Overclock.net › Forums › Graphics Cards › NVIDIA › How To Flash A Different BIOS On Your 1080 Ti.
New Posts  All Forums:Forum Nav:

How To Flash A Different BIOS On Your 1080 Ti. - Page 21

post #201 of 326
To me it seems that for those that are Shunt modded, the best BIOS' would be

"Palit GTX 1080Ti GameRock" Version 86.02.39.00.77 " (300Target 350Max)
https://www.techpowerup.com/vgabios/191499/palit-gtx1080ti-11264-170331



I haven't tried it but based on Max wattage am leaning to trying these.


Thoughts?
Edited by KraxKill - 4/30/17 at 10:48am
post #202 of 326
Quote:
Originally Posted by KraxKill View Post

To me it seems that for those that are Shunt modded, the best BIOS' would be

"Palit GTX 1080Ti GameRock" Version 86.02.39.00.77 " (300Target 350Max)
https://www.techpowerup.com/vgabios/191499/palit-gtx1080ti-11264-170331



I haven't tried it but based on Max wattage am leaning to trying these.


Thoughts?

YMMV, but do a search on palit bios and see what other users are saying about it in this thread and the 1080 Ti thread.

You're trading more TDP for worse performance. In the end, on timespy and firestrike you might just end up scoring the same but draw more power, more heat, etc.

2100 Mhz, 6055 Memory



This is consistent for me with everything.


Above is stock bios.

For me, I asked myself. For timespy and firestrike in particular, is it worth getting %1-2% worse performance in all the circled areas in purple just so I can get 1%-2% more performance in green? Probably not.

In regards to playing video games, I never ever hit any power limits, not the normalized one. So why bother taking that 1%-2% hit?

I guess if you want to experiment with it and see if it gets you 1% more in a benchmark so you can go up in the rankings roster, then it'll be worth toying with definitely.
Edited by SlimJ87D - 4/30/17 at 11:34am
post #203 of 326
Quote:
Originally Posted by SlimJ87D View Post

YMMV, but do a search on palit bios and see what other users are saying about it in this thread and the 1080 Ti thread.

You're trading more TDP for worse performance. In the end, on timespy and firestrike you might just end up scoring the same but draw more power, more heat, etc.

2100 Mhz, 6055 Memory



This is consistent for me with everything.


Above is stock bios.

For me, I asked myself. For timespy and firestrike in particular, is it worth getting %1-2% worse performance in all the circled areas in purple just so I can get 1%-2% more performance in green? Probably not.

In regards to playing video games, I never ever hit any power limits, not the normalized one. So why bother taking that 1%-2% hit?

I guess if you want to experiment with it and see if it gets you 1% more in a benchmark so you can go up in the rankings roster, then it'll be worth toying with definitely.


I disagree and my post here shows otherwise.

http://www.overclock.net/t/1624521/nvidia-gtx-1080-ti-owners-club/7830#post_26059668

I don't think there is any performance difference between any of the BIOS bin for bin volt for volt. The difference is the power draw and if you hit pwr your card is missing on power it wants to draw. There is no magic here. That's wasted performance and induces frame judder as a result of trying to stay under the limit.

I would only suggest this for people that can hit 2126 an up becuase you are right games will not always make you draw that power and hit pwr but I haven't seen any evidence to suggest it performed lower or better than the FE bios. Mine certainly hits right where the FE bios hits at lower volt bins and higher at volt bins that used to induce pwr.

I measured my performance with induced pwr and the bios made that go away and the scores went up. At lower bins. Where the draw is under the limit of both BIOS' there was no measurable difference but in instances where the pwr limit is induced, it's certainly a measurable improvement.
Edited by KraxKill - 4/30/17 at 7:16pm
post #204 of 326
Quote:
Originally Posted by KraxKill View Post

I disagree and my post here shows otherwise.

http://www.overclock.net/t/1624521/nvidia-gtx-1080-ti-owners-club/7830#post_26059668

I don't think there is any performance difference between any of the BIOS bin for bin volt for volt. The difference is the power draw and if you hit pwr your card is missing on power it wants to draw. There is no magic here. That's wasted performance and induces frame judder as a result of trying to stay under the limit.

I would only suggest this for people that can hit 2126 an up becuase you are right games will not always make you draw that power and hit pwr but I haven't seen any evidence to suggest it performed lower or better than the FE bios. Mine certainly hits right where the FE bios hits at lower volt bins and higher at volt bins that used to induce pwr.

I measured my performance with induced pwr and the bios made that go away and the scores went up. At lower bins. Where the draw is under the limit of both BIOS' there was no measurable difference but in instances where the pwr limit is induced, it's certainly a measurable improvement.

I disagree wink.gif

I have a Shunt moded 1080Ti FE, i flashed my card with the Palit BIOS you mentioned, cuz i wanted to check if i can get rid of the power limit which i have in Time Spy game test 2.
I can confirm that my card power limits a bit less than before( with Palit bios just 2 times a dip for 1 second or so) but i can also see that im getting less average FPS with that bios in comparison to stock FE bios.

So 2100MHz Core and +850MHz on the Vram is giving me 65.80 +/- 0.05 @ 1062mV and Stock BIOS
Palit BIOS is giving me with exact the same settings 65.30 +/- 0.05
Thats 0.5 FPS difference in Time Spy Game Test 2
Even with a higher Bin (2113MHz) i can't reach the Stock BIOS performance, averging at 65,50 +/- 0.05 than.

So i would definetly say there is a performance difference between those BIOS versions.
Quote:
Originally Posted by KedarWolf View Post


For bench tests Asus seems the way to go.

Which ASUS do you mean the FE one or the STRIX?

I flashed the STRIX one on my FE card and couldn't get a display signal on Displayport, had to reflash other bios with iGPU

Edit: maybe worth to mention: If you not DDU after flash reboot and Windows is automatically installing NVdia drivers you pretty much excatly lose 0.8 FPS in Time Spy Game Test 2 ( at least in my case). The Point is its MANDATORY to DDU your drivers after flashing and rebooting.
Edited by Hulio225 - 5/1/17 at 8:16am
    
CPUMotherboardGraphicsRAM
i7 7700k Asus Maximus IX Apex 1080 Ti Founders Edition G.Skill DDR4 B-Die 
Hard DriveHard DriveCoolingOS
Samsung Evo 960 Nvme Western Digital RED Custom Loop Windows 10 
MonitorKeyboardPowerCase
Samsung 4K Logitech G15 Fractal Design Tesla R2 650W Enthoo Primo 
MouseAudioAudioAudio
Logitech G500 Asus Xonar STX Beyerdynamic DT 1990 PRO Headphones Blue Snowball Mic (Alu Edition) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7 7700k Asus Maximus IX Apex 1080 Ti Founders Edition G.Skill DDR4 B-Die 
Hard DriveHard DriveCoolingOS
Samsung Evo 960 Nvme Western Digital RED Custom Loop Windows 10 
MonitorKeyboardPowerCase
Samsung 4K Logitech G15 Fractal Design Tesla R2 650W Enthoo Primo 
MouseAudioAudioAudio
Logitech G500 Asus Xonar STX Beyerdynamic DT 1990 PRO Headphones Blue Snowball Mic (Alu Edition) 
  hide details  
Reply
post #205 of 326
Quote:
Originally Posted by Hulio225 View Post

I disagree wink.gif

I have a Shunt moded 1080Ti FE, i flashed my card with the Palit BIOS you mentioned, cuz i wanted to check if i can get rid of the power limit which i have in Time Spy game test 2.
I can confirm that my card power limits a bit less than before( with Palit bios just 2 times a dip for 1 second or so) but i can also see that im getting less average FPS with that bios in comparison to stock FE bios.

So 2100MHz Core and +850MHz on the Vram is giving me 65.80 +/- 0.05 @ 1062mV and Stock BIOS
Palit BIOS is giving me with exact the same settings 65.30 +/- 0.05
Thats 0.5 FPS difference in Time Spy Game Test 2
Even with a higher Bin (2113MHz) i can't reach the Stock BIOS performance, averging at 65,50 +/- 0.05 than.

So i would definetly say there is a performance difference between those BIOS versions.
Which ASUS do you mean the FE one or the STRIX?

I flashed the STRIX one on my FE card and couldn't get a display signal on Displayport, had to reflash other bios with iGPU

Edit: maybe worth to mention: If you not DDU after flash reboot and Windows is automatically installing NVdia drivers you pretty much excatly lose 0.8 FPS in Time Spy Game Test 2 ( at least in my case). The Point is its MANDATORY to DDU your drivers after flashing and rebooting.

Are you seriously trying to say that 1/3 or .03fps is a measurable difference? We're talking well within the margin of error for the bench itself. Your services running at the time and the **** windows decides to do in the background during your test. Not to mention temperature influenced IPC performance and everything else involved. Simply running a test on fresh boot vs a hour after boot will post different results.

If you really want to test do this. Downclock your card to stock so you're not running at the limit. Limit the voltage for a reasonable bin below your stock boost clock using the v curve. Make sure that both bios hit the same clock and Vbin and you have no chance of pwr limiting at all. As in a blank GPUz graph. Then test the bios' Make sure that your load temps are within 2C of each other between runs as momentary temp throttled will affect your score hence selecting a lower frequency an Volts that make this less probable. Make sure your card, coolant if you are on water and radiator are desaturated completely between your tests.

I don't understand the physical instructions involved, unless somebody can suggest otherwise a reason one BIOS would score higher than another "bin for bin" "voltage for voltage". That is litterely all it's doing. Picking a vbin and feeding it to the card at a particular clock. I see no way for one BIOS (unless you are experiencing environmental" throttling to affect the cards performance.

What exactly else do you think the BIOS is doing that allows for variations?

Your reduced performance at the higher bin could be anything from error correction to memory distanilizarion at the higher bin etc.

Volts in at equal hz is equal W consumed. There is nothing extra special beyond that that the gpu understands. If you get your card to consume more wattage at your top bin without hitting temp throttles ofcorse you are gaining performance capacity it's just science.

Again 2000 at 1.05v is 2000 at 1.05v regardless of what BIOS is handing out that instruction. What you are suggesting is that X on BIOS 1 does not equal X on BIOS 2.

It's going to take a lot more splainin than .03fps to illustrate this. I haven't seen a true valid comparison by anybody yet. And I illustrated with my card hitting PWR vs not hitting PWR my that scores improved and I began to scale favorably at higher bins. Which makes logical sense as I get an extra 30watts of power to play with.

Keep in mind if you can't dissipate 30w of extra heat in that FE cooler, you will loose performance due to extra heat but if you can keep below or within a single Boost 3.0 temp throttle points on both the Palit bios and the FE bios you will gain performance capacity. There is just no other way around this.

In order for what you say to be true, the Palit Bios would have to be eating 30w as a snack making it 10% less efficient at the same voltage a variable so ridiculous given the basics of what a bios does that I can't think of a way for this to be true. And for it to loose performance, it would have to be even less efficient.
Edited by KraxKill - 5/1/17 at 9:21am
post #206 of 326
Quote:
Originally Posted by KraxKill View Post

Are you seriously trying to say that 1/3 or .03fps is a measurable difference? We're talking well within the margin of error for the bench itself. Your services running at the time and the **** windows decides to do in the background during your test. Not to mention temperature influenced IPC performance and everything else involved. Simply running a test on fresh boot vs a hour after boot will post different results.

If you really want to test do this. Downclock your card to stock so you're not running at the limit. Limit the voltage for a reasonable bin below your stock boost clock using the v curve. Make sure that both bios hit the same clock and Vbin and you have no chance of pwr limiting at all. As in a blank GPUz graph. Then test the bios' Make sure that your load temps are within 2C of each other between runs as momentary temp throttled will affect your score hence selecting a lower frequency an Volts that make this less probable. Make sure your card, coolant if you are on water and radiator are desaturated completely between your tests.

I don't understand the physical instructions involved, unless somebody can suggest otherwise a reason one BIOS would score higher than another "bin for bin" "voltage for voltage". That is litterely all it's doing. Picking a vbin and feeding it to the card at a particular clock. I see no way for one BIOS (unless you are experiencing environmental" throttling to affect the cards performance.

What exactly else do you think the BIOS is doing that allows for variations?

Have you raised your thermal target?

I find when I put like a 90 degree thermal target that is when you notice the difference. If you don't do that, your BIOS probably is thermal limiting prior to voltage limiting.
post #207 of 326
Quote:
Originally Posted by KraxKill View Post

I disagree and my post here shows otherwise.

http://www.overclock.net/t/1624521/nvidia-gtx-1080-ti-owners-club/7830#post_26059668

I don't think there is any performance difference between any of the BIOS bin for bin volt for volt. The difference is the power draw and if you hit pwr your card is missing on power it wants to draw. There is no magic here. That's wasted performance and induces frame judder as a result of trying to stay under the limit.

I would only suggest this for people that can hit 2126 an up becuase you are right games will not always make you draw that power and hit pwr but I haven't seen any evidence to suggest it performed lower or better than the FE bios. Mine certainly hits right where the FE bios hits at lower volt bins and higher at volt bins that used to induce pwr.

I measured my performance with induced pwr and the bios made that go away and the scores went up. At lower bins. Where the draw is under the limit of both BIOS' there was no measurable difference but in instances where the pwr limit is induced, it's certainly a measurable improvement.

Personally, I'm getting worse performance in actual game benchmarks like Rise of the Tomb Raider. At the end of the day, I use this card to play video games. I know I've used it mostly for benchmarking as of late, but I've since stopped chasing numbers to a certain extent. Only doing research to verify things for the community.

My scores in video games matter more to me. If the Palit bios scores worse in most gaming scenarios but gives me slightly better scores if and when I would be power limited, it's just not worth it to me. With the shunt mod, I am yet to be power limited while gaming, ever.

I can do 2152 Mhz at 1.075v in a superposition benchmark without power limiting. It'll limit in Timespy and Firestrike and just for a tiny bit, literally like 2 tiny segments. But I have yet to play a game that pushing my card anywhere close to 100% where 105% for my card is where the TDP normalized is hit. At most I've seen it hit 95% and I still have 10% more headroom for a game to hit the normalized TDP limit.

So if I'm speaking to someone that's reading this thread that just wants to game. I recommend testing the Palit bios in actual game benchmarks like Rise of the Tomb Raider, Batman Arkham Knight, GTAV, etc. Test the stock and palit bios thoroughly in realistic scenarios in what you would experience while gaming and go with what scores better.

If benchmarking is a hobby of yours, I recommend doing the above but with your benchmark software of your choice.

These are my recommendations.
Edited by SlimJ87D - 5/1/17 at 9:05am
post #208 of 326
Quote:
Originally Posted by KraxKill View Post

Are you seriously trying to say that 1/3 or .03fps is a measurable difference? We're talking well within the margin of error for the bench itself. Your services running at the time and the **** windows decides to do in the background during your test. Not to mention temperature influenced IPC performance and everything else involved. Simply running a test on fresh boot vs a hour after boot will post different results.

If you really want to test do this. Downclock your card to stock so you're not running at the limit. Limit the voltage for a reasonable bin below your stock boost clock using the v curve. Make sure that both bios hit the same clock and Vbin and you have no chance of pwr limiting at all. As in a blank GPUz graph. Then test the bios' Make sure that your load temps are within 2C of each other between runs as momentary temp throttled will affect your score hence selecting a lower frequency an Volts that make this less probable. Make sure your card, coolant if you are on water and radiator are desaturated completely between your tests.

I don't understand the physical instructions involved, unless somebody can suggest otherwise a reason one BIOS would score higher than another "bin for bin" "voltage for voltage". That is litterely all it's doing. Picking a vbin and feeding it to the card at a particular clock. I see no way for one BIOS (unless you are experiencing environmental" throttling to affect the cards performance.

What exactly else do you think the BIOS is doing that allows for variations?

I've spent hours, literally hours and hours comparing ASUS, Stock, MSI bios weeks ago. I literally compared 2032, 2050, 2062, 2088, 2101, 2114 Mhz and I varied the voltages from 1.042v to 1.093v, ever voltage inbetween.

I have a entire folder filled with benchmark results, I can take a screenshot when I get home if you want me to. But I spent hours and hours doing this research (of course it wasn't a total waste, I was working while the benchmarks ran), and I can't telly ou why because I don't write bios, but I can show you all my results. The stock bios just ran better, the closes bios to having similar to consistent results was the SC2 bios, but the stock bios does slightly better than it.

Here's just some of them.



These are my findings, if you have data, techniques or methods to make the Palit bios perform better than the stock bios then please share them. It only benefits us all.

But read through this thread and the other thread, other people have said the Palit bios gave them worse performance.
post #209 of 326
Quote:
Originally Posted by KraxKill View Post

Are you seriously trying to say that 1/3 or .03fps is a measurable difference? We're talking well within the margin of error for the bench itself. Your services running at the time and the **** windows decides to do in the background during your test. Not to mention temperature influenced IPC performance and everything else involved. Simply running a test on fresh boot vs a hour after boot will post different results.
?

Have you even read my numbers properly?

10 runs with palit bios in range of 65.30 (+/- 0.05) <- this is the range in wich each run is with palit run so min 65.25 and max 65.35
10 runs with stock bios in range of 65.80 (+/- 0.05) <- this is the range in wich each run with stock bios is so min 65.75 and max 65.85

this is repeatable all the time and thats not within the margin of error anymore, its a clear difference!!!

that is a difference of 0.5 FPS repeatable and reporducable all the time

Edit: In addition you are getting worse scores with a newer driver than before.... im getting better scores across the board with the new nvidia driver (time spy, fire strike etc.)... go figure
Edited by Hulio225 - 5/1/17 at 9:18am
    
CPUMotherboardGraphicsRAM
i7 7700k Asus Maximus IX Apex 1080 Ti Founders Edition G.Skill DDR4 B-Die 
Hard DriveHard DriveCoolingOS
Samsung Evo 960 Nvme Western Digital RED Custom Loop Windows 10 
MonitorKeyboardPowerCase
Samsung 4K Logitech G15 Fractal Design Tesla R2 650W Enthoo Primo 
MouseAudioAudioAudio
Logitech G500 Asus Xonar STX Beyerdynamic DT 1990 PRO Headphones Blue Snowball Mic (Alu Edition) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7 7700k Asus Maximus IX Apex 1080 Ti Founders Edition G.Skill DDR4 B-Die 
Hard DriveHard DriveCoolingOS
Samsung Evo 960 Nvme Western Digital RED Custom Loop Windows 10 
MonitorKeyboardPowerCase
Samsung 4K Logitech G15 Fractal Design Tesla R2 650W Enthoo Primo 
MouseAudioAudioAudio
Logitech G500 Asus Xonar STX Beyerdynamic DT 1990 PRO Headphones Blue Snowball Mic (Alu Edition) 
  hide details  
Reply
post #210 of 326
Quote:
Originally Posted by Hulio225 View Post

Have you even read my numbers properly?

10 runs with palit bios in range of 65.30 (+/- 0.05) <- this is the range in wich each run is with palit run so min 65.25 and max 65.35
10 runs with stock bios in range of 65.80 (+/- 0.05) <- this is the range in wich each run with stock bios is so min 65.75 and max 65.85

this is repeatable all the time and thats not within the margin of error anymore, its a clear difference!!!

that is a difference of 0.5 FPS repeatable and reporducable all the time

Edit: In addition you are getting worse scores with a newer driver than before.... im getting better scores across the board with the new nvidia driver (time spy, fire strike etc.)... go figure

Did you ever run game benchmarks at all? Like Rise of the Tomb Raider, GTAV, Batman Arkham Knight, etc?

As you, I have found the Palit bios to score worse. But kudos and +REP for doing 20 test! That is dedication.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA
Overclock.net › Forums › Graphics Cards › NVIDIA › How To Flash A Different BIOS On Your 1080 Ti.