***M18x - NVIDIA SLI (Driver Questions, Performance, Problems) Discussion Thread***

Hi, I could not find the 80.04.5B.00.02 with OV I already have this one Dell 680m - 80.04.5B.00.02_'OCedition'_revised_00 but I want to find the OV one compatible with Secure Boot. W8 user here.
Can you PM me or tell me where I can find it?
Thank you!

 

Your best bet would be to ask over at T|I on the relevant thread: NVIDIA Kepler VBIOS mods - Overclocking Editions, modified clocks, voltage tweaks - Page 269

I'm not up to date on svl7's mod's - he might have an over volted secure boot friendly vBios. You could also ask there or here: http://forum.notebookreview.com/ali...alienwarem18x-r2-gtx-680m-sli-7970m-cf-2.html

 

Any consensus on the best Nvidia drivers for the 680/780 SLI solutions? I am running the 326.80 beta that came out a week ago and they just don't seem to be very good in-game.

 

Unless something has changed, I would say 314.22 is still the best go-to driver for 680M SLI. I'm not sure about 780M SLI.

 

I'm having a hard time finding a driver to play the Vanilla Crysis 1 with for my 780m. Anyone having a recommendation? I have tried the 326.80 and now using the 326.19 to no avail. The game is suffering significant input lag. Most other games are running flawlessly.

 

Dangit, I'm back to overclocking being completely bupkis for core. I thought it was a driver issue with the 326.x beta at the time so I installed 320.49 and everything was ok for awhile but now once again core will not go past the stock 850 in the TI vbios. RAM has always been ok but no joy on the core anymore. I even back-leveled to the Dell 311.48's but no change. Has anyone yet seen a real explanation for this one? I know several other folks have seen the same thing but wonder if a hard fix has ever been discovered.

EDIT - no idea if this really had anything to do with it but I disabled CPU overclocking in the BIOS and afterward I was again able to overclock both GPU cores. I then set the BIOS back to overclocking enabled and am still holding a good GPU core OC along with XTU and ThrottleStop disciplining my CPU. Perhaps the act of 'toggling' the BIOS had something to do with it? Man I hate this BIOS.

 

Greetings to everybody.

I have an Alienware M18xR1 with Nvidia Sli 580m, in June 2013 I had a problem of screen flickering and after being in DELL to repair (where they seems to have changed GPU and SLI bridge) it seems to be solved, but the pc came back from assistance with original video drivers and I was informed verbally to use only these drivers because the new others damage.
Unfortunately, these drivers are the 269.03 year 2011 and are outdated, the site DELL for my pc also offers the 306.97 (year 2012), but do not work because the image dances and indeed the assistance does not recommend them, but why then offer them?
I tried to install the 314.22 that I have read to be stable around M18x but the PC crashes.

Obviously the problem is serious because I'm with updated graphics cards for games of 2011 and not for those following, for example Borderlands2 (which is not exactly brand new) does not work Physx.

I bought an Alienware PC because I thought it was the ultimate portable gaming in Italy, and instead I find myself with a PC inefficient after less than two years after purchase.

Why DELL does not provide for a system upgrade in order to keep updated the Nvidia cards, if a client sells a pc of this kind that costs thousands of Euros, should be supported properly.

I am really disappointed by this and I do not know if purchase again an Alieware, if anyone has any suggestions I would be grateful.

Pardon my English but I speak rarely.

 

You speak it very well, as far as I can read!

Did you try a complete clean install of the drivers? - I mean uninstall the old ones, run CCleaner and Driver Sweeper/Driver Fusion and install a fresh download?

Welcome to NBR, by the way.....

 

Thanks but do not deserve it at all.

Yes, I did a clean install of the new drivers but the problem persists.

I repeat that the DELL technicians insist that I have to keep the old drivers, I think it is no one correct answer.

What can I do?

 

Hmmm. You mention the PC crashes....is that during driver install or during gaming on that driver? - I'm just 'sideways thinking' here, but have you checked the fans and heatsinks for dust/debris....if the FPS are jumping all over the place, the gpu clocks might be throttling down due to heat....any idea of your temperatures in-game and at idle???

As for being told to use only Dell drivers, that's what Dell will tell you - it is not the case that Nvidia drivers will not work, just the case that Dell employees can only recommend Dell drivers, for obvious reasons.

 

Idea Storm

 

So, here is my bit of storey,

ok here comes another m18xr2 with some weird issues

had originally latest dell VBIOS 80.04.5B.00.02
Experienced some weird issues on 3dmark firestrike and cloudgate, etc GPU was setting its clocks to whatever it was its max (was 719mhz on original vbios) but GPU load fluctuated between 76-86% etc never more. This only happened if I set my GPU's in SLI Enabled mode. If I set it to single GPU mode GPU utilization was 99% no problem.

Thought that could have been driver issue, tried all, didnt help(yes clean driver uninstall, driversweeper, ccleaner, reinstall etc)

Thought must be the VBIOS is doing something, did grab svl7 VBIOS of the same version, but did not solve the problem. Tried OV/OC nothing helped.
Temps are stable solid sub 65c etc.
Also we tried few different versions of this VBIOS still no joy... however, I come to think, either:

My PSU is faulty?
My GPU's are on there last legs (unlikely, not stressed before)?
My SLI cable problem?
My motherboard problem?

After short run just to visualize what I experience here you can see what I mean:



EDIT:

Also here is msiAB monitoring screenshot of crysis3 everything maxed out to very highest setting (v-sync OFF) OC to 901MHz GPU / OV 0.975v (monitoring @ 100ms refresh intervals)



EDIT#2:

with help of svl7 I've downgraded to older vbios unlocked by svl7 (from 80.04.5B.00.02 to 80.04.33.00.32)

didnt help...




EDIT #3:

well, here is my single GPU score's on 3dmark cloudgate



performs pretty nice and even load if single GPU only... maybe my secondary GPU is no good?


any ideas????

Tell me what do you think what can it be?

 

hey ultraGSM

i dont oc GPUs myself at all but it could be many things, including the ones you have mentioned. you can test your power supply and see how much power its drawing from the outlet but theres no telling just how much of it is going into your machine although I must add if you use any other under powered dell PSU, alienware will tell u its not enough at boot up.

GPU could likely be faulty but if you tried single GPU, test on both card (single) and got no problem then probably the SLI cable. there are total of four combination, card A in slot A, card A in slot B and B in A/B to test if MXM port (mobo) is faulty.

if there are no issues then its likely vbios acting up? which is also the mobo, or perhaps power just isn't going into your graphics card in SLI for whatever reason. 65C seems low, unless of course your fan is on full blast.

lastly if all tested fine I'd say it's your cpu. a problem CPU can cause all kind of trouble, since everything is processed by the chip if you have a faulty weird chip its likely that.


edit: I only OC CPU but there are settings preventing current/voltage going into the chip/machine, theres probably a similar thing for graphics card in bios settings, maybe ask Mr. Fox or johnkss they'd know more.

 

Yea I know about the PSU, I'm only using OEM alienware 330W PSU's and have two of those, one at home and one in the office, Im back to office on Monday again so thats the only time I can test PSU theory, hope thats all can be wrong...

If its GPU fault... I would see that pretty easy removing primary GPU card leaving secondary one only which would act as primary and only in that case, and if it works flawlessly as a card in its own MXM slot, cards and slots are fine, right? Just trying to disassemble the GPU heasinks from Cards yet, still low mileage OEM TIM and pads yano...

Well, I may just hit the hard road after the above tests and tr downgrade to oldest VBIOS and try one by one upping one version every time and hope for the best version to work...? Yea I dont like auto fan control for benchmarking I just went in to hwinfo and set all fans to 3800rpm to be on a safe side

Well CPU is OEM and it performs well and cool enough, if I give it abuse as media decoding BD-RIP and etc etc it has never ever froze or BSO'd on me or any unexpected behaviour whatsoever... weird... I would like to count CPU out of this. I am rning stock CPU no any CPU OC for this matter, it runs mid 45-60c on regular use and when gaming can go as high as 90-95 while on desk doing crysis3 all maxed out (auto fan settings)

I totally feel something like that too - it could by system bios is preventing something too....

really need masters to chime in here on this

Thanks for the input!

BR

 

I doubt that you are having a hardware problem. I suspect an issue with software or configuration settings.

Is 3DMark the only benchmark with this issue, or do you see it with other benchmarks and games?

Here are a few possibilities:

• Be sure you are using a profile (Use the advanced 3D image settings) instead of the option to "Let the 3D application decide"
• Be sure to set everything in the global profile to focus on performance instead of power savings or quality
  - in particular confirm the global profile has "Prefer maximum performance" (not Adaptive) for the Power management mode
• Be sure you are using a modified profile for each of your games and benchmarks that is optimized
  - for 3DMark you must manually select AFR2 for good results
• Be sure you do not have Windows Update KB2670838 installed - uninstall it if you do

If you have already tried the above suggestions, try using stock clocks and see if the erratic behavior improves. Your overclock may be near the edge of what stock voltage can support. Try a vBIOS with 1.050v or 1.087v instead of stock voltage. For +140 core overclock 1.050v is probably enough.

 

^^^ oh man, "Like" button is not big enough to say how thankfull I am for you givin me those tips quickly, I got sorted with GPU's usage dipping on and off and managed to get pretty good score after altered those settings and uninstalled update pointed out!

just wanna share what ive got with these settings after:

CPU - TS OC only 43x4 cores
GPU - modified vbios with 1.025OV
GPU CORE - 1GHz
GPU RAM - Stock
FAN @ 3800rpm with hwinfo



now time to apply proper VBIOS and latest unlocked system bios, and let the game begin!

Massive thanks fellow mr.Fox and +Rep+ goin your way!

 

That's awesome, Bro. Excellent results. You're welcome. +1 back to you.

 

yep, this is why its best to wait for some expert oc'ers for tips. I was gonna say to reset cmos and try again lol glad you solved it

 

Hi Everyone:

I have an alienware m18x r2, i7 3840 QM, CPU 2.8GHz, 680SLi. 32gb ram.

I have been having no issues with my laptop until 1 month ago. Whenever I try to play any game, there will be a slow down where it is unplayable- Frame rate will drop to 4-5 per second. If I hit the escape button on my keyboard and resume game, it will go to normal. I have to do this like almost every 5 mins.

I have replaced my fan and heat sink in my processor with Dell's tech people. Is there something I am doing wrong? For Nvidia drivers, I even installed the most recent uptodate drivers and I also tried changing from sli to single driver. it's still doing the same thing.

 

Hello and welcome to the NBR Alienware Community. You have a really nice laptop. We will try to help you resolve the issues.

When was the last time you removed the GPU fans and thoroughly cleaned out the fans and GPU heat sink radiators? This is a leading cause of GPU performance issues.

Set up OSD as described in the thread linked below and let us know what kind of CPU and GPU temperatures you are seeing while gaming and maybe we can draw a correlation to the drop in performance you are experiencing.

http://forum.notebookreview.com/ali...mperatures-game-benchmark-screen-display.html

 

Thanks Mr Fox with the help. I am going to try to setup it and will post screenshots. I wonder if it's a software glitch because it's strange that hitting the escape button would help with the stuttering I am experiencing.

 

Anything is possible. If you escape or alt-tab to desktop the CPU and GPU can cool very rapidly. You need to have some way to identify what the maximum CPU and GPU temperatures are. This can be deceiving if you rely on a real-time desktop display of current temps due to the rapid change. The evidence of overheating can vanish by the time you reach the desktop. If temps are the underlying issue you need to either be logging the data (to CSV file or something like the HWiNFO64 sensor table) or watching it real-time, in-game, with OSD to determine what is happening.

 

Anyone knows a stable nVidia driver version for the AW18 with GTX 780M SLI? I really want to update the driver that comes from dell as it is really outdated but I've heard there are only a few versions that are stable for the AW18.

The games that I usually play:
-Final Fantasy 14 ARR
-Bioshock Infinite
-Tomb Raider
-Crysis 3

 

pangers94, 331.40 is working outstanding for me. It worked fantastic with 680M SLI and I am using it now with 780M SLI on the 18 and M18xR2 and getting superior results with it.

 

Once again, Mr.Fox to the rescue !. Thank you for this info brother!

After updating to the new driver, I notice a significant increase in the frame rate of all the games. I get around 10-15 frames increase. Couldn't be happier with the new driver

 

Great news! Glad I could help. +1 Rep

For anyone else that needs it, modded INF attached for 331.40 beta for M18xR2 with GTX 780M with on-the-fly SLI mode

 

hi Gents,

ive just took delivery of an m18 with 4700m cpu and dual gtx780m sli, 24gig ram, 750gb hdd, 64gb ssd as cache.

everything i play just stutters, i have the power adapter connected, i have sli enabled... is there anything that im missing that i need to check...

msi afterburner is picking up both gpu's with good temps, clock speeds etc. the fps counter seems to be 100's which seems wrong considering the games are unplayable....

im sure this is a setting that im missing... i have been searching around but not found alot

any help would be very gratefully received

thanks alot

Ian

 

Are you using Wifi? Trying switching to LAN and see if the problem persists.

The Broadcom wifi card has some problems with stuttering, I didn't exactly experiment with it that much, since I change to Intel AC very early on. The Intel AC has its fair share of driver problems too.

 

hi

thanks alot for the reply, i will try lan, but i have used an intel 6330n wireless card in my old m17... i think i might try this in my new m18.. i really need wireless to work..

thanks alot

Ian

 

Oh my god I have to take it apart just to change the WiFi card....

Now I'm scared!!!

Sent from my GT-I9500 using Tapatalk 4

 

Maybe this might help - I can't vouch for it's effectiveness as I don't have an 18.....quoted from pg. 240 of the http://forum.notebookreview.com/ali...al-alienware-18-owners-lounge-thread-240.html

 

OK here goes my question now. I've strange issues with GPU Load utilization while in SLI. Regardless of clock's/ CPU settings stock or OC'ed still same issues.

If I try run my GPU as single GPU "NON-SLI" it will be all dandy nice even load utilization at any clocks 99% full utilization constant troughout the tests/benches, now if I load system in SLI mode, not that the second GPU works weird, primary kinda mirrors the spikes of the secondary GPU LOL

Ok, so I have went trough alot of conclusions and theories, no - PSU is not bottlenecking yet, while all on stock I monitored it pulled about 250W from the wall, if OC'ed I can make my 330W brick go past 420watt border (amazed by that btw!) so, I tried always OV the GPUs same and never get even close to 98-99% constant load on both GPU's, sometimes more stable more steady load on primary GPU but hell-noh no way for the secondary!!! So I tried one trick to apply little more OV on secondary GPU vs primary GPU and I did notice that secondary GPU stablized little more load but nowhere even close to somewhat constant 97-98% utilization... just weird

I ran firestrike on single GPU trougout the tests clocks were up to par and utilization was glass-smooth 99% during tasks, right after I have switched to SLI primary GPU load fluctuates between 96-99% and so does the second GPU but more between 94-97% and just very odd time spikes to 99% for split of a second...

tried 314.22 drivers most the time as they took OC'ed 680M SLI of mine pretty nicely, now been trying to do some pulls with 331.xx latest beta and still get same poor utilization in SLI.

Can you, guys with 680m SLI setup, try run nvinspector monitoring of GPU load utilization and report what GPU load are you getting?

BTW Im runing SVL7 unlocked adjustable voltage VBIOS (version seen in screenshot), moved down from UEFI-able bios down one level as I was recommended to do, no complaints here, but prior downgrading I been running the latest unlocked NV vbios from DELL 680m and still had same issues.

Another Idea/Question - can it be that the "adjustable-on-the-fly-voltage" vbios be not so stable as fixed voltage vbios? By means - not always providing super-constant voltage during load? just shooting in the dark ofc...

Also followed Mr.Fox's advice on NV-CP settings, but not much of a help in resolving my GPU load... or maybe Im trying to fix something that aint broken??? bcause that would be a bummer



and here is what I meant exactly about EVEN NICE LOAD on single-GPU mode vs SLI mode

 

Might be something with the vBIOS. I would recommend using a fixed voltage on the 680M. I got consistently better results in benchmarks with fixed voltage when running 680M SLI. Other than this and the previous suggestions, I don't know what might be causing the issues unless it is the AC adapter running out of juice. (This will cause odd behavior and a decline in performance if you are pushing your system right up to the verge of tripping the AC adapter.)

 

the load instability occurs in sli mode only, regardless of clocks, even at stock all gpu and cpu. like I mentioned earlier. can you do one benchmark pull sample and post screenshot of nvinspector gpu load graph monitoring of your 780M sli setup? stock sli and oced. I've two 330w psu's but as I need them both intact I won't be doing mod

out of these now, will hunt around for cheap pair of spare ones

 

OK, here are stock and OC 780M SLI 3DMark11 test 1, 2, 3 and Fire Strike with NVIDIA Inspector graphs. All four tests at 4.3x4 on CPU and GPUs are with 1.025v (stock is 1.0v for 780M). Is this what you are asking for? I don't remember your other post requesting this. If you need something else, let me know. Hope this helps.

3DMark11 Tests 1-3 w/ stock GPU clocks	Fire Strike w/ stock GPU clocks
3DMark11 Tests 1-3 w/ GPU overclocked	Fire Strike w/ GPU overclocked

 

Is it true that even without the SLI cable (or with SLI disabled), one could still dedicate the 2nd GPU to Physx? Without the faulty SLI cable in, I can dedicate the 2nd GPU to Physx, but the option for enabling SLI doesn't show.

 

Yup from memory when I tried doing the same thing SLi then is not an option anymore. I think this is because you are dedicating one GPU for physics only as its sole purpose. Personally I think most of the modern day CPUs handle physics well enough already however if the game has specific physX enhanced gaming features and eye candy then it doesn't hurt to play around with settings.

Cheers.

 

Well I wouldn't even know about this if I didn't have a broken SLI bridge. And the SLI bridge wouldn't have broken if I didn't have to get the mainboard replaced twice. And actually had I found out earlier on how to reset the BIOS (or make least make it able to boot), I wouldn't even need to replace the mainboard, not even once. And without that the SLI bridge would be fine and I will never learn about the fact with Physx. I didn't expect it to work similarly in a laptop as it does on desktop.

 

New to the forums, but thought this might be the place to ask for help. M18x R2 new windows 8 install. Have 680m sli, I have intel graphics installed and I have my gtxs' enabled, both show up in bios, and of course they show up in device manager as the basic monitor display. I have tried inf. modification on several different drivers. 331, 327, and 307 I could not get to work at all. 305.46 I could get to install but get code 43 in device manager. I assume I am not doing the mods right.

PCI\VEN_10DE&DEV_119F&SUBSYS_05501028&REV_A1\4&1D1DEA7A&0&0008

I took the 119F and input it just like Mr. Fox's guide says. I am no elitist on this just know enough to probably be hazardous to my laptop. Any help would be much appreciated.

 

Welcome to the forums!

Firstly, I'm not sure why you are attempting to modify drivers for graphics cards that are officially supported for your machine? - 680m's were an option for the M18x R2 and should not need any driver modification for them to install correctly. Maybe you should start out again fresh...uninstall any Nvidia drivers you have and ensure that you clean any old files left behind with CCleaner and Driver Sweeper/Driver Fusion. Then you can install fresh drivers without issue....hopefully. If unsure about the cleaning aspect, just ask for further clarification on what to do.

Also, you mentioned a code 43 error - do you have any screen artefacts going on? - sometimes, this can relate to a shorting of the backplate on the gpu itself - see this thread to rule out that particular possibility: http://forum.notebookreview.com/ali...gtx-580m-error-code-43-revision-ff-error.html

Don't worry, I'm sure it is something that we can easily fix for you!!

 

Please note that PCI\VEN_10DE&DEV_119F&SUBSYS_05501028 is the hardware ID for GTX 780M installed in an M18xR2, not GTX 680M.

The hardware ID for GTX 680M would be PCI\VEN_10DE&DEV_11A0&SUBSYS_05501028 in an M18xR2 and PCI\VEN_10DE&DEV_11A0&SUBSYS_048F1028 in an M18xR1.

As Brother Steve points out, an M18xR2 with GTX 680M SLI should not require any kind of modification. Something else is going on and that link provided may provide a clue if one or both support plates are grounded out on the PCB. Did these GPUs come standard or did you upgrade to 680M SLI? Or, is this problem a new development with the same video cards that have always worked correctly until now?

 

Thank you for the quick replies! I figured it out finally, indeed it needed inf. files modded, I just had to put the 119F in the 11ao spot instead, just started really looking at ID path and notice its backwards of what other people had wrote. This was a standard sli with win 7, I upgraded to win 8 pro and didn't know this would even be an issue. Thanks again!

 

So, you have 780m not 680m?

 

I think what he meant was he inserted 119F by mistake when it needed to be 11A0. But still, I used the same NVIDIA driver for Windows 7, Windows 8 and Windows 8.1 with 680M and no INF mod was required for the M18xR2. If 119F works, then yeah... that's a 780M hardware ID.

 

I am sorry that I was not clear before, Running 18x r2 680m sli. I used the latest NVidia drivers this morning 331.58. I Upgraded to win 8 pro and did a NEW install. When I went to install drivers for the 680ms, It said no compatible equipment. Changed the 11AO to the 119F in the 3 strings with the 0550. At first I kept trying to change the 0550 to the 119F and failed several times with 4 different drivers. If you look at my device path this will make sense.

PCI\VEN_10DE&DEV_119F&SUBSYS_05501028&REV_A1\4&1D1DEA7A&0&0008

In the actual driver software its labeled as

%NVIDIA_DEV.11A0.0550.1028% = Section187, PCI\VEN_10DE&DEV_11A0&SUBSYS_05501028


So the only way I got this to work was to put it in as

%NVIDIA_DEV.119F.0550.1028% = Section187, PCI\VEN_10DE&DEV_119F&SUBSYS_05501028

I don't know if this is how this is suppose to work or not but it is Hope this might help

 

Well, it's great that you got them installed - kinda strange that you had to mod the files for them to install though, as the 680m's hardware ID should not require changing on the R2. Very strange.

 

That is what I thought, I couldn't find anyone else having this problem. My next question is now when new drivers come out I will probably have to do this every time I go to update. Anyone have any idea why my gpus would have different device path's or how to change them? Surely their was not a mix up and I got 780m sli? If the device path is that of a 780m. I guess I could take the back plate off and look. When I get home from work.

Just a thought though, I downloaded the driver from NVidia and I think I remember seeing 780 in the nvdm file. So that driver should of worked for the 780m or the 680m before being modded?

 

Mr. Fox you were exactly right! It turns out the 680m sli I thought I had is actually 780m sli. I would of never even thought twice about it if you hadn't of posted Fox Thank you very much! So just had to change a few things around and getting ready to see what my pc score is now!

 

Interesting to note, the 331 Nvidia drivers enable full boost activity during sli, as well as OC on gtx 680M on the m18xr2 even on stock OEM vbios

 

Has anyone run a 3dmark Firestrike on a 780sli? What is usually the score for stock clocks?

Sent from my SGH-T889 using Tapatalk