Nvidia Shares GeForce RTX 4060 Performance Numbers

Nvidia GeForce game ready driver update
(Image credit: Nvidia)

Nvidia has published official benchmark results of its upcoming GeForce RTX 4060 graphics card just a week ahead of its launch on June 29. The new $299 Ada Lovelace-based graphics card is shown to be across-the-board faster than its predecessor based on the Ampere architecture, but there is catch: the newcomer shows its most significant advantages with AI frame generation enabled. Without it, it is merely 20% faster, according to Nvidia.

 

(Image credit: Nvidia)

Nvidia's GeForce RTX 4060 graphics card is based on the AD106 GPU with 3072 CUDA cores enabled that has peak FP32 compute throughput of 15 TFLOPS, which is just 15% higher compared to GeForce RTX 3060 with its 13 FP32 TFLOPS. But the AD106 has noticeable advantages over GA106 in the form of massively improved ray tracing performance (+40%) and Tensor compute throughput (+137%). The latter can be used for AI, advanced DLSS 3 upscaling, and AI image generation workloads. We'll see if that's enough to make it one of the best graphics cards.

(Image credit: Nvidia)

We can make guesses why Nvidia decided to balance its Ada Lovelace microarchitecture the way it balanced it, but it is obvious that the company will use benefits — DLSS 3 and AI image generation — that is has to outshine predecessors and competitors in games.

That said, it is not particularly surprising that Nvidia demonstrated its GeForce RTX 4060 with DLSS 3 and image generation enabled in as many games as possible, showing rather dramatic performance gains compared to its GeForce RTX 3060. This is indeed a major improvement of the new GeForce RTX 4060 graphics card as it can enable high framerates with all the eye candy enabled in the latest games, something the GeForce RTX 3060 just cannot do.

(Image credit: Nvidia)

The company admits that far not all games support AI frame generation and this is where its GeForce RTX 4060 is only 20% faster than its predecessor. A 20% improvement is still not bad, only it is just not something one would expect from a new generation product based on the all-new architecture. 

Nvidia considers lower power consumption of its GeForce RTX 4060 as another advantage of its new board as it will allow to save some money. Yet, this advantage is less obvious than performance gains.

(Image credit: Nvidia)

Leading graphics cards manufacturers like Asus, Colorful, Gainward, Galax, Gigabyte, Inno3D, KFA2, MSI, Palit, PNY, and Zotac will be releasing the GeForce RTX 4060 graphics cards starting June 29. Nvidia's recommended price for GeForce RTX 4060 boards is $299, but expect products that will carry different price tags as well.

Anton Shilov
Freelance News Writer

Anton Shilov is a Freelance News Writer at Tom’s Hardware US. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.

  • Metal Messiah.
    Nvidia's GeForce RTX 4060 graphics card is based on the AD106 GPU with 3072 CUDA cores enabled that has peak FP32 compute throughput of 15 TFLOPS, which is just 15% higher compared to GeForce RTX 3060 with its 13 FP32 TFLOPS. But the AD106 has noticeable advantages over GA106 in the form of massively improved ray tracing performance (+40%) and Tensor compute throughput (+137%). The latter can be used for AI, advanced DLSS 3 upscaling, and AI image generation workloads.

    No, the RTX 4060 will utilize the AD107-400-A1 GPU, the fully enabled AD107 Ada die. The same configuration is used by the GeForce RTX 4060 Laptop GPU. Not 106.
    Reply
  • Metal Messiah.
    I will say this again.

    No point in counting DLSS 3 performance boost here, only pure raw rasterized performance if we really want to compare the 4060 with previous gen cards which lack proper support for DLSS3. It is moot.

    Nvidia claims the card offers a 20% improvement over the GeForce RTX 3060 12 GB in traditional rasterization and bigger gains when RT and DLSS are factored in.

    So, based on Nvidia's specs/claims, since we are getting up to a 70% increase (with DLSS 3) and a 20% increase (without DLSS) over the RTX 3060, that says that DLSS 3 is bad or unfair metric for comparison. A lot of disparity here.
    Reply
  • Giroro
    DLSS numbers between generations are completely useless.
    Nvidia is obscuring the real performance of these cards. That's a bad sign.
    Reply
  • PlaneInTheSky
    $300 for a GPU that is already outdated with its pitiful 8GB VRAM seems like an awful deal.

    I just can't stand this treadmill race of increasing hardware prices anymore. The idea of buying an 8GB GPU for $300 just to have to buy a 16GB one in a year to keep up with the latest games seems awful.

    The price of PC hardware has made me reject $60 AAA games and embrace $10-$30 games instead.

    Playing less graphical intense games is
    -cheaper
    -they take up less space on my SSD
    -they are faster to boot up and generally have less performance issues
    -they don't include microtransactions like "AAA" games
    Reply
  • Amdlova
    I want to see some real benchmarks @JarredWaltonGPU make this 4060 suffer!
    Reply
  • Elusive Ruse
    DLSS3 numbers are invalid as long as the overwhelming majority of games don't support it and it comes with a huge input lag penalty. At this point I'm more excited about Intel's next generation of GPUs than anything Nvidia has to offer. There I said it :ROFLMAO:
    Reply
  • bdcrlsn
    Nvidia relying so heavily on DLSS and AI is concerning, because whose to say that each generation following increasingly relies on upscaling technologies at the expense of traditional rasterization rendering.
    Reply
  • PlaneInTheSky said:
    $300 for a GPU that is already outdated with its pitiful 8GB VRAM seems like an awful deal.

    I just can't stand this treadmill race of increasing hardware prices anymore. The idea of buying an 8GB GPU for $300 just to have to buy a 16GB one in a year to keep up with the latest games seems awful.

    The price of PC hardware has made me reject $60 AAA games and embrace $10-$30 games instead.


    So you sell the PC... buy an XBox Series X or PS5 and call it a day.

    When it comes to cutting edge PC gaming... you have to pay the premium. A cutting edge GPU should be expensive...


    PlaneInTheSky said:
    Playing less graphical intense games is
    -cheaper
    -they take up less space on my SSD
    -they are faster to boot up and generally have less performance issues
    -they don't include microtransactions like "AAA" games

    If that's how you wanna justify it... can't really agree about the taking up less space and being faster to boot... but if something loading 1 second faster makes you sleep better... hey... that's cool.

    As for microtransactions... you don't have to click the buy button. I don't.
    Reply
  • Alvar "Miles" Udell
    Sus results (as all first party numbers are). The Witcher 3 with AI frame generation performing at 60fps, when the 4060 Ti without DLSS (and RT) performs at 180fps? That's a pretty pathetic result for a $300 GPU.
    Reply
  • hotaru251
    The company admits that far not all games support AI frame generation and this is where its GeForce RTX 4060 is only 20% faster than its predecessor. A 20% improvement is still not bad, only it is just not something one would expect from a new generation product based on the all-new architecture.
    they also ignore fact its ram limited compared to the last gen meaning anythign that uses that ram is also better.

    frame gen is basically raytracing early days....less than 1% of stuff support it.

    so in 99% of workloads its gonna be maybe 20% better...oh but if the application wants more than 8gb ram also betetr for last gen...


    can already see GN's "biggest fail" award going to this or the ti version.
    Reply