NVIDIA today unveiled its next-gen GeForce RTX 40 series graphics cards for enthusiasts and high-end gamers, starting with the RTX 4080 and RTX 4090. Company CEO Jensen Huang gave us the first official look at the capabilities of the Ada Lovelace architecture, including DLSS 3.0, Dual-NVENC, and Shader Execution Reordering (SER).
The age of RTX ray tracing and neural rendering is in full steam and our new Ada Lovelace
Jensen Huang, NVIDIA CEO
architecture takes it to the next level. Ada provides a quantum leap for gamers and paves the way for creators of fully simulated worlds. With up to 4x the performance of the previous generation, Ada is setting a new standard for the industry.
Jensen threw big numbers at the audience, claiming a somewhat unrealistic 4x performance leap over the existing RTX 30 “Ampere” parts. I’m not saying that NVIDIA is lying, but they aren’t being completely honest either. We’re probably looking at the ray-traced performance of the next-gen RTX 40 series GPUs, and that too when paired with DLSS 3.0.
NVIDIA RTX 4080 and 4090: Specs, Prices, and Release Dates
Let’s talk about the specifications, pricing, and launch dates. The specs are more or less set in stone by now with leakers sharing every little detail on the next-gen GPUs pretty much every other week:
GPU | GA102 | AD102 | RTX 4090 | AD103 | RTX 4080 | RTX 4070 Ti (AD104) | RTX 4070 |
---|---|---|---|---|---|---|---|
Arch | Ampere | Ada Lovelace | Ada Lovelace | Ada Lovelace | |||
Process | Sam 8nm LPP | TSMC 5nm | TSMC 5nm | TSMC 5nm | |||
GPC | 7 | 12 | 11 | 7 | 7 | 5 | 5 |
TPC | 42 | 72 | 64 | 42 | 40 | 30 | 30 |
SMs | 84 | 144 | 128 | 84 | 80 | 60 | 60 |
Shaders | 10,752 | 18,432 | 16,384 | 10,752 | 9,728 | 7,680 | 7,680 |
TP | 37.6 | ~100 TFLOPs? | 83 TFLOPs | ~50 TFLOPs | 47 TFLOPs? | ~35 TFLOPs | 35 TFLOPs? |
Memory | 24GB GDDR6X | 48GB GDDR6X | 24GB GDDR6X | 16GB GDDR6X | 12GB GDDR6X | ||
L2 Cache | 6MB | 96MB | 72MB | 64MB | 48MB | ||
Bus Width | 384-bit | 384-bit | 256-bit | 192-bit | |||
TGP | 350W | 600W | 450W | 450W | 285-340W | 300W | 285W |
Launch | Sep 2020 | Sept 22? | Sept 22? | Q1 2023? |
The RTX 4090 will leverage the partially enabled AD102 die with 16,384 FP32 cores spread across 128 SMs, 64 TPCs, and 11 GPCs. The GPU will be paired with 24GB of GDDR6X 24Gbps memory across a 384-bit bus, complemented by 72MB of L2 cache. As for the TBP, we’re looking at an upper limit of 450W. Coming to the price, the RTX 4090 will cost $1,499 (US), the same as its predecessor. NVIDIA plans to launch the Ada flagship on the 12th of October.
The RTX 4080 will be notably slower than the RTX 4090, leveraging the smaller AD103 graphics core. With the RTX 4080 16GB, you get 9,728 shaders spread over 80 SMs paired with 16GB of GDDR6X 22.5Gbps memory across a 256-bit bus and an L2 cache of 64MB. The 12GB variant will cut the core count to 7,680 cores and the bus width to 192-bit. The RTX 4080 16GB will cost $1,199 with a TGP of 340W while the 12GB model will start at $899 with a TGP of 285W. These two are set to hit retail in the first half of November.
As you can see, NVIDIA has priced the RTX 4080 16GB in the same bracket as the RTX 3080 Ti and the RTX 4080 12GB higher than both the variants of the RTX 3080. The RTX 4090 is the only part to retain the pricing of its predecessor, likely due to the impending launch of the Radeon RX 7900 XT. The volatile prices of contemporary GPUs are the primary reason (other than profits) for the increased pricing of the 4080, a card that will be the go-to SKU for most enthusiast gamers.