Why one of my R9 270 is hotter than others?

[hobbyandfree]

Hi.
I have 3x Gigabyte R9 270( without X).  But one of them was hitting 88C but others were hovering around 75C even though I set target temp 75C. Then I removed 1 of them( 88C one) and decreased intensity.  Strange part is: One of remaining 2 GPU jumped to 84C from 75C while other one is still around 74C.
I don't understand what is going on.   My cgminer setting: --temp-target 75 --auto-fan

[hostmaster]

check air space and how much space between cards i guess its the middle one which is hotter.

[hobbyandfree]

I removed middle one.  Why R9 270 is ignoring temp target(75C) I set on cgminer settings?

[hostmaster]

I removed middle one.  Why R9 270 is ignoring temp target(75C) I set on cgminer settings?

driver issue.

[xbudahx]

I have the same cards and the cgminer temp cutoff does not work, at least in linux.

You need risers for the cards, with 3 on one mb and no risers 85 is somewhat normal.

[gnomegawd]

monday card and friday cards boys...

learn...

[Androidicus]

Ok, 'old' thread but...

I have 2 XFX R9 270 cards on one mobo (Asrock ProBTC H61) on powered USB style risers.

Xubuntu 13.10 - cgminer 3.7.2

GPU O runs around 8-12C hotter than GPU 1 and gets sick and dies when using same settings that GPU 1 handles comfortably... I have to set GPU 0 at, for example -I 18 when GPU 1 is on -I 19 etc.

We're talking temps of GPU 1 53C, GPU 0 63C

These are running open (no case) and no significant factors affecting cooling.

These cards were ordered separately from 2 diff suppliers about 2 weeks apart. If GPU 0 (guilty one) can run for 24 hours on -I 18 and cgminer left to sort all other settings does this prove that the card is ok but has some other factor i.e. BIOS version that causes it to run hotter etc?

We're talking difference of 45 - 50 k/hs between settings that it'll run (392k/hs) on against settings that cause it to DIE (445k/hs) (in cgminer). GPU 1 will run at 58-60C at 445 k/hs all day...

Mysterious

[Androidicus]

Just as an update:

Simply using:

-I 18 -g 1 -w 256 --thread-concurrency 0 --gpu-fan 100 --no-submit-stale

Returns 435k/hs on each card at GPU0 66C | GPU1 55C

I sometimes feel that messing about with all the 'tweaks' might squeeze 10-15 extra k/hs but causes other issues...

My 2 MSI 7790s have been running for weeks on end at combined average of 500 k/hs for the 2 at 68c/64c