RX 460 crashing in games

Adec001

Commendable
Jan 22, 2017
58
0
1,660
Hello everyone,

Recently I purchases Gigabte RX 460 Windforce edition, that doesnt require any 6 pin power connectors, thinking it would be perfect for my budget oriented build.

Now, the card runs very well in most games I tried (Fallout 4, Overwatc, Doom, Darksiders 2 etc.)
But in the past few months I experienced maybe 2 or 3 driver crashes and 1-2 BSODs.

I always monitor its activity with MSI Afterburner, checking temps and all that stuff.
Only thing that was weird to me was its power usage. It wasnt going over 50W, even though its 75W rated, even on full load, it was always under 50W.

Now, I am using on old OEM mobo Medion MS 7646, so I though maybe its PCI socket was limited.
Here is its manual:

http://msi-ftp.de/Manuals/7646-englv1.0-MedionP6300.pdf

It doesnt say anything about PCIE slot power tho.

Lastly I checked with GPU-Z and indeed it says, board power limit of 48W
(EDIT:as it has been pointed out below 48W is GPU board rating, not MOBO)

http://gpuz.techpowerup.com/17/07/14/8q5.png

Now guys, how do I workarround this, because replacing card or mobo is a no-go for now.
How do I prevent card from crashing when trying to use more than 48W. My bet was to underclock the core and memorry, am just not sure how much?

Any advice is very much appreciated.



Rest of my configuration:

AMD II X4 B97 3.2Ghz
Gigabyte RX 460 2gb Windforce
Medion MS 7464 mobo
8Gb of RAM
Corsair cx 400W
Random 320GB HDD
 

Adec001

Commendable
Jan 22, 2017
58
0
1,660


Are you sure 10% will be enough? By some logic 10% of 75W is only 7.5W less, excue me if this isnt the way it works, I never really messed arround with it.

Also is that the only thing I should do? No underclocking?

Thank you
 

RobCrezz

Expert
Ambassador


Worth a try as a starting point, the power limit on the card should be lower than 75w as standard anyway.
 

Board power limit means GPU board power limit (not motherboard pcie slot power limit).
You can try updating BIOS of your GPU.

Those crashes might be temperature/OC related. What temperatures are you getting during gaming? Try reducing OC.
 

Adec001

Commendable
Jan 22, 2017
58
0
1,660


Its definitely not temp related because like I said I carefully monitor all that stuff, and it never got over 60C during load. Asside of its 1% factory OC, I never OC-ed the GPU, now, I was just running some benchmarks trying to replicate the crash. Didnt have any luck, but what I did notice is:

During Unigine valley benchmark, GPU power usage was always 47-48W, so, Its maximum value I guess, But core clock was having hard time keeping those 1212Mhz,jumping anywhhere between 1150-1200 card is rated to. I did a slight downclock to 1120Mhz, and power consumption dropped to 42W max, but core clock stayed 1120Mhz trhoughout 3 whole runs.
(meassured with MSI afterburner)

Any idea, what to make out of that? I was certain, this would explain board power limitation, since underclocking ensured core clock stability with less than 48W drawn. Idk tho, please help :(
 

Adec001

Commendable
Jan 22, 2017
58
0
1,660


No I didnt, and to be honest, I am very scared to do it anyway, because of the possibility of bricking my card.
 
BIOS update of motherboard can potentially brick it. Then it's problematic to revive it
(unless motherboard has specific revival mechanisms like dual-bios and other).

BIOS update of graphics cards is quite different. Even if you brick it, to revive it you just have to flash in the old BIOS.
So - save old BIOS of your graphics card before flashing it and you're quite safe
(GPU-Z allows saving GPU BIOS).
 

Adec001

Commendable
Jan 22, 2017
58
0
1,660


Ok, didnt know that :D
I might give it a go, if the crashes continue. I did underclock it for 80Mhz for now, and will see how it goes. Also, new update is out, so I will try that too.

Thanks for the help, I will update thread if something changes.

On the side note, where would I find official Bios update?
 

RobCrezz

Expert
Ambassador


Careful. If something goes wrong with a graphics card flash it can be bricked too and unable to reflash.
 

RobCrezz

Expert
Ambassador


Its possible to brick it from a failed flash and it no longer show up as a device.
 

Adec001

Commendable
Jan 22, 2017
58
0
1,660
Its ok, I dont think I would be updating bios anyway. I cant really risk bad flash, since I dont have any other gpu to connect, and I cant afford to mess up my new card :/

I have tried downclocking the card 90mhz. Experienced one freeze in Fallout 4 in past 5 days,
(As I said problems do occur very rarely and impossible to replicate)
However, this time driver didnt crash and recover as it happened before when fallout freezes occured, it just froze, and I was able to go to desktop and end process. What I found in event viewer was error in fallout4.exe, no mention of gpu or drivers, so Idk really.

I think next I will try is to increase power to +20% in MSI afterburner, since card does sometimes struggle to keep its 1212mhz clock under load. Will post back if any crashes occur with power increased.
 


I agree. Reduce the power limit in increments, starting with (say) 10%. If that's not enough, simply go lower.

Does GPU-Z or Afterburner allow you to see the cards consumption in watts?

That'll give you a clue as to how near you are to the limit of the PCIe slot.
 

Adec001

Commendable
Jan 22, 2017
58
0
1,660
Thank you for replies, but as it was pointed out above, PCI slot isnt limited in power,aside from standard 75W, thus there is no point in reducing card's power consumption. Its already using well beyond that 75W limit, meassured in MSI afterburner its limited on 48W, when power limit is at 0.
I believe I read somewhere that other ppl were having problems with card power throttling with stock power limit, and increasing it, helped them stabilise that boost clock.

Some weird things happened in Overwatch though, where card is only being like 66% under load, ( I play with everything on low for max fps), and my CPU is working much harder, usually at 80-100% to maintain 75 fps, which is max for my monitor. On that settings card is only using 35W max even in heavy fights, and my brother stil had one crash and one BSOD few days ago, so am really not sure what is going on anymore...
 
The card can ofc theoretically draw more than 48 W from the PCIe slot. It's the same with a slot 'rated' at 75 W.

However if the slot is indeed 'rated' at 48W by the manufacturer (some Dell main boards are limited to less than 75 W for example), then there is no guarantee that the power delivery above 48 W will be 100% stable.

For example, I modded the bios of a 750Ti to allow up to 90 W through the PCIe slot. This is 15W above the rating. In my case it has worked out fine. But in your case it might be an issue, as the delta is much larger.

I would reduce the power limit in increments until you get full stability under heavy loads. Then whatever that is (it might be only -10%, it may be more) is what you'll have to live with.

If you are having problems with Overwatch, try testing the different power limits out with other games. You might have a separate issue with Overwatch going on for all we know, and if you remove it from the equation (for now) you can make some progress on finding stability.


 

Adec001

Commendable
Jan 22, 2017
58
0
1,660


Ok, board's PCI slot is definitely not limited to 48W, I did found that some OEM boards had 25W or 30W limitations on their PCI slot, but this does not seem to be the case. Already went through manual, no mention of it, checked the board for written indicator. There is none.

So, in theory, card should be absolutely fine on this board. I even contacted guys on NCIX, to confirm compatibility, and they were positive.

I got the Minidump from the last BSOD my brother had on overwatch few days ago. I cant really make up lot from it. I used Blue screen view to open it. And 2 files were highlited as cause.

If you guys have idea, what is it about:

http://prntscr.com/fwrxd7

Unfortunately rare crashes on Fallout4 dont leave any traces beside Fallout4.exe file crash in event viewer.

I did have another theory, that somehow MSI afterburner and Global Wattman were interfeering with each other. So I uninstalled MSI completely and will use Wattman for modifications.
 

Adec001

Commendable
Jan 22, 2017
58
0
1,660


Nah, Fallout froze once yesterday, even with reduced clock speed. It happened after enemy appeared behind door, and actually scared the sh*t out of me, but it wasnt a heavy load fight. Once frozen I could still see all info from MSI overlay, clock speed was at rated 1212Mhz, usage was 60%, temp 61C, and wattage at 35W. Really weird. No driver crash tho, just app froze.