VERY hot Gigabyte Geforce 6800
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
VERY hot Gigabyte Geforce 6800
I installed a fanless Gigabyte 6800 yesterday. Today I was playing Half-Life2 for a while and my monitor went snap just like if you unplug the VGA cable while it's running. It was kinda scary.
I have an Asus P4C800E Deluxe that "talks" when it encounters an error and it took me a while to realize what the sound I was hearing was. It looks like the system shut itself down because it was above a temperature threshold. My Zalman CPU cooler just throws hot air right into the AGP card so that can't be good. http://www.zalman.co.kr/eng/product/vie ... ode=005009
I touched the heatsink on my 6800 and it was painfully hot. I can still feel it a bit on the tips of my fingers.
SO: After I opend up my case the fans slowed down to a reasonable rate and everything seems to be back to a normal temperature.
Right now while typing this with the side of the case open my Asus utility is saying CPU-36deg, MB-33deg and V-Tuner 2 says the GPU is at 58 degrees. I don't know why that giant heatsink doesn't even touch the 6800's ram chips?
QUESTION: I have no fans in this case other than the 2 in the Enermax power supply and the Zalman CPU cooler. I only had a Geforce 2 MX up until now so cooling wasn't an issue.
What would be the best way to cool down this video card?
I'm thinking of just cutting a giant hole in the back of my case right above the AGP slot (in line with the CPU) and putting a Nexus fan blowing out.
A 120mm fan would only fit if I attached it outside the case and a 92mm would fit with a bit of cutting.
OR would a large 120mm Nexus installed in the side of my case be better? Blowing in or out?
I have an Asus P4C800E Deluxe that "talks" when it encounters an error and it took me a while to realize what the sound I was hearing was. It looks like the system shut itself down because it was above a temperature threshold. My Zalman CPU cooler just throws hot air right into the AGP card so that can't be good. http://www.zalman.co.kr/eng/product/vie ... ode=005009
I touched the heatsink on my 6800 and it was painfully hot. I can still feel it a bit on the tips of my fingers.
SO: After I opend up my case the fans slowed down to a reasonable rate and everything seems to be back to a normal temperature.
Right now while typing this with the side of the case open my Asus utility is saying CPU-36deg, MB-33deg and V-Tuner 2 says the GPU is at 58 degrees. I don't know why that giant heatsink doesn't even touch the 6800's ram chips?
QUESTION: I have no fans in this case other than the 2 in the Enermax power supply and the Zalman CPU cooler. I only had a Geforce 2 MX up until now so cooling wasn't an issue.
What would be the best way to cool down this video card?
I'm thinking of just cutting a giant hole in the back of my case right above the AGP slot (in line with the CPU) and putting a Nexus fan blowing out.
A 120mm fan would only fit if I attached it outside the case and a 92mm would fit with a bit of cutting.
OR would a large 120mm Nexus installed in the side of my case be better? Blowing in or out?
-
- Posts: 255
- Joined: Thu Jun 05, 2003 9:45 am
- Location: CA
i have been exchanging email with gigabyte support regarding this card. i have been getting an odd error:
reg:0x26,Temperature_Measurement_MachineJP4=Reg_Val4=-1degree of c
but i can't corelate it with any oddities in the card's behavior and can't tie the message to the card (don't have the tools to check for error message strings on windows).
gigabyte support insists that the card's processor is safe as long as it's under 100'C (mine idles at 57-58'C and doesn't go above 66-68'C under 3D graphics load). i am using a DVI monitor and have seen no issues with the image except rare rendering problems with one piece of software, which i fix by slightly resizing the window it's running in (which suggests it's not heat-related).
another thing gigabyte support insists on is that this card requires a PSU of minimum 480W. my machine has a 300W nexus NX3000. may be they are right, may be this PSU is too feeble for this card. but i don't see any evidence of strain being put on the PSU (the PSU fan doesn't spin up, for instance).
btw, the card initially spiked to 72-74'C under load, but following the suggestions made in other threads, i had made minor changes that dropped the card's full load temperature by at least 6'C. here is a related discussion
reg:0x26,Temperature_Measurement_MachineJP4=Reg_Val4=-1degree of c
but i can't corelate it with any oddities in the card's behavior and can't tie the message to the card (don't have the tools to check for error message strings on windows).
gigabyte support insists that the card's processor is safe as long as it's under 100'C (mine idles at 57-58'C and doesn't go above 66-68'C under 3D graphics load). i am using a DVI monitor and have seen no issues with the image except rare rendering problems with one piece of software, which i fix by slightly resizing the window it's running in (which suggests it's not heat-related).
another thing gigabyte support insists on is that this card requires a PSU of minimum 480W. my machine has a 300W nexus NX3000. may be they are right, may be this PSU is too feeble for this card. but i don't see any evidence of strain being put on the PSU (the PSU fan doesn't spin up, for instance).
btw, the card initially spiked to 72-74'C under load, but following the suggestions made in other threads, i had made minor changes that dropped the card's full load temperature by at least 6'C. here is a related discussion
-
- Posts: 158
- Joined: Wed Jan 05, 2005 11:34 pm
I only have a 300W Enermax EG301P-VE and was wondering about that but I don't know if it's a problem.grandpa_boris wrote: another thing gigabyte support insists on is that this card requires a PSU of minimum 480W. my machine has a 300W nexus NX3000. may be they are right, may be this PSU is too feeble for this card. but i don't see any evidence of strain being put on the PSU (the PSU fan doesn't spin up, for instance).
btw, the card initially spiked to 72-74'C under load, but following the suggestions made in other threads, i had made minor changes that dropped the card's full load temperature by at least 6'C. here is a related discussion
I'm DEFINITELY gonna need to get some airflow going in this case.
Instead of using simple slow turning fans I think I would rather have 2 controlable case fans blowing air in and a large case fan in the back blowing out that I can control.
I don't care if it makes noise while I'm playing a game but I'd like to make it quiet any other time by turning them way down like I can with my Zalman CPU cooler.
What kinda controler could I use to control these fans?
I think putting a 120mm in the side of my case blowing towards the AGP card and a smaller rear blowing case fan in the back would be perfect because it would blow directly onto the video card and exhaust out the back. This way I could run them so slow that they would barely be audible when I don't need them.
Even being able to turn OFF the fans when I don't need them would be good.
While I'm playing a game I would just crank them up as needed.
My power supply already ramps WAY up while I'm playing games anyway so what's a few more fans.
I guess this post would be better suited on the case cooling forum now 'cause it's just as much about case cooling and fan controling than video card cooling (even though the video card is the culprit).
I saw a review of 6600GT cards (I know you have a 6800 passive) on another site (don't recall where) that criticized the Gigabyte for heat problems (was not able to overclock because of it). The reviewer determined that the problem was with the mounting of the HSF which was done with some kind of adhesive, and the HSF was a little loose causing overheating problems. Apparently, once a HSF with adhesive is bumped and becomes loose, and will not reset on the GPU properly like one with normal paste. This is just speculation as to what might be causing your problem, but it influenced me to stay away from the Gigabyte cards (active or passive) and put a VF700 on a Leadtek 6600GT.
Last edited by m0002a on Sun May 15, 2005 12:16 pm, edited 1 time in total.
IMO if you are going to maintain the configuration you'll soon fry something, probably the graphics card first. If you could afford the card, why don't you get a nice case with two 120 fans. I would suggest 1,200 rpm ones, and use a fan controller to lower them to 900-1,000 under normal use and full 1,200 while gaming.
Yup, I've been thinking of getting a new case that's ready for 120mm fans.Tzupy wrote:IMO if you are going to maintain the configuration you'll soon fry something, probably the graphics card first. If you could afford the card, why don't you get a nice case with two 120 fans. I would suggest 1,200 rpm ones, and use a fan controller to lower them to 900-1,000 under normal use and full 1,200 while gaming.
I wonder how hot this card is supposed to be under normal use?
My MSI 6600GT dissipates about 5W more than a vanilla 6800 (according to xbitlabs, if you want to believe them). I have an Aerocool VM-101 on it and it runs at the same temps as with the stock MSI cooling solution. Right now 40C and under load upto 64C (recorded by Rivatuner, when returning to desktop from FarCry it only shows 54C). That's in Antec SLK 3000B case with two 120 mm fans (at rpms as in my previous post).
You could use a Zalman VF-700 on your 6800 and get away with it for the card, but since you have so limited airflow you may end up overheating the CPU and PSU.
You could use a Zalman VF-700 on your 6800 and get away with it for the card, but since you have so limited airflow you may end up overheating the CPU and PSU.
-
- Posts: 255
- Joined: Thu Jun 05, 2003 9:45 am
- Location: CA
your load temperatures are close to what i am getting with my passive 6800 (according to gigabyte's vtuner2). i do have two 120mm fans running at 5V moving air through the case, though.Tzupy wrote:My MSI 6600GT dissipates about 5W more than a vanilla 6800 (according to xbitlabs, if you want to believe them). I have an Aerocool VM-101 on it and it runs at the same temps as with the stock MSI cooling solution. Right now 40C and under load upto 64C (recorded by Rivatuner, when returning to desktop from FarCry it only shows 54C). That's in Antec SLK 3000B case with two 120 mm fans (at rpms as in my previous post).
You could use a Zalman VF-700 on your 6800 and get away with it for the card, but since you have so limited airflow you may end up overheating the CPU and PSU.
for what it's worth, i tried openning up the case and placing another 5V 120mm fan right over the card, simulating a blowhole with the same fan right over the card and some kind of ducting or partitioning to keep the heat from flowing to the CPU. the card's temperature under load was in the mid-40s, which is a lot more palatable.
another observation: while openning the PCI slot covers below the card dropped the temperature under load by 5-6'C, the air wafting out of those opennings is barely warmer than ambient and much cooler than what the PSU fan is pushing out. this suggests that the open slots under the card eliminate a cul-de-sac zone in the air flow, which was probably causing the hot air to accumulate and circulate over the card's lower heatsink. would this have worsened the temperature gradient in the heatpipe? i have no instruments to check this.
i am considering adding a blowhole with a slow 120mm fan to cool the card (should be still quieter than any actively cooled cards), but the ammount of effort may be sufficent to justify simply getting a new case (i am not in love with the sonata and 2-3 hours of my time modding the case would cost a lot more than buying a new case). but then may be i should be getting a new motherboard and a newer, faster, cooler CPU? and a new CPU heatsink? and a new PSU? and may be replace the AGP card with a PCI-X?... that's what always stops me from making any changes until my systems are 3-4 generations behind the state of the art.
Would you happen to know what IS a normal temperature for this passive 6800 while it's not under load and under load?grandpa_boris wrote: your load temperatures are close to what i am getting with my passive 6800 (according to gigabyte's vtuner2). i do have two 120mm fans running at 5V moving air through the case, though.
I did some testing tonight with the side off and a big 9" Home Hardware fan blowing at it and I got the GPU temp down to 41. It hovers around 57-60 without the fan and after coming out of Half-Life 2 (without the giant fan but the case side off and my ZALMAN CPU cooler turned all the way up to 2500 RPM) while running VTuner2 it showed 65. Taking the side off my case seems to have helped allot.
I think my computer reset itself last night because the videocard heatpipe was pumping too much heat into the nearby CPU.
I still need some big case fans I can turn up and down as needed though. I'm not gonna keep my side open all the time. Maybe some ducting to keep the heat away from the CPU would help too.
-
- Posts: 255
- Joined: Thu Jun 05, 2003 9:45 am
- Location: CA
i have the same question, actually. i asked gigabyte's support if 100'C upper safe limit on vtuner2 was real. i am not sure they understood the question. their email reply saiddentaku wrote: Would you happen to know what IS a normal temperature for this passive 6800 while it's not under load and under load?
hm. VIA's EPIA is supposed to withstand temperatures above 90'C, but i doubt many other active components around the CPU or GPU will be as happy with the heat.It should reach around that temperature when it's gaming or intensive 3d applications are being process.
if the card is idling, it should not be generating as much heat. when my system is idling (folding, actually, but the power saving mode kicks in after 5 minutes or so and turns off the video), the GPU seems to hover around 57'C (134'F). the ambient temperature is somewhere around 74'F. it's puzzling that nvidia, who now have experience with mobile GPUs that shut off parts of their circuitry that isn't in-use to save power, hadn't built their latest generation of chips with power conservation in mind and didn't provide for video circuitry going into low power mode when idling.dentaku wrote: I think my computer reset itself last night because the videocard heatpipe was pumping too much heat into the nearby CPU.
Good to know you're getitng 57'C too. It seems to hover around 57-60 normally. I'm assuming it goes up to 70 or so while playing HL2.grandpa_boris wrote: if the card is idling, it should not be generating as much heat. when my system is idling (folding, actually, but the power saving mode kicks in after 5 minutes or so and turns off the video), the GPU seems to hover around 57'C (134'F).
Like I said before, I doubt my 3.2Ghz P4 like having that kind of heat next to it and neither does my motherboard chipset (or whatever it's called with the bifg black ASUS heatsink on it).
-
- Posts: 255
- Joined: Thu Jun 05, 2003 9:45 am
- Location: CA
on my system, the CPU temperature is 41'C at this moment EDIT: without FaH, CPU is at 37'C/100'F. realistically, because of FaH, my system never idles -- FaH kicks in to eat up 100% of CPU and drives its temperature up to mid-50's. this is probably 4'C higher than it was before i added this 6800 card. the GPU's temperature according to vtuner is 57'C. both temperatures were 10'C higher while running heavy graphical load. at 51'C, the CPU is well within the safety limit for this chip (P4/2.8, pre-HT, and it's 70'C, i think?)dentaku wrote:I doubt my 3.2Ghz P4 like having that kind of heat next to it and neither does my motherboard chipset (or whatever it's called with the bifg black ASUS heatsink on it).
(i think i could get the temperatures down another 2-3'C or more by cutting a wider openning into the bottom of sonata's front bezel, a mod highly recommended by other sonata owners)
check the CPU temperatures you are getting with and without load (MBM used to be the tool of choice around here, but software links page now recommends something that doesn't agree with my browser's security settings and displays a blank page, so i have no opinion on it). your machine's CPU may not be ice-cold, but it may be well within tolerance.
Last edited by grandpa_boris on Tue May 17, 2005 11:49 pm, edited 2 times in total.
-
- Posts: 255
- Joined: Thu Jun 05, 2003 9:45 am
- Location: CA
wow! that just isn't right. are you sure your card's heatsink didn't break the thermal coupling to the chip? earlier in this thread, m0002a mentioned a reference to a review that panned a related card for having an inadequate HS/GPU adhesive. i can't see how your card can run that hight unless it also had its GPU/HS thermal bond broken. when i had my sysetm's case open and a 120mm fan blowing at the card, the idle temps were well below 50 (i think it went below 40, but i am having a senior moment and openning up the case again is most inconvenient).rei wrote:gah! i just hit 105c, even with the 120mm fan pointed at it, i had to tilt the fan back so it's blowing at underside of card
now maxing out at 80s like it should be :/
so in my view, something is very, very wrong with either your card or the way its temperature is being measured.
Definitely, that looks quite wrong.grandpa_boris wrote:so in my view, something is very, very wrong with either your card or the way its temperature is being measured.
Today I took the fan from my old Zalman PIII cooler (http://www.zalman.co.kr/eng/product/vie ... =005009010NP FAN & Bracket : FB123)and pointed it directly at the 6800 heatpipe with the side off and I was playing HL2 Garry's Mod for quite a while and it was only at 50'C.
This simple little fan has made a HUGE difference.
SO: I'm assuming just putting a 120mm NEXUS into the side of my case and pointing it at the AGP card will help my temperatures. Of course maybe it won't work because I won't have the side open anymore. Maybe a rear exaust will also be needed if I want to keep my side closed?
I'm idling at 43'C right now.
It'll have to do while I figure out what kinda fan controller to get and what kind of fans to buy.
i have it blowing out back (with pci slot underneath it open) at full blast, around 1100-1200rpm on the nexus
maxes out at 90c load still.
i need to take it out to take down serial # etc, but do you think i should still get a rma from gigabyte for possible thermal bond problems? :/
got that back exhaust 92mm nexus running too.
correction: it was running 100+ like that with nexus at low, i turned it high and tilted it down so the bottom is nearer to back of case and it's now blowing upwards to bottom of card
i woulda thought covering both bottom and top of card was enough
i had a THIRD fan mounted directly over the card with fb123 and 80mm nexus, but the fan noise was getting ridiculously
i got it to 45-50 idle then but still 85 load
do you think it's running too hot?
i woulda thought covering both bottom and top of card was enough
i had a THIRD fan mounted directly over the card with fb123 and 80mm nexus, but the fan noise was getting ridiculously
i got it to 45-50 idle then but still 85 load
do you think it's running too hot?
-
- Posts: 255
- Joined: Thu Jun 05, 2003 9:45 am
- Location: CA
YES. gigabyte support seems to think that 100'C at full load is fine, but in my view that's ridiculous. a 40'C swing under load seems excessive. my card has a 10'C swing and that's with a fan placed significantly farther than yours (4-5" away, i estimate) and running at 5V (wich i think is 1200-1300 RPM). what you have in your photo should give you better temperatures than my configuration gives. i'd say RMA it.rei wrote: i got it to 45-50 idle then but still 85 load
do you think it's running too hot?
I also have a Gigabyte 6800 NU that is passively cooled. It idles around 60C and hits 80C while gaming with ambient around 80F. Its in an Antec 3000B Case with a Tri-Cool fan on low in the back and a Zalman 7000 AlCu on the processor. I also have a third fan, a 12mm Papst at 5V, in the front for hard drive cooling.
I think the heatsink may be separated from the chip or the temperature readings may be off. Either way, I would RMA it. I think there was a review of high end video cards awhile back that mentioned many of the cards tested had large heatsinks that could easily come loose from even a small bump or drop. The gigabyte was supposedly one of the better ones since the two heatsink halves are screwed together, clamping down on the board (one of the reasons i choose it). Nevertheless, your card is probably loose or the thermal material isn't working as it should. Unless you're willing to have lots of fans running at full power to cool it, I would RMA the card. Tell them it's not cooling properly and overheats.
I think the heatsink may be separated from the chip or the temperature readings may be off. Either way, I would RMA it. I think there was a review of high end video cards awhile back that mentioned many of the cards tested had large heatsinks that could easily come loose from even a small bump or drop. The gigabyte was supposedly one of the better ones since the two heatsink halves are screwed together, clamping down on the board (one of the reasons i choose it). Nevertheless, your card is probably loose or the thermal material isn't working as it should. Unless you're willing to have lots of fans running at full power to cool it, I would RMA the card. Tell them it's not cooling properly and overheats.
taped up the sides with construction paper and i think i bettered temps by 5-10 degrees if the two fans are maxed. no air leaking out the side now.
so game temps go up to 80-85-90 max now and the case has a non-disturbing hum from the two fans when gaming. if i turn both fans down to min, it'll idle in windows at 55 and the system is essentially silent.
would you consider these temps acceptable/safe?
so game temps go up to 80-85-90 max now and the case has a non-disturbing hum from the two fans when gaming. if i turn both fans down to min, it'll idle in windows at 55 and the system is essentially silent.
would you consider these temps acceptable/safe?
not sure if this is off topic for this but if you want to monitor your load temps get rthdribl
http://www.daionet.gr.jp/~masa/rthdribl/
it run a windowed real time renderer so you can watch your software temp gauge while under load.
if your games don't corrupt or crash while playing then i wouldn't worry about the temps. usually a vid card will show anamolities well before the temps needed to become a paperweight.
http://www.daionet.gr.jp/~masa/rthdribl/
it run a windowed real time renderer so you can watch your software temp gauge while under load.
if your games don't corrupt or crash while playing then i wouldn't worry about the temps. usually a vid card will show anamolities well before the temps needed to become a paperweight.
Slightly off topic, but I have a gigabyte 6800gt which had similar temp trouble. I took off the stock heatsink and found it was making very poor contact with the chips. Changed to Artic Cooling unit and was amazed by the temp drop... 35 idle and max under heavy load...55 celcius, and I can't hear it with case closed.
Ps... I used artic silver on gpu and ram chips
Ps... I used artic silver on gpu and ram chips
sorry forgot to say those temps are with 6800gt overclocked to beyond 6800ultra speeds.....man I love that cooler, here is the link
http://www.arctic-cooling.com/vga2.php?idx=40
http://www.arctic-cooling.com/vga2.php?idx=40
-
- Posts: 255
- Joined: Thu Jun 05, 2003 9:45 am
- Location: CA
i had checked the idle temperature on my system first thing this morning, right after it turned on the display and before the video chip had a chance to heat up, and it was @ 56'C. so your idle temperatures match what i have. (my system *may* run hotter when "idle" than when in use because of F@H heating up the CPU far more than what my normal use of the system would do, so that may warm up the video card, although i doubt it.rei wrote:so game temps go up to 80-85-90 max now and the case has a non-disturbing hum from the two fans when gaming. if i turn both fans down to min, it'll idle in windows at 55 and the system is essentially silent.
would you consider these temps acceptable/safe?
but i am not getting temperatures above 68'C while running graphical load, though. which games are you using? it is possible that i am simply not driving my system as hard.
i will definitely try using the renderer suggested by Katharsis to see how the system reacts.