After the embarrassing “unlaunch” of the GeForce RTX 4080 12GB, the Ada Lovelace lineup looks more like a one-man army than a family. The RTX 4090 is a solid GPU priced at $1,599 (a little on the high side), and then you have the RTX 4080 16GB, which is half as fast but costs $1,199 (just 25% cheaper). Meanwhile, the scraped 12GB variant (likely launching as the RTX 4070 Ti) @ $699-799 is slower than the RTX 3090 Ti and on par with the RTX 3080 Ti.
Well, it turns out that NVIDIA had completely different plans for the RTX 4080 and 4070 at the beginning:
According to Tom from MLID, board partners were told that the AD103 die that now powers the RTX 4080 16GB was supposed to become RTX 4070, much like the previous generation. That means NVIDIA nuked the original RTX 4080 (based on the AD102) and promoted the 4070 as a replacement with 20-25% less performance. This would align with what various insiders on Twitter have been claiming over the past year and explain why there is such a massive delta between the RTX 4080 and 4090.
| GPU | GeForce RTX 4090 | GeForce RTX 4080 16G | RTX 4070 | GeForce RTX 3090 Ti | |
|---|---|---|---|---|---|
| GPU Name | AD102 | AD103 | AD104 | AD104 | GA102 |
| SMs | 128 | 76 | 60 | 46 | 80 |
| FP32 Cores | 16384 | 9728 | 7680 | 5888 | 10240 |
| TMUs/ROPs | 512 / 176 | 320 / 112 | 240 / 80 | 184/64 | 320 / 112 |
| Tensor/RT Cores | 512 / 128 | 304 / 76 | 240 / 60 | 184/46 | 320 / 80 |
| Base Clock | 2230 MHz | 2210 MHz | 2310 MHz | 1365 MHz | |
| Boost Clock | 2520 MHz | 2510 MHz | 2610 MHz | 1665 MHz | |
| Memory Buffer | 24 GB G6X | 16 GB G6X | 12 GB G6X | 12GB G6X | 24 GB G6X |
| Bus | 384-bit | 256-bit | 192-bit | 192-bit | 384-bit |
| Memory Speed | 21Gbps | 23Gbps | 21Gbps | 21Gbps | 19Gbps |
| Bandwidth | 1008 GB/s | 736 GB/s | 504 GB/s | 504GB/s | 912 Gbps |
| L2 Cache | 72MB | 64MB | 48MB | 36MB | 6MB |
| TBP | 450W | 320W | 285W | 250W | 350W |
| MSRP | $1599 | $1199 | $899 | $649? | $1199 |
| Launch | Oct 2022 | Nov 2022 | 5th Jan 2023 | 3rd June 2021 |
Source:
