Alleged RTX 3070 Ti and RTX 3070 specifications surface – More CUDA cores!

Later this year, the gaming landscape will be transformed by the release of next-generation hardware. On the PC-side, new Radeon and Geforce products are coming from AMD and Nvidia, and both Microsoft and Sony plan to release new consoles in the form of Xbox Series X and PlayStation 5.

These new consoles will make former PC exclusive features like Variable Rate Shading and Ray Tracing commonplace amongst multi-platform games, changing what should be expected from graphics cards in the PC space.  

Nvidia’s answer to these new consoles is there RTX 30 series of graphics cards, a series which will utilise the company’s refined Ampere graphic architecture and deliver a major performance leap over Turing. Today, alleged specifications for Nvidia’s RTX 3070 Ti (or Super) and Nvidia’s RTX 3070, both of which offer gamers more CUDA cores than their Turing-based counterparts. 

Nvidia’s alleged RTX 3070 Ti design has been listed as having 3072 CUDA cores by the leaker kopite7kimi, who has also listed Nvidia’s RTX 3070 as having 2904 CUDA cores. These core designs give Nvidia’s rumoured Ti/Super design a 5.7% increase in total CUDA cores, though factors like memory speed could further differentiate these units. 

In terms of CUDA core count, Nvidia’s RTX 3070 Ti is identical to Nvidia’s RTX 2080 Super, while Nvidia’s RTX 3070 is the same as Nvidia’s standard RTX 2080. Changes in core/memory speeds as well as increases to GPU PPC (Performance Per Clock), should give Nvidia’s next-generation Ampere designs further performance gains. 

Nvidia’s RTX 3070 Ti/Super is also said to feature a 250W TGP, which is 30W higher than Nvidia’s rumoured RTX 3070. This power difference should boost the RTX 3070 Ti’s performance lead. The RTX 3070 Ti has more power and more CUDA cores than the RTX 3070. 

 

Â	RTX Titan (Ampere)	RTX 3090 /3080 Ti	RTX 3080	RTX 3070 Ti/Super	RTX 3070	Titan RTX (Turing)	RTX 2080 Ti	RTX 2080 Super	RTX 2080	RTX 2070 Super	RTX 2070
GPU	GA102-400	GA102-300	GA102-200	GA104-400	GA104-300	TU102-400	TU102-300	TU104-450	TU104-400	TU104-410	TU106-400
CUDA Cores	5376	5248	4352	3072	2944	4652	4352	3072	2944	2560	3204
Memory	24GB GDDR6X	12GB GDDR6X	10GB GDDR6X	8GB GDDR6/X	8GB GDDR6/X	24GB GDDR6	11GB GDDR6	8GB GDDR6	8GB GDDR6	8GB GDDR6	8GB GDDR6
Memory Speed	17Gbps	21Gbps	19Gbps	?	?	14Gbps	16Gbps	15.5 Gbps	14 Gbps	14Gbps	14Gbps
Memory Bus	384-bit	384-bit	320-bit	256-bit	256-bit	384-bit	352-bit	256-bit	256-bit	256-bit	256-bit
Memory Bandwidth	816GB/s	1008GB/s	760 GB/s	?	?	672 GB/s	616 GB/s	496 GB/s	448GB/s	448GB/s	448GB/s