M17x R2 7970m CrossfireX Ultimate Installation and Tweaking Guide

Doing the test inside, so guess 20-22 DegC apx?

Did some fun tests last winter when my computer was sitting outside in -10

 

Hey people. I need a little help. My 5870m died on my R2 and R2 doesn't have integrated video card. I bought 7970m upgrade monkey but my luck lol it has 015.022 vbios. Im sending it back. Can anybody give me advice if I can install 7970m with older vbios and black screen?

 

We have 30c in greece. I cant wait for temps to drop here

Maybe your hdmi out or vga out will work so you can flash the correct vbios

 

Do NOT try to flash an older vbios to that card. Return it to upgrade monkey and request an earlier vbios revision card. Flashing a lower vbios to that card will brick it.

 

Thanks guys. I called dell about my problem i should have done it like a month ago lmao. it is actually motherboard that is broke. They don't have it and told me to go ztronics. After fixing it i'm planing to put 2x7970m.

 

Hi AlienHack.
Ive done 28X now no OC and OV in BIOS.
Its not very stable and had to do the test twice elevating the rear of the PC 2 cm last time to be sure it wasnt temps taking it out the first time.Only internal cooling.
Ended up at 80 deg C max 24 ambient inside here now. Maximum turbo core clock 3 724 MHz.


Had 110 TDP and 70 TDC. CM ticked and CCM ticked. (also SM)

Will go back to 75mV over. Since temps let me do it with ease right now.

 

@GroGan: You know at this point down the timeline, they may just end up having to offer you a replacement m18x or AW18... with probably very expensive cards

I'm so jelly

 

No need to be, could easily be my CPU not giving it all since i dont get you guy's scores.

But i did a good job on the cooling when putting it together after the last pasting, and got a great compound.

 

Ok, I need some suggestions guys.

For about a week now my cards have begun a really nasty behavior, where running crossfire doing anything 3D will result in my cards doing absolutely MASSIVE throttle. Typically my cards would start to to warm up, you'd see a bit of that good ol' crossfire FPS, then suddenly BAM, 60fps > 20fps for about 1-2 seconds, then the cards start pumping again, 4-5 seconds pass, and WHAM. Back to 10-20. This is with Windows 7 32/64-bit, Windows 8, Windows 8.1, 13.8beta/beta2, 13.10beta and even the 13.15 windows 8.1 preview drivers.

Sometimes the performance stabilizes somewhat for a while and I can play for an extended period of time fluently, but other times the cards will just get stuck at an ultra-low frequency and require system restart or disabling/re-enabling crossfire to work correctly.

Single-GPU is flawless of course. Another thing I've noticed is that running any app with RadeonPro and checking the "Always use highest performance clocks while gaming" will increase my average and max fps in a pretty big way (and often overheat my cards resulting in thermal shutdown).

None of this helps me with the cards massively downclocking/throttling.
Here's an example of the behavior attached in a hwinfo log.

EDIT: Here's the log. https://mega.co.nz/#!zZhUDYrR!KxAWIwcJdVtU4vFLp5xzdjxG5uxIQYsrRn2G4YtbaQ4

Can anyone make heads or tails of this behavior?

Also, i'm thinking maybe I just use too many pads? I padded almost everything on the cards. What pad layouts do you guys use?

 

Have you tried opening 2 gpuz instances , then selecting the sensor tabs so you can chek each cards speed temps usage etc and then running a game. Then when the throttling hapens alt tab and see what the gpuz windows show. Was your card speeds lowered? The temps? Maybe just one of the cards drops speed...

About the thermal pads, I just use some very old and overused pads on the mems, they are broken and dirty from swaping on varius cards (I had them on my old 5870s). They dont even cover all the mem chips. I dont use pads on power modulators.
And mu gpu temps are 72-84 C (ambient 30C) when gaming. So I dont think that overuse of pads is the best thing to do.

 

Yea, I'm guessing that's true. Unfortunately, I've checked the logs. Both cards drop clocks as well as voltages. I'll re-apply pads and paste tonight... may be simply retarded cooling on my part.

 

Did you check with double gpuz as I proposed?also what temps do they have when they throttle? My cards never throttle up to 90C where they just shut down

 

Got the card this evening. Cant install until after i arrive from offshore next week. Short trip.
My cooler is polished to mirror shine using crome polish. Paste waiting at home. Cooling pads for memory in the mail.

Hope you manage to cool you GPU's again Sangemaru. (Problem even when using assisted cooling? (cooling fan tray))


Actually happy to see that 3Dvantage tests done with my R2 without extra cooling, i am very close to what they have on top of the list for CPU speed, but heat is worse than 3Dmark 11. (21300)
27X gives me 99 deg and probably slowdown. My extra fan help should keeping temps lower.

 

Ok, so they're throttling, but not really because of heat. The throttling occurs at any temp (50, 60, it doesn't matter).
I've had instant shutdowns just idling in windows, for no reason. I'll be seeing if GPU-z reports anything special, but I don't see what to do...
Ironically enough, performance seems to even out once the cards get hotter....

 

Try removing all drivers with amd driver removal and installing again from the begining

remember the problems i had with the second card that stayed stuck at 3d speedS? well after installing the latest drivers it works perfect

 

Unfortunately I've gone way beyond the call of duty with multiple windows reinstalls, driver reinstalls, gpu swaps, etc... damn those CF cables for being so expensive, I'd pick one up, maybe that's where the problem lay, but that means getting into food money, and we're not halfway through the month yet...

Good luck mate Take your time, don't rush, and upgrade that bad boy. That Vantage score is very very good, and yes, it does heat up quite a bit.

 

is it that expensive?
well try just drinking water for 2 days... you know for cutting expenses

 

It's about 45 EUR, which is around half my monthly wage after expenses. Soooooo, yea

EDIT: Well at least I can apparently still game. Tomb Raider on ultimate settings 1920x1200 bench says 57ps average (this is with enhanced AA Adaptive MLAA Edge-detect). Game never goes below 50fps, so smooth.

 

WHHHaatattt??? 90Euros is your salary? and i thought things were bad here in Greece!!! is that the normal monthly salary or just yours?

 

Nah man, mine is 400, but after expenses (rent, loan, food, utilities etc.) i have about 90 left. I have to find some time to look upon greener pastures and hope to gtfo

 

oh... well salaries are like this here too.. 400-600 Euros.... thats why i work double jobs


by the way, i sold myoculus rift dev kit, it was sold in just 4 minutes on ebay and bought this in its place
PC Gaming on a 21:9 LG Monitor with GTX Titans in 2-Way SLI - YouTube
!! ( the dell one actually but its the same panel)

although its a IPS screen , its colors still dont compare with our alienwares rgb screen...

 

Yea, looks cute, but I wouldn't move on to screens for just 5 extra inches The RGBLED panel is too good.
Graphic artists at my studio stare at it like they haven't seen colors before.

 

exactly as you say it, whenever someone looks at the RGB panel , even myself, we feel like we have never seen real color in another monitor.


About the 21:9 screen, its not the extra inches! its the aspect ratio!!! it is a very very long screen and so it gives me extra viewing space on horizontal plane. Its natural resolution is 2560x1080!! compared to 1920x1080 that usually screens are just imagine what the extra 640 pixels on the horison give! when playing 1st person games or simulators the wide view is great!! in strategy games also... and its an ips panel, good colors (not like rgb though) and great viewing angles.. i actually bought it because i was disatisfied by the rift. i bought it to play star citizen with it but its resolution is too low for gaming right now. so i wanted something else to get me in the game (feeling as much inside the game as i can )

 

guys that have the m17rx2 and the rgb led screen, i opened a new thread about screen questions i have if you could help answer my questions i would be great full

http://forum.notebookreview.com/accessories/732073-rgb-led-vs-ips-colors.html#post9370126

 

Because all the members posting in the current thread have the m17x R2 with the RGB LED panel, i think that the above link interests you all.
After a lot of searching i did the conclusion is that our alienwares screen gives comperable colors (color range of Adobe RGb ) similar to the newest professional AH-IPS panels sold by dell and other manufactures for 1300-1500$ !!!!

Just apreciate the rarity of this 6 years old screen (our alienware RGB LED screen that is) that is comperable in color range to the newest technological advancement of IPS and AH-IPS panels )

*actually IPS panels were doscovered back in the 1990's but they recently came back in commercial use as apple and its retina screens use and advertise these IPS panels. So everybody is raving for this new technology in screens called IPS when in truth its a 20 years old thing...

Also i want to mention that the newest AH-IPS panels i talk about above actually are IPS panels With RGB LED backlight!! like our alienware screen's backlite!!

 

So my crossfire seems to have stabilized in a weird place. On the one hand, the clock drops still occur but they are not as frequent as before, and I can actually play most games semi-reliably. Also, I haven't had anymore alt-tab crashes in any game for a while now. When the game crashes, it just crashes to desktop (and it's very rarely - usually if I'm playing a youtube video at the same time).

On the other hand, there are still clock drops and I get occasional screen shake and this weird black thin vertical line pattern made of a line that tries to simulate a checkerboard pattern. Usually in conjunction with the screen shake, and it occurs just in some games, and just in some areas in those games (SC2 menus and some maps, Tomb raider in some wide-angle scenes, Metro:LL in menus only, etc.).

So... it's sort of usable now. I'd still like to get to the bottom of this though... I like my experiences seamless.

 

Seems like a quiet before the storm

 

sangermaru , might i ask you, why dont you try to flash my vbios and give it a try? from the first day i used it i have totally no problems, not a single crash to desktop or slowdowns or anything.
You can try flashing the second card first to check that its working properly and then go for the 1st too.

just tell me and i will send you the edited one with the 1V vddc
you dont lose anything to try.

 

Hey guys, sorry for going awol. I've just been incredibly busy recently and haven't had the chance to reply to some of your interesting posts. I still haven't done the PSU mod as yet, although I have all the components ready and waiting. I just need to find the time.

@Alienhack I totally agree man. This screen is probably one of the main reasons most R2 users aren't giving up their machines for newer models. Plenty of other reasons too... like 10K 3Dmark11 scores Although the viewing angle of these screens aren't good the colour's are amazing. I've also grown used to the unnatural eye-popping colours and when I load up the screens EDID data for colour temperature from within CCC, everything becomes too dull in comparison. I'm not sure how accurate this EDID data is though, I don't use it anyway. From what I know screen calibration is dependant on the lighting in your environment as well so its a bit of a schlep for a mobile screen and besides I don't do professional design work. Unnatural eye-popping colours ftw!

@sangemaru Are you still running 0.975v on both your cards? What TS settings are you using for gaming? Your symptoms are really weird and the only thing I can think of its power related (you used imsolidstate's exact method ie. chip transplant?) or maybe one of the cards is going bad... I really hope its not door number 2 I know you might not want to read the following but it could also possibly be a hiccup on the power circuits on your motherboard. It may be the reason why your XM was running really hot when the 240W was exerting itself trying to provide more power. Now you're running the 330W and the power is cleaner but it may be too much for the circuit to handle. This could be total nonsense though, I'm just thinking out loud. Also menu screens dont normally draw excessive amounts of power... so it probably is complete nonsense

 

Glad to have you back mate!
Yes our screens are a thing of beauty. Even my new 21:9 ips dell screen comes nowhere close...
Although its blacks and viewing angles are top! On the colors department its good compared to other screens bud the alienwares screen has spoiled me!

I sagermaru should try software solutions before thinking about hardware. Maybe his crossfire cable is bad or something. But trying my bios cant hurt I think.

by the way have I told you that since I received your edited vbios I have been keeping my alienware running 24/7 and both cards mining litecoins. They have made back 50 euros so far
and they are always totaly stable

 

Yea, I'm trying a fresh crossfire cable next. If you guys know where I can find any for around 30 eur that would be great, i don't want to wait another 2 weeks. I've asked my techie buddie and he suggests the same, since it's too weird for each card to works perfectly on its own but hiccup during cross.
I don't know about viewing angle problems for you guys, mine are great.
Leave me your vbios bro, I'll try it (the 1V version).
Yea, I'm currently still at 0.975 stock, nothing's changed since the days they ran perfect.

 

Tuning IN.
Bro, and from where did you got your cards from ?

Getting CF soon...

 

Go for it flingin About the cards... NBR, ebay ... Where else? I did get one of mine from a countryman, so there's that.

Also, I just about despise my cooling system at this point. I don't know if the screws are losing force, or what the hell is going on, but I'm back down at reaching 100C temps at 60W TDP again. This is driving me insane. I'll be doing a retention mod tomorrow. Hope I don't crack the die.

 

Man that sucks, must be something with the heatsink !

The heatpipe must have lost its Gas/liquid charge inside.

I usually just place line of IC diamond in the middle and let the screws do the job by just tightening them in 1.2.3.4 order, and i do not screw them to 100% straight away, just by 25% at a time to ge a good spread.

AND EVERYTIME my paste job is just great this way, can do TDP 90 and temps will be 85 max......

i would try to change this heatsink bro, they are not expensive on ebay, Get one, and GOOD LUCK !

 

I already replaced the heatsink once this month This one worked fine a few last week, and all of a sudden, WHAM. Crap temps. I'll use your method, maybe it'll go better. What thickness pads are you using for the coil cubes above the CPU? For me, 1mm seems far too thin, so I've put 2x 1mm pads stacked at the moment, maybe they're pushing my heatsink up or something

 

that may be the issue, 2mm may be too thick, i believe i have 1.5mm Phobya Thermal pad Ultra 7W/mk

 

Alienware AMD HD DELL 7970M 2GB GDDR5 -for M15x, M17x , M18x ,upgrade monkey | eBay

thats where i got my cards

 

Unfortunately, the card currently listed there is 022 vbios. Won't work. He needs something older.

 

Already got 2 clevos from Eurocom.

Starting to think about PSU mod....but....

Is there any post that will say: "i modded my psu and it is working Super Duper'' ?

All feedback i got so far is that people are soldering/resoldering ID chips with different results

Is there a 100% working Mod out there ?

 

I have the mod working 100%. The troubles I have currently have nothing to do with the PSU. I'm not getting PSU shutdowns, nor do I get throttling due to low power. Follow the instructions here: 330W power supply for M17x update » imsolidstate

If your chips are in a different package form from the SOT-23 imsolidstate has, I can guide you.
Do the mod.
Trust me

 

Bin flikkin waiting for 12 days now for cooling pads.
Just re-ordered from watercooler in GB.
Just had to get the Phobya. Didnt know they had 7w/mk pads.

So eager to start im almost about to paste it instead while waiting 3-5 days for the pads :-(

 

If you have leftover pads from your previous card/cards, feel free to use them (without stressing the cards too much) until you get the phobyas. Don't be too enthusiastic about the 7W/mk ones. They're quite inflexible and brittle compared to their 5W/mk ones I'm still using 5-year-old ripped up dell thermal pads ) I have some special stuff coming in from germany. 14W/mk fujipolys.

 

As I told you guys my pads are the ones I used on my 5870m s . Old , dirty and torn appart. But they keep these cards running stable like hell...

so I dont think that the pads are everything. Also using pads on the power regulators might not be a good idea... as the extra heat from them moves to the heatsink which then heats up and draws less heat from the gpu itself.
these pwm regulators are meant to get warm..

 

Don't worry about it, it's not like the heat from them is dumped directly into the heatpipe. Most of that heat enters the heatsink and is then dumped into the machine itself
Anyway, the coils and the rows of 8 mosfets each above them all need to be cooled, along with a few small ic's on each side. Not cooling them can result to instability in the long run.

 

If they are 14W7mk, they almost double any heat compound.
Good ones have 8, GELID has 8.5, and is the highest ive found.
Could be great for even CPU/GPU even if slightly thicker than paste.

 

Like I said... special stuff
Using it instead of paste is a bad idea though, since the whole point of using paste is to only fill the microgaps in between the copper and the die. Even those pads don't compare to the conductivity of copper, thus resulting in no point whatsoever in using them for chips with big TDP's.

 

the reason i said that is because each heatsink has a thermal load it can cool

i dont know what the thermal limit of our heatsinks but dont forget that they were made for 5870m not 6970m. SO we are probably reaching our heatsink's limits. so there is no reason to load it with even more thermal Watts from the pwm chips.

the more closer you are to the heatsinks thermal limit (in watts of heat you feed it) the worse it will desipate heat and your whole system (graphic card - heatsink) will get hoter overal.

so i prefer not to feed my heatsink more watts than it is needed. these pwm chips can cool themselves, even on desktop mobos they are rarely actively cooled.



PS.Question - on the alienware m18x are the 7970s' pwm chips cooled with thermal pads? if not then the designers of the system knew what they did and so we dont have to cool these pwm chips either..

just for reference. now that the ambiemts temps have droped in Greece to 25C , my cards can game hard and be around 65-70C... just imagine in winter with 18C ambient.. now if that isnt cool...

 

The Fujipoly is not as soft as the normal heat pads either. Shore00 80 hardness.

 

That's true, but compared to the 5870m's, the 7970m's are much cooler. Remember those beautiful 90C+ temps? Don't worry about the heatsinks.

Yea, i know But I really want to try them. I want those chips (especially the vrams) as cool as possible, because I want me some vRAM OC

 

So ive seen the installation instructions for heatpads.
What thickness is perfect on the RAM for instance? Not to make it increase gap on GPU`?
Such a thin layer will easily be made to thick and prevent GPU to hit heat sink.
The thicker ones will be compressable.

Anyone have a better cooling drawing than what they show in the KIT drawings by upgradeyourlaptop for instance. Any components not needing cooling compared to these showings? Not sure where the regulators are for instance. :-(