Jump to content

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


Recommended Posts

I just recently started dabbling again. Can someone tell me what the max "safe" voltage is for 780m (in sli if that matters)? I am having trouble going past +100 and +250. Temperatures are fine, but I get black screen/driver crashes at anything beyond that. Would that be a voltage issue? Currently at stock volts using stock drivers - with SVL7's wonderful bios!

Thanks for your help!

Link to comment
Share on other sites

I just recently started dabbling again. Can someone tell me what the max "safe" voltage is for 780m (in sli if that matters)? I am having trouble going past +100 and +250. Temperatures are fine, but I get black screen/driver crashes at anything beyond that. Would that be a voltage issue? Currently at stock volts using stock drivers - with SVL7's wonderful bios!

Thanks for your help!

I don't have a 780M, but know a little about it. The stock voltage is 1.00V at 850Mhz isn't it? I was going to say don't go any higher than 1.05V, which is what I run my card at, but my card uses close to 100W at 1.05V, so your 780M at 1.05V would probably use close to 1.5 times more Watts than mine based on the number of shaders your card has. So, if yours was at 1.05V, then I estimate each card would use about 135W. MXM slots are supposedly rated for 100W, but I don't know how dangerous (if at all) it is to push them above the 100W ceiling for extended periods. I've seen people bench 780M cards at up to 1.1V, so they would be using over 150W per card, but that's only for a short time. I should think 1.025V would be OK for extended use provided temperatures are good, you definitely want less than 90 degC, and ideally less than 80degC. This is just my opinion though, based on the things I've learned & read, so it's your judgement call.

Link to comment
Share on other sites

I don't have a 780M, but know a little about it. The stock voltage is 1.00V at 850Mhz isn't it? I was going to say don't go any higher than 1.05V, which is what I run my card at, but my card uses close to 100W at 1.05V, so your 780M at 1.05V would probably use close to 1.5 times more Watts than mine based on the number of shaders your card has. So, if yours was at 1.05V, then I estimate each card would use about 135W. MXM slots are supposedly rated for 100W, but I don't know how dangerous (if at all) it is to push them above the 100W ceiling for extended periods. I've seen people bench 780M cards at up to 1.1V, so they would be using over 150W per card, but that's only for a short time. I should think 1.025V would be OK for extended use provided temperatures are good, you definitely want less than 90 degC, and ideally less than 80degC. This is just my opinion though, based on the things I've learned & read, so it's your judgement call.

Thanks for your response, Robbo!

Yes, you are right on the stock numbers. I want to give it a little nudge for gaming - I have read through a lot of threads of benching but not many really give good insight on what they use for gaming situations in terms of overclock or higher voltage. Perhaps someone can point me at what they run? I know the whole "no 2 chips are the same" thing applies...I just want a realistic goal that CAN be ran for longer periods of time. Can someone point me in the right direction? Thanks

Link to comment
Share on other sites

Thanks for your response, Robbo!

Yes, you are right on the stock numbers. I want to give it a little nudge for gaming - I have read through a lot of threads of benching but not many really give good insight on what they use for gaming situations in terms of overclock or higher voltage. Perhaps someone can point me at what they run? I know the whole "no 2 chips are the same" thing applies...I just want a realistic goal that CAN be ran for longer periods of time. Can someone point me in the right direction? Thanks

Maybe someone else can give you some ballpark overclocks for that. In the meantime why not just try upping the voltage a notch or two & then seeing how high you can push the clocks without crashing or artifacting, while keeping an eye on temperatures. Then you could post back here saying what your stable overclocks were for gaming, which will help others like yourself when they read this thread.

Link to comment
Share on other sites

Maybe someone else can give you some ballpark overclocks for that. In the meantime why not just try upping the voltage a notch or two & then seeing how high you can push the clocks without crashing or artifacting, while keeping an eye on temperatures. Then you could post back here saying what your stable overclocks were for gaming, which will help others like yourself when they read this thread.

Well, I am a bit of a newbie at this. Last night anytime I raised it over +130/+300 - even bumped voltage to 1.025 - heaven crashed/black screened (ie 135/300) I just know that even for synthetic benches a lot got higher than that. So I am concerned I am doing something wrong - like I am not sure at what point I should start tinkering with voltage. When it was crashing, I was at around 68-70C. I know temperature isn't everything but it just was odd to me that I have seen numbers of like +200/+600 for benching and I can't even do that. Again, I know no two will oc the same - it just seems kind of like a huge margin. That is why I was looking for insight.

Also, I was using EVGA Precision. Not sure if that will make much of a difference.

Link to comment
Share on other sites

I realize I'm slightly late to this party, but I have an MSI gt60 with the 680m, and i've been reading through this thread getting tips on the best way to flash the vbios. Can anyone who has done this before with my card tell me whats best to start off with voltage wise?

Link to comment
Share on other sites

Well, I am a bit of a newbie at this. Last night anytime I raised it over +130/+300 - even bumped voltage to 1.025 - heaven crashed/black screened (ie 135/300) I just know that even for synthetic benches a lot got higher than that. So I am concerned I am doing something wrong - like I am not sure at what point I should start tinkering with voltage. When it was crashing, I was at around 68-70C. I know temperature isn't everything but it just was odd to me that I have seen numbers of like +200/+600 for benching and I can't even do that. Again, I know no two will oc the same - it just seems kind of like a huge margin. That is why I was looking for insight.

Also, I was using EVGA Precision. Not sure if that will make much of a difference.

Hi, ok, here's a description of a good approach & methodology to overclocking, which I've cut & pasted from a previous post of mine (so didn't have to type it all out again!):

1)Determine your max stable overclock of the core at stock voltage. Increase Core Clock by 100Mhz, run 3DMark11 through till end, if no visual artifacts or crashes and if temperatures are ok then increase core clock by a further 50Mhz and repeat process. Keep repeating this process until you see artifacts or crashes, at which point back down to your previous stable overclock & do further more vigorous stress testing. Best way to do this is run a game that pushes the GPU to a constant 100% GPU utilisation (GPU load in GPUz) - eg TombRaider, Far Cry 3 are good for this - if the game is stable for 1 hour you've probably found a stable core overclock.

2)Now determine your Max Stable Memory Overclock. Leave your core at your maximum stable overclock for this process. Increase Memory Clockby 200Mhz, and do the same testing procedure as above using 3DMark11to work out an initial stable max overclock. You can increase the memory clock in bigger chunks in NVidiaInspector, because there's a peculiarity with GDDR5 memory in NVidia Inspector whereby a 200Mhz increase is actually equal to a real 100Mhz increase- it's do with a very technical fact that I don't fully understandwith GDDR5 being 'quad-pumped' (4 times faster than DDR3 at any given frequency) - (you'll see evidence of what I'm talking about when you view your memory clock in GPUz, where it displays the REAL memory clock). Anyway, in NVidia Inspector, increase the memory by 200Mhz the first time you test, then in 100Mhz chunks thereafter. Each time you complete the 3DMark11 test view the GPU score, if it's not increasing anymore as you raise the memory clock, then stop your memory overclocking where it is. This is because GDDR5 has memory error correction. As memory overclock increases the rate at which errors occur outpace the rate of error correction, thereby resulting in a lower or not increased GPU score, so overclocking the memory beyond that point is futile & only serving to greater stress your card. Once you've reached your max stable overclock in 3DMark11 then do that 1 hour of gaming like your did for the core clock (make sure you have your core at your max overclock when you do this too). If it's stable, then you've now reached your max overclock for both the core & memory at stock voltage.

That above process is the same if you decide to increase the voltage, but the voltage only affects the GPU core, so don't bother tweaking the memory overclock if you raise the voltage, because by raising the voltage you're not supplying anymore voltage to the VRAM. So, if you overvolt, leave the memory at your max stable Mhz, and then start tweaking the GPU core Mhz in the same way I described in the steps above, but in smaller increments (not 100Mhz increments).

Use NVidia Inspector to overclock, I believe svl7 has specifically designed his vBIOS to work in tandem with NVidia Inspector. (I also think it's the best overclocking utility in my experience).

  • Thumbs Up 1
Link to comment
Share on other sites

I realize I'm slightly late to this party, but I have an MSI gt60 with the 680m, and i've been reading through this thread getting tips on the best way to flash the vbios. Can anyone who has done this before with my card tell me whats best to start off with voltage wise?

I think the lowest voltage one, and then if temperatures are good, and you want/need more performance then flash the next voltage up & repeat.

Link to comment
Share on other sites

Just a question: the OC possibilities are predetermined since the beginning depending on the quality of the chip, or using a good thermal paste and dissipation system it's always possible to achieve good results?

Potential OC abilities are mostly & largely predetermined by the quality of the chip, although cool running chips are more stable than hotter ones, so if you've got a good cooling system with a chip running at 65 degC then this might reach a higher stable OC than the same chip on a cooling system at 85 degC.

Link to comment
Share on other sites

  • Bios Modder
I flash my card 780m but it didn't work. card was turned to be unknown and cannot use

my laptop is cyberpower fang

Are you make backup of vbios from your card before flashing?

Link to comment
Share on other sites

I don't know if you are interested, but I have a MSI GT60 here with the 880m in it if you need a vbios dump from it.

I only gave it a +600 core clock limit tweakable with msi afterburner, nvidia inspector etc. Tell me if you need anything else like increased voltage, disabled boost or whatnot.

880m gt60 backup modified.zip

Link to comment
Share on other sites

  • Bios Modder
I only gave it a +600 core clock limit tweakable with msi afterburner, nvidia inspector etc. Tell me if you need anything else like increased voltage, disabled boost or whatnot.

[ATTACH]11220[/ATTACH]

+600 limit! For 880? Why? Are you seen the stock core clock in GTX 880M? Are you read this page: Nvidia GTX 880m review

Modification of vbios for 780 and 880 is not simple. It is very difficult and requires very accurate selection of combinations of voltage, Power Limit, Power Target and core frequency.

Link to comment
Share on other sites

+600 limit! For 880? Why? Are you seen the stock core clock in GTX 880M? Are you read this page: Nvidia GTX 880m review

Modification of vbios for 780 and 880 is not simple. It is very difficult and requires very accurate selection of combinations of voltage, Power Limit, Power Target and core frequency.

Yep, you're right. I just gave it a high limit in case he wants to experiment. I will be working on the power target and voltage later since I was in a rush to go eat lunch. It would also help if I had a tester. I am just changing one thing at a time because I don't want to screw up and destroy anything with an unfamiliar card :)

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.