After the first synthetic result was shared, an alleged gaming performance result of the NVIDIA GeForce RTX 4090 graphics card has now been unveiled, showing a 2.2x gain over the RTX 3090 with ray tracing and DLSS enabled at 4K resolution.
NVIDIA GeForce RTX 4090 graphics card over twice as fast as RTX 3090 under control, reportedly with ray tracing and DLSS enabled at 4K resolution
The NVIDIA GeForce RTX 4090 graphics card will be the green team’s flagship for 2022. There have been various leaks and rumors surrounding the card throughout the year, but as far as we can tell the launch is more or less expected around Q4 2022. With 3-4 months to go, the rumors are now shifting the focus from specs to performance numbers.
3090 in Control with the same settings makes 72 FPS on Avg, so 2.2x gain pic.twitter.com/UdK2kCEXhO
– Hassan Mujtaba (@hms1193) July 19, 2022
While we’ve heard of an alleged synthetic performance benchmark for the NVIDIA GeForce RTX 4090 showing an almost 2x performance increase in the 3DMark Time Spy Extreme benchmark, an alleged gaming benchmark has now been shared by @XpeaGPU stating that the graphics card is tested and delivers over 160+ FPS in Control at Ultra (RT + DLSS) in 4K. The specific model tested is reported as an “AD102 High-Power Draw” variant, which could indicate that it is an overclocked model.
We get a score of up to 2.2x compared to the NVIDIA GeForce RTX 3090 in high preset with RT and DLSS. So if we ran the previous flagship in the Ultra preset we would get up to 2.5 x gain which is just ridiculously large. Our benchmark also had the DLSS factor set to the quality preset, and we don’t know what the allegedly leaked benchmark had DLSS set to. Quality mode favors IQ over FPS, while Performance mode does the opposite and lowers IQ a bit.
Overclocked 3090 Ti +100 / +1200
Native / DLSS 1440p / DLSS 1253p / DLSS 1080p
~49 FPS / ~86 FPS / ~102 FPS / ~120 FPS pic.twitter.com/MQeRtMsXuJ
– Hassan Mujtaba (@hms1193) July 19, 2022
So it looks like you’re expecting 2x to 50% better performance for the RTX 4090, depending on the DLSS version used for the alleged benchmark.
The NVIDIA GeForce RTX 4090 was expected to deliver a 2-2.5x performance boost, but that was only expected in synthetic benchmarks. If playing performance also comes close to those numbers, Ada Lovelace may be one of the green team’s biggest leaps in performance we’ve seen to date. All in all, if this result is correct, then avid gamers will definitely be in for a treat as 4K gaming at 100+ FPS becomes a reality.
Many Thanks. I don’t know how to discuss this 800w beast. In fact, its circuit board is very small, but its cooler is very large. 😱😱😱 Waiting for your good news. https://t.co/6nKJHVtwc1
— kopite7kimi (@kopite7kimi) July 19, 2022
In related news, Kopite7kimi states that an AIB model of NVIDIA GeForce RTX 4090 graphics card can score 20,000 points in the same 3DMark Time Spy Extreme benchmark. That’s a slight 2x gain over the RTX 3090 Ti and a 70-80% jump over the current-gen flagship RTX 3090 Ti graphics card.
I think an OC version of RTX 4090 from AIC can reach 20000 points in TSE Graphics. I will take it for granted.
— kopite7kimi (@kopite7kimi) July 19, 2022
NVIDIA GeForce RTX 4090 “Rumoured” Specs
The NVIDIA GeForce RTX RTX 4090 graphics card is expected to be powered by the best AD102-300 GPU, but only the “Ti” variant will feature the full chip. The GeForce RTX 4090 will use a slightly stripped down configuration.

The NVIDIA GeForce RTX 4090 will use 128 SMs of the 144 SMs for a total of 16,384 CUDA cores. The GPU comes with 96MB of L2 cache and a total of 384 ROPs, which is just insane. Clock speeds are not yet confirmed, but considering that the TSMC 4N process is used, we expect clock speeds to reach or even exceed the 3.0 GHz range. The core count is up 56% over the RTX 3090, so many things like clocks, cache, architecture, and various other technologies boost performance by such a huge margin.
As for memory specs, the GeForce RTX 4090 is expected to rock 24GB GDDR6X capacities clocked at 21Gbps over a 384-bit bus interface. This provides a bandwidth of up to 1 TB/s. This is the same bandwidth as the existing RTX 3090 Ti graphics card, and as far as power consumption goes, the TBP is said to be rated at 450W, meaning TGP might be lower. The card is powered from a single 16-pin connector that delivers up to 600W of power. It’s likely that we’ll see custom designs in excess of 500W, as we’ve seen with the RTX 3090 Ti.
The graphics card is expected to launch in October 2022, followed by the rest of Ada Lovelace’s line of GPU-powered graphics cards.
NVIDIA GeForce RTX 4090 Ti & RTX 4090 “Preliminary” Specifications:
Graphics card name | NVIDIA GeForce RTX 4090 Ti | NVIDIA GeForce RTX 4090 | NVIDIA GeForce RTX 3090 Ti | NVIDIA GeForce RTX 3090 |
---|---|---|---|---|
GPU name | Ada Lovelace AD102-350? | Ada Lovelace AD102-300? | Amp GA102-350 | Amp GA102-300 |
process node | TSMC4N | TSMC4N | Samsung 8nm | Samsung 8nm |
The size | ~600mm2 | ~600mm2 | 628.4mm2 | 628.4mm2 |
transistors | open | open | 28 billion | 28 billion |
CUDA cores | 18432 | 16128 | 10752 | 10496 |
TMUs / ROPs | open / 384 | open / 384 | 336 / 112 | 328 / 112 |
Tensor / RT cores | TBD / TBD | TBD / TBD | 336 / 84 | 328 / 82 |
base clock | open | open | 1560MHz | 1400MHz |
boost clock | ~2800MHz | ~2600MHz | 1860MHz | 1700MHz |
FP32 calculation | ~103 TFLOPs | ~90 TFLOPs | 40 TFLOPs | 36 TFLOPs |
RT TFLOPs | open | open | 74 TFLOPs | 69 TFLOPs |
Tensor TOPs | open | open | 320 TOPS | 285 TOPS |
storage capacity | 24GB GDDR6X | 24GB GDDR6X | 24GB GDDR6X | 24GB GDDR6X |
memory bus | 384-bit | 384-bit | 384-bit | 384-bit |
memory speed | 24.0 Gbps | 21.0 Gbps | 21.0 Gbps | 19.5 Gbps |
bandwidth | 1152GB/s | 1008GB/s | 1008GB/s | 936 Gbps |
TGP | 600W | 450W | 450W | 350W |
Price (RRP / FE) | $1999? | $1499 US? | $1999 | $1499 |
Home (Availability) | 2023? | October 2022? | March 29, 2022 | 09/24/2020 |
Products mentioned in this post