NVIDIA's GeForce RTX 4090 is the first gaming graphics card to achieve over 100 TFLOPs of compute performance. You can also read our full review of the card here.
Breaking The 100 TFLOPs Barrier! NVIDIA GeForce RTX 4090 Becomes The Fastest Gaming Graphics Card For Compute & Fastest Gaming Graphics Card, Period!
Breaking the 100 TFLOPs barrier is no easy feat. Before today, NVIDIA's fastest gaming graphics card, the GeForce RTX 3090 Ti, only delivered 40 TFLOPs of compute horsepower. With the launch of the GeForce RTX 4090, we get close to the 100 TFLOPs barrier but not officially. NVIDIA states that the GeForce RTX 4090 Founders Edition offers 83 TFLOPs at default settings. This means that the card is 17 TFLOPs shy of that 100 TFLOPs mark.
So we decided it was time to test how far we can push the NVIDIA GeForce RTX 4090 Founders Edition with some overclocking. To get to 100 TFLOPs, we first pushed the power limit and temp limit slider all the way to the max and upped the Core and Memory clocks by +275 and +1100 MHz, respectively. This wasn't enough as the card was being limited by its power design. That is when we landed our hands on MSI's latest Afterburner which allowed us to raise the core voltages. At 100%, we saw some performance regression so we had to stick with +55% which showed us some good results.
With the overclock applied on our NVIDIA GeForce RTX 4090 graphics card, we saw a maximum GPU core clock of 3150 MHz on the AD102 Ada GPU, a maximum power draw of 547W and our temps peaked at 69C. All of this was done on air and with no exotic liquid cooling, chillers or LN2 were used.
And behold, we saw the magical number of not 100 but almost 101 TFLOPs right in front of our eyes. To put things into perspective, this is a 22% compute boost over the stock RTX 4090 and a 2.5x compute performance boost over the RTX 3090 Ti. The AD102 GPU also ripped apart the data-center-focused Hopper H100 GPUs by offering over 50% better FP32 performance. Ada Lovelace is truly a game changer and we can definitely see it become a popular compute and AI graphics card when Quadro variants of the said chip launch as the RTX 6000 ADA and L60.
FP32 Compute Horsepower Comparisons (Higher is Better)
NVIDIA GeForce RTX 4090 'Official' Specifications - $1599 US Pricing
The NVIDIA GeForce RTX 4090 will use 128 SMs of the 144 SMs for a total of 16,384 CUDA cores. The GPU will come packed with 72 MB of L2 cache and a total of 176 ROPs which is simply insane.
As for memory specs, the GeForce RTX 4090 will feature 24 GB GDDR6X capacities that will be clocked at 21 Gbps speeds across a 384-bit bus interface. This will provide up to 1 TB/s of bandwidth. This is the same bandwidth as the existing RTX 3090 Ti graphics card and as far as the power consumption is concerned, the TBP is rated at 450W. The card will be powered by a single 16-pin connector which delivers up to 600W of power. Custom models will be offering higher TBP targets.
The NVIDIA GeForce RTX 4090 GPU officially hits retail shelves tomorrow when NVIDIA and custom card partners' designs become available to the public. You can check out our review here.
NVIDIA GeForce RTX 40 Series Official Specs:
Graphics Card Name | NVIDIA GeForce RTX 4090 | NVIDIA GeForce RTX 4090 D | NVIDIA GeForce RTX 4080 | NVIDIA GeForce RTX 4070 Ti | NVIDIA GeForce RTX 4070 | NVIDIA GeForce RTX 4060 Ti | NVIDIA GeForce RTX 4060 |
---|---|---|---|---|---|---|---|
GPU Name | Ada Lovelace AD102-300 | Ada Lovelace AD102-250 | Ada Lovelace AD103-300 | Ada Lovelace AD104-400 | Ada Lovelace AD104-250 | Ada Lovelace AD106-350 | Ada Lovelace AD107-400 |
Process Node | TSMC 4N | TSMC 4N | TSMC 4N | TSMC 4N | TSMC 4N | TSMC 4N | TSMC 4N |
Die Size | 608mm2 | 608mm2 | 378.6mm2 | 294.5mm2 | 294.5mm2 | 190.0mm2 | 146.0mm2 |
Transistors | 76 Billion | 76 Billion | 45.9 Billion | 35.8 Billion | 35.8 Billion | 22.9 Billion | TBD |
CUDA Cores | 16384 | 14592 | 9728 | 7680 | 5888 | 4352 | 3072 |
TMUs / ROPs | 512 / 176 | TBD | 320 / 112 | 240 / 80 | 184 / 64 | 136 / 48 | TBD |
Tensor / RT Cores | 512 / 128 | 456 / 128 | 304 / 76 | 240 / 60 | 184 / 46 | 136 / 34 | TBD |
L2 Cache | 72 MB | 72 MB | 64 MB | 48 MB | 36 MB | 32 MB | 24 MB |
Base Clock | 2230 MHz | 2280 MHz | 2210 MHz | 2310 MHz | 1920 MHz | 2310 MHz | 1830 MHz |
Boost Clock | 2520 MHz | 2520 MHz | 2510 MHz | 2610 MHz | 2475 MHz | 2535 MHz | 2460 MHz |
FP32 Compute | 83 TFLOPs | TBD | 49 TFLOPs | 40 TFLOPs | 29 TFLOPs | 22 TFLOPs | 15 TFLOPs |
RT TFLOPs | 191 TFLOPs | TBD | 113 TFLOPs | 82 TFLOPs | 67 TFLOPs | 51 TFLOPs | 35 TFLOPs |
Tensor-TOPs | 1321 TOPs | TBD | 780 TOPs | 641 TOPs | 466 TOPs | 353 TOPs | 242 TOPs |
Memory Capacity | 24 GB GDDR6X | 24 GB GDDR6X | 16 GB GDDR6X | 12 GB GDDR6X | 12 GB GDDR6X | 8-16 GB GDDR6 | 8 GB GDDR6 |
Memory Bus | 384-bit | 384-bit | 256-bit | 192-bit | 192-bit | 128-bit | 128-bit |
Memory Speed | 21.0 Gbps | 21.0 Gbps | 23.0 Gbps | 21.0 Gbps | 21.0 Gbps | 18.0 Gbps | 17.0 Gbps |
Bandwidth | 1008 GB/s | 1008 GB/s | 736 GB/s | 504 GB/s | 504 GB/s | 288 GB/s (554 GB/s Effective) | 272 GB/s (453 GB/s Effective) |
TBP | 450W | 425W | 320W | 285W | 200W | 160-165W | 115W |
Price (MSRP / FE) | $1599 US / 1949 EU | 12,999 RMB (China-Only) | $1199 US / 1469 EU | $799 US | $599 US | $399-$499 US | $299 US |
Price (Current) | $1599 US / 1859 EU | 12,999 RMB (China-Only) | $1199 US / 1399 EU | $799 US | $599 US | $399-$499 US | $299 US |
Launch (Availability) | 12th October 2022 | 28th December 2023 | 16th November 2022 | 5th January 2023 | 13th April 2023 | 24th May / 18th July 2023 | 29th June 2023 |