Nvidia GTX Titan - modified VBIOS for more control and better overclocking

[KingOFSorrow]

yo yo svl7 can u make the OC SC version with 1006mhz GPU clock?

[aylan1196]

No idea how you managed to mess up like this, but reading your post you probably not the brightest person around. Don't blame me for your stupidity, I can guarantee you it has nothing to do with the vbios. If done properly recovery is a matter of minutes.

Take a break, we don't need disrespectful users here.

[svl7]

Sounds like an issue with MSI AB... not sure. Or maybe it's a setting in AB, just like in the Asus GPU tweak tool where you first need to set a checkbox in order to get to max voltage.

[KingOFSorrow]

Ok, here is another one (I really like this bios manages the clocks...): I have two cards and when I flash my second titan (in the slot 3) default voltage gets always increased by 12.5mv. On deafult Titan bios the default voltages on both cards are the same +0mv that is. Why is that happening? That, and I 'd like to repeat my request on custom gpu clocks on the SC bios? You may Pm me if that requires a fee or a donation. That is ok. I can't pm anyone because I don't have 5 posts yet.

[rage fuury]

Hi svl7, great work from your part it seems...Still, don't understand why the power limit was increased up to 350W...Could this card to suck more than 300W- is that posible? May be is more useful to set the Power Limit at 300W so people could calculate consume without making adjustments...

Best regards

[rage fuury]

@fommof :hey fo... your second Titan have a lower ASIC- that's why it boost at a lower freq with more voltage ... My non SC Titan (Asic 66%)stock bios boost to 980MHz with 1.162 V instead of 993MHz with 1.15V like (almost) everybody's else... So, it's normal behaviour, just that your first one was of better quality - too bad you had to RMA it

[booya]

Maybe it's a stupid question, but what if someone will run a few Titans @ 1.212V 24/7 at high load (up to 300 watts TDP for each card) but with water cooling? Is it safe? I think yes, but what about 1.25V? This voltage it's almost like a ghost - never seen.

[rage fuury]

@booya: 1.25V is not possible unless a hard-mode... 1.215V is the maximum voltage for Kepler (hard coded in the bios) and this value was unlocked in SLV's OC BIOS (with stock Titan bios you can go only up to only 1.2V with tools like EVGA Precision X, MSI Afterburner and NVidia Inspector); Titan had an artificial limit of max 250W from a consume/energy perspective - this bios move up this limit to 300W as far as I understand it... You should be OK

[Solar Trans]

Hey! I downloaded your BIOS today and flashed on my 1 TITAN and it was awesome! I would like to request that in the Superclocked edition one you raise the minimum voltage some, i.e. to what nVidias default stock is. I request this because MSI Afterburner is the only OC software out of all the ones I've tried that will keep your voltage tweak and apply it on startup, and with 875mV being the minimum it will only let me offset it by 100mV to 975mV. That is nowhere near the possible max on the card, and it would be awesome if you could raise the base voltage on the BIOS (If you really want it lower programs let you lower it past stock). I'm trying to run at a voltage offset of 187, and because Afterburner doesn't go that high (and no, it is not an adjustable setting like GPU Tweak) I have to manually do it in GPU Tweak on startup and go into AB for and OSD in games.Please make this adjustment I'd be very grateful! (or direct me to a method where I could do it myself)

[booya]

rage fuury

>1.25V is not possible unless a hard-mode.

Apparently yes But I read about it and some users had higher clocks with that kind of BIOS.

>with stock Titan bios you can go only up to only 1.2V with tools like EVGA Precision X, MSI Afterburner and NVidia Inspector

We can use KeplerBiosTweaker to unlock power limit, I did that, so the limit will be exactly like we set it.

But someone sad than this card (like some other NV cards) has hardware protection, so under Furmark and with even unlocked BIOS that protection will be activated when the card will try to consume... how many watts? 350? 400? I tryed to load them only at 300 watts (3DMark11, first test with medium preset). I'm afraid to use Furmark.

I also tryed "Nvidia GTX Titan - 'OC edition' VBIOS". It's OK, but I have 3-sli in surround mode and using EVGA Prec. for overclock. Situation:

Firefox is running (Adobe flash is working in a backbround) and the cards are at a higher frequency (about 700 MHz @ 0.862-0.887V - each card has a different voltage cause of dif. ASIC, I think so). We starting a game that loads the cards for about 70-80% each. And there is no any frequency and voltage increase, so it's like OK, but I can feel a little input lag, because anyway the cards a loaded pretty hard. I think it's a bug, but not a BIOS fault.

So now using this BIOS like before (mine) - manualy switching between two states - P8 (600/810 MHz @ 0.862-0.887V ) and P0 (1150/3004 MHz @ 1.212V). It`s not ideal but we can be sure that in 3D mode we will have all the power. Also the is no need for higher frequencies than ~600/800 MHz in most other situations (browsing, movies etc.). Only for folding, encoding etc. that will use CUDA.

[aylan1196]

hi svl7 any new version on the way ?? v.3 : )

[OccamsRazor]

rage fuury

>1.25V is not possible unless a hard-mode.

Apparently yes But I read about it and some users had higher clocks with that kind of BIOS.

>with stock Titan bios you can go only up to only 1.2V with tools like EVGA Precision X, MSI Afterburner and NVidia Inspector

We can use KeplerBiosTweaker to unlock power limit, I did that, so the limit will be exactly like we set it.

But someone sad than this card (like some other NV cards) has hardware protection, so under Furmark and with even unlocked BIOS that protection will be activated when the card will try to consume... how many watts? 350? 400? I tryed to load them only at 300 watts (3DMark11, first test with medium preset). I'm afraid to use Furmark.

I also tryed "Nvidia GTX Titan - 'OC edition' VBIOS". It's OK, but I have 3-sli in surround mode and using EVGA Prec. for overclock. Situation:

Firefox is running (Adobe flash is working in a backbround) and the cards are at a higher frequency (about 700 MHz @ 0.862-0.887V - each card has a different voltage cause of dif. ASIC, I think so). We starting a game that loads the cards for about 70-80% each. And there is no any frequency and voltage increase, so it's like OK, but I can feel a little input lag, because anyway the cards a loaded pretty hard. I think it's a bug, but not a BIOS fault.

So now using this BIOS like before (mine) - manualy switching between two states - P8 (600/810 MHz @ 0.862-0.887V ) and P0 (1150/3004 MHz @ 1.212V). It`s not ideal but we can be sure that in 3D mode we will have all the power. Also the is no need for higher frequencies than ~600/800 MHz in most other situations (browsing, movies etc.). Only for folding, encoding etc. that will use CUDA.

Hi Booya, what bios are you talking about? ive stumbled once on a engineering bios (made by TIN if im not mistaken) but i didnt have my titan then, but there was a warning that it could damage the card!

cheers

Ed

[fommof]

@rage fuury

@fommof :hey fo... your second Titan have a lower ASIC- that's why it boost at a lower freq with more voltage ...

Hey...the prob was not the fact that it was boosting lower than my first one and with higher voltage, that was expected since every chip is different. The issue was that using svl7's SC bios i was not able to set card at 1.137V like with my first one which is odd since it's like the card didn't "honor" the bios...strange but it's already RMA'd anyway, i am getting the (2nd) replacement today...

Still, don't understand why the power limit was increased up to 350W...Could this card to suck more than 300W- is that posible? May be is more useful to set the Power Limit at 300W so people could calculate consume without making adjustments...

I have seen peak reported TDP at 96% (1163Mhz, 1.1212V, +250 mem, #1 Titan) using the TI bios running Sleeping Dogs loops. The 96% in TI bios translates to 336W peak. I have also seen peek reported TDP at 98% running 3dMark13 Firestrike 1st test loops (don't have the gpu/mem/voltage figures handy at the moment, #2 Titan for sure) which translates to a whooping 343W peak.

While i can never be sure because of the lack of precision of my measurements, my killowatt readings seemed to follow the increment of the power consumption/TDP. When it was hitting +340W it almost reminded me my GTX580 at 950Μhz and 1.15V (all other setup exactly the same as then) so i'd say that these figures are not very far from reality.

IMHO of course and in my experience...

[fommof]

@svl7

Remember this?

Got my SC Titan replacement on Monday (RMAd for clicking/ticking fan noise) and noticed something interesting...

My "original" SC boosted at 1058Mhz using 1.15V, the replacement boosts at 1045Mhz at 1.162V. Nothing strange so far, different cards boost at different freqs and voltages.

The strange thing is that with my "original" Titan and the TI bios i had full voltage control all the way from 1.137V (default value at 928Mhz) to 1.212V. With the replacement card and the same bios i can't set the voltage at anything less than 1.162V and 1.162V is the default value for the same 928Mhz. Still i can set it anywhere from 1.162V to 1.212V.

Strange though, i thought the 1.137V was kind of "universal" base voltage in this bios...i uninstalled/installed the Precision...

I don't remember uninstalling/installing the drivers though but i can't see how this would fix it since i was going back and forth from the stock bios to TI in my first Titan without having reinstall the drivers. I just missed running the card at 928Mhz and 1.137V in not demanding games (like F1 2012 maxed) and keep the card cool and quiet.

Well, received the replacement of the replacement of my original Titan (lol, long story short, the #3 Titan).

Using your bios (SC V2) this one starts at 1.175V no matter what (and 928Mhz gpu freq as usual). The default max boost and voltage for this card with the stock bios is 1019Mhz at 1.15V.

I still don't get it...is the starting/base voltage of your bios independent from the quality of each GPU itself (i.e always 1.137V no matter what) or does each individual GPU dictates the starting/lowest voltage depending on the quality of the chip?

[cayman]

wonderful work. I'll try this on after I finished the sub-zero water cooling 2 way SLI rig. normal water cooling is just not enough for Titan. pull out a big fish tank , use it as a cooling pool filled with icy water. right now just have to get a waterproof motherboard done. when I finish the rig, will return the benchmark numbers with this bios mod, Thx a lot

[rage fuury]

@svl7

Remember this?

Well, received the replacement of the replacement of my original Titan (lol, long story short, the #3 Titan).

Using your bios (SC V2) this one starts at 1.175V no matter what (and 928Mhz gpu freq as usual). The default max boost and voltage for this card with the stock bios is 1019Mhz at 1.15V.

I still don't get it...is the starting/base voltage of your bios independent from the quality of each GPU itself (i.e always 1.137V no matter what) or does each individual GPU dictates the starting/lowest voltage depending on the quality of the chip?

[rage fuury]

well, fommmof, I stilldon't get it how the card could suck more than 150W (8 pins power)+75W (6pins power)+75W (PCIExp.)=300W ...

[fommof]

well, fommmof, I stilldon't get it how the card could suck more than 150W (8 pins power)+75W (6pins power)+75W (PCIExp.)=300W ...

The figures you posted are theoretical. You think in practice that 8 and 6 pin can't provide any more juice? What limits them? The 6/8pin wires themselves? Think about it...

Unfortunately i am not able to measure the consumption of the vga alone, i can only measure the the consumption of the whole system and find the relative difference. Not an accurate methodology, works way better for the cpu alone,.

Still if you think backwards and take for granted that the absolute limit is the 300W (which i disagree with unless there is a hardware limiter at 300W in the card itself, which again if there is i haven't hit it) then something is very wrong about the TDP readings. If you think that's the case then do the following experiment:

1)Flash your card with the stock bios

2)Set the power to 106% (265W), run an app that does NOT power throttle the card (you might want to set the gpu offset to -100), run it until your gpu temperature is stabilized then check the max TDP reading (lets say that is 95% which translates to 240W)

3)Flash your card with the TI bios

4)Leave the power to 100% (350W), run the same app until your gpu temperature is stabilized then check the max TDP reading.

5)If the actual max reported TDPs are close and both BELOW 265W then chances are the readings are fine. If for instance you got 69% max TDP with the TI bios then that translates to 241.5W which agrees with the reading you got with the stock bios, it's almost dead on so readings look fine. Now If for instance TI bios reports 80% which translates to 280W then you can't trust the readings.

My personal experience is that the readings are close (EDIT: except on idle, with the stock bios 2 out of my three Titans were reporting idle power which doesn't make any sense and this i can easily check with a wattometer and compare my system idle with and without the card by using the igpu), but you can try and find it out yourself if you like.

[MADman]

Hi,

first off all, big thanks for this great work, this BIOS is really great

But maybe you can help me with a problem. I have a 2-way Titan SLI which runs fine with Stock BIOS OCed at 1163 MHz with 1,212V and 106% Powertarget (done with Precision X). One notch more an it doesn't run stable anymore. With Aircooling I was at 74°C at this point, so I expected to have reached the Power Target since there was enough room at temp target. The Memory runs at Stock clock. If I do the slightest overclocking of the memory it won't run stable anymore at 1163 MHz GPU (which is stable without Mem OC). Another hint that I have reached the Power target. Ok, as far as good...now the next step

Then I applied this great BIOS, which have may advantages for my setup, but the cards are still behaving the same. This point I don't quite understand because I should have much more room for OC now because of the increased Power Target in this BIOS (if not at the GPU, so at least at the Memory I thought). That seems strange to me and I was not expecting this.

Then I applied Watercooling, now the Cards ran with the same settings at 40°C...but the cards are still behaving exact the same as before. With a greater Power Target I should be able to do some Memory OC, even if the GPU were at its limit. But this isn't the case. If I put down the GPU clock, I can do some Memory OC (again a hint for a Power Target Problem).

Can somebody give me an explanation for this observations, maybe I overlooked an important point. Or misunderstood something

I have verified, that both cards are correctly flashed with the modded BIOS and I use the 320.49 Driver atm.

It would be great if sombody has an idea, why my cards dont want to move just a little step farther, even with watercoolung and 350W Power Target.

Thanks in advance.

[Brian]

Hi,

first off all, big thanks for this great work, this BIOS is really great

But maybe you can help me with a problem. I have a 2-way Titan SLI which runs fine with Stock BIOS OCed at 1163 MHz with 1,212V and 106% Powertarget (done with Precision X). One notch more an it doesn't run stable anymore. With Aircooling I was at 74°C at this point, so I expected to have reached the Power Target since there was enough room at temp target. The Memory runs at Stock clock. If I do the slightest overclocking of the memory it won't run stable anymore at 1163 MHz GPU (which is stable without Mem OC). Another hint that I have reached the Power target. Ok, as far as good...now the next step

Then I applied this great BIOS, which have may advantages for my setup, but the cards are still behaving the same. This point I don't quite understand because I should have much more room for OC now because of the increased Power Target in this BIOS (if not at the GPU, so at least at the Memory I thought). That seems strange to me and I was not expecting this.

Then I applied Watercooling, now the Cards ran with the same settings at 40°C...but the cards are still behaving exact the same as before. With a greater Power Target I should be able to do some Memory OC, even if the GPU were at its limit. But this isn't the case. If I put down the GPU clock, I can do some Memory OC (again a hint for a Power Target Problem).

Can somebody give me an explanation for this observations, maybe I overlooked an important point. Or misunderstood something

I have verified, that both cards are correctly flashed with the modded BIOS and I use the 320.49 Driver atm.

It would be great if sombody has an idea, why my cards dont want to move just a little step farther, even with watercoolung and 350W Power Target.

Thanks in advance.

Did you also raise your voltage to max? I didn't see you mention that in your post. Also keep in mind that with this bios, your frequency is now pegged at 1163 MHz during 3D operations unlike the stock bios where it adjusts based on power targets/tdp.

Edit: nevermind, you did mention the voltage.

[MADman]

Did you also raise your voltage to max? I didn't see you mention that in your post. Also keep in mind that with this bios, your frequency is now pegged at 1163 MHz during 3D operations unlike the stock bios where it adjusts based on power targets/tdp.

Yes, the Voltage is always at 1,212V when I test. I double-check that everytime

[Brian]

Yes, the Voltage is always at 1,212V when I test. I double-check that everytime

If that's the case then you may just have a poor overclocking and leaky card. A hard mod to raise the voltage past 1.212v may be the only thing you can do. Also I'm blind and tired, I see that you did mention the voltage. Alternatively you could set 1150 as your core and then raise the memory. The difference in performance between a card that does 1200 core vs 1150 mhz is at most 1-2 fps.

[MADman]

Yes, that would be an idea. But my main problem is that the max clockability of my cards does not change even after changing so much external parameters like power target and cooling. That my cards have exactly the same clockability with 350W Power Target and Watercooling like with Stock BIOS and Stock Air cooling is still confusing me. I did not expect a major improvement but no change at all seems quite strange to me.

But even if there is no solution for this "mystery", I'm still very happy with this BIOS, because I have much more control over my clock rates as before

And I apprecciate you help, of course Thank you very much.

[Arnoud van Lieshout]

Can someone help me? Run at 1150/+400 @ titan > EVGA precision 4.1 / SC bios from here.

It throttle back to 930mhz if i tunr of kboost.

Problem is when i turn Kboost on it idles 65 degrees!!!!!!

help...

appreciated.

//

figgered out how to solve it... Now i use Nvidia Inspector and works super!!

EVGA and afterburner just dont cut it ..... and throttle :S. its redicolous they cannot make a proper piece of software. because Inspector keeps ur clocks temps and voltage nice low on idle...

i donated 5 dollar;)

[lonethrasher]

Glad I read your post. Was getting ready to dl precision and got Nvidia Inspector instead. What temps are you running at under load?