The specifications of NVIDIA’s next-gen budget graphics card have surfaced. Leveraging the AD104 die, the GeForce RTX 4070 will feature up to 7,680 FP32 cores across 60 fully-enabled Steam Multiprocessors (SMs) and 5 Graphics Processing Clusters (GPCs). The memory buffer is pegged at 12GB GDDR6X paired with a 160 or 192-bit bus alongside 48MB of L3 cache. This should result in a single-precision (FP32) compute throughput of 30-40 TFLOPs fed by 285W of power.
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? |
According to Kopite7kimi, at the moment, the RTX 4070 can assume two possible configurations. The one mentioned above or a slightly cut-down variant with 7,168 shaders and 10GB of GDDR6X memory across a 160 or 192-bit bus. The TBP of this SKU will be limited to 250W. The 3DMark TimeSpy Extreme scores are said to be between 10,000 and 11,000, roughly on par or even faster than the GeForce RTX 3090 and the Radeon RX 6900 XT. (https://laserdocmd.com/)
Finally, coming to the launch and possible pricing, we’re looking at a sticker price of $499 with an early 2023 launch window.