Nice guide tron2^![]()
Just remember they are catering for the size of a desktop CPU spreader and not a naked GPU die. From the guide the 3.5mm bead looks like the appropriate amount for our GPU's.
-
-
One more update from me. There is no IC Diamond 7 paste to buy in my country to buy and i can't wait anymore to come from ebay. I will apply Artic cooling MX-4 and if temps will be too high I will repaste core with IC7.
-
-
-
Any updates slavinjo?
As for me I went to A09 motherboard bios as I still got the pink hue and then white screen at bios screen bootup randomly someday's 3 days apart other times 1 day apart.
I also tried rjit bios and although it works unfortunately, no AMD driver works with this bios tried 3 of them too. Also something else, it has maximum brightness this bios and you cannot change it even with fn + F4/F5. Seems this bios and a few others cause the screen to be incorrectly detected or functioning. Since then, gone back to original bios the card came with and running amd 13.6 beta 2 drivers. Games run splendidly.
In order for the bios issue not to be an issue I have discovered closing the screen and putting the computer to sleep is the best option instead of shutting down and then awakening tomorrow to see the bug and then having to connect vga separate monitor which is a pain in the bum really. -
Hey guys, my thermal pads haven't come yet, but today they COULD arrive, if ebay arrival estimation time was correct. However I made preparations and bought Artic Silver 5 thermal paste. I talked to my friend and he gave me a one syringe of Shin Etsu paste. However, neither of us don't know which "model" of Shin Etsu paste it is. It is in a blue syringe with blue top and white plunger. Does anyone know which Shin Etsu paste "model" is this? I can't find any info on internet about the matter. And which one to use for paste job AS5 or this Shin Etsu?
-
Hello again. I finally installed the card and managed boot into windows once. But my laptop keeps restarting. I guess i did great paste job there
-
What do you mean keeps restarting?
Is it bsod? -
No, it's just instant shutdown. While laptop is "cold" i manage to boot into windows for about 1 minute and then it just shuts down. After restart i can't even boot into windows because shutdown happens while booting. It also shuts down in BIOS screen, so i guess it is thermal shutdown. I placed pads properly and used artic silver 5 on gpu core (1 grain of rice ammount).
-
Wow... that is weird.
And you repasted correctly? For it to shutdown in BIOS could mean something else.
What 7970m card are you using may I ask? -
Yep, very weird. There are no light indicators that something is wrong, no beeps, everything seems normal and then shutdown. Card is Dell card, with original bracket, and I didn't change screws from my old 4870, they fit perfectly.
-
Update from my last post. Overheating was happening because of wrong screws, original screws were not good, so i used another ones. So I managed to install drivers and then the show started - again overheating which caused shutdowns. So, after i turn my laptop on idle temp is 39C. OK, browsing with few tabs (with Flash) temps go up to 75C and sometimes causes shutdown. Furmark and Witcher 2 - shutdowns. HDMI works great, but i noticed when connecting HDMI temps go up to 70C on idle, is that normal? HWInfo is set to 4500 rpm on everything more than 60C. How to cool this damn thing?
-
Hmmm
Well that is positive news slavinjo.. :thumbsup:
I think you need to properly use icd7 to do the pea method to paste the GPU.
This way the paste will properly attach the heatsink to the GPU from it's thickness alone. From memory you were using as5? This could be too thin and not properly contacting both surfaces, which is my initial hypothesis..
Your idle temps are way too high. You should be aiming for 30-35C idle. To me it's your thermal paste. Go with ICD7 and never look back to the crappy arctic silver 5.
One other thing about the vbios I wonder if other 7970m users can share.
I went to this link:
http://www.techpowerup.com/vgabios/...=&model=HD+7970M&interface=&memType=&memSize=
and saw that the vbios in my clevo 7970m is actually this one (the orginal bios that came with the card). I also recall the eurocom is exactly the same as this one too:
Dell HD 7970M 2048 MB 2012-03-13 15:09:00 015.017.000.001.042717 PCI-E
Those running the rjtech vbios which version does it match with on that page?
Curious and also the first one linked is not what came with my clevo card.... weirdness to say the least and not game enough to try it too.
It is this one --->Clevo HD 7970M 2048 MB 2012-06-05 22:36:00 015.021.000.002.042768 anyone tried this? -
OK, i will order ICD today. I will also try to repaste with Shin Etsu, which is supposedly a great paste, but i don't know which "model" of paste i have. Which card do you have tron^2, dell or clevo? If you have Clevo, does you HDMI working (audio and video)? Did you also notice temperature jumps when HDMI is enabled? Is that normal behaviour?
-
-
Thanks mate, I will post here on my success / failure
-
And here I am again
Well, I have news. It seems i managed to make some progress with Shin Etsu paste. But i think I have fan issues. After boot idle temp is 34-35C. I tried 3DMark 11 and scored P5423 (is that OK, i don't have full version of 3DMark), max temp was 84C. But when I finish anything I do and leave laptop idle temps don't fall to 35C, instead they fall to 55. I think i did paste/pad job good because keyboard and plastic cover over heatsink are HOT. I think problem is fan. It's strong as medium fart. I could swear that with my old 4870m fans were working twice as strong. In hwinfo it says gpu fan speed 35% all the time, but i can see rpm's change as I wanted. This confuses me, 35%? Any what is max rpm, mine goes to 4100. Any ideas? -
I think you might be yet another victim of manufacturing dissimilarity between our beloved R2 Heatsinks.
Some may be better, some may be worse, leading to our ''perfect'' results vary.
It is sometimes good to show somebody how to do it, in here we can only explain, step by step, which also may cause some....miss-understanding.
Not for the first time i am doing this on NBR, and i would like to offer you to talk with you on skype, if you want, i can help you to reach better temps.
All assuming, your Heatsink stands up to the challange
BTW, throw shin Etsu out a window, Get ICD 7. -
And I just want to share what i experienced: shutdowns only happen in Witcher 2, in Skyrim temps with ultra settings never went over 80C. I also noticed temperature drops by 10C if HDMI is disconnected.
Anyway, I will PM you flingin and thank you very much for your offer! -
Its the paste. I am sorry but it aint up to it for this modification.
-
If it's the Shin Etsu X-23D paste you're using its an excellent paste. Its a pain to apply but if you leave it in some boiling hot water for about 10 mins before application its the same as any other application. I have no experience with ICD but from my experience with X-23D I would recommend it highly to anyone.
Also very important to remember is your application method. Top tier pastes are within +-3C difference and X-23D is up there with the best of them. A good paste job with a crappy paste can be as good as a mediocre job with a great paste.
What card did you have in there before the 7970M? From what sangemaru mentioned the nVIDIA cards had a slightly different heatsink than the ATi cards and the thermal pads he used had to be 1mm on the VRM inductors (square thingys) and 0.5mm on the vRAM to allow the heatsink to make proper contact with the GPU die.
Try looking at the card from the side after screwing the heatsink down. This should give you an idea of where the fault lies. I normally do this without paste and use my phones flash as a flashlight to see if the surfaces are making proper contact. I shine the light from behind the components to check if any light is coming between the heatsink and component surface.
It could also be that your heatpipe/s on your heatsink is damaged. I've read recently that an easy way to test this is to fill a bowl with very hot boiling water and dip the GPU contact end in the bowl and then immediately check the opposite end of the heatpipe (near the fins). There should be almost no delay for the heat transfer. If it takes a while to heat up its most likely that one or both of heatpipes are damaged and the gas/liquid has evaporated. You should then replace the heatsink. -
It is Shin Etsu, but I don't know which "model" - I suspect it's some crappy OEM paste. Although, I got MUCH better results with it than AS5 (4-5°C). I ordered IC7, it should arrive at the end of this week or the next one.
I had 4870m before but i got original Dell xbracket with 7970m, just needed to replace screws. I did check paste/pad job as you described and checked if the connection is good in front of the window, everything was fine.
About the heatsink - i tested it also, i left keyboard and plastic cover unscrewed, heated up the card, and touched the sink. It seems to work normal, heat is spreading quite fast, but if applying ICD7 does not work I will buy new heatsink. I use 1mm Phobya pads and on one VRM chip I had to put 0.1mm transistor thermal pad on Phobya pad because there was not good grip between pad and heatsink (like sandwich: phobya pad - transistor thermal pad - heatsink). -
It could well be the heatsink. ICD is a very good paste but its highly unlikely going to give you a 15C-20C drop in idle temps. A drop that significant is most likely going to be related to heatsink and/or heatsink installation.
I was under the impression that the Dell screws worked on the 7970M Dell x-bracket, irrespective if it came from the 4870M or 5870M. Why don't they work? You wouldn't even have to remove the c-clips.
Is your heatsink resting firmly on the metal square surrounding the GPU die on all sides? -
At first i had such huge idle temps because ther original screws had "limiters" so I could not screw them enough to make good contact with thermal pads applied. So i removed those, went to hardware store and bought same diameter, 5mm long to fit into the bracket holes. Now i have idle temps 32-40C (without HDMI connected) and around 50C with HDMI and up to 83 on load. So i hope to get 3-5C less temperatures with ICD which will reduce shutdowns which occur when i play Witcher 2 and the scene is fire or fog for 5 minutes(then card overheats and shutdowns).
-
Ah okay that makes much more sense now. Sorry, I don't think I read all of your posts. I thought you were idling at 55C.
Your temps seems more or less in line with the temp range of the 7970M. If you are running a Dell card the shutdowns could be related to the 80C-85C shutdown bug. What you can do to reduce this from happening is undervolt. You will most likely drop at least 5C and be under 80C on load. One of my cards is running default clocks at 0.975v and the other 1.00v without any problems or throttling.
The higher temps with HDMi connected could be related to the card clocking to its intermediate clocks which are around 450Mhz for the GPU core if I remember correctly. Try running GPUshark in the background and connect your HDMi while monitoring your clocks to confirm. -
Yep, it is a Dell card and it causes shutdowns over 85C. I can also confirm that connecting HDMI rises GPU temp by 10C, at least in my case (can anyone else confirm or test that?) - i used GPUShark and HWInfo to monitor that. About the voltage - VDCC readings in HWINfo is min 0.825V and max 1.05V.
My next move is repaste with IDC7 and hope it will give 3-4C which will hopefully stop those shutdowns and keep temps under 80C. If that fails i will undervolt a bit. -
While connected through HDMi check if your GPU core clock has increased from 300Mhz. That would explain the higher temps.
Sounds like good plan of action. Best of luck :thumbsup: -
Excellent work slavinjo. The screws make sense and now idle is very good.
My voltages interstingly are 0.825 v idle and 0.975 load.
I never gone past 70c.
This is with icd7. I am currently overseas for several weeks. I will continue to watch this thread. I found eurocom vbios the best for my clevo 7970m i dont get as many white screens of death on bootup with this bios but i still get them. Hope my screen lasts long. -
One of my cards also runs at 0.975v stock. I think its related to a poor quality VRM but I've seen users report this for both Dell and Clevo cards. It doesn't seem to affect performance at all so the cooler temps and lower power usage is a bonus especially since I'm running them in xfire with the 240W PSU.
@tron2^ Have you managed to try to get your card working with the Dell vBIOS? I don't think its wise running that Eurocom vBIOS if its the same version that flingin and rsgeiger used. It worked for them for a while as well before their screen was stuck permanently on that white screen. -
It seems temps are getting a bit lower with no repaste job. Maybe Shin Etsu needs some time to kick in
I will wait until ICD7 arrives and make a Witcher 2 test and if laptop holds without shutdown I won't even do repaste.
I had no whitescreen issues, not a single time. When I come home I will check VBIOS version and write it here if someone is interested (i remember it begins with 0.15...... or something like that).
@Nospheratu - wow how did you manage to use 2 cards with 240W PSU? Any modifications, removing CD drive...? I thought that can't be done on R2. Respect! -
Flingin bios would give me bsod could get into safe mode, the rjtech worked but no amd drivers work with it tried heaps and only original bios which looks exactly the same in gpuz as eueocom but performs slightly better ie less bleach screens. Interstingly no more red tinge on alien logo got that 1st time installed so hope my screen isnt a ticking time bomb. I never get bleach in windows both 2d and 3d run superbly.
I have tried them all. Bare in mind i have a v1.1G card clevo whereas most others i tjink have the v1.2G look for the sticker underneath card.
Its all very weird. I have had white bleach screen on bios alien bootup probably 7 times now each solved with the vga cable thingy. Without that i would be cactus like flingis with his 780m.
Edit: at frankfurt airport -
I haven't done anything out of the ordinary actually. I think its maybe cos my cards run at low voltage by default. I'm running my CPU at stock Dell settings which is 24/24/24/24 Multi's and 62/62 TDP/TDC. I've only had one shutdown in Crysis 3 but that was when I boosted TDP to 90W.
I also think I have a good PSU as I recorded up to 274W power draw in Crysis 3. I am planning on using the 330W though to give my CPU a little more overclock headroom. Check the power draw i recorded here.
-
all i would say is your RGB screen days are counted.....sadly...white bleach is nothing good.... -
I aim to prove that the bleach screen is an anomaly and life with it can go on without hardware failure.
Wish me luck
In the greek islands ps with these ambient air temps our alienware's would struggle..fortunately humans more tolerable of heat.
I may when i get back put the 680m and buy a 920xm in alienware and go for world record score and then place the amd 7970m in my clevo. Should be ok i think???thoughts ideas? -
...and yes, i wish you all the best with your screen bro, mine died in 3 months, together with mobo. -
-
Thnx guys 4 the feedback/suggestions.
I will aim to break 8000 points with 3dmark11 with max overclock on 920xm and 680m in alienware m17x. It is my destiny/goal. -
Hi guys. I repasted with ICD7 and it is better. In Crysis 3 temps never hit 80C. But in Witcher 2 temps keep going over 86 and of course i get shutdowns. So I will try undervolting but as I never did this I have few questions
1.)Is it possible to undervolt a card without flashing BIOS? Are there any tools/programs with which i can remove slider or enter new value to undervolt?
2.)Does undervolting affect performance?
3.)What should I be aware of? Can I damage card?
4.) Can someone try to play Witcher 2 to see is it only me or this game really pushes card so hard? -
1. Yes. See my post here for instructions. I have successfully undervolted using ASUS GPU Tweak on both my cards. Although it only works for me with 13.4 Cats and below.
2. Only if you go too extreme. If your card cannot maintain the clocks you set at the lower voltage you will experience throttling. If you set the correct undervolt for the stock clocks for example you wont notice any performance difference except the lower temps.
3. Try 0.975v at stock clocks as quite a few cards are able to manage that. If that doesn't work out and you experience throttling try increasing by 0,0010v at a time and test. If you are stable you could try going lower but I would advise against that. 0.975v is already great for temps. By running a lower voltage you can actually extend the life of your card as you are running cooler and the components are running below their max power spec. -
I will try it as soon as get home. I have Catalyst 13.6 beta 2. Is it really necessary to use 13.1 drivers?
-
No, you can use the latest WHQL driver which is 13.4 as mentioned in my link.
You can try 13.6 or 13.8 as well. Maybe it doesn't work in my config, it may work for you. Another alternative is using Afterburner with the -xcl config and then using TRIXX to adjust voltage. I don't use it as TRIXX only adjusts voltage for my primary card and changes my 2D voltage as well, this may not be the case for you. -
I tried undervolting. Strange thing is I put value of .975 in Trixx but HwInfo shows 900. Is that normal? Sometimes Afterburner just freezes my pc, is that normal too? And is Trixx config lost after restart? After restart HwInfo shows voltage like it was before using Trixx. Ah yes, one more question - is it possible to "undo" -xcl switch in Afterburner?
-
Is that HWinfo reading max voltage after loading the GPU with a game? I've noticed my cards sometimes idle at 0.900v. It's totally unpredictable though.
Yes, Afterburner does that a lot. Disabling ULPS may help. I think its got to do with AMD's buggy power state management.
TRIXX can be set to start on windows load and to restore the clocks you've set. Look at the options under the settings tab.
You can undo the -xcl changes by re-installing your driver. There might be another easier method but I don't know of it. My advice would be to use ASUS GPU Tweak after running Afterburners -xcl switch. -
No, HwInfo reads it right after apply voltage in Trixx. But when i start game HwInfo shows max voltage 1.050, like Trixx didn't "freeze" desired voltage of 975 (or 900 which Trixx sets it).
-
Yeah, that's similar to the issue with TRIXX that I mentioned. It changes my 2D volts. I was hoping your system would work with it.
ASUS GPU Tweak is the only utility that works for me for undervolting successfully and correctly without messing with 2D clocks/volts. -
I will try it. But I also bought air conditioning unit for room where laptop is
-
Hey all. I did a simple bottle cap mod and everything seems OK now
-
Aristotelhs2060 Notebook Virtuoso
Does anyone know if those here will work in crossfire on R2?
Alienware M17X M18X VGA Upgrade KIT AMD HD8970M 4GB DDR5 MXM 3 0B | eBayreborn2003 likes this. -
Yes they should workreborn2003 likes this. -
Aristotelhs2060 Notebook Virtuoso
reborn2003 likes this.
And another.. 7970m in M17x R2..Installation with photos
Discussion in 'Alienware 17 and M17x' started by flingin, Jul 16, 2012.