyone else comes across this issue, i also up the power limit percentage to +20. I guess for some strange reason having the splitters on a 8 pin was causing issues. Yet it would crash without any splitters at all on both connections too. So i am guessing one of these 6+2 rails are wonky. Still have the molex splitters for the 6 pin connection on it without issue. Glad it wasn't the card, it is so much better in every way then my old 550Ti. I will upgrade this ol PSU soon as it is a old part(3+ years running strong now) and probably just time to send her out to pasture and into a spare pc build or backup unit.
@fckr ran 3DMark through all the test after a bit of a cool down earlier Made it through without any crashes again. Thanks for the help, i would of never thought to tinker with the different rails and what not. Itll do for now for sure!
@fckr ran 3DMark through all the test after a bit of a cool down earlier Made it through without any crashes again. Thanks for the help, i would of never thought to tinker with the different rails and what not. Itll do for now for sure!
-
Reply to crabbypatties
m
0
l
fkr
September 7, 2014 10:19:55 PM
jimthenagual
September 8, 2014 11:11:11 AM
fkr said:
so the antec only puts out 18 amps on the 12v rail so that equates to 216 watts and that is not enoughthecoolmax looks like this
+3.3V@30A, +5V@30A, +12V1@20A, +12V2@20A, +12V3@25A, +12V4@25A, -12V@0.8A, +5VSB@6.0A
it says your 3 and 4 12v rail put out more power. you may want to try and use those rails or power leads to connect to the GPU or a combonation of one of each
at 25 amps that is 300 watts per rail and that should be plenty, actually 2 20 amp rails should be plenty also
btw amps * vollts = watts
25 amps * 12v rail = 300watts
I don't believe it works quite like that. The +12v rails refer to internal circuits, and do not indicate how many amps/watts are being put out on the individual PCI-e cables. The PCI-e cables will put out what is asked of them individually, up to the PSU's rated maximum.
So according to here, that PSU is putting out 800W on the +12v rails total. That means there is a total of 800W/12v equals 66A available on the combined +12v rails. Not 90A, like it would be if you simply added the Amps on all the +12v rails individually.
Each PCI-e cable in combination with any other parts using 12v will all share the total available watts/amps available, which in this case is 800W and 66A on the combined +12v rails.
-
Reply to jimthenagual
m
0
l
jimthenagual
September 8, 2014 11:24:29 AM
fkr
September 8, 2014 2:49:06 PM
jimthenagual said:
fkr said:
so the antec only puts out 18 amps on the 12v rail so that equates to 216 watts and that is not enoughthecoolmax looks like this
+3.3V@30A, +5V@30A, +12V1@20A, +12V2@20A, +12V3@25A, +12V4@25A, -12V@0.8A, +5VSB@6.0A
it says your 3 and 4 12v rail put out more power. you may want to try and use those rails or power leads to connect to the GPU or a combonation of one of each
at 25 amps that is 300 watts per rail and that should be plenty, actually 2 20 amp rails should be plenty also
btw amps * vollts = watts
25 amps * 12v rail = 300watts
I don't believe it works quite like that. The +12v rails refer to internal circuits, and do not indicate how many amps/watts are being put out on the individual PCI-e cables. The PCI-e cables will put out what is asked of them individually, up to the PSU's rated maximum.
So according to here, that PSU is putting out 800W on the +12v rails total. That means there is a total of 800W/12v equals 66A available on the combined +12v rails. Not 90A, like it would be if you simply added the Amps on all the +12v rails individually.
Each PCI-e cable in combination with any other parts using 12v will all share the total available watts/amps available, which in this case is 800W and 66A on the combined +12v rails.
http://www.jonnyguru.com/forums/showthread.php?t=3990
this is actually written by jonny guru and it explains everything.
-
Reply to fkr
m
0
l
jimthenagual
September 8, 2014 2:58:13 PM
Yep, read that many times. Has nothing to do with the point I am making however. He is talking about the difference/advantages of multiple +12v rails. My point is, those rails are internal circuits.
You said:
+3.3V@30A, +5V@30A, +12V1@20A, +12V2@20A, +12V3@25A, +12V4@25A, -12V@0.8A, +5VSB@6.0A
"it says your 3 and 4 12v rail put out more power. you may want to try and use those rails or power leads to connect to the GPU or a combonation of one of each"
It doesn't really make sense. You can't say if it has 4 different +12v rails that it means the PSU has 4 pci-e connectors, and that each of those 4 cables are putting out the watts/amps as written on the PSU sticker. The number of PCI-e connectors has nothing to do with the number of +12v rails. There are many examples of single rail PSUs with multiple PCI-e cables, and vice versa, plenty of examples of PSUs with multiple +12v rails but only one PCI-e cable. When we talk about +12v rails, we are talking about internal circuitry, not the capacity of the individual PCI-e cables.
This is the point I am trying to make. Sorry, not trying to be insulting or anything like that. Just wanting clarification for the benefit of readers.
You said:
+3.3V@30A, +5V@30A, +12V1@20A, +12V2@20A, +12V3@25A, +12V4@25A, -12V@0.8A, +5VSB@6.0A
"it says your 3 and 4 12v rail put out more power. you may want to try and use those rails or power leads to connect to the GPU or a combonation of one of each"
It doesn't really make sense. You can't say if it has 4 different +12v rails that it means the PSU has 4 pci-e connectors, and that each of those 4 cables are putting out the watts/amps as written on the PSU sticker. The number of PCI-e connectors has nothing to do with the number of +12v rails. There are many examples of single rail PSUs with multiple PCI-e cables, and vice versa, plenty of examples of PSUs with multiple +12v rails but only one PCI-e cable. When we talk about +12v rails, we are talking about internal circuitry, not the capacity of the individual PCI-e cables.
This is the point I am trying to make. Sorry, not trying to be insulting or anything like that. Just wanting clarification for the benefit of readers.
-
Reply to jimthenagual
m
0
l
fkr
September 8, 2014 3:40:04 PM
-
Reply to fkr
m
0
l
crabbypatties
September 8, 2014 4:45:22 PM
So why is it now working for me is what i am wondering? I still have the molex splitter on the 6 pin btw. 6+2 in the 8 pin, no splitter. Has to be one of the 6+2 rails is wonky you reckon? No doubt one of the weirdest things i have ever come across building pc's.This would be a good time to have a voltometer thingymajigger me thinks.
Weirdest ever which was en ragingly silly being a lose screw trapped behind motherboard, you can imagine everything i went through only to find that was the reason for no post lol.
Weirdest ever which was en ragingly silly being a lose screw trapped behind motherboard, you can imagine everything i went through only to find that was the reason for no post lol.
-
Reply to crabbypatties
m
0
l
fkr
September 8, 2014 5:22:46 PM
jimthenagual said:
Yep, read that many times. Has nothing to do with the point I am making however. He is talking about the difference/advantages of multiple +12v rails. My point is, those rails are internal circuits. You said:
+3.3V@30A, +5V@30A, +12V1@20A, +12V2@20A, +12V3@25A, +12V4@25A, -12V@0.8A, +5VSB@6.0A
"it says your 3 and 4 12v rail put out more power. you may want to try and use those rails or power leads to connect to the GPU or a combonation of one of each"
It doesn't really make sense. You can't say if it has 4 different +12v rails that it means the PSU has 4 pci-e connectors, and that each of those 4 cables are putting out the watts/amps as written on the PSU sticker. The number of PCI-e connectors has nothing to do with the number of +12v rails. There are many examples of single rail PSUs with multiple PCI-e cables, and vice versa, plenty of examples of PSUs with multiple +12v rails but only one PCI-e cable. When we talk about +12v rails, we are talking about internal circuitry, not the capacity of the individual PCI-e cables.
This is the point I am trying to make. Sorry, not trying to be insulting or anything like that. Just wanting clarification for the benefit of readers.
my thought was just that since i saw the different amperage I figured maybe one of those filters that set the amperage went bad and another pcie power lead would help. I woud need to see a [H] article to know what 12v went to where
this is a single 12v rail power supply review by tech powerup
http://www.techpowerup.com/reviews/EVGA/SuperNOVA_P2_10...
this is a power supply with 6 12v rails
http://www.techpowerup.com/reviews/Enermax/EMR1350EWT/2...
the picture on the bottom of page 2 shows that each connector is available to a seperate 12v rail as they are labeled on the package
the benefit of a multi rail is that one can fail and not blow the entire power supply as OCP can be triggered on every rail.
I beleive [H] lost some testing equipment recently while testing a 130 amp 12v rail
-
Reply to fkr
m
0
l
fkr
September 8, 2014 5:34:33 PM
crabbypatties said:
So why is it now working for me is what i am wondering? I still have the molex splitter on the 6 pin btw. 6+2 in the 8 pin, no splitter. Has to be one of the 6+2 rails is wonky you reckon? No doubt one of the weirdest things i have ever come across building pc's.This would be a good time to have a voltometer thingymajigger me thinks. Weirdest ever which was en ragingly silly being a lose screw trapped behind motherboard, you can imagine everything i went through only to find that was the reason for no post lol.
connectors are know to cause issues with power delivery and it is not recommended unless you have to of course. I personally have never had an issue.
do you have the model number of your power supply or a link to the exact one. not really needed just bored at the moment and kind of curious
-
Reply to fkr
m
0
l
crabbypatties
September 8, 2014 8:08:19 PM
fkr said:
crabbypatties said:
So why is it now working for me is what i am wondering? I still have the molex splitter on the 6 pin btw. 6+2 in the 8 pin, no splitter. Has to be one of the 6+2 rails is wonky you reckon? No doubt one of the weirdest things i have ever come across building pc's.This would be a good time to have a voltometer thingymajigger me thinks. Weirdest ever which was en ragingly silly being a lose screw trapped behind motherboard, you can imagine everything i went through only to find that was the reason for no post lol.
connectors are know to cause issues with power delivery and it is not recommended unless you have to of course. I personally have never had an issue.
do you have the model number of your power supply or a link to the exact one. not really needed just bored at the moment and kind of curious
http://www.newegg.com/Product/Product.aspx?Item=N82E168...
Have it had it for a good while now, at least 3 years but never needed both 6+2 rails. The 550Ti never really pushed the psu but hey i was playing WoW like a crackhead, didnt need much
-
Reply to crabbypatties
m
0
l
Related resources
- XFX R9 280x Black Edition - Instant Crash Forum
- R9 280x Toxic Issues (constant black screens/crashes/freezing) Forum
- Solvedis the XFX Radeon R9 280X Double Dissipation Black Edition 3GB backwards compatible to pci express 2.0? Forum
- SolvedR9 280x XFX or GTX 770 MSI Twin Frozr 2gb Forum
- Solvedr9280 x sapphire or r9 280x xfx black?? Forum
- SolvedXFX R9 280x low performance on almost all games. Forum
- SolvedR9 280x XFX 3D gaming problem? Forum
- Solvedwhats better ? XFX R9 280X OR GIGABYTE GTX760 Forum
- SolvedXFX r9 280x 60 hertz monitor Forum
- SolvedWill the XFX r9 280x fit in the Zalman u3 Z9? Forum
- Solvedxfx r9 280 vs xfx r9 280x Forum
- Solvedxfx R9 280x or asus r9 280? Forum
- SolvedWhere to send rebate form of XFX r9 280x Forum
- SolvedXFX pro 550w and Sapphire R9 280x Toxic Forum
- SolvedSapphire Radeon R9 280X 3GB vs x2 XFX Radeon HD 5970 Black Edition 2 GB Forum
- More resources
Read discussions in other Graphics & Displays categories
!