Jump to content

NVIDIA Kepler VBIOS mods - Overclocking Editions, modified clocks, voltage tweaks


Recommended Posts

I don't think you understood my post, it's nothing to do with your system BIOS, so I was saying that by you trying to reflash your SYSTEM BIOS that you weren't going to get anywhere for the reasons I explained. So, I'm not surprised you're back at square one (at the same point as before). Have a read of my last post (post #4388), those are the options that are open to you. It's a pain for you I'm sure, I'd be very frustrated in your position, but there are solutions & they're summarised in my previous post.

Hi,

Ok , I Understood you post . there is Two main point

1)Return to Dell saying that it's just stopped working (without providing the details!). --- This is already in Process

2)Some independent computer shops can flash vBIOS to cards. ---- This is my Second option.

Thanks.:disturbed:

Link to comment
Share on other sites

Hi,

Ok , I Understood you post . there is Two main point

1)Return to Dell saying that it's just stopped working (without providing the details!). --- This is already in Process

2)Some independent computer shops can flash vBIOS to cards. ---- This is my Second option.

Thanks.:disturbed:

Hi, that's ok, that's right, and someone else also mentioned a soldering option, but that seems risky unless you know what you're doing & have the tools. If I was you I would have investigated the cost & effort of Point #2 you listed, and then if that proved too expensive/unworkable then I would have pursued Point #1. Don't tell them you flashed a modified vBIOS though - not the Dell guys! But...if they ask you if you flashed a modified vBIOS, then that's your call what you tell them.

Link to comment
Share on other sites

Way to much going on in here. :D.

@Anil Kumar Your best bet is to send your card to some one for flashing and have them ship it back or go to a computer store that has a machine laying around that can flash your card. For get soldering. Your card isn't hard bricked it's soft bricked. Very easy to recover from with the right tools.

Link to comment
Share on other sites

Well I just cleaned my heatsinks, so much dust...haha solved the problem on 3.31 I am back to 12650 gpu score on 3dmark11, however when I updated to 3.37 it went down to 11500 for throttling over 87, its annoying because of my fan profile, the slave gpu is a bit hotter on the p370em but average around 82-85 when gaming for a longtime, but the fan doesn't kick on to high until 93 so in benchmarks it throttles it on the newest driver...I just ordered new P375EM heatsinks, hopefully should solve problem.

If it's based on the GPU's getting hot, then you could repaste to prevent them throttling. Strange that you say the latest drivers would make that happen. I don't think that is controlled in the drivers, I think the temperature throttling point is set in the vBIOS. I think it's 87 degC, that's what is says in NVidia Inspector for the one I'm using (never gets that hot though, so don't know for sure if it would throttle at that temperature). How hot are your GPU's getting? (And in my experience the driver doesn't make the GPU run hotter or colder either).

- - - Updated - - -

Ah, that definitely sounds like a heatsink mounting manufacturing defect to me (on the GPU). I'm sure you know that the heatsink surface needs to sit flat against the GPU core, and from what you've described it sounds like there is a gap in some places where it's not making contact. The heatsink should make contact with the GPU even if there is no paste applied - the paste is there just to fill in the microscopic gaps & holes in the heatsink and GPU surfaces (because paste transfers heat better than a tiny air gap where those microscopic holes would be).

Link to comment
Share on other sites

Well I just cleaned my heatsinks, so much dust...haha solved the problem on 3.31 I am back to 12650 gpu score on 3dmark11, however when I updated to 3.37 it went down to 11500 for throttling over 87, its annoying because of my fan profile, the slave gpu is a bit hotter on the p370em but average around 82-85 when gaming for a longtime, but the fan doesn't kick on to high until 93 so in benchmarks it throttles it on the newest driver...I just ordered new P375EM heatsinks, hopefully should solve problem.

Cool, or try using a 3rd party tool to set up custom fan profiles, like HWInfo64 for instance. That way you might be able to keep below the 87 degC throttling point. Just be careful & certain that your applied profiles are working properly because I've heard of stories where fans have been accidentally disabled using HWInfo and then GPUs burning out etc.

Link to comment
Share on other sites

With the modified vbios your card runs effectively faster than with the stock version, therefore you have to expect higher temps. New thermal paste would be a first approach to this issue.

The issue I'm talking about isn't the temp itself, its the fact the fans don't spin up until a much higher temp than they did with the stock vbios. Once they kick on it can cool itself OK, but that's currently happening at 87'c, so it gets there much faster under load than if they'd come on earlier. I can force them to spin up with the old fn+1 trick, but having to do that all the time is a bit annoying. I only really see those extreme temps when running benchmarks which hammer the hell out of it (around 80'c is the most I tend to see when actually gaming).

I'm currently using I-C Diamond 7 paste, and before I applied this stuff I was seeing 90+ core temp on stock clocks, never mind the insane OC I'm now running with your modded vbios, so I think the paste job is pretty good.

Link to comment
Share on other sites

I have been have some discussions over in the Alienware section. After LOTS of testing I determined that my GPU was simply running too hot to be pushed any harder. I did a hardware mod that involved cutting a big ol' hole in the bottom cover to let more cool air into the GPU fan.

That helped a lot.

Today I repasted with MX-4. Before I did the repaste I took the heatpipe to my garage and used 1200 wet or dry along with a 1-2-3 block on the 1" end to flatten and smooth the die contact surface of the heatpipe. I also slightly bent the springs on the heatpipe to apply a little more pressure on the GPU. The results were quite good.

I ran this at 1153 core, 1475 memory, 1.0875v. Temps did hit 92c so I am at my limit again. I saw a Kill-a-watt load of 278w at the wall with my 230w AC adapter.

Again, cooling is my limiting factor... although I am not sure how much more my AC adapter can take!

3DMark 11: 9650

NVIDIA GeForce GTX 880M video card benchmark result - Intel Core i7 4900MQ,Notebook P17SM-A

SagerNP8278GTX880M1153-14753DM11-9650_zps6b2bbbf5.png

SagerNP8278GTX880M1153-14753DM11-9650Sensors_zpsdcb3b517.png

20140629_193112_12999_zps7f667257.jpg

Link to comment
Share on other sites

Very nice man!

I hit a thermal wall with my 9377 at the same point

http://www.3dmark.com/fs/2327316

Trying to run 3DMark 11 next blew my adapter lmao

The overall is lower than it would have been had I not disabled the turbo and locked the CPU at 3.1GHz. I knew it would overload the adapter if I had the turbo on.

Link to comment
Share on other sites

@deadsmiley, that's a good result! So, how much did your heatsink smoothing & mx4 paste lower temperatures by? If you've done a like for like comparison at the same overclock or at stock - what are the temperature differences before & after? Would be interesting to know just how much difference the heatsink smoothing/spring bending/mx4 made. (Although can't ascertain which one of those 3 things made the most difference, might just have been a bad paste job before).

Link to comment
Share on other sites

@deadsmiley, that's a good result! So, how much did your heatsink smoothing & mx4 paste lower temperatures by? If you've done a like for like comparison at the same overclock or at stock - what are the temperature differences before & after? Would be interesting to know just how much difference the heatsink smoothing/spring bending/mx4 made. (Although can't ascertain which one of those 3 things made the most difference, might just have been a bad paste job before).

Robbo,

I think that getting the heatsink *flat* so it makes the best contact possible with the GPU die is critical to achieving the best cooling that the heatpipe can offer. Getting the surface smooth helps. Remember that the purpose of thermal paste is to close up the air gaps between the GPU and the cooler. The surface of my Foxcon heatpipe was definitely not flat. Others on the this forum have experienced the same thing on previous model Celvos. I cannot speak for Alienware products, but I suspect that the surface that contacts the GPU is better (in terms of flatness and surface finish) than Clevo's model. I it would be cool (pun intended) if one of our Alienware brothers could check out their GPU heatsink the next time they have it off and report back. While the big hole cutout in the bottom cover made a difference in cooling. I think that flattening the heatsink surface made a bigger difference by far. I have no numbers to back this up, just general feelings from lots of testing and past experiences with desktop GPU cooling systems. I am kicking myself for not looking at the heatpipe sooner. Part of the reason I didn't look into it was because I wasn't sure how to approach the process of getting the thing flat, or at least getting it on a flat enough surface to check it. This is because the heat sink has rivets that protrude beyond the mating surface of the cooler, which makes it a challenge to get something in the space to work on that surface (a picture here would be nice).

As far as thermal paste goes, I really can't say. I am pretty sure I had a good AS5 paste job previous to using MX-4. It was my second attempt with AS5 on this machine. The first one I didn't apply enough paste and temperatures were very hot under gaming loads and worse while benchmarking.

Springs are tough to quantify too. I paused for a moment considering whether I wanted to test with just the heatsink refinishing and AS5, then with MX-4, then with MX-4 and spring tweaking. I decided that I simply didn't want to take the time involved to do that so I did the three things all at once. NOT scientific and we don't get to learn much except that some combination (or all of them) yielded good results. The springs themselfs are not robost. One of them bent at the rivet (the weakest point) and gave me an "oh shit" moment. I wanted them all to apply force equally. I have no idea if I succeeded in that or if tweaking the springs brought any positive or negative results.

One other thing I need to stress is this. My AC adapter was pulling 278w at the wall. It's rated for 230w output. Assuming 85% efficiency, we are looking at 236w. I really can't push any further without risk of popping my AC adapter. I really don't want to go back and test at this level without having a spare AC adapter on hand. I use this machine for work (self-employed). It's my money-maker. I probably shouldn't be monkeying with it like I am!

Proceed at your own peril! I am not responsible for your system if you fuck it up, void your warranty, explode it, implode it, create a tear in the space time fabric, etc... :)

Link to comment
Share on other sites

Deadsmiley, you asked about the surface finish of Alienware GPU heatsinks. The one on mine is a matt surface (but free from any big grooves), so it's not a mirror finish, but I don't think any laptop heatsinks have a mirror finish. When you say you were just trying to make your heatsink 'flat' by polishing it, you mean smoother rather than flat don't you? I guess your heatsink was sitting 'flat' on your GPU core before the polishing, and the polishing was just to get a smoother surface with less minute imperfections right?

Link to comment
Share on other sites

Robbo,

No sir. The die contact area was not flat. It had high spots and low spots. I am more concerned with it being flat and less so about it being shiny. My heatsink is not shiny after I worked on it. It has a matte finish. The surface is smoother along with being flatter. It seems that warped heatsink contact areas are a common problem with the Clevo.

Looky here:

http://forum.techinferno.com/clevo/5275-%5Bhardware-mod%5D-clevo-p150-p170-cooling-system-solutions-here-3.html#post97277

Link to comment
Share on other sites

Ah, I see, others have experienced the Clevo heatsink to be not flat. Well, it looks like you did a good job of making it flat - what with your gaming temperatures dropping from the 80's to the 60's (in that thread you linked)! That's one bendy heatsink (or was)!

  • Thumbs Up 1
Link to comment
Share on other sites

I have a Clevo/Sager 680M SLi cards. I'm using the custom ROM "Clevo 680m - 80.04.67.00.01 'OCedition' revised_01".

Which numbers should I put in the "MSI Afterburner" software to get the best performance?

Core Clock can go all the way to +405

Memory Clock can go all the way to +1000

I'm using the default, as of now!

Link to comment
Share on other sites

I have a Clevo/Sager 680M SLi cards. I'm using the custom ROM "Clevo 680m - 80.04.67.00.01 'OCedition' revised_01".

Which numbers should I put in the "MSI Afterburner" software to get the best performance?

Core Clock can go all the way to +405

Memory Clock can go all the way to +1000

I'm using the default, as of now!

Dump AfterBurner for nVidia Inspector as a first move, then increase core in small increments until it crashes or freezes, then back that off by about 10-15MHz and you have your core clock.

Do the same thing with the memory but I would knock 25-50MHz off your max overclock that you achieve on the memory. While a core overclock is unlikely to kill a card, a memory overclock is an entirely different story.

If your thermals allow, you can use the voltage sliders to increase voltage to the cards and push them further. Definitely use something like HWMonitor or HWINFO64 to track your temperatures.

Whatever you do to the primary card, choose the second card in the dropdown box and apply the same settings.

Once you find a stable overclock that you're happy with, save the Inspector profile to your desktop and you can apply it any time you want from that shortcut.

AfterBurner and Precision X have random issues on systems while nVidia Inspector is generally quite stable, hence my recommendation to use that instead.

Every card is different, hence why you have to find the overclock on your own. Also when you are testing for your maximum overclock, it is best to set the other back to its base setting (so if you are overclocking the memory, put the core back to stock), find your max overclock for the other then work on finding their maximum stable point together.

  • Thumbs Up 3
Link to comment
Share on other sites

Dump AfterBurner for nVidia Inspector as a first move, then increase core in small increments until it crashes or freezes, then back that off by about 10-15MHz and you have your core clock.

Do the same thing with the memory but I would knock 25-50MHz off your max overclock that you achieve on the memory. While a core overclock is unlikely to kill a card, a memory overclock is an entirely different story.

If your thermals allow, you can use the voltage sliders to increase voltage to the cards and push them further. Definitely use something like HWMonitor or HWINFO64 to track your temperatures.

Whatever you do to the primary card, choose the second card in the dropdown box and apply the same settings.

Once you find a stable overclock that you're happy with, save the Inspector profile to your desktop and you can apply it any time you want from that shortcut.

AfterBurner and Precision X have random issues on systems while nVidia Inspector is generally quite stable, hence my recommendation to use that instead.

Every card is different, hence why you have to find the overclock on your own. Also when you are testing for your maximum overclock, it is best to set the other back to its base setting (so if you are overclocking the memory, put the core back to stock), find your max overclock for the other then work on finding their maximum stable point together.

That sounds like good advice to me! Although, once you've found your max stable core clock, then I'd personally leave the core clock at max stable overclock when working out your max stable memory overclock. Possible that a high core overclock might limit the memory clock, and because you probably want to maximise your core overclock to maximise performance then you may as well have core overclock present went determining your memory overclock. I'll also add that any voltage increases you apply in software are only applicable to the core, it will not change the voltage of the memory modules, therefore increasing voltage will only increase your core overclock & won't increase the highest stable memory overclock.

Link to comment
Share on other sites

That sounds like good advice to me! Although, once you've found your max stable core clock, then I'd personally leave the core clock at max stable overclock when working out your max stable memory overclock. Possible that a high core overclock might limit the memory clock, and because you probably want to maximise your core overclock to maximise performance then you may as well have core overclock present went determining your memory overclock. I'll also add that any voltage increases you apply in software are only applicable to the core, it will not change the voltage of the memory modules, therefore increasing voltage will only increase your core overclock & won't increase the highest stable memory overclock.

Yeah I slapped it together pretty quickly but my reasoning for finding the stable overclock together is because of those cases where you actually take a performance hit from overclocking the memory as opposed to increasing the core. I guess for the sake of being simple, doing them together works best (and that's what I usually do, I am impatient though).

Link to comment
Share on other sites

Hi!

Just bought used MSI GT70 with GTX780M. And of course I want to unlock that VBIOS for some overclocking but not sure which bios to use?

There's so many to choose from! There's the plethora of 680M vBIOS and then there's the 880M, and the 860M too, but for maximum compatibility I would recommend the sole 780M vBIOS! You might want to google if there's been any incompatibilities using the 780M vBIOS with your particular laptop model though, but I don't think you need to google for compatibility for the other vBiOS's! ;-)

- - - Updated - - -

Yeah I slapped it together pretty quickly but my reasoning for finding the stable overclock together is because of those cases where you actually take a performance hit from overclocking the memory as opposed to increasing the core. I guess for the sake of being simple, doing them together works best (and that's what I usually do, I am impatient though).

Yeah, for your 880M that might be limited by heat development (in some laptops), then possible you can get better gains by just overclocking the memory & not running a maximal core overclock - especially if the game is bandwidth limited. For the 680M though (the card the OP has), I reckon you'd get maximum return from maximising your core overclock.

Link to comment
Share on other sites

Dump AfterBurner for nVidia Inspector as a first move, then increase core in small increments until it crashes or freezes, then back that off by about 10-15MHz and you have your core clock.

Do the same thing with the memory but I would knock 25-50MHz off your max overclock that you achieve on the memory. While a core overclock is unlikely to kill a card, a memory overclock is an entirely different story.

If your thermals allow, you can use the voltage sliders to increase voltage to the cards and push them further. Definitely use something like HWMonitor or HWINFO64 to track your temperatures.

Whatever you do to the primary card, choose the second card in the dropdown box and apply the same settings.

Once you find a stable overclock that you're happy with, save the Inspector profile to your desktop and you can apply it any time you want from that shortcut.

AfterBurner and Precision X have random issues on systems while nVidia Inspector is generally quite stable, hence my recommendation to use that instead.

Every card is different, hence why you have to find the overclock on your own. Also when you are testing for your maximum overclock, it is best to set the other back to its base setting (so if you are overclocking the memory, put the core back to stock), find your max overclock for the other then work on finding their maximum stable point together.

Is it safe to test until it "crash or freezes"? nVidia Inspector doesn't have the on-screen display that shows the info, I do need to use MSI Afterburner for that, but I leave the OC things OFF in the MSI, is this way good?

A good way to try the OC is running softwares like 3DMark and see if everything runs OK?

And what's the difference among "Clevo 680m - 80.04.29.00.01 'OCedition' revised_01", "Clevo 680m - 80.04.33.00.10 'OCedition' revised_01", "Clevo 680m - 80.04.67.00.01 'OCedition' revised_01", which one should I use?

I'm using the 80.04.67.00.01, and I feel that the GPU2 temperature is a bit higher with this custom ROM comparing to the default ROM.

My GPU:

post-26895-14494997885983_thumb.png

Link to comment
Share on other sites

I have a question about vbios modding. I have an m18x r2. I flash my gpus easily using nv flash and the command nvflash -4 -5 -6 and then the bios name. Is this the same procedure for single gpu machines such as the msi gt70? Mt girl has a msi gt 70 and I'm thinking of flashing her 770m with the modded on but I don't know if it's the same procedure as the multi gpu m18x. Thanks guys

Link to comment
Share on other sites

I have a question about vbios modding. I have an m18x r2. I flash my gpus easily using nv flash and the command nvflash -4 -5 -6 and then the bios name. Is this the same procedure for single gpu machines such as the msi gt70? Mt girl has a msi gt 70 and I'm thinking of flashing her 770m with the modded on but I don't know if it's the same procedure as the multi gpu m18x. Thanks guys

Same command. The only time you need to worry about using any other command is if the system has an on board nVidia card (old school M17xR1 for example) or cards that have two GPU cores and have a PLX as a result.

The default command flashes all nVidia cards found, one or more.

Link to comment
Share on other sites

Hi!

Just bought used MSI GT70 with GTX780M. And of course I want to unlock that VBIOS for some overclocking but not sure which bios to use?

I didnt buy after all, it was a fake dealer and almost lost my money on that, that was close.

So, i ordered xmg p304 from mysn.de, now waiting for that.

Link to comment
Share on other sites

Same command. The only time you need to worry about using any other command is if the system has an on board nVidia card (old school M17xR1 for example) or cards that have two GPU cores and have a PLX as a result.

The default command flashes all nVidia cards found, one or more.

Hey dude, take a look at my post #4422, please.

Link to comment
Share on other sites

  • Brian featured this topic

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use. We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.