something not right w/ second 280x

O

Orddie

2[H]4U
Joined
Dec 20, 2010
Messages
3,369
Hey all,

Got my new 280x yesterday. Plugged it up and everything went normal until i started to run cgiminer.

i have two xfx 280x black editions.

MSI afterburner is showing gpu0 at 53c and gpu1 at 81c. The concern is that the CPU usage on gpu1 is going from 60% to 99%. Any need for concern?

my command line is in a bat file is
setx GPU_MAX_ALLOC_PERCENT 100

C:\cgminer-3.7.2-windows\cgminer.exe --scrypt --thread-concurrency 8192 -I 13 -g 2 -w 256 --lookup-gap 2 --gpu-engine 1027,1027 --gpu-memclock 1500,1500 --expiry 1 --scan-time 1 --queue 0 --no-submit-stale -u test-p password -o stratum+tcp://usa1.X.com:3333 --gpu-fan 95,95 --api-allow W:10.0.1.3,127.0.0.1 --api-listen --shaders 2048,2048 --temp-cutoff 88,88 --gpu-powertune -20,-20

msi.jpg

img_3124.jpg

img_3125.jpg
 
Last edited:
Any reason to have powertune at -20? I found that I could relieve bouncing hashrates and temperature by actually bringing the powertune to +20, and removing my undervolt. I was using a 6950 however.
 
jjandrob said:
MSI afterburner is showing gpu0 at 53c and gpu1 at 81c. The concern is that the CPU usage on gpu1 is going from 60% to 99%. Any need for concern?
Click to expand...

It's throttling for temps. Those xfx throttle in the neighborhood of around 83-85c core temp (or could just be throttling to a correlated VRM temp).

File said:
Any reason to have powertune at -20? I found that I could relieve bouncing hashrates and temperature by actually bringing the powertune to +20, and removing my undervolt. I was using a 6950 however.
Click to expand...

That's because it allows you to exceed stock power and temp usages. Reading about Powertune is actually pretty informative as to it's design.

Edit: Looks like you are using an Extreme4 or so. The spacing causes GPU1 (closes to the CPU) to suck heat off the back of GPU0 around the VRM area and dump is straight onto the VRM area of GPU1. I had a similar issue. Cleared up with box fans lol.
 
Last edited:
the window fan is blowing 0 degree air onto the cards.

the fan on top of the cards should be taking the outside air an blowing into the intake of the cards.

what do you suggest? I had a second 20mm fan @ the back of the cards? i was thinking of getting a PCI express riser cable and move it out.
 
Looks like the GPU fans are directly blowing into the window fan if I read the picture right.
Any way to rotate the rig so those are not blowing against each other?
That one small fan on top may just be pulling the hot air from the card outlets back through the gap between.
 
i thought the fans on the cards are intake fans. the big black fan on the back, is pulling air away into the room.

the fan on top is blowing air down into the what i though was the intake fans for the card.
 
jjandrob said:
the window fan is blowing 0 degree air onto the cards.

the fan on top of the cards should be taking the outside air an blowing into the intake of the cards.

what do you suggest? I had a second 20mm fan @ the back of the cards? i was thinking of getting a PCI express riser cable and move it out.
Click to expand...

My solution was two box fans, one blowing across, the other blowing down, works like a treat, but burns power. I have risers en route to my house, so I'll be able to fine out if spacing works better.
 
You may want to try swapping the two cards around - there may be a problem with the hot one. You shouldn't have that much difference in temps considering you are piping cold air directly into the machine.
 
what kind of hashrate are you getting with those XFX black cards? I have one and haven't really gotten it to perform like it should yet. I have bad RAM heat issues too and mine is hanging with a big box fan blowing over it. I believe there may be some kind of design issue at play here.
 
cgi miner says im getting 1.39mh. When i was doing a single card, it was sitting at 714k/h

hash.jpg
 
What are those drops to, 0%? Odd.

Does the card do this when alone? If you swap the cards, does it still do this or does it switch to the other card? (That would probably indicate a MB issue)
 
Yakk said:
What are those drops to, 0%? Odd.

Does the card do this when alone? If you swap the cards, does it still do this or does it switch to the other card? (That would probably indicate a MB issue)
Click to expand...

Thanks for the response.

the drops are actually to 67%, then back up to 97, then 100%. It only looks like 0% since the graph is maxed out and shrunk in view. The card that's doing it was the card that was running alone for over a week before the second card arrived. The new card is the one running at 59c. I do not think the original card has an issue now.... least i hope not!

I think as others have said here... i have a cooling issue. 2night i will throw a few extra 20mm fans at the issue and c what happens.
 
mkrohn said:
what kind of hashrate are you getting with those XFX black cards? I have one and haven't really gotten it to perform like it should yet. I have bad RAM heat issues too and mine is hanging with a big box fan blowing over it. I believe there may be some kind of design issue at play here.
Click to expand...

I'm getting 720kh/s on mine. The VRMs do get hot. The outer shell of the card assembly is what cools the VRMs. It's kind of like a three piece system. Better than XFX previous attempts but still a lackluster design for high density/compact design. They perform great if you have generous amounts of space and airflow.

Check out the pictures here. http://www.hardware-360.com/xfx-radeon-r9-280x-dd-black-edition-oc-3gb-review/4/
 
I'll have to spend some time in the basement and figure out exactly which one the XFX one is. I've been meaning to get down and try the pcie 1x jumper thing somebody on here showed me so that I can get my risers working properly
 
jjandrob said:
i thought the fans on the cards are intake fans.....
Click to expand...

Not an expert but I believe this is reversed. The card fans usually throw the heat out of the back of the PC. Plus air will leak out around the shrouds as well.
But I could be wrong.... :)
 
when i place my hand by the back of the cards, i feel a good bit of air coming out.
doing the finger trick, i feel air going in the fans's and out the front & back of the card.

so i guess 1 vote for intake, and 1 vote for exhaust. maybe i should do a thread pool. ha
 
jjandrob said:
when i place my hand by the back of the cards, i feel a good bit of air coming out.
doing the finger trick, i feel air going in the fans's and out the front & back of the card.

so i guess 1 vote for intake, and 1 vote for exhaust. maybe i should do a thread pool. ha
Click to expand...

These third party coolers do in fact dump more heat INTO the case than out of it. airflow out the back of the card is quite a bit stronger than what should be tossed OUT of the case. This crap is why the reference coolers are desired for mining.
 
jjandrob said:
if i get a PCI express riser, should it be powered?
Click to expand...

there's quite a bit of debate on this. I'm just having problems with risers in general it seems. the x16 ones I have all work great but the x1 is really goofy
 
mkrohn said:
there's quite a bit of debate on this. I'm just having problems with risers in general it seems. the x16 ones I have all work great but the x1 is really goofy
Click to expand...

i was looking at... PCI-E PCI Express 16X to 16X cable. but it comes w/ a 4pin power. What i do not want is to run the risk of shorting the card if I do or do not plug the power in to the cable. Why would you need to power the cable anyway? Sorry for the noob questions, but i thought power was brought into the card via the 4+8 pin connection in back.

Thanks!
 
jjandrob said:
i was looking at... PCI-E PCI Express 16X to 16X cable. but it comes w/ a 4pin power. What i do not want is to run the risk of shorting the card if I do or do not plug the power in to the cable. Why would you need to power the cable anyway? Sorry for the noob questions, but i thought power was brought into the card via the 4+8 pin connection in back.

Thanks!
Click to expand...

well on some cards they need some added juice that the motherboard normally provides. On well designed cards that get the power they need from the PSU cords hooked straight to it you will not have any need for the powered cords.

Do the 8 pin or 6 pin calculation on what your cards need and what the PSU is sending to it. I personally only have powered risers but apparently the boards I have need me to do the pin short to get the 1x risers to work and I haven't gotten around to doing that quite yet. I think I'll pull the asrock Z87 rig off the shelf and to a spot I have better lighting before I try to do the pin shorting trick.
 
jjandrob said:
if i get a PCI express riser, should it be powered?
Click to expand...

With my limited experience with risers, it depends on your motherboard. I am running 3 cards on each board right now using unpowered 16x-16x risers. They work fine. However, on a different board altogether, I cannot even get one card to run using a riser... I have not tried using powered ones with that board yet.

The unwritten rule I've been hearing from various places around the interwebs is you can run unpowered risers for anything up to 3 cards, anything more than that need to be powered otherwise you risk burning out your board since your cards will start drawing more power than your board can supply through the PCI-e slots.
 
Carbon_Rod said:
With my limited experience with risers, it depends on your motherboard. I am running 3 cards on each board right now using unpowered 16x-16x risers. They work fine. However, on a different board altogether, I cannot even get one card to run using a riser... I have not tried using powered ones with that board yet.

The unwritten rule I've been hearing from various places around the interwebs is you can run unpowered risers for anything up to 3 cards, anything more than that need to be powered otherwise you risk burning out your board since your cards will start drawing more power than your board can supply through the PCI-e slots.
Click to expand...

AH! that explains why they supply the power risers. awesome m8. Thanks!
 
odd that MSI afterburner see's GPU 1 as GPU1 but GPU0 in cgiminer as GPU2.
 
all of you are awesome! added a fan to the setup and now temps are in the 70's with both CPU's at 99% usage.

Thanks so much all!
Solved1.jpg
 
Looks like you could save the fan life on GPU0 by adding the fan control command line and setting the temp up to 70C maybe.
Not sure if that would allow GPU1 temps to rise again but it might be worth a try.
I actually set my three GPUs to 75C control.
Allows one to run at only 50% fan, one at 70% fan and the hottest at 85% (pretty much full blast).
 
You should try undervolting your cards.

My 280x runs 1020 / 1500 @ 1.019v (from the stock 1.200v). Power consumption went down on my rig by ~70 watts, plus my fans/VRM don't need to work nearly as hard since heat production is way down. Card stays at 67 C on 64% fan speed.

Doesn't sound like a lot, but since it's 16 cents per KW/H here it's another $8 off my electric bill to boot. Between all of my cards, I save ~220w by undervolting, so the savings are measurable (~$250/year in mining power use reduction)
 
canna said:
You should try undervolting your cards.

My 280x runs 1020 / 1500 @ 1.019v (from the stock 1.200v). Power consumption went down on my rig by ~70 watts, plus my fans/VRM don't need to work nearly as hard since heat production is way down. Card stays at 67 C on 64% fan speed.

Doesn't sound like a lot, but since it's 16 cents per KW/H here it's another $8 off my electric bill to boot. Between all of my cards, I save ~220w by undervolting, so the savings are measurable (~$250/year in mining power use reduction)
Click to expand...

how would i do that? i tried a few things while the app was running but it stayed at 1.200v
 
MSI Afterburner if you are on windows. Linux, I believe you are out of luck currently.
 
_r2h said:
MSI Afterburner if you are on windows. Linux, I believe you are out of luck currently.
Click to expand...

cgminer has a switch "--gpu-vddc" where you can specify voltage
 
canna said:
What brand/model of card?
Click to expand...

XFX 280x black

_r2h said:
MSI Afterburner if you are on windows. Linux, I believe you are out of luck currently.
Click to expand...

i am on windows. The option for volts is grayed out... even though the box the cards came in shows that voltage is unlocked...

bigted said:
cgminer has a switch "--gpu-vddc" where you can specify voltage
Click to expand...

tried that... such is --gpu-vddc 1.019,1.019. Looking at the settings.. it still showing 1.200v
 
I have an XFX black 280x as well and have not had any luck changing the voltage. The gigabyte 280x I have is also voltage locked. My Asus is not locked and works nicely.
 
mkrohn said:
I have an XFX black 280x as well and have not had any luck changing the voltage. The gigabyte 280x I have is also voltage locked. My Asus is not locked and works nicely.
Click to expand...

on the box it says that the voltage is unlocked.... what did xfx do.
 
mkrohn said:
I have an XFX black 280x as well and have not had any luck changing the voltage. The gigabyte 280x I have is also voltage locked. My Asus is not locked and works nicely.
Click to expand...

My Gigabyte R9 280X is voltage locked, but you can change the voltage using a BIOS editor, which is what I did.

GPU-Z shows it as a 7900 series, but it's actually a 280x.

xfPVV0F.png


V7BE tool: http://www.techpowerup.com/forums/threads/vbe7-vbios-editor-for-radeon-hd-7000-series-cards.189089/

Note: I'm not responsible if you F up your card by doing it wrong. I'd advise you only proceed if you have a backup bios switch that you can swap to.
 
mkrohn said:
I have an XFX black 280x as well and have not had any luck changing the voltage. The gigabyte 280x I have is also voltage locked. My Asus is not locked and works nicely.
Click to expand...

What are you trying to use to change voltages?
 
You must log in or register to reply here.
Back
Top