Why are the VRMs on some of the 280x's running in the 100C range?!?!

[Gazza1]

For all of you that have throttling 280x's from running scrypt that bounce from 64-99% it is because the VRMs are running over 100C!

Last year's 7970s did not have this problem!

So after buying some heatsinks yesterday and getting ready to put them on, guess what I found when I opened a card up?  Heat sinks already on the card...

So what is going on?!?!?!  MSI released a fix for their cards.  SAPPHIRE GET YOUR ASS IN THE GAME AND FIX THIS NOW.

Just to be clear I have many 280x Sapphire OC running spaced apart by a window that has been 0 degrees outside with a box fan exchanging the air and the VRMs are still overheating

And for those that may be confused VRM temps are not core temps.  Use GPUZ or another monitor for reading VRM temps.




Alright I took a couple screen shots for you guys.

Both rigs are near a wide open window with box fans exchanging air with the freezing outside.  The 7970 rig is at 80% fan speed, the 280x is at 100% and the 280x rig is also on the bottom shelf with access to more of the cool air from outside since it is a wire shelving unit and there is plenty of space below it for airflow.  Now if VRM temps are in the 100's when it's 0 degrees outside, what can you imagine it will be in the summer!!!

Also, the throttling isn't so bad on this rig (ATM).  The rig in the living room has hashrates throttling to the 500s because it is warmer.

Here is a system with last years 7970s.  All Sapphire OC


Here is a system with 280x's.  All Sapphire OC

[Snard]

My Sapphire Toxic is performing great at full load. Its been going full boat for the past 24 hours and doesnt go above 72 C.

[Gazza1]

You are right.  I have a toxic that performs great too.  This is for the cards that have throttling issues

[Snard]

Ahh my bad.  Still not sure why i cant get the temp of the vrms of the toxic. 

[stormtide]

Hmm.. I got 2 Sapphire 280x Vapor-X each pushing about 760KH/s, both hoover around 60-70C, never been higher. I love these cards.. All my 7950's struggle to keep under 80C..

[DrJimmy]

I have two Sapphire Dual X 280X VRM at 106 and 96 C been running for almost a month. I hope it doesn't fry.

[Gazza1]

I have two Sapphire Dual X 280X VRM at 106 and 96 C been running for almost a month. I hope it doesn't fry.

I hear ya man

Hmm.. I got 2 Sapphire 280x Vapor-X each pushing about 760KH/s, both hoover around 60-70C, never been higher. I love these cards.. All my 7950's struggle to keep under 80C..

This is for the VRM temps not core temps.  VRM temps can be read in GPUZ

[stormtide]

Ok, I'm looking for a way to get VRM temps in Debian... seems hard to find an app that will do it

[stormtide]

I got 1 280x in a windows box, ran it at 730KH/s and yes VRAM temp is 103C for #1 and 101C for #2. I don't know is this a problem though?

If they were to die, I guess I would just return them. I mean they do have warranty?

EDIT: I also noticed something funny, when running the GPU-Z tools the hashrate actually drops about 40KH/s for me. Simply closing it put's it back at 730KH/s.

[Gazza1]

Yeah 100C is definitely too hot.  Do your cards throttle at all?  Hashrate bouncing up and down and not a solid constant 730ish?  They start throttling right around 105C it seems, which is AMDs way of saying it's too hot.  Hashrates end up dropping to the 500s and stay that way.

The Sapphire are under 2 year warranty, I imagine if they don't fix this they are going to be swamped with RMAs eventually.  Let's hope it doesn't come to that.

Yes hashrates drop sometimes when monitoring apps are open.

Last years model 7970 of the same card the VRMs run in the 70s-80s.

I don't know what they did different.  I pulled off the heatsinks on the 280x and the thermal padding is very thick which probably isn't helping.  The thick padding almost looks like it would just smother the VRMs instead of helping them.  I might try and remove the heatsinks and padding on one of them again and plug the card in to test.  In a bit.  I'm fooling around with some other things atm.  

[Nullu]

What are the stock ramsinks made out of? Because if they're aluminium, then heat capacity is an issue. Copper has a higher capacity to transfer heat than aluminium does, and it's a lot cheaper for card manufacturers to use aluminium.

[Gazza1]

It is indeed aluminum.

[MisO69]

Hmm.. I got 2 Sapphire 280x Vapor-X each pushing about 760KH/s, both hoover around 60-70C, never been higher. I love these cards.. All my 7950's struggle to keep under 80C..

Yes, I have 6 of those. They have the Hynix memory and older bios version. I also have 18 of the ones with problems. They have Elpida memory and newer bios. The Bioses are not compatible because of the different memory manufacturers. I managed to get a tweaked bios from the litecoin forums that gets the hashrate to a steady 700 but its still shy by 60. FFS! Sapphire you had such a great rep. I hope they fix this with a bios update. Otherwise I'll have to never buy sapphire again.

[MisO69]

For all of you that have throttling 280x's from running scrypt that bounce from 64-99% it is because the VRMs are running over 100C!

Last year's 7970s did not have this problem!

So after buying some heatsinks yesterday and getting ready to put them on, guess what I found when I opened a card up?  Heat sinks already on the card

So what is going on?!?!?!  MSI released a fix for their cards.  SAPPHIRE GET YOUR ASS IN THE GAME AND FIX THIS NOW.

Just to be clear I have many 280x Sapphire OC running spaced apart by a window that has been 0 degrees outside with a box fan exchanging the air and the VRMs are still overheating

And for those that may be confused VRM temps are not core temps.  Use GPUZ or another monitor for reading VRM temps.

Do you guys think there is a generic waterblock that could fit this card? Th reason I ask is because I have no idea what I'm going to do when the temps get higher. AC is likely but liquid cooling would be the ultimate.

[Nullu]

For all of you that have throttling 280x's from running scrypt that bounce from 64-99% it is because the VRMs are running over 100C!

Last year's 7970s did not have this problem!

So after buying some heatsinks yesterday and getting ready to put them on, guess what I found when I opened a card up?  Heat sinks already on the card

So what is going on?!?!?!  MSI released a fix for their cards.  SAPPHIRE GET YOUR ASS IN THE GAME AND FIX THIS NOW.

Just to be clear I have many 280x Sapphire OC running spaced apart by a window that has been 0 degrees outside with a box fan exchanging the air and the VRMs are still overheating

And for those that may be confused VRM temps are not core temps.  Use GPUZ or another monitor for reading VRM temps.

Do you guys think there is a generic waterblock that could fit this card? Th reason I ask is because I have no idea what I'm going to do when the temps get higher. AC is likely but liquid cooling would be the ultimate.

Let me know the card model number and I'll check.

[Nullu]

It is indeed aluminum.

That'll certainly be part of the problem. Replacing them with copper heatsinks should help.

[Vickytor]

100C on the VRMs is pretty hot indeed, even tho they are guaranteed to work till 120C. But I don't think 90C+ is going to do them any good.
My ASUS cards also have VRM heatsinks and temps are 70C-ish.

[Gazza1]

Updated with more pictures.

I would like to see if they are going to fix the problem so I can avoid having to do anything that will void the warranty.  I have emailed them with a very large earful, so we will see.

If they do nothing, I will rip off the heatsinks and put copper ones on.  And I as well will stop buying Sapphire and move to Asus, MSI, or Gigabyte.

Also, this is the card Model #: 100363BF4L

[BrewCrewFan]

Ehhhh I donno. Can you return the card and get another brand? I know with my xfx cards they put stickers on the screws now and if broken they will void the warranty. Which sucks because it never used to be like that and in fact they did not mind at all. I guess too many idiots that did not know what they were doing fucked up enough cards to put the end to that.

The only thing I can think of is poor contact. No way should they get that hot . Usually that is the lead cause to the over heating like that.

[Gazza1]

yeah man, the heat is absurd with the cold temps around it.  I thought about returning them all but with the cost of shipping and the few days/week or so of not having them running it just seems like I'd lose more money than it's worth.  Plus I bought most of them when they were 300 bucks. Even with throttling hashrates I think that still makes them worth it for 300 bucks, but the heat is still a nono.

If they can't release a bios that fixes it I'll slap copper all over them and just switch brands from now on.   Actually I'm pretty sure I'm switching brands anyway.