NVIDIA RTX 4070 Possibly Downgraded to AD106, A 35% Decrease In Core Count Expected
Hardware leaker Kopite on Twitter has confirmed that NVIDIA has scrapped all plans of launching the AD104 based RTX 4070. The model, in particular, is AD104-275 which was rumored to be canceled before Ada’s launch. However, this tweet confirms that NVIDIA is no longer planning to use this GPU (AD104-275).
RTX 4070 Updated Specifications
The ‘original’ RTX 4070 is no longer under consideration by NVIDIA as per Kopite. Now this can mean one of 2 things. The RTX 4070 will either be downgraded or upgraded; sadly, the former appears to be probable. The RTX 4070 Ti (Rebranded RTX 4080 12GB) will use the AD104-400 GPU which does not leave much room for the 4070 for an upgrade.
We won't see the original RTX 4070(AD104-275).
— kopite7kimi (@kopite7kimi) December 1, 2022
Harukaze also had reservations regarding NVIDIA’s latest move. In a tweet, the leaker suspects that NVIDIA may push the RTX 4070 to AD106. This matches our aforementioned theory.
에이 설마 AD104가 아니라 AD106을 쓴다거나 그러겠어?
에이 설마……아니지?
— 포시포시 (@harukaze5719) December 1, 2022
Using Google Translate, the tweet reads:
Hey, could it be that AD106 is used instead of AD104? Ayy, can’t it be… isn’t it?~Harukaze
Conclusion
If this leak along with Harukaze’s guess turn out to be true, then AMD will have a clear chance to dethrone NVIDIA in the budget segment. It’s too early to make guesses on our own, so we should wait for NVIDIA’s call at CES 2023.
SKU | Chip | FP32/CUDA | SMs | Max Clock | Cache | Memory Bus | VRAM | Memory Spec | Speed (Gbps) | TDP | |||
RTX 4000 Titan | AD102-450 | 18432 | 144 | 3.0GHz+? | 96MB? | 384/382-bit | 48GB | GDDR6X | 24 | ~800W | |||
RTX 4090 Ti | AD102-350 | 18176 | 142 | 3.0 GHz? | 96MB | 382-bit | 24GB | GDDR6X | 24 | 600W? | |||
RTX 4090 | AD102-300-A1 | 16384 | 128 | 2.52GHz | 96MB | 384-bit | 24GB | GDDR6X | 21 | 450W+ (TGP) / 660W (Max TGP) | |||
RTX 4080 Ti | AD102 | 14848 | 116 | 2.7 GHz? | 80MB? | 320-bit | 20GB | GDDR6X | 23 | 420W | |||
RTX 4080 (Variant 1) | AD103-300-A1 | 9728 | 76 | 2.505GHz | 64MB | 256-bit | 16GB | GDDR6X | 22.5 | 320W(TGP)/ 516W (Max TGP) | |||
RTX 4070 Ti (4080 12GB Revamped) | AD104-400-A1 | 7680 | 60 | 2.61GHz | 48MB | 192-bit | 12GB | GDDR6X | 21 | 285W (TGP) /366W (Max TGP)? | |||
RTX 4070 | AD104-300? | 7168 | 56 | 2.61GHz? | 48MB? | 192-bit/160-bit | 12GB? | GDDR6X | 21? | 250W | |||
RTX 4070 (Rebranded 4060) | AD106 | 4608 | 36 | ? | ? | 160-bit? | 10GB | GDDR6X | 21? | 250W | |||
RTX 4060 Ti | AD104 | 6656 | 52 | 3.2 GHz? | 40MB? | 160-bit | 10GB | GDDR6X (?) | 18 | 200W | |||
RTX 4060 | AD106 | 4352 | 34 | 3.0 GHz? | 32MB | 128-bit | 8GB | GDDR6X (?) | 18 | 180W |