Jump to content

GPU Boost is a broken power saving feature developed by Nvidia. 

It shouldn't be called GPU Boost, but GPU limit.

It's holding your card back from its full potential, by throttling down clocks and voltages when it feels like it.

It will only boost when your card is cool enough or whatever the fuck it decides, AKA throttling/limiting your card when it feels its too hot. 

Throttling can start as low as the mid 60c's in cases, it's actually retarded.

 

They also use a 1 size fits all voltage table which has plagued many users with stability issues, where the voltage table provided didn't always provide sufficient voltage for the current clock.

This happens to people who HAVEN'T overclocked at all.

They experience TDR's and crashes, it happens most commonly in less demanding games like Starcraft or whatever you would get hundreds of fps in, where the clocks and voltages would plummet due to there not being a demanding enough load placed on the card, causing stability issues.

 

Basically, fuck GPU Boost and fucking Nvidia for developing and not implementing it properly, disable that shit with a BIOS edit or K-Boost in EVGA Precision X.

You never want fluctuating voltages and clocks when you're actually gaming, unless your cooling solution is pure dog shit and you actually will overheat without it.

I just leave my GPU on prefer maximum performance in the drivers.

Link to post
Share on other sites

My card has a clock speed of 1590mhz.

It's unlikely yours is higher, if it is, congrats man!

 

GPU Boost will throttle my card down and cause it to crash at stock clocks. Disabled I have no problems.

I've read hundreds of posts from different users experiencing the same issues, looking for ways to fix/disable the broken feature.

K-Boost and BIOS edits are the only options, no easy power saving disable option like Intel provides with EIST or C-States etc.

 

Also, I'm sorry you don't understand what you're talking about, but that's not reason to just say everything I've said is bullshit.

How about proving me wrong instead of saying I'm wrong, that would be a nice fucking change on these forums.

You could have a defective card.

Link to post
Share on other sites

Yes. Because some cards are naturally better than others because of the quality of the silicon and other factors. The advertised clock speed is a conservative estimate.

I get that but my main concern is stability, for example my GTX 970 G1 is rated for 1392Mhz @ about 1.2ish volts but if my card was "better" and it ran at 1450Mhz with the same voltage how can they guarantee that exact card will be stable? Surele some cards will be but not every card will.

Link to post
Share on other sites

I get that but my main concern is stability, for example my GTX 970 G1 is rated for 1392Mhz @ about 1.2ish volts but if my card was "better" and it ran at 1450Mhz with the same voltage how can they guarantee that exact card will be stable? Surele some cards will be but not every card will.

the idea behind GPU boost is that it doesn't boost your card to unstable levels and doesn't go over stock voltage. there's an algorithm

Current PC: Origin Millennium- i7 5820K @4.0GHz | GTX 980Ti SLI | X99 Deluxe 

 

Link to post
Share on other sites

Just be happy y'all even have GPU boost. If a high end AMD gpu(390 and up) even tried to go 100mhz over stock without voltage, you'd have an artifacting mess and crashes all over.

Current PC: Origin Millennium- i7 5820K @4.0GHz | GTX 980Ti SLI | X99 Deluxe 

 

Link to post
Share on other sites

Great minds thinking alike and shit lol. I'd bet my boot drive it does.

It probably does because GPU boost needs to determine the temps first. My friend's old reference 980Ti had an ASIC score of 60% and was hitting 84c+ at stock and could barely boost lol.

Current PC: Origin Millennium- i7 5820K @4.0GHz | GTX 980Ti SLI | X99 Deluxe 

 

Link to post
Share on other sites

I just leave my GPU on prefer maximum performance in the drivers.

So do I, this doesn't disable GPU Boost.

 

You could have a defective card.

I don't, I have a card that needs a little bit more voltage for stability and GPU Boost says "I don't give a fuck, Imma lower it so much you'll be unstable"

Link to post
Share on other sites

I just bought a Gigabyte 980ti Xtreme (air), and I was quite shocked when I opened MSI afterburn and saw that the default clock was running at 1430mhz with memory at 8009mhz.  I thought it was a fluke in the software and had to verify in GPUZ, same thing. Also, just for an FYI, my ASIC is 76.1 and stock voltage is 1.18. 

 

I am pretty new to overclocking, but trying to make sure I test for all the bugs while still in my initial grace period.  I tried overclocking a little bit, and it will go up to 1528 core and 8150 memory without adjusting the power at all. The highest I have tried, without adjusting the power all, was 1550, and while it doesn't crash, I get a couple green screen flickers every once and awhile (Firestrike demo version).  Not a lot, but obvious enough.  I noticed that the stock fan setting was letting it get into the 70s without even really ramping up, so I did a custom, even though I dont know what I am doing, just made sure it was twice as high and steep.  It didn't really increase the noise all that much, but the temps stay in the 40s to 50s now. 

 

One question is how to test to see if it might have throttle issues if the only game I have is CS GO.  Another question would be what the general safe voltage range is so I can test its true potential?  Im really happy with the card, so I don't want to push my luck, but am curious to see what it can do.  Also, is there any resource that explains a bell curve for ASIC and/or max clocks?  Would be curious to know how my card falls into the overall spectrum.

 

Thanks guys. 

|[ i7 6700k ][ MSI z170a XPower Gaming Titanium MB ][ 16gig G.Skill Trident Z 3000mhz ][ Gigabyte Xtreme GTX 980ti ][ Samsung Evo & Intel SSDs ][ Corsair 540 air ][ XFX Pro 750w Platinum ][ 24" Asus 144hz ]|

Link to post
Share on other sites

Hmm, I might be lucky who knows. Most of my friends run Nvidia and the developer I used to work for worked with Nvidia and this is the first I've ever heard of an issue like this. Thermal throttle in blower style coolers, that I know about. But this, no.

I think its actually a combination of GTX 970 G1 Gaming not being binned correctly (or at all-ridiculously low asic score with mine and it refuses to overclock+only boosts to 1356MHz, and hits 73oC under full load with ambient temps at 21oC)+drivers. I've found that any driver before and after 355.98 causes my GTX 970 to run unstable.

           .;ldkO0000Okdl;.                michael@SUSE-BlackBox
        .;d00xl:^''''''^:ok00d;.            OS: openSUSE 20260405
      .d00l'                'o00d.          Kernel: x86_64 Linux 6.19.11-1-default
    .d0K^'  Okxoc;:,.          ^O0d.        Uptime: 2d 21h 52m
   .OVVAK0kOKKKKKKKKKKOxo:,      lKO.       Packages: 6556
  ,0VVAKKKKKKKKKKKKK0P^,,,^dx:    ;00,      Shell: bash 5.3.9
 .OVVAKKKKKKKKKKKKKk'.oOPPb.'0k.   cKO.     Resolution: 3840x1080
 :KVAKKKKKKKKKKKKKK: kKx..dd lKd   'OK:     DE: KDE
 lKlKKKKKKKKKOx0KKKd ^0KKKO' kKKc   lKl     WM: KWin
 lKlKKKKKKKKKK;.;oOKx,..^..;kKKK0.  lKl     GTK Theme: Breeze-Dark [GTK2], Breeze [GTK3]
 :KAlKKKKKKKKK0o;...^cdxxOK0O/^^'  .0K:     Icon Theme: breeze-dark
  kKAVKKKKKKKKKKKK0x;,,......,;od  lKP      Disk: 13T / 22T (60%)
  '0KAVKKKKKKKKKKKKKKKKKK00KKOo^  c00'      CPU: AMD Ryzen 7 5800X3D 8-Core @ 16x 4.55295GHz
   'kKAVOxddxkOO00000Okxoc;''   .dKV'       GPU: AMD Radeon RX 6700 XT (radeonsi, navi22, ACO, DRM 3.64, 6.19.11-1-default)
     l0Ko.                    .c00l'        RAM: 13127MiB / 48094MiB
      'l0Kk:.              .;xK0l'          
         'lkK0xc;:,,,,:;odO0kl'             
             '^:ldxkkkkxdl:^'    

 

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×