GIGABYTE GTX 9xx H2O/AIR BIOS Tweaking ?(ô?ô)? - Page 10 - Overclock.net - An Overclocking Community

Forum Jump: 

GIGABYTE GTX 9xx H2O/AIR BIOS Tweaking ?(ô?ô)?

Reply
 
Thread Tools
post #91 of 7850 (permalink) Old 03-22-2015, 03:01 PM - Thread Starter
Maximum_Unleashed
 
Laithan's Avatar
 
Join Date: Mar 2015
Location: United States
Posts: 3,697
Rep: 500 (Unique: 265)
Quote:
Originally Posted by OniLink View Post

It is not "locking". Setting top 3 sliders to max doesn't not remove the voltage/freq scaling and you are not sending more voltage to lower bins.
- The voltage for each bin is determined by bin's own voltage slider.
- 1st slider is overall max voltage and must be set to 1312 mV.
- 2nd slider is the one that affects the software slider voltage alteration. Based on how the 2nd slider is set, the software slider can affect max voltage, min voltage or doing nothing at all.
Imho you should just not care about it and set it to 1312-1312 (software slider will be useless but who cares anyway).

I'm not giving up on the quest for 1312.5V (or to prove it is no more effective than 1281.3V wink.gif ) so I've been doing a bunch more testing. biggrin.gif
With your help, I've learned some more things about voltage. I wanted to help illustrate exactly what we are talking about if it isn't super clear for everyone.


     ↑  Using MSI Afterburner as the example (same for other O/C apps), you can see that I've been able to configure the slider to allow +100 mV.

Does this mean this configuration allows more voltage than 0-68mV? NO!. wth.gif

It basically just means that we've identified specifically where in the BIOS the range of the maximum "extra voltage" can be edited.
The MAXIMUM VOLTAGE is always going to be the same as programmed in the BIOS, it is just giving you either a 0-68 or a 0-100 "slider" range to choose from.

Oni makes a good point that if you are going to just max out the slider anyway, why not just max it out in the BIOS. I am sure if we took a vote, the "PREFERRED" answer would be split 50/50. There could be those that simply don't want to run their GPU with the additional voltage all the time unless they are benching. Heat could be an issue in the summer as mentioned earlier. On the other hand, Overclocking software uses memory and can crash and setting all O/C settings directly in the BIOS is something people do. I think that in the end, regardless of which ever you prefer the maximum GAME STABLE overclock WILL NOT CHANGE so it is just a personal preference.

The BIOS options for adjusting the +mV 'slider' within a GPU overclock utility are basically:
(1) Normal "0 - 68mV" slider range
(2) Expanded "0 - 100mV" slider range (or any custom value between 1 and 100)
(3) Disabled (This occurs when the minimum and maximum value on the 2nd slider is the same).




     ↑  You can see that I've configured the minimum value ONLY of the 2nd slider with a minimum value of 1175 which provides the 0-100mV range in MSI afterburner's voltage slider. The maximum value does not appear to impact the voltage slider! So REGARDLESS of what you maximum value is there, the MINIMUM value is what dictates the range.

Some examples of the minimum values and the results:
1200mV BIOS = +75mV Slider range
1175mV BIOS = +100mV Slider range
IDENTICAL TO MAX VALUE = SOFTWARE VOLTAGE CONTROL DISABLED

Consider it a "People's choice" option thumb.gif

Quote:
Originally Posted by OniLink View Post

- 3rd slider is the bin selector; the only way to get to the last bin is to have the min value of this slider same as max value of CLK 74. Otherwise you should (not sure) get PerfCaps.


Unfortunately I could not identify any settings above a maximum of 1281.3V without a PERFCAP unless the 3rd slider minimum and maximum values were identical. exclamation.gif

When testing a BIOS with the 3rd min/max voltage both set to 1312.5mV I was able to see a jump in voltage as I increased the slider but again, I don't know WHAT the actual maximum value was.

Testing with 1312.5mV and a 0-100mV voltage slider configured
+ 0mV = 1.237V
+20mV = 1.262V
+40mV = 1.275V - This is the first notch where 1.275V is shown via software. It must be truly only 1.275V as the PERFCAP was still NOT removed.
+70mV = 1.275V (MAX SOFTWARE CAN REPORT) - This is the first notch where the PERFCAP is REMOVED.

This is again proving without question that we are sending MORE THAN 1.275V to the GPU...as the Perfcap isn't removed until +70mV are added

It does not answer the $1 million question
..does it send 1281.3mV or 1312.5mV?!?
The voltages aren't all that far from each other really anyway..

Editorial note: When testing 1312.5V at 4K, really pushing the GPU I have frozen solid needing a hard reset. That has never happened with 1281.3V, for whatever that may be worth and it could also be coincidence.

For those that want to test with the 1312.5mV BIOS', here are two:
980G1_F3_Onistockmem.zip 136k .zip file
The BIOS Oni shared with us, I just modified the memory back to default speed.

1312test.zip 136k .zip file
Derived from V4BETA but with voltage increased to 1312.5mV


Oni, what is your max GAME STABLE overclock with your BIOS?
Your ASIC is slightly better than my best one.. I am curious if with V4BETA you lose any GAME STABLE clocks.
Attached Files
File Type: zip 980G1_F3_Onistockmem.zip (136.4 KB, 62 views)
File Type: zip 1312test.zip (136.4 KB, 49 views)

GIGABYTE GTX 9xx G1 GAMING BIOS Tweaking
  ̿̿ ̿̿ ̿̿ ̿̿̿'̿'\̵͇̿̿\з=༼ຈل͜ຈ༽=ε/̵͇̿̿/'̿'̿ ̿ ̿̿ ̿̿ ̿̿  
░▒▓│ FORKAY‼ «  » WΘΘT ‼ │▓▒░
Xeon E5-1680 V2 (IVY-E) Inside


Laithan is offline  
Sponsored Links
Advertisement
 
post #92 of 7850 (permalink) Old 03-22-2015, 03:49 PM
New to Overclock.net
 
OniLink's Avatar
 
Join Date: Jun 2014
Location: Italy
Posts: 89
Rep: 6 (Unique: 4)
Quote:
Originally Posted by Laithan View Post

The BIOS options for adjusting the +mV 'slider' within a GPU overclock utility are basically:
(1) Normal "0 - 68mV" slider range
(2) Expanded "0 - 100mV" slider range (or any custom value between 1 and 100)
(3) Disabled (This occurs when the minimum and maximum value on the 2nd slider is the same).

3) I thought that but after re-reading Zoson post, eh says "When equal to 1, slider shows +100, but only increases minimum voltage while under PerfCap such as Util. This is why the voltage slider seems to do nothing.". So it does something and that something is the min voltage but it's not clear under which conditions: any PerfCap or just Util (=idle) ?

4) Not tested but may be working for people who wants to enable/disable their overlock, maybe because of bad room temps (summer) or whatever:
Set 2nd slider max value to something lower than the 1st slider (which should be 1312 mV anyway). In theory, if you do that the software slider will regulate the maximum boost voltage and if you set the CLK bins voltage ranges correctly, then you will be able to artificially create a Vrel PerfCap and so limiting the frequency too.
So basically you can OC/OV or UC/UV with one single slider rapidly.
Just an idea, need testing if someone is interested.
(the 2nd slider min value should be set the same as max value; afaik 2nd max has no influence, it just has to be greater or equal to 2nd min).

The strange part to me is the fact that software slider was "only" +100 mV with 2nd slider min set tom 1175. In theory it should have +137 mV (from 1175 to 1312).
Maybe MSI and other software are limited to +100 because of the sensor limit of 1275? No idea.
Quote:
Originally Posted by Laithan View Post

Unfortunately I could not identify any settings above a maximum of 1281.3V without a PERFCAP unless the 3rd slider minimum and maximum values were identical. exclamation.gif

This is to be expected. The bin selection occurs by 3rd slider min value. So if your last bin has 1312, you have to set the 3rd min to 1312 too. And because 3rd max must be equal or greater than 3rd min, they happen to be the same.
In theory you could set 3rd max above 1312 but it will not change anything.

However, I went back to give a look to stock BIOS and in that case 3rd min does NOT match the max voltage of the top bin used by stock BIOS boost.
Conclusion: 3rd slider should be the bin selector but further investigation is needed. However since we aim to bin 74, we should not bother with it at all and leave it to 1312-1312 tongue.gif

Quote:
Originally Posted by Laithan View Post

Testing with 1312.5mV and a 0-100mV voltage slider configured
+ 0mV = 1.237V
+20mV = 1.262V
+40mV = 1.275V - This is the first notch where 1.275V is shown via software. It must be truly only 1.275V as the PERFCAP was still NOT removed.
+70mV = 1.275V (MAX SOFTWARE CAN REPORT) - This is the first notch where the PERFCAP is REMOVED.

This is again proving without question that we are sending MORE THAN 1.275V to the GPU...as the Perfcap isn't removed until +70mV are added

This confirms that 2nd min influences both software slider value and the effective voltage sent to card. This can be used for the point 4 I suggested above.
Quote:
Originally Posted by Laithan View Post

It does not answer the $1 million question
..does it send 1281.3mV or 1314.5mV?!?
The voltages aren't all that far from each other really anyway..

I asked you some posts above but maybe you haven't read: do you know the measuring points? I can take the DMM and try it out but I don't know the points on the G1.
Quote:
Originally Posted by Laithan View Post

Oni, what is your max GAME STABLE overclock with your BIOS?
Your ASIC is slightly better than my best one.. I am curious if with V4BETA you lose any GAME STABLE clocks.

I have no idea right now tongue.gif The fact is that when I OCed first time, the card was under air and with old BIOS version.

I suspect that lower temps and new BIOS improved the stability quite a lot.
Temps because higher temps = higher electrical resistance. This should increase the voltage in theory but also increases the current loss.
Well, it's just a guess, I'm hitting again my physics knowledge limit.

The idea that the stability improved a lot comes from the fact that if I remember correctly, I was unable to run 1531-8000 with neither artifacts nor crashes before watercooling/updated BIOS.
I may probably do another benchmark OCing tomorrow but currently I'm not playing PC games much (damn Monster Hunter tongue.gif ) so I can not assure a game stable OC :/
If it may help, I'm folding with my card all the time when I'm at the PC. I've read that folding is kinda "OC-sensitive" so may it be a sign of stability.
OniLink is offline  
post #93 of 7850 (permalink) Old 03-22-2015, 07:51 PM - Thread Starter
Maximum_Unleashed
 
Laithan's Avatar
 
Join Date: Mar 2015
Location: United States
Posts: 3,697
Rep: 500 (Unique: 265)
Quote:
Originally Posted by OniLink View Post

3) I thought that but after re-reading Zoson post, eh says "When equal to 1, slider shows +100, but only increases minimum voltage while under PerfCap such as Util. This is why the voltage slider seems to do nothing.". So it does something and that something is the min voltage but it's not clear under which conditions: any PerfCap or just Util (=idle) ?

4) Not tested but may be working for people who wants to enable/disable their overlock, maybe because of bad room temps (summer) or whatever:
Set 2nd slider max value to something lower than the 1st slider (which should be 1312 mV anyway). In theory, if you do that the software slider will regulate the maximum boost voltage and if you set the CLK bins voltage ranges correctly, then you will be able to artificially create a Vrel PerfCap and so limiting the frequency too.
So basically you can OC/OV or UC/UV with one single slider rapidly.
Just an idea, need testing if someone is interested.
(the 2nd slider min value should be set the same as max value; afaik 2nd max has no influence, it just has to be greater or equal to 2nd min).

We're able to (the BIOS posted above) have a 'default' perfcap and then it doesn't go away until +70mV is added, which is accomplished without the need to reduce the max value of the 2nd slider to be lower than the max value of the first. This is also the case with V4BETA (+40mV needed to remove). In other words, when you want performance mode and you max out the voltage you won't have perfcaps but when you want to lower temps and conserve power and lower voltage you will have perfcaps again (just like the stock OEM BIOS). We have this already added in V4BETA, did you mean something different?

Honestly, I have no clue what the max value of the 2nd slider does (stock OEM BIOS' do NOT set it to the max voltage of the card).. but it doesn't appear to have anything to do with the minimum value... Do you know specifically what it is used for?

Quote:
Originally Posted by OniLink View Post

The strange part to me is the fact that software slider was "only" +100 mV with 2nd slider min set tom 1175. In theory it should have +137 mV (from 1175 to 1312).
Maybe MSI and other software are limited to +100 because of the sensor limit of 1275? No idea.

I was pretty surprised that math (relative to the max value) didn't seem to have anything to do with the results.
For this particular 'reverse engineering' I first tried to adjust the MAX value of slider #2 leaving the min value the same..
After flashing and testing I could see right away that the changes made NO difference to the +mV in MSI Afterburner (and confirmed with NVIDIA Inspector also)

Then of course I realized that I needed to change the minimum value of slider #2 instead which is when I started to see the changes in the MSI voltage slider.
(I don't know if folks realize but I am flashing my cards like 20 times a day lol)

Quote:
Originally Posted by OniLink View Post

This is to be expected. The bin selection occurs by 3rd slider min value. So if your last bin has 1312, you have to set the 3rd min to 1312 too. And because 3rd max must be equal or greater than 3rd min, they happen to be the same.
In theory you could set 3rd max above 1312 but it will not change anything.

However, I went back to give a look to stock BIOS and in that case 3rd min does NOT match the max voltage of the top bin used by stock BIOS boost.
Conclusion: 3rd slider should be the bin selector but further investigation is needed. However since we aim to bin 74, we should not bother with it at all and leave it to 1312-1312 tongue.gif

That makes sense, and we don't see perfcaps when the values are identical.
The interesting thing is that my V4BETA also doesn't have perfcaps (neither did V3) and they don't follow that theory.
Stock OEM values for 2nd slider max, seem to indicate that there might be some relativity to the max voltage and that value..
In testing, lowering it or putting at max doesn't seem to change much...


thumbsupsmiley.png
Quote:
Originally Posted by OniLink View Post

This confirms that 2nd min influences both software slider value and the effective voltage sent to card. This can be used for the point 4 I suggested above.
I asked you some posts above but maybe you haven't read: do you know the measuring points? I can take the DMM and try it out but I don't know the points on the G1.
I have no idea right now tongue.gif The fact is that when I OCed first time, the card was under air and with old BIOS version.

Oops, sorry no I don't know where I was going to look them up. I blindly bought a DMM thinking "hey WTH I need to have one anyway"..
and as soon as I ordered it I was like.. umm.. my waterblocks are full cover with added backplates. doh.gif
I still need one but..lol

Quote:
Originally Posted by OniLink View Post

The idea that the stability improved a lot comes from the fact that if I remember correctly, I was unable to run 1531-8000 with neither artifacts nor crashes before watercooling/updated BIOS.
I may probably do another benchmark OCing tomorrow but currently I'm not playing PC games much (damn Monster Hunter tongue.gif ) so I can not assure a game stable OC :/
If it may help, I'm folding with my card all the time when I'm at the PC. I've read that folding is kinda "OC-sensitive" so may it be a sign of stability.

Reminded me of something else I wanted to share..
When testing a modified BIOS I noticed that increasing the memory from +0 to even +300 places a strain on the available TDP.
Fire up Kombuster and notice the TDP going over 100%, put the memory back to +0 and it drops to high 90's.
Both of us have also increased the TDP in the BIOS and it still happens.

I can't speak about folding workloads personally, I also hear they can be demanding but let me tell you, 4K testing is brutal on GPUs...
For those of you that aren't aware, you can fit (4) 1080P screens inside ONE 4K screen! smile.gif
So yes, it is testing 4x the workload of 1080P and going to validate anyone's GAME STABLE overclocks for real drunken.gif

GIGABYTE GTX 9xx G1 GAMING BIOS Tweaking
  ̿̿ ̿̿ ̿̿ ̿̿̿'̿'\̵͇̿̿\з=༼ຈل͜ຈ༽=ε/̵͇̿̿/'̿'̿ ̿ ̿̿ ̿̿ ̿̿  
░▒▓│ FORKAY‼ «  » WΘΘT ‼ │▓▒░
Xeon E5-1680 V2 (IVY-E) Inside


Laithan is offline  
Sponsored Links
Advertisement
 
post #94 of 7850 (permalink) Old 03-22-2015, 09:52 PM - Thread Starter
Maximum_Unleashed
 
Laithan's Avatar
 
Join Date: Mar 2015
Location: United States
Posts: 3,697
Rep: 500 (Unique: 265)
Check this out, these may be the resistors that are limiting to 1281.3V wink.gifLINK (Click to show)
"You can go as high as 1.7-1.85V with this kind of modification, so it's suitable for extreme overclocking.
But beware, safe voltages even on LN2 are rather around 1.5-1.6ish range, which is still enough to get decent overclocking on most of the cards. 1700-1800MHz clockspeeds should not be an issue.

Resistance at marked point on non-modded card is about 90-110 ohm with resistors removed, and about 4-8 ohm with resistors present."

GIGABYTE GTX 9xx G1 GAMING BIOS Tweaking
  ̿̿ ̿̿ ̿̿ ̿̿̿'̿'\̵͇̿̿\з=༼ຈل͜ຈ༽=ε/̵͇̿̿/'̿'̿ ̿ ̿̿ ̿̿ ̿̿  
░▒▓│ FORKAY‼ «  » WΘΘT ‼ │▓▒░
Xeon E5-1680 V2 (IVY-E) Inside


Laithan is offline  
post #95 of 7850 (permalink) Old 03-23-2015, 03:57 AM
New to Overclock.net
 
OniLink's Avatar
 
Join Date: Jun 2014
Location: Italy
Posts: 89
Rep: 6 (Unique: 4)
Quote:
Originally Posted by Laithan View Post

We're able to (the BIOS posted above) have a 'default' perfcap and then it doesn't go away until +70mV is added, which is accomplished without the need to reduce the max value of the 2nd slider to be lower than the max value of the first. This is also the case with V4BETA (+40mV needed to remove). In other words, when you want performance mode and you max out the voltage you won't have perfcaps but when you want to lower temps and conserve power and lower voltage you will have perfcaps again (just like the stock OEM BIOS). We have this already added in V4BETA, did you mean something different?

Honestly, I have no clue what the max value of the 2nd slider does (stock OEM BIOS' do NOT set it to the max voltage of the card).. but it doesn't appear to have anything to do with the minimum value... Do you know specifically what it is used for?

Nope, it's the same but yesterday I started typing, then went back reading, then searched for info, then looked at my and stock BIOS and I got confused in all of this XD
Anyway the 2nd max seems useless after all. I suggest leaving it at 1312 and it will be fine.
Does 2nd min+software voltage match theused CLK max voltage? In otherwords: can you precisely aim to the wanted CLK with the software slider?

Quote:
Originally Posted by Laithan View Post

I was pretty surprised that math (relative to the max value) didn't seem to have anything to do with the results.
For this particular 'reverse engineering' I first tried to adjust the MAX value of slider #2 leaving the min value the same..
After flashing and testing I could see right away that the changes made NO difference to the +mV in MSI Afterburner (and confirmed with NVIDIA Inspector also)

Then of course I realized that I needed to change the minimum value of slider #2 instead which is when I started to see the changes in the MSI voltage slider.

Still, there are 37 missing mV from sw slider. If you set the 2nd min to 1212, does the sw slider still goes up to +100 or it stucks at +63 (or any other value) ?

Quote:
Originally Posted by Laithan View Post

That makes sense, and we don't see perfcaps when the values are identical.
The interesting thing is that my V4BETA also doesn't have perfcaps (neither did V3) and they don't follow that theory.
Stock OEM values for 2nd slider max, seem to indicate that there might be some relativity to the max voltage and that value..
In testing, lowering it or putting at max doesn't seem to change much...


thumbsupsmiley.png

Actually yout V4 beta is set that way O.o I mean, you have 3rd min same as CLK74 Vmax. Did you mean something different?
Stock BIOS is the most mysterious one by far XD

Quote:
Originally Posted by Laithan View Post

Oops, sorry no I don't know where I was going to look them up. I blindly bought a DMM thinking "hey WTH I need to have one anyway"..
and as soon as I ordered it I was like.. umm.. my waterblocks are full cover with added backplates. doh.gif
I still need one but..lol

Same as me XD I bought one because I wanted to measure the real V and because "hey! I don't have one at home!" and now I have the waterblock backplate on the card XD
However I should be able to remove the backplate without removing the waterblock.
For the next card, I should consider the voltage measuring points that some manufacturers put on the border of the card.

Quote:
Originally Posted by Laithan View Post

Reminded me of something else I wanted to share..
When testing a modified BIOS I noticed that increasing the memory from +0 to even +300 places a strain on the available TDP.
Fire up Kombuster and notice the TDP going over 100%, put the memory back to +0 and it drops to high 90's.
Both of us have also increased the TDP in the BIOS and it still happens.

I can't speak about folding workloads personally, I also hear they can be demanding but let me tell you, 4K testing is brutal on GPUs...
For those of you that aren't aware, you can fit (4) 1080P screens inside ONE 4K screen! smile.gif
So yes, it is testing 4x the workload of 1080P and going to validate anyone's GAME STABLE overclocks for real drunken.gif

I don't know Kombuster that much, but I guess it is an extreme loadout like Furmark. If that's the case, then there is no point using it (as you agreed in another thread).
Agree that 4K may be the real "brutal" testing for stability. However it is not the best one for artifacts because you can't see thw whole scene unless you have a 4K display.
I guess that 4K FireStrike Extreme demo loop should be used for stability and 1080/1440 (depending on your screen) FireStrike Extreme demo loop for artifacts.
In that regard, Unigine Heaven (all options maxed out, resolution of your monitor) is very good for artifacts, especially the night scenes.

Quote:
Originally Posted by Laithan View Post

Check this out, these may be the resistors that are limiting to 1281.3V wink.gifLINK (Click to show)
"You can go as high as 1.7-1.85V with this kind of modification, so it's suitable for extreme overclocking.
But beware, safe voltages even on LN2 are rather around 1.5-1.6ish range, which is still enough to get decent overclocking on most of the cards. 1700-1800MHz clockspeeds should not be an issue.

Resistance at marked point on non-modded card is about 90-110 ohm with resistors removed, and about 4-8 ohm with resistors present."

I've found that too time ago but I'm not brave enough (nor soldering skilled enough) to try it tongue.gif
And btw, the limit is 1312, not 1281 XD
OniLink is offline  
post #96 of 7850 (permalink) Old 03-23-2015, 08:02 AM - Thread Starter
Maximum_Unleashed
 
Laithan's Avatar
 
Join Date: Mar 2015
Location: United States
Posts: 3,697
Rep: 500 (Unique: 265)
Quote:
Originally Posted by OniLink View Post

Does 2nd min+software voltage match theused CLK max voltage? In otherwords: can you precisely aim to the wanted CLK with the software slider?
Still, there are 37 missing mV from sw slider. If you set the 2nd min to 1212, does the sw slider still goes up to +100 or it stucks at +63 (or any other value) ?

No it doesn't match. In fact, I think there is always the same value of "extra voltage" no matter what you do. The min value of slider #2 just permits you to divide that same additional voltage into a smaller, or a larger range. In other words, if you have 1 gallon of water, you can put the water in to (4) containers, each 1/4 of a gallon or you could put the water into (8) containers, each 1/8th of a gallon. No matter how to divide up the water, you still only have 1 gallon (oversimplified for my point).

I've seems the "total voltage" never changes regardless if you set it at +75mV max capacity or +100mV max capacity or leave it stock..What changes is just the location on the +mV slider where the PERFCAP is REMOVED. Expanding the voltage slider to +100mV actually requires me to give +70mV before perfcaps are removed where the default value I only need +40mV.. This proves it simply adjusts the point where a certain voltage is applied and "stretches it out".

To answer your question, we know that 1175 on the minimum slider #2 = +100mV MSI overvolt slider.
Lowering the value, does nothing. Increasing the value immediately starts to reduce the amount of +mV from 100.
That is why we see that +1200 which is an increase from 1175, actually reduced the slide range from +100 to +75.
You can actually create a very tight value that would probably more accurately represents the actual changes.
In other words, we are limited to the bins and we cannot fine tune voltage down to the 1mV, so really the MSI overvoltage slider should be more like a "4-5 setting switch" than a slider.

Quote:
Originally Posted by OniLink View Post

Actually yout V4 beta is set that way O.o I mean, you have 3rd min same as CLK74 Vmax. Did you mean something different?

The value is actually 1218, not 1281 (yet I read it as 1281 a few times myself). 1218 is indeed lower than 1281 but I may have misunderstood.

Quote:
Originally Posted by OniLink View Post

I don't know Kombuster that much, but I guess it is an extreme loadout like Furmark. If that's the case, then there is no point using it (as you agreed in another thread).

It is like furmark but the biggest difference is that Kombuster can trigger boost clocks properly and Furmark cannot. I only use Kombuster as a 'quick check' to see what my boost clocks and TDP are. I don't use Kombuster for stability or performance testing at all. Kombuster is more like GPU-Z for me, just checking things in between actual 3D tests.

Quote:
Originally Posted by OniLink View Post

Agree that 4K may be the real "brutal" testing for stability. However it is not the best one for artifacts because you can't see thw whole scene unless you have a 4K display.
I guess that 4K FireStrike Extreme demo loop should be used for stability and 1080/1440 (depending on your screen) FireStrike Extreme demo loop for artifacts.

You can actually run 4K tests without a 4K monitor. I think you mentioned DSR earlier which is one way.
DSR will RENDER the image at 4K (or higher) just downscale it to your display. That is one way.
3D Mark Ultra will also do the same thing if you don't have a 4K display, it will render the test in 4K but downscale.

Quote:
Originally Posted by OniLink View Post

And btw, the limit is 1312, not 1281 XD

cheers.gif

GIGABYTE GTX 9xx G1 GAMING BIOS Tweaking
  ̿̿ ̿̿ ̿̿ ̿̿̿'̿'\̵͇̿̿\з=༼ຈل͜ຈ༽=ε/̵͇̿̿/'̿'̿ ̿ ̿̿ ̿̿ ̿̿  
░▒▓│ FORKAY‼ «  » WΘΘT ‼ │▓▒░
Xeon E5-1680 V2 (IVY-E) Inside


Laithan is offline  
post #97 of 7850 (permalink) Old 03-23-2015, 10:56 AM
New to Overclock.net
 
OniLink's Avatar
 
Join Date: Jun 2014
Location: Italy
Posts: 89
Rep: 6 (Unique: 4)
Quote:
Originally Posted by Laithan View Post

No it doesn't match. In fact, I think there is always the same value of "extra voltage" no matter what you do. The min value of slider #2 just permits you to divide that same additional voltage into a smaller, or a larger range. In other words, if you have 1 gallon of water, you can put the water in to (4) containers, each 1/4 of a gallon or you could put the water into (8) containers, each 1/8th of a gallon. No matter how to divide up the water, you still only have 1 gallon (oversimplified for my point).

I've seems the "total voltage" never changes regardless if you set it at +75mV max capacity or +100mV max capacity or leave it stock..What changes is just the location on the +mV slider where the PERFCAP is REMOVED. Expanding the voltage slider to +100mV actually requires me to give +70mV before perfcaps are removed where the default value I only need +40mV.. This proves it simply adjusts the point where a certain voltage is applied and "stretches it out".

To answer your question, we know that 1175 on the minimum slider #2 = +100mV MSI overvolt slider.
Lowering the value, does nothing. Increasing the value immediately starts to reduce the amount of +mV from 100.
That is why we see that +1200 which is an increase from 1175, actually reduced the slide range from +100 to +75.
You can actually create a very tight value that would probably more accurately represents the actual changes.
In other words, we are limited to the bins and we cannot fine tune voltage down to the 1mV, so really the MSI overvoltage slider should be more like a "4-5 setting switch" than a slider.

So Nvidia somehow locked the software slider to 100 mV range, otherwise with 2nd min at 1175 and 1st at 1312 the slider should go up to 137.
Why Nvidia is putting so many limits is beyond me.

Quote:
Originally Posted by Laithan View Post

The value is actually 1218, not 1281 (yet I read it as 1281 a few times myself). 1218 is indeed lower than 1281 but I may have misunderstood.

So the bin selection is again a mystery. Every time I think I finally understood how BIOS works, you find a way to prove me wrong biggrin.gif

Quote:
Originally Posted by Laithan View Post

It is like furmark but the biggest difference is that Kombuster can trigger boost clocks properly and Furmark cannot. I only use Kombuster as a 'quick check' to see what my boost clocks and TDP are. I don't use Kombuster for stability or performance testing at all. Kombuster is more like GPU-Z for me, just checking things in between actual 3D tests.
You can actually run 4K tests without a 4K monitor. I think you mentioned DSR earlier which is one way.
DSR will RENDER the image at 4K (or higher) just downscale it to your display. That is one way.
3D Mark Ultra will also do the same thing if you don't have a 4K display, it will render the test in 4K but downscale.
cheers.gif

I know but with downscaling you are loosing "pixels" which may be where artifacts appear. This is why I want a 1:1 ratio if I'm looking for artifacts.
Stability wise, going for 4k is fine. I would test 1440p too just to simulate a closer-to-reality scenario.
OniLink is offline  
post #98 of 7850 (permalink) Old 03-23-2015, 12:11 PM
New to Overclock.net
 
maddogfargo's Avatar
 
Join Date: Dec 2014
Posts: 40
Rep: 5 (Unique: 3)
@ Laithan

Thanks for sharing your work on this 980 BIOS with me. Was helpful.

I was able to create a custom set of 3 BIOS for the Gigabyte G1 Gaming GTX 970. It required tightening the tweaks on the voltage tables.

I know this thread is for the 980 version of the card, but if anyone coming here has the 970, my BIOS can be found posted in Zoson's thread, page 357, here:

https://www.overclock.net/t/1517316/extract-and-flash-gtx-970-and-980-firmware-zosons-h2o-and-air-custom-bios/3560#post_23700324

Just sharing back. Hope this helps.
maddogfargo is offline  
post #99 of 7850 (permalink) Old 03-23-2015, 01:31 PM
InstantAli3n
 
cookieboyeli's Avatar
 
Join Date: Sep 2013
Posts: 2,002
Rep: 150 (Unique: 110)
Quote:
Originally Posted by maddogfargo View Post

@ Laithan

Thanks for sharing your work on this 980 BIOS with me. Was helpful.

I was able to create a custom set of 3 BIOS for the Gigabyte G1 Gaming GTX 970. It required tightening the tweaks on the voltage tables.

I know this thread is for the 980 version of the card, but if anyone coming here has the 970, my BIOS can be found posted in Zoson's thread, page 357, here:

https://www.overclock.net/t/1517316/extract-and-flash-gtx-970-and-980-firmware-zosons-h2o-and-air-custom-bios/3560#post_23700324

Just sharing back. Hope this helps.

Thanks! I can't afford a 980 but will eventually get a 970 G1.

Have you modeled this bios after the 980 V4 beta?
cookieboyeli is offline  
post #100 of 7850 (permalink) Old 03-23-2015, 05:25 PM - Thread Starter
Maximum_Unleashed
 
Laithan's Avatar
 
Join Date: Mar 2015
Location: United States
Posts: 3,697
Rep: 500 (Unique: 265)
Quote:
Originally Posted by OniLink View Post

So Nvidia somehow locked the software slider to 100 mV range, otherwise with 2nd min at 1175 and 1st at 1312 the slider should go up to 137.
Why Nvidia is putting so many limits is beyond me.

I think they realize that people could start clocking Maxwell to 1800+ boost and it would be pointless to buy a Titan X wink.gif
In the mean time, we are limited to 1281.3V wink.gif (LMAO!!)

Quote:
Originally Posted by OniLink View Post

So the bin selection is again a mystery. Every time I think I finally understood how BIOS works, you find a way to prove me wrong biggrin.gif

I'd prefer to look at it as us all getting closer to the truth thumb.gif
Quote:
Originally Posted by maddogfargo View Post

@ Laithan

Thanks for sharing your work on this 980 BIOS with me. Was helpful.

I was able to create a custom set of 3 BIOS for the Gigabyte G1 Gaming GTX 970. It required tightening the tweaks on the voltage tables.

I know this thread is for the 980 version of the card, but if anyone coming here has the 970, my BIOS can be found posted in Zoson's thread, page 357, here:

https://www.overclock.net/t/1517316/extract-and-flash-gtx-970-and-980-firmware-zosons-h2o-and-air-custom-bios/3560#post_23700324

Just sharing back. Hope this helps.

I wouldn't be doing this if others gained no benefit so I am glad to hear it! smile.gif

GIGABYTE GTX 9xx G1 GAMING BIOS Tweaking
  ̿̿ ̿̿ ̿̿ ̿̿̿'̿'\̵͇̿̿\з=༼ຈل͜ຈ༽=ε/̵͇̿̿/'̿'̿ ̿ ̿̿ ̿̿ ̿̿  
░▒▓│ FORKAY‼ «  » WΘΘT ‼ │▓▒░
Xeon E5-1680 V2 (IVY-E) Inside


Laithan is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off