Slashdot Mirror

← Back to Stories (view on slashdot.org)

Nvidia Scanner Brings One-Click Overclocking To Its GeForce RTX Graphics Cards (pcworld.com)

Posted by BeauHD on from the push-it-to-the-limit dept.

Nvidia's new "Scanner" tool for the company's newest GeForce RTX 2080 graphics cards will provide one-click overclocking. PCWorld reports: Nvidia Scanner isn't actually a tool you can download. Instead, it's an API that developers can implement, similar to how current GeForce overclocking software relies on Nvidia's NVAPI. Tom Peterson, Nvidia's director of technical marketing, says all of the major overclocking programs will implement Scanner. You simply press the Test button, and the software starts walking through your graphics card's volt frequency curve, running arithmetic tests all the while. If the overclock starts pushing too far, Nvidia Scanner will discover a math error before your card crashes. When that happens, Scanner ramps up your card's voltage and starts testing again. After about 20 minutes, Scanner will have a complete understanding of your RTX card's capabilities, and automatically generate an overclocking profile built to squeeze as much performance as possible out of it without crashing. Easy-peasy. PCWorld's Brad Chacos mentions a demonstration where "Nvidia's Tom Peterson showed Nvidia Scanner pushing the GeForce RTX 2080 -- which ships with a 1,710MHz boost clock -- all the way to 2,130MHz at 1,068mV."

18 of 39 comments (clear)

  1. That's no Overclocking by klingens · · Score: 1

    That is at most the usual nvidia branded boost in different clothes.

    nvidia is the worst company by far regarding any kind of video card overclocking: it hurts their bottomline.
    Nvidia were the first to introduce signed VGA-BIOSes so no more Mod BIOS.
    Nvidia were the first to introduce to introduce hard powerlimits, so one has to solder resistors on the PCB to increase this limit.

    1. Re:That's no Overclocking by JohnStock · · Score: 2

      By the very description of what they are doing, that IS overclocking.

  2. One click? by Anonymous Coward · · Score: 1

    Back in my day we just pressed the TURBO button to crank my rig up from 16Mhz to 33Mhz.

    1. Re:One click? by KiloByte · · Score: 1

      The turbo button was actually a misnomer, it was actually a "slow button" to allow for backwards compatibility with older games

      Not on oldest machines, where the default was 4.77MHz but you could switch to an 8MHz mode that was "not recommended to use for long, and may cause instability". It was Ctrl-Alt-Minus rather than a physical button.

      Later on, when actual physical buttons appeared, one of machines I had featured a button that popped in/depressed itself when you pressed the Ctrl-Alt-Minus combination; kid me considered that effect insanely cool.

      --
      The creatures outside looked from Alt-Right to Antifa; but already it was impossible to say which was which.
    2. Re:One click? by TheRealQuestor · · Score: 2

      Pretty sure they were all LEDS back then :) At least all the 8088's I had with the turbo button were

    3. Re:One click? by Dutch+Gun · · Score: 1

      The first time I saw a PC actually turn it's own power off, I thought that was voodoo. A button that physically changes state on its own beats that for sheer geekiness.

      --
      Irony: Agile development has too much intertia to be abandoned now.
  3. Them errors by DigiShaman · · Score: 2

    Not long enough for testing. Even after 20 minutes of passing, the card would be right at that threshold to where if even the ambient temps got hotter hours later, it could throw an error and lead to data / result corruption.

    I much prefer having some safety distance to ensure running integrity.

    --
    Life is not for the lazy.
    1. Re:Them errors by Mashiki · · Score: 1

      Those are for consumer GPUs in which errors don't matter if they are not very common

      Remember the massive TDR(driver reboot) problem with nvidia cards back in the 500/600/700 days? No? It was because their drivers were undervolting during low draw states, even during gaming in order to reduce the amount of heat that GPU was producing. This caused the infinite TDR/blackscreen problem that went on for years, until nvidia finally admitted that yes, it wasn't the customers PC's, it was us. It was bad enough they were paying to have people ship their PC's directly to nvidia for testing.

      --
      Om, nomnomnom...
    2. Re:Them errors by thegarbz · · Score: 1

      Even after 20 minutes of passing, the card would be right at that threshold

      How do you know already what this unreleased software does and does not do?

    3. Re: Them errors by Mashiki · · Score: 1

      I have a 780 and still have this problem: my driver reboots randomly. No rhyme or reason behind it.

      Two things you can do. You can get the "bin-ID" off the GPU and then look up through the database to see what the cards core vmm/gpu voltages should be. Or use rivatuner, and up the core voltage by just a tiny bit then lock it so that the drivers can't change it. You shouldn't have anymore problems. The "bin-ID" is the number that the GPU was dumped into when QC testing, in general a lower number = higher quality, while higher means lower quality. That doesn't always translate into a stable card though.

      Check guru3d there's a big thread on it that can help you solve the problem.

      --
      Om, nomnomnom...
  4. Re:House fire cards run hot enough as it is by tepples · · Score: 1

    In some cases, the marginal money through completing a task using fewer hours of labor or gaining a greater advantage in an esport makes up for the marginal use of electric power.

  5. all the way to 2,130MHz at 1,068mV by SpankiMonki · · Score: 1

    "Nvidia's Tom Peterson showed Nvidia Scanner pushing a cherry-picked GeForce RTX 2080 -- which ships with a 1,710MHz boost clock -- all the way to 2,130MHz at 1,068mV."

    FTFY

  6. Re:I'd prefer something that tests for undervolt by retchdog · · Score: 1

    you can do that, y'know. even in linux.

    --
    "They were pure niggers." – Noam Chomsky
  7. I want ALL the flops! by elrous0 · · Score: 1

    More flops! I want to bathe in flops! I want flops for breakfast, lunch, and dinner. I want to marry flops and make flop babies. I want to cheat on flops, get left by flops for a younger guy, then get slapped with flop alimony that I can't possibly afford--leading me to end my life by jumping off a flop bridge.

    ALL YOUR FLOPS!

    --
    SJW: Someone who has run out of real oppression, and has to fake it.
  8. Re:House fire cards run hot enough as it is by thegarbz · · Score: 1

    How much inefficiency is introduced by running them faster?

    No worse than buying a better memory card in the first place.

  9. Re:I'd prefer something that tests for undervolt by thegarbz · · Score: 1

    I'd rather have it run a bit slower at a lower voltage to use less power and create less heat and fan noise.

    If your card isn't doing that already you have something fundamentally broken in your computer.

    Have you monitored your graphics card at load and at idle? I mean my GPU core clock is currently at 120MHz and 0.62V and no I didn't miss a zero in the speed. If I actually run a game it spikes to 1860MHz and 1.03V. At idle if you're creating heat or fan noise then there's something broken in your GTX1080.

  10. Re:I'd prefer something that tests for undervolt by Khith · · Score: 1

    If you want to undervolt to make your Nvidia card work less during games or any sort of GPU-intensive app, get MSI Afterburner and choose one of the options below. Note that I'm using my 1080Ti as an example.

    Make a Voltage/Frequency curve with Ctrl-F. The lowest you can manually set is 800 mV, and with that I set my frequency to 1632 MHZ. This is dependent on the card type and also the specific card that you have, so start lower than those if you try this.

    Slide the Power Limit downward, and it will bring the voltage and frequency down. I've seen it as low as 693 mV. This is a much easier and safer way to underclock and undervolt, though the numbers will shift around somewhat, unlike the first option.

    Both of these methods will let your card run cooler and quieter in games (and it's pretty much necessary if mining cryptocurrency), and when your card is idle it'll still drop down to its idle voltage and frequency. (139 MHz and 650 mC) I recommend creating a profile for each one and testing them to see which works best for you. Test for stability too! If things crash or show visual artifacts, dial down the frequency. This is mainly an issue if you're locking at a specific voltage and frequency like in the first method. You can also mess around with the memory clock and drop as low as -502 MHz

    Also make sure to cap your FPS if you're running a game that normally goes well above your refresh rate. You don't need to run Minecraft at 300 FPS. Combine that with the above methods and your card won't be doubling as a space heater.

  11. "automatically generate..." by Red_Chaos1 · · Score: 1

    "...an overclocking profile built to squeeze as much performance as possible out of it without crashing."

    I've heard these words before. I believed them. I ran the tests. The computer hard locked. Every time.

    I no longer believe it when I see/hear these words.