4870 VRM temps?

I was hoping members here may be able to post up there 4870 vrm/pwm temps or even shed some light on safe temps.
The reason I ask is a user on another forum, posted there concerns over the apparently high temps whilst running Furmark 120c+ and around the 80's whilst gaming. He was on a core only block with sinks on ram/vrm and with a 120mm fan cooling the area. I upgraded my version of everest to 4.50.1448 so I could see GPU VRM temps and ran Furmark and I saw temps of 126c. I also seem to be getting around the same 80's in game.
However I was at the same time monitoring the VRM's with a infared thermometer and the rear most ones recorded 60c and the front most where 50c. I am running a mcw60 with sinks on ram/vrm/and the chips in front of the vrm's. I am curious as I never took any readings when the the stock HSF was fitted and as usual do not entirely trust software for temps. If you can help great and thanks in advance.

In this case I would say go with the IR thermometer, although they do feel hot even in idle. Why dont you trust your thermometer?

I was going to ask the exact same question :)

My VRM temp reached 100-105*C when running ati tool ("live" cube, not the scan aritfact as it is less stressful/hot for the card).

I just have heatsink on the small chips (MOSFET ?), they get hot but I still can touch the heatsinks for few seconds without getting burned.

The Vitek chip gets burning hot though, I can't touch it even for a brief instant (actually I got burned when I tried :p:). I need to find a northbridge heatsink and hacksaw it to put on these chips ...

I did it this way http://img363.imageshack.us/my.php?i...scf3447yq3.jpg
The stock PWM Cooling from the Accelero S1 fits on the 4870 Series with a bit of help, i added MX2 on the MOSFETs, and the PWM cooling touches the whole S1 cooling and because the S1 cooling fins are wide and if you're blowing directly on them all pars are being cooled, on both cards my PWM is 40 idle 60 3 hours of Crysis.

Quote:

---

Originally Posted by v0dka

In this case I would say go with the IR thermometer, although they do feel hot even in idle. Why dont you trust your thermometer?

---

It is not a case of not trusting IR thermometer more a case of getting more readings to make a more educated decision on what I am going to do or not do about it.

Quote:

---

Originally Posted by Kondik

I did it this way http://img363.imageshack.us/my.php?i...scf3447yq3.jpg
The stock PWM Cooling from the Accelero S1 fits on the 4870 Series with a bit of help, i added MX2 on the MOSFETs, and the PWM cooling touches the whole S1 cooling and because the S1 cooling fins are wide and if you're blowing directly on them all pars are being cooled, on both cards my PWM is 40 idle 60 3 hours of Crysis.

---

It would be nice to see a picture of the VRM cooling solution before the HSF was fited if you have any.

The HD 4870 stock cooling is one big red metal plate exept GPU where is a big copper cube, and from the bottom side of the HS is a thermal duck tape or something like that. It's pretty useless even if it looks good, 87.5 ~ 92.5 degrees on VRM and 85 on RAM modules. Because its one solid plate the row of 5 Mosfets and 3 Pulses + 8 Memory Modules are heating pretty extreme so Imo you need to cool them separatley.

Quote:

---

Originally Posted by Kondik

The HD 4870 stock cooling is one big red metal plate exept GPU where is a big copper cube, and from the bottom side of the HS is a thermal duck tape or something like that. It's pretty useless even if it looks good, 87.5 ~ 92.5 degrees on VRM and 85 on RAM modules. Because its one solid plate the row of 5 Mosfets and 3 Pulses + 8 Memory Modules are heating pretty extreme so Imo you need to cool them separatley.

---

Sorry I think you misunderstood me, what I was referring to was the Artic Cooling pwm cooling part that you fitted prior to installing the Accelero. A overhead shot of that would have been great. I agree that to remove the red plate from the stock cooler is one way to do it, but like you I preferred the idea of the components being separately cooled and not contributing to each others heat load.

Ah yes the AC one, it was easy, there is a hole at the top of the 5 MOSFET's put one screw there [ with a rubber circle to not to short circuit it ] and the other hole will be 1 mm off the card, so I put through another screw with a rubber circle so just the half of the screw/rubber circle head is holding the Heatsink but it's pretty tight and sits there tighter than jeans on Pamela Anderson. I'm too lazy to put one of the cards out and take a pic of it. But it's pretty easy. The only thing is don;t short circuit the board with screws or the Pulses with the MOSFET HS you will screw there, and a direct access to fresh airflow will keep it pretty low. Don't leave it just on passive.

Did you polish the corners of that PWM cooler? Cause i tried to fit that too and noticed that it seems abit high cause the middle ones aint touching the cooler as good as the ones in the end. Anyway i tried it as you have in your picture, kinda exact like that with heatsinks of the others ones and 2x 120 fans.. and i got worse VRM temp then using the OEM backplate and accelero. You using Everest to check your VRM temp btw? Cause 60c in crysis after 3 hours is kinda really nice ^^

heres what i did to my 4870.

http://img514.imageshack.us/img514/9600/099vq7.jpg
By rangerone7669, shot with Photosmart M415 at 2008-07-25

http://img514.imageshack.us/img514/3167/098nz0.jpg
By rangerone7669, shot with Photosmart M415 at 2008-07-25

http://img401.imageshack.us/img401/4362/vrmworksdm0.jpg
By rangerone7669 at 2008-07-28

i have a 120mm fan strapped onto the hr03-gt, and another 120 blowing on the power circuit area of the card.
notice 97-97% gpu usage, so i'm running at load atm.

any questions feel free to ask.

Ok, from the look of that screen shot GPUZ has been monitoring for about 45 sec's, you also appear to be running F@H now if I am right as per my screen shot F@H does not appear to actually draw any current what so ever really.
http://www.r5jones.aquiss.com/Pictures/fah%23.jpg

Try using Furmark, that will do it after say a 1min run and the games I have found to be the most taxing so far are, Crysis(no surprise there),Mass Effect both of these push the VRM temps upto the 90's after a while.

Ye use furmark pls.. its another world when it comes to pushing your GPU and VRM temp.

Another odd thing is when running Furmark any resolution for me up to 1680x1050 0xAA gets me up to 126c very quickly, however using 4 or 8xAA appears to run much cooler just touching 90c. Very strange as you would have thought 4 or 8 AA would place greater strain on the card.:confused:

ye you right same here. that was strange

Could other people try this and report there findings?

only got abt 10 minutes till my folding finishes. then i'll shut it dow and do some testing with fur.

mine went to 126c after a few minutes aswell.

http://img70.imageshack.us/img70/3917/126kv8.jpg
By rangerone7669 at 2008-07-29

i let it run for a few minutes and it never went above 126c. i havent seen anyone get a temp reading higher than 126c, perhaps that value is nor being read correctly? i touched everything and nothing was hot, a bit warm but not hot.

why are you guys basing this soley on everest? its not ati and it sure as hell cant read cpu temps correctly so why would it read this correctly? it probably isn't and you guys are freaking for no reason. I use a chopped stock cooler with the fan still on there and with my voltmodded (896/1125) card it never crashes, ever.


forget Everest and Furmark and start using your cards for GAMES (I know its hard to comprehend :rolleyes:)

my thoughts exactly! i dont think everest is reading correct values and it games just fine.

hey sniipe hows that vmod working out? i picked up some 10ohm resistors today, gonna try to solder it up later this week. i've never tried to solder anything so small, but i think i can do it. lolz

its nice, I would suggest going for 20 ohm instead of 30, if your going to be watercooling, i think im going to be snapping one of the resistors on each card off my setup because i am volt limited to 896. also i would like to try some RAM penciling but im a bit too lazy to drain my loop ;) eventually ill get to it but so far i am very happy.


I am not happy with this latest driver (8.7) hopefully 8.8 will give me the performance i need to get some real benching in :rolleyes: (read: vantage is broken in CF for me no matter how many times I re-install everything :\ )

Quote:

---

Originally Posted by SNiiPE_DoGG

its nice, I would suggest going for 20 ohm instead of 30, if your going to be watercooling, i think im going to be snapping one of the resistors on each card off my setup because i am volt limited to 896. also i would like to try some RAM penciling but im a bit too lazy to drain my loop ;) eventually ill get to it but so far i am very happy.


I am not happy with this latest driver (8.7) hopefully 8.8 will give me the performance i need to get some real benching in :rolleyes: (read: vantage is broken in CF for me no matter how many times I re-install everything :\ )

---

gonna stick with air on the gpu for now. what would the core voltage be roughly with 20ohm resistor?

Quote:

---

Originally Posted by SNiiPE_DoGG

why are you guys basing this soley on everest? its not ati and it sure as hell cant read cpu temps correctly so why would it read this correctly? it probably isn't and you guys are freaking for no reason. I use a chopped stock cooler with the fan still on there and with my voltmodded (896/1125) card it never crashes, ever.

---

I am basing this soley on Everest because as far as I know nothing else reports those temps. I am not saying it is correct far from it, as I found out using my IR thermometer, however I can not put aside the fact that under certain situations the VRM temps are maybe going out of hand, maybe because I am using a core only block and the normal heatsinks I have used on other cards are not man enough.
I thought that this was be good information to gather so maybe more informed people than myself would contribute.
Also the cooling I have on my card is what Swiftech is now pretty much recommending for w/c cooling this card on their site and if it is not man enough then maybe it saves a few people from wasting money and or burning up their card. I believe Dnottis's card died the other night maybe because of this who know's?

Quote:

---

Originally Posted by SNiiPE_DoGG

forget Everest and Furmark and start using your cards for GAMES (I know its hard to comprehend :rolleyes:)

---

Regarding playing games, that is all I use my card for, I actually do very little benching these days and probably use my PC for about 5hrs gaming per day. So I find it a little patronising for your comment above and it was not really needed or wanted. Maybe that's just your style.:shrug: Just my thoughts I do not want to get in to a p*****g match here!

Quote:

---

Originally Posted by SNiiPE_DoGG

forget Everest and Furmark and start using your cards for GAMES (I know its hard to comprehend :rolleyes:)

---

Have to agree somewhat with your comment. However, I never have known a benchmark to crash the plethora of cards I have owned in the past. Love these diag LEDS.

Granted the games I play (COD4, KO), have no issues at all. I have not tested Oblivion, BS, FSX, DC4.

I'll be grinding out my 4870 shortly, my PSU finally arrived. Woohoo. :D

Quote:

---

Originally Posted by holwill

I am basing this soley on Everest because as far as I know nothing else reports those temps. I am not saying it is correct far from it, as I found out using my IR thermometer, however I can not put aside the fact that under certain situations the VRM temps are maybe going out of hand, maybe because I am using a core only block and the normal heatsinks I have used on other cards are not man enough.
I thought that this was be good information to gather so maybe more informed people than myself would contribute.
Also the cooling I have on my card is what Swiftech is now pretty much recommending for w/c cooling this card on their site and if it is not man enough then maybe it saves a few people from wasting money and or burning up their card. I believe Dnottis's card died the other night maybe because of this who know's?



Regarding playing games, that is all I use my card for, I actually do very little benching these days and probably use my PC for about 5hrs gaming per day. So I find it a little patronising for your comment above and it was not really needed or wanted. Maybe that's just your style.:shrug: Just my thoughts I do not want to get in to a p*****g match here!

---

i wasn't personally talking to you man, no hard feelings :)

i was just making the point that when my card(s) are voltmodded and benching i never crash and all i have is a chopped stock cooler on there. it just seems like all this fuss is for nothing when you can install the stock cooler (maybe chopped, maybe just red plate for ram and all if you want your undeserved warranty) and everything will be fine ;)

I understand what you are saying and all is cool :up: