GIGABYTE GTX 9xx H2O/AIR BIOS Tweaking ?(ô?ô)? - Page 39 - Overclock.net - An Overclocking Community

Forum Jump: 

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

Reply
 
Thread Tools
post #381 of 7850 (permalink) Old 05-08-2015, 04:33 PM
New to Overclock.net
 
facboy's Avatar
 
Join Date: Oct 2014
Posts: 99
Rep: 0
Curious to know what effect you've found changing the voltage tables at the base clock bin...been experimenting and on mine it seems changes to eg Bin 53 (1228mhz at stock) have no effect on the voltage. I can move the voltages up a couple of bins (eg Bin 53 now has the values that used to be in 55, 54 has the values from 56, etc), but the card always runs 1228Mhz at 1.062V regardless. the upper bins have some effect because if i lower the voltages on the upper bins then the card will boost higher because those bins become available.
facboy is offline  
Sponsored Links
Advertisement
 
post #382 of 7850 (permalink) Old 05-08-2015, 05:55 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 facboy View Post

Curious to know what effect you've found changing the voltage tables at the base clock bin...been experimenting and on mine it seems changes to eg Bin 53 (1228mhz at stock) have no effect on the voltage. I can move the voltages up a couple of bins (eg Bin 53 now has the values that used to be in 55, 54 has the values from 56, etc), but the card always runs 1228Mhz at 1.062V regardless. the upper bins have some effect because if i lower the voltages on the upper bins then the card will boost higher because those bins become available.

Sure let me see if I can help and cover some other details while on the subject. biggrin.gif

Here is a basic breakdown of the voltage table, as I configure it in V7BETA. In a nutshell, it has a lot to do with reading / testing and seeing how others set the table and there are MANY ways to modify the table, and many of them are successful. Here's my thought process:

CLK 0 through CLK 25
This range, having the lowest voltage of the table is of course for when the GPU is idle, or minimal load. The default when you install NVIDIA's driver, is to configure your GPU for "adaptive" power management. If you set to "prefer maximum performance" power management, these tables will never even be used. The reason I set all the minimums to 825mV in this range is because there were some unique instances where real low voltages were causing some issues (known issue). This also reduces the voltage range to the core GPU speed during a spike (adaptive power management will spike to the DEFAULT GPU CORE Mhz first, and then your GPU will boost above the DEFAULT CORE speed as normal). I increased the max voltage for these clocks very little.

CLK 26 through CLK 54
This is where the voltage curve really begins. Similar to above, if "prefer maximum performance" is selected, these tables won't be used either. This is basically the section of the table that is specific to adaptive power management. . Voltage slightly increased to help stability during boosting/returning to idle.

CLK 55
This is the idle CLK when "prefer maximum performance" is selected. Voltage is 1.075v, clock speed is 1291 Mhz. To answer your question, voltage doesn't play a big role here yet because your GPU is simply not being pushed enough, and the design mechanics of BOOST 2.0 are comfortable using this voltage, even at a slightly different clocks. The GPU isn't being really being used yet and there is a wide voltage variance when not under load (meaning the same voltage will work with many speeds simply because the GPU isn't being pushed). There is a voltage range for every CLK and the voltage I see here is closer to the minimum of what CLK 55 provides. The BIOS defines the base clock of the GPU and the card knows what voltage to apply based on the table and BOOST 2.0 internal mechanics. The voltage becomes more predictable and consistent under load.

CLK 56 through CLK 74
When placed under load, this is where BOOST 2.0 occurs. All these CLKs are BOOST clocks. It will boost to the highest CLK that it can based on a few different factors. If the heat is low and good solid/stablepower is available your GPU should remain at CLK 74, the maximum boost speed. In the event you are starved for power or heat builds or even too much voltage, it will work backwards, dropping back to whatever CLK it feels it is now "satisfied" with. Modifying your BIOS can help manage this behavior.

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


Laithan is offline  
post #383 of 7850 (permalink) Old 05-08-2015, 11:04 PM
New to Overclock.net
 
facboy's Avatar
 
Join Date: Oct 2014
Posts: 99
Rep: 0
Quote:
Originally Posted by Laithan View Post



CLK 26 through CLK 54
This is where the voltage curve really begins. Similar to above, if "prefer maximum performance" is selected, these tables won't be used either. This is basically the section of the table that is specific to adaptive power management. . Voltage slightly increased to help stability during boosting/returning to idle.

CLK 55
This is the idle CLK when "prefer maximum performance" is selected. Voltage is 1.075v, clock speed is 1291 Mhz. To answer your question, voltage doesn't play a big role here yet because your GPU is simply not being pushed enough, and the design mechanics of BOOST 2.0 are comfortable using this voltage, even at a slightly different clocks. The GPU isn't being really being used yet and there is a wide voltage variance when not under load (meaning the same voltage will work with many speeds simply because the GPU isn't being pushed). There is a voltage range for every CLK and the voltage I see here is closer to the minimum of what CLK 55 provides. The BIOS defines the base clock of the GPU and the card knows what voltage to apply based on the table and BOOST 2.0 internal mechanics. The voltage becomes more predictable and consistent under load.

I guess it starts at 55 because this is where you have set the TDP and 3D base clocks on the common tab? so on a vanilla BIOS it would start at 53 (1228Mhz)? The reason I ask is that if I edit the vanilla BIOS and change the voltage values at bin 53 (eg move both sliders up 3 notches) it doesn't actually change the voltage at all when running at the base clock...it stays at the value it was before I edited the BIOS. Before you ask, obviously I am flashing the edited BIOS to the card, I'm not a complete idiot smile.gif. I know it's having some effect because I lower the voltage ranges in the upper boost bins (like 65+) so that my card boosts higher.
facboy is offline  
Sponsored Links
Advertisement
 
post #384 of 7850 (permalink) Old 05-09-2015, 05:22 AM - Thread Starter
Maximum_Unleashed
 
Laithan's Avatar
 
Join Date: Mar 2015
Location: United States
Posts: 3,697
Rep: 500 (Unique: 265)
Yes, that is specific to the BIOS we've been tweaking here in this thread. I did raise the factory overclock to 1291Mhz on the GPU core clock, which will determine which CLK on the voltage table it will use and you're right, factory would use a lower CLK.

The behavior you see on the base clock voltage is normal because it is a voltage range and the card isn't pushed enough to call for any higher voltage yet, even if you allow for more. This was what I was referring to in my last post you quoted.

You could force a certain voltage if the min/max were the same but there's really no point in doing that at idle clocks, I would focus more on the last 10-15 CLKs with regard to raising voltages.
wink.gif

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


Laithan is offline  
post #385 of 7850 (permalink) Old 05-09-2015, 08:35 PM - Thread Starter
Maximum_Unleashed
 
Laithan's Avatar
 
Join Date: Mar 2015
Location: United States
Posts: 3,697
Rep: 500 (Unique: 265)
Anyone else testing with V8BETA, the 1.312v version of V7BETA?

Any results yet? teaching.gif

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


Laithan is offline  
post #386 of 7850 (permalink) Old 05-10-2015, 11:21 AM
New to Overclock.net
 
DeathAngel74's Avatar
 
Join Date: Aug 2014
Location: Next to the evil monkey in your closet!
Posts: 5,380
Rep: 267 (Unique: 141)
DeathAngel74 is offline  
post #387 of 7850 (permalink) Old 05-11-2015, 12:56 PM
New to Overclock.net
 
artins90's Avatar
 
Join Date: Feb 2013
Posts: 39
Rep: 2 (Unique: 2)
Guys I have a WF3 OC with air cooling, when I OC to 1541 I get throttled back to 1528 after a while even with the power limit set to 320 W. The weird thing is that according to GPUZ I don't reach 100% power usage, the highest it goes is 78% and then it throttles. The main problem I have is that when it throttles to 1528 the vcore drops down from 1.28 to 1.1930 which is not enough to keep the gpu stable at that frequency and I get artifacts from time to time and it seems to be disregarding the voltage tables.

I made my own version of the bios, the GPU came out of the box with bios v F2 so I picked F3 as base.

I will attach both the GPU z log and the custom bios I made to this message, I hope someone can help me solve my issue, ideally I would like to keep the GPU at 1540-50 with the least power limit possible (I fried the vrm of my old 7950 after using it at + 50% power limit for a couple of years and I don't want to repeat the experience) GPU-ZSensorLog.txt 496k .txt file OC2.zip 136k .zip file
Attached Files
File Type: txt GPU-ZSensorLog.txt (495.9 KB, 6 views)
File Type: zip OC2.zip (136.5 KB, 8 views)
artins90 is offline  
post #388 of 7850 (permalink) Old 05-11-2015, 05:22 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 artins90 View Post

Guys I have a WF3 OC with air cooling, when I OC to 1541 I get throttled back to 1528 after a while even with the power limit set to 320 W. The weird thing is that according to GPUZ I don't reach 100% power usage, the highest it goes is 78% and then it throttles. The main problem I have is that when it throttles to 1528 the vcore drops down from 1.28 to 1.1930 which is not enough to keep the gpu stable at that frequency and I get artifacts from time to time and it seems to be disregarding the voltage tables.

I made my own version of the bios, the GPU came out of the box with bios v F2 so I picked F3 as base.

I will attach both the GPU z log and the custom bios I made to this message, I hope someone can help me solve my issue, ideally I would like to keep the GPU at 1540-50 with the least power limit possible (I fried the vrm of my old 7950 after using it at + 50% power limit for a couple of years and I don't want to repeat the experience) GPU-ZSensorLog.txt 496k .txt file OC2.zip 136k .zip file

Hi there!

Throttling is one of the areas we've been focusing our attention on. The BIOS posted on the OP (V7BETA is the latest) is actually a gigabyte F3 BIOS also, but for a 980. You can download it and then copy all the settings on every tab, over to your BIOS. The ONLY exception is that one of your PCI-e power connectors is a 6-pin, and some other power/tdp adjustments that I can make for you or you can make it yourself and I can tell you what to change. You should enjoy the results, we've had some happy 970 users stop by and toss one back with us here. biggrin.gif

Quote:
Originally Posted by DeathAngel74 View Post

throttling only stops when I set 1.312v.

You are running your own custom BIOS though right? As you know we've been able to get rid of the perfcaps even @ 1281.3mV so I suspect some more experiments could be done wink.gif.

So far, just like before, I'm not personally seeing any difference at all with 1312 vs 1281. I'm still looking for someone boosting over 1600 stable that can test with 1312, because it may be that the extra voltage only really helps with a really good ASIC (Mine are 72).

The main purpose of the 1312 testing was to see if max overclocks increased with the additional voltage. thumb.gif

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


Laithan is offline  
post #389 of 7850 (permalink) Old 05-11-2015, 05:34 PM
New to Overclock.net
 
DeathAngel74's Avatar
 
Join Date: Aug 2014
Location: Next to the evil monkey in your closet!
Posts: 5,380
Rep: 267 (Unique: 141)
DeathAngel74 is offline  
post #390 of 7850 (permalink) Old 05-11-2015, 05:57 PM
New to Overclock.net
 
artins90's Avatar
 
Join Date: Feb 2013
Posts: 39
Rep: 2 (Unique: 2)
I figured out what was wrong, this is my first Nvidia card and things work quite differently from AMD cards. Basically I thought that the value inside the voltage table for each clock speed was already set; let's say CLK 74 was set to 1.128 I though the the GPU would automatically switch to that frequency once I set the clock in afterburner but that's not how it works, basically that is only the max value and the GPU doesn't use it unless you push the voltage slider in afterburner all the way to the right.

I was basically voltage starving the GPU, now with the voltage slider all the way to the right GPU z doesn't report Vrel anymore and the GPU doesn't downclock. However, the bios still behaves in a strange way, when Vrel occurs shouldn't it drop to the CLK voltage table directly below the unstable one? Mine goes straight to 1.1930 v, it picks this arbitrary voltage out of nowhere, none of the voltage tables in my bios are set to 1.1930.

Could it be the last voltage value of the list, the one without a CLK assigned? It is set to 1.200 and is very close to 1.1930.

How do you guys determine which is the lowest stable voltage with nvidia gpus?
With my 7950 I used OCCT error checking and I gradually lowered the voltage until the first error showed up and then I used to set the voltage on step above the one reporting errors to test it again for a longer period of time until I found the lowest stable voltage without errors. That card had only 1 clock to worry about but here there are entire tables to take care of, do you know any convenient way to find the best voltage for each CLK table?

I wasn't getting noticeable artifacts at 1.21 and 1.23 do you think a 75.5 ASIC card can achive such low clocks at 1530-1550 mhz?
artins90 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