P170HM / X7200 and the 6970m

Hello,

as some may know, I am not happy with the 6970m crossfire solution... many problems and BSOD´s....
So I ordered the GTX 485m, still waiting for the heatsinks, though.

Since I change the cards, I thought about what I was going to do with the 2 6970m cards.
I talked to a friend of mine about this and since he wanted a new laptop anyway, the idea was born: Get a P170HM Barebone for him and take one of the 6970m cards, buy it "cheap" from me and build it in the P170HM.
To make a long story short: He ordered one, got cheap RAM and the 2630QM from ebay and yesterday it finally arrived...
I took the slave card out of my X7200 and we built it in the P170HM. Was a lot of fun, if you put a barebone together and got it working...

Now to the thing that amazes me:
In the P170HM we can overclock the memory of the 6970m up to 1080 MHz and get good results in 3DMark Vantage while in X7200 I couldn´t go past 930 MHz or it crashed... Now I tried it with the single left 6970m in the X7200 again and I get the same crash...
And: Installed Red Alert 3 on the P170HM... guess what: On my X7200, I got BSOD´s when I enabled high shaders... on the P170HM, nothing happens! It just runs fine...?

I am not finished with testing, but these results where so strange that I had to post it here...
Only 2 things I can imagine:
1. Only one of my cards has problems with overclocking and some other things. In crossfire, it was clear that the problems will show up since both cards are used. BUT: I once changed master/ slave cards and did Red Alert 3 with just 1 card enabled on both configs and had the BSOD on both configs!
2. The X7200 can´t handle the 6970m well. There have to be some issues with it... ( I use the latest Bios, the correct vBios...)

And now to the last problem, where I need some help...

The 6970m in both machines idles at 250/ 900.
It goes down to 150/ 100 if I disable Powerplay (btw using 11.6 CCC). but then it won´t play any 3D application with the normal settings and stay downclocked...

I read that maybe the display settings could be the problem, but I enabled only "1 Desktop" and it is the right one...
That is on the X7200 and P170HM...

BTW, could someone upload the vBios of the 6970m for the P170HM? Now I am using the one from the X7200, but I think the vBios from the P170HM is different as I remember some GPU-Z screenshots back then...

So any ideas about this strange thing? Just curious if I´ll get problems with the GTX 485m SLI, too. Then maybe my motherboard is not working correctly...

 

Use this thread to learn how to control your clocks.

I've attached the BIOS.

 

Thanks for the vBios! +1

All right, I´ll try ATT to see if this works... nevertheless it should downclock at idle to 150/100 without any tools, or is this a common problem with the 6970m?

 

Even without ATT, your GPUs should normally idle at 100/150, and will go up to 300/900 if you open a video.

250/900 is very odd, and you should never have to mess around with Powerplay.

 

My machine runs at 250/900 if it is plugged in to an external monitor. Are you running dual screens?


I do, however, downclock manually using ATT (back to 100/150).

 

That´s what I read somehere - but I don´t use dual screens... I selected "Show Desktop only on 1" and "1" is my Laptop screen, there is no dual screen...

 

That is odd. Let's see how the 485Ms perform. Another issue could be the Crossfire cable too.

BTW, the only difference between the X7200 VBIOS and the P170HM VBIOS would be that Crossfire will be disabled in the P170HM VBIOS.

Also, I may have a copy of Red Alert 3 and C & C 3 laying around somewhere in my house. I'd like to test to see if I have the same problems as you had with the blue screens and such.

 

As a person with your previous gen laptop, I will say this. Stuff works weirdly. Very weirdly. Most things are fine, but a few programs seem to be unable to grasp the concept of a desktop CPU/chipset coupled with mobile RAM and GPUs. That's the only explanation I have. Also, this may be a moot suggestion, but have you tried clean-wiping and reinstalling Windows? It can help. I don't know why, but every single time I've installed windows on my laptop, it's worked differently. Same disc, same OS. This is really all I can add; me and my poor 280M.

 

Yes, it´s very strange. Thanks for your responses!

I tried the X7200 vBios, the vBios from Kevin_Jack and a Dell vBios that was uploaded by someone over in the Alienware forum.
The Dell vBios gave the best results in Benches, but the fans seemed to run a little more then with the Clevo vBios...
The problem with C&C3 was solved by 11.6 and seemed to be a driver related problem that anyone should have experienced, the Red Alert 3 problem still remains only in my X7200...

Once I had to install the OS new, because my first SSD was broken and I had to replace it. The problem stayed the same, in that case...
In the first place, I used a I7-960 and right now a I7-970, the problems were the same with both CPU´s.

Maybe you are right, D2 Ultima, and it´s the combination of Desktop parts/ mobile parts... that would explain why Alienware users doesn´t seem to have any problems like that...

 

I wish there were more people that seemed to have a D900F in this forum, but if there are, they're not active, or it's not in their sig. I can't even get a response when I ask how new drivers work with the 280M in those new-driver-released threads. Hell, the entire 26x.xx driver series gave me artifacts and freezes on Vista x64 SP2.

What drivers are you using for everything else in your machine EXCEPTING your video card? If possible I'd tell you to try stock clevo drivers for everything. Doesn't really make a performance hit, your CPU is so overly powerful that it'll destroy anything even with a generic PnP driver generated by Windows. But really, your GPU does strike me as odd. Did you try setting up the power options? How you want it to idle out/etc? There's no point doing anything other than maximum power savings for anything; the clocks will throw up to maximum past a certain demanding point, anything else just wastes power/battery. See if that can help you with your 250/900 idle-out clocks as well.

That's all I can offer; I only switched to Windows 7 last Tuesday night/Wednesday morning, I was on Vista the whole time before; and I can already tell after using Vista for two years, EVERYTHING works differently. It's the same OS shell, different GUI, but things just... work differently.

Oh right, I just remembered. I want you to install Playclaw, and enable the CPU and GPU overlays, with all their options. They'll tell you per-thead load and per-core temperature real-time in-game for your CPU, and per-card load, frame-buffer load and temps in-game. My normal is ~100% load ~50% frame buffer load. If your ATI card is anywhere less than that, then something's definitely wrong in either how your motherboard is making use of your card (as you've tried the card in another machine, we can completely rule out the card itself) and or game software truly cannot marry desktop and mobile technologies.

Just my <s>little two cents</s> paragraph to try.

 

Thanks for all your help!
The problems were there with stock drivers, I then updated some drivers in the second place...
Maybe it is the RAM, who knows...? I really don´t believe that, because the problems are always related to the ATI driver... I will only try this out if my GTX 485m SLI won´t work properly in that notebook, too...

I think I know what you mean: Even X7200 users are very rare... maybe 4-5, and no one with 6970m seems to be active...
That´s a little bit weird and make me think that either they are all happy with their machines and don´t post because of that or they have problems like me and "hiding" from the public, because when you buy such an expensive machine you want to go out and tell everyone "look, this great machine can this and this, and it is mine!" . But now they can´t .

I played around with all kinds of power settings, the windows one and the AMD one, but the only think that helps with the 250/900 is Ati Tray tools or you turn off powerplay, but in the last case the clocks won´t go high on 3D apps anymore...

That´s an interesting thing with that "Playclaw", never heard before - but I will try out and post what I´ll find out .
Thanks! +1

 

Well good luck; feel free to tell me anything that might be going on. I've had this thing for approximately two years now, so I'm a little accustomed to its quirks. As for your stock drivers, I don't know, I often leave everything stock excepting the GPU and maybe the wireless card/audio drivers. I can't push my i7-950 to max load with my day-to-day usage no matter how I try; so if a driver can squeeze more performance out of it I wouldn't know. I don't need it hehe.

There were a lot of people who used the D900F though, I'm sure they're still on the forum, but probably upgraded to a new machine and sold the old one on e-bay or something.

As for playclaw, you can go here and download it yourself. The trial version will give you almost all the features of the full version; excepting your screenshots/videos will have a watermark and the teamspeak plugin will only show the name of 1 person speaking at a time. Just set up the overlay options and see.

Also, it'd be a good thing to try it out on the P170HM and see what results it shows for the 6970M in similar games; if your machine is struggling and your load/frame buffer load is the same as the P170HM's but the 170 is performing better, something's very seriously wrong; as well as if your machine cannot max out load/frame buffer but the P170HM can. I would actually suggest RMA-ing the machine at that point, because there'd have to be a fault in your motherboard or something for that to happen.

 

Hello,

I am French student and I hope that you could understand my english.

Here is my laptop review (in French) : P4G / [REVIEW] Clevo P170HM avec i7 2720QM ES, HD 6970M et OCZ Vertex 2

My HD 6970M works originally with this frequences :

With Powerplay and Power supply:
2D mod : 250 900 0,95V
3D mod : 680 900 1,1V

With Powerplay and batterie:
2D mod : 250 900 0,95V
3D mod : 400 900 1,0V

Without Powerplay :
Always 100 150 0,9V

VBios with RBE :



Uploaded with ImageShack.us

Strange..

I think I have the same graphic card revision. I can't overpass 890 1010 1,1V with Sapphire trixx otherwise BSOD. Have you tested to overclock with several softwares? Because I haven't got the same limits with Sapphire Trixx, ATI tray tools and ATI graphics overdrive.

I obtain a very stable OC at 850 1000 1,1V.

 

Yes, you seem to have the same strange idle clocks as me...
I changed almost every value in the vBios editor and that means nothing as long as powerplay is enabled... always 250/900 in 2D. So powerplay has nothing to do with the settings in the vBios, I guess.
Only methods that works is Ati Tray Tools (There: Hardware. Overclocking). I run everything but 3D on 100/ 150 and even can watch videos without any problems. I made 2 profiles (100/150 - 680/ 900) and set them to automatic switching between 3D and 2D. Works without problems so far. The good thing is, now the card idles between 40-45°, before that with 250/ 900 it idled at about 60°!

I overclocked the cards with Overdrive, or flashing the vBios or MSI Afterburner...
Nearly always getting the same results...

 

I have the new Sager NP7282 but with the Nividia 485m GTX SLI. I was looking at the ATI cards but I was weary with the compatibility with Autocad since my daily function for this machine was that. The only gripe I have with mine is the Intel 990x Idles around 50C on 3 cores and 45 on the other 3. When the temps get up over 50 the fan kicks in and it keeps alternating which is really annoying.

I downloaded the modded bios but don't know if it is current with the stock one I already have on this machine. I still say with one autocad drawing open it shouldn't Idle this high, my room temperature is around 20C.

 

Undervolt it using XTU.

http://forum.notebookreview.com/sag...-np7280-overclocking-undervolting-thread.html

 

I have it undervolted and turbo boost turned off. Voltage is set to 0.9 and the clock is at 1.6 Ghz and still it idles at 50C all day long. For gaming I turned up the clocks to 3.46 Ghz and the temps dont go higher than 65C.

 

You won´t believe this!

The 6970m in the P170HM is DEAD!
Just today I wanted to give it to my friend, did some last testing and did play something. Everything was all right.
Then next game was loading, screen stayed black... startet the latop new, and at the time the desktop would show up, it stayed black...
After uninstalling ATT from Safe Boot, I was able to go into Windows again - for about 2 minutes. Then the screen startet flickering and got black...
I did a new Windows installation, all went fine, but after a new start the Laptop didn´t post! No bootscreen, nothing...
I inserted one of the GTX 485m that is waiting to work in my X7200 (still waiting on heatsinks, 4 weeks now...), and the P170HM posted! So, the 6970m is dead obviously...

But the best is yet to come:
Turned on my X7200, it bootet, the Desktop showed up and a few seconds later, the screen turned black!
That was the moment I became hysterical
Second attempt, SAME result...
I then did boot in safe mode and uninstalled Ati Tray Tools - then did a reboot and everything is fine right now!
Did some testing and it works...

So, maybe ATT with it´s profiles destroyed one 6970m and was close to destroy the second one???
Or was it just accidently...? Maybe the card was faulty as described in the other thread... from a bad batch...???

After this, I will not use ATT anymore! Of course I don´t know if it had something to do with it, but I won´t risk that anymore...
The temps were all right, never reached 80°!

And will definitely stay with nVidia the next time - no more AMD.
Never had problems with their cards (except my 9800m GTX, but I overclocked in very hard and it was stressed and failed...)

 

^ This is why I like nVidia XD

Hehe, but in all seriousness, I'm not surprised. And many many people used the 6970M and had no problems in other machines; if your cards were from a bad batch I'd believe it; as you probably bought them together. Maybe that's why it acted so weirdly. I hope you get your GTX 485M heatsinks soon; I know how annoying it is to have something you spent money for and be unable to use it...

I suggest RMA-ing them if possible, if you get two new 6970Ms, you could simply sell them off as new, or ask for a money-back RMA if it's within 30 days (somehow I don't think yours is though). If you get two new ones and you don't really care, just sell them off, or put one in your friend's P170HM and then sell the backup. It'd be brand new cards, not "used", so you should be able to get full price for them.

Now... who wants to give me a P170HM? You can has my D900F in return hehe...

 

ha ha thats hilarious, but tragic. Ive had a similar experience with my dell d820, where the graphics card burnt out, but it was embedded in the motherboard, nvidia geforce 7400 go(piece of crap). luckily i bought it through the dell epp which came with a 3 year warranty, but they gave me a refurb motherboard. by the end i had gone through 5 motherboards, one due to a technician forgetting to replace the thermal paste. what a pain. anyways it works fine now, but just ordered the laptop in my sig, with the 485m, so it should be reliable and about 3283% percent increase in performance(calculated from notebook check).

 

Sorry to hear about your card loss Tirenz

But how come 6970M used in Alienware dont have problem like 6970M produced by Clevo .... i dunt think it was AMD's fault i think this was Clevo's fault...

CMIIW

 

because the m18x probably doesn't use the dedicated cards as muchdue to optimus, and because they payed the premium of an alienware maybe they get checked better, same thing goes for imacs

 

First I got ES cards and then - I think in April - I got "regular" cards, and I remember that they both had different vBios and one card even was locked, I had to unlock it before flashing it. So I think the cards came from different batches, but both are from Clevo.
Maybe the card in the X7200 will last longer, but who can say... since it really happened out of the blue, it can hit anyone...anytime...

 

There's seriously nothing ATT can do, to damage hardware.

Did you remember to select the higher clocked profile, before starting the game? Were you using an external monitor?

NO ONE is hitting 890MHz on the core. That is a huge attempt, and I recommend you don't try it anymore.

 

I set it to automatic clocks - games in 680/ 900 and 2D in 100/ 150. Worked fine for a couple of days...
I don´t know what the reason was, but I know for sure that ATT did run and then the card was defective (in P170HM), and I installed ATT at X7200, too and after a few days of using it, it startet to load only a black screen, about the same time the other card in P170HM died. I immediately uninstalled ATT and after that the black screen was gone...
I am not saying that it was the reason, I can not imagine how this could be done, too (maybe the Voltage...???), but since I had TWO strange experiences at nearly the same time and both times ATT was new, I don´t want to take a risk...
Of course I do believe more that the card just failed, too, but as I said, don´t want to risk it and there definitely was a problem with ATT, or else there would have not been the black screen, at least...

 

what were your overclocking settings on ATT? Are you sure they weren't set to something outrageous that would lock your system up? a voltage setting of 80, for example, is not .8V its .08V.

ATT is a well-coded piece of software. The caveat is that it gives you ample freedom to create overclocking settings that aren't stable.

I agree with Kevin_Jack ... I don't think ATT can destroy your hardware, at least I've never heard of it. Sure it can lock your system up if it an unstable setting, but completely fry it? i dunno ...

 

He never tried messing with it like that; he just had issues. He even put the card in a P170HM and had it outperform what he got in his x7200. He only used ATT to check and make sure I believe; he had weird idling clocks. Besides, the card in the P170HM died first; and that came from his x7200.

 

Man, this is strange. Not gonna lie, but I'm kinda spooked now that something may happen to my cards since they're Clevo cards and not the Dell ones.

BTW, have you tried undervolting the cards? Because I hear alot about undervolting these cards in this forum, and now I'm starting to see bugged/dead systems with these cards right after...

 

I undervolt my card, which leads to amazing results in temps.

By default it is set to 0.9V on low clock speeds (100/150) but I undervolt to 0.8. By doing so I idle at 40 degrees C.

By default it is set to 1.1V on High clock speeds (680/900) but I undervolt to 0.95. By doing so I reach max temps of 77 degrees C on extreme load (games on ultra settings). Before undervolting I was reaching 87 degrees C. a 10 degree difference !

 

Could you post a short guide of undervolting?

 

this thread has lots of info on undervolting: http://forum.notebookreview.com/sager-clevo/576540-6970m-undervolting.html


Basically, download and install Ati Tray Tools.


From there, you can create overclocking profiles in which you can adjust the clock speeds as well as the voltage.

I have three profiles saved:

Low -- light tasks, browsing, work, max 720p videos -- 100/150/0.8V

Medium -- 1080p videos -- 250/900/0.8V

High -- Gaming -- 680/900/0.95V

You have to manually change your profiles, but I don't really mind it. Gives me more control of my card. You can set up hot keys with ATT to change your clock settings, so I use ctrl+shift+l, ctrl+shift+m, ctrl+shift+h.

 

Yes, it is strange, and I wouldn´t have thought of ATT that would causing the damage if the 6970m of my X7200 wouldn´t start getting problems shortly after using ATT and the dead of the 6970m in the P170HM shortly after using ATT...

I undervolted the card and it ran fine... for a couple of days... thought it would solve my idling temp problems with not going below 250/ 900. That was the main reason, but after I read about the undervolting, I tried that, too.

Right now, for testing, I put the 6970m card that was left in X7200 in the P170HM and works fine so far... (as the first card did...) my friend for whom I bought the P170HM is very sad that this caused so many troubles for me, but I never thought that my 6970m would die out of the blue...
Really funny but strange thing btw: The second card in the P170HM reaches exactly 12k on the GPU score in 3DMark Vantage, and only 11550 points in the X7200.