Last Updated on July 12, 2024 by Nahush Gowda
NVIDIA has finally announced its RTX 40-series “Super” refresh with three new high-end GPUs: the RTX 4070 SUPER, 4070 Ti SUPER, and 4080 SUPER. These upgrades replace the existing RTX 4070 Ti and 4080 while offering better performance for maxed-out 1440p and 4K gaming with ray tracing, thanks to boosted specs and DLSS 3 tech.
The new graphics card will launch in a staggered manner throughout January with RTX 4070 Super being the first one to enter the market on January 17. It will be followed by the RTX 4070 Ti Super (stupid nomenclature, by the way) on January 24 and finally, the RTX 4080 Super on January 31.
Here’s a breakdown of each new GPU announced.
RTX 4070 Super
- Price: $599
- Available on: January 17
Specifications
- Chip: AD104 (5nm TSMC process)
- Streaming Multiprocessors (SMs): 56 (out of 60)
- CUDA Cores: 7,168
- Tensor Cores: 224
- RT Cores: 56
- Memory: 12GB GDDR6X
- Memory Bus: 192-bit
- Memory Speed: 21 Gbps
- Memory Bandwidth: 504 GB/s
- Base Clock: 1980 MHz
- Boost Clock: 2475 MHz
- Total Graphics Power (TGP): 220W
- Display Outputs: 1x HDMI 2.1, 3x DisplayPort 1.4a
The RTX 4070 SUPER uses the same “AD104” GPU as in RTX 4070. This 5nm powerhouse packs 56 streaming multiprocessors (SMs), squeezing more performance out of the same chip as the standard RTX 4070. This translates to a hefty boost in processing power, with 7168 CUDA cores, 224 Tensor cores, and 56 RT cores ready to tackle demanding games and high-fidelity ray tracing.
Unlike the original RTX 4070 which left some potential untapped, the SUPER variant unlocks the full 80 ROPs within the silicon, maximizing its graphics capabilities. However, the memory sticks with the familiar 12GB of 21 Gbps GDDR6X across a 192-bit bus, which is still plenty for most high-end gaming needs.
This performance boost comes at a slight cost: the RTX 4070 SUPER needs a bit more juice, bumping its total graphics power (TGP) to 220 W. So, say goodbye to single 8-pin PCIe connectors on custom designs – the SUPER embraces the newer 12VHPWR standard.
It’s quite annoying to see Nvidia using DLSS 3 frame generation to show higher performance gains comparatively. According to Nvidia, the RTX 4070 Super is up to 55% faster than the RTX 3070 in Guardians of the Galaxy and Shadow of the Tomb Raider. We will just wait for third-party reviews and benchmarks to accurately tell how fast it is.
RTX 4070 Ti Super
- Price: $799
- Available on: January 24
Specifications
- Chip: AD103 (5nm TSMC process)
- Streaming Multiprocessors (SMs): 66 (out of 80)
- CUDA Cores: 8,448
- Tensor Cores: 264
- RT Cores: 66
- Memory: 16GB GDDR6X
- Memory Bus: 256-bit
- Memory Speed: 21 Gbps
- Memory Bandwidth: 716 GB/s
- Base Clock: 2310 MHz
- Boost Clock: 2610 MHz
- Total Graphics Power (TGP): 285W
- Display Outputs: 1x HDMI 2.1, 3x DisplayPort 1.4a
The RTX 4070 Ti SUPER ditches the maxed-out “AD104” for the “AD103” chip. This translates to a monster spec upgrade that’s the star of the show in this refresh. The RTX 4070 Ti SUPER boasts a generous 16GB of GDDR6X memory with a 256-bit memory bus and 716 GB/s of bandwidth.
It also gets 10% more streaming multiprocessors (SMs), pushing the count to 66 out of 80 on the “AD103.” This translates to a hefty 8,448 CUDA cores, 264 Tensor cores, and 66 RT cores. While base and boost clock speeds stay similar to the original 4070 Ti, the SUPER demands a bit more juice. Its 285W TGP (total graphics power) means custom designs ditch the single 8-pin PCIe connector for the newer 12VHPWR standard.
The marketing material shows that the RTX 4070 Ti Super is up to 45% faster than the RTX 3070Ti in Guardians of the Galaxy. Again, this doesn’t mean much to us and we just need to wait for the third-party benchmarks.
Related: Best Graphics Card of 2023: Top GPUs in every price range
RTX 4080 Super
- Price: $999
- Available on: January 31
Specifications
- Chip: AD103 (5nm TSMC process)
- Streaming Multiprocessors (SMs): 80
- CUDA Cores: 10,240
- Tensor Cores: 320
- RT Cores: 80
- Memory: 16GB GDDR6X
- Memory Bus: 256-bit
- Memory Speed: 23 Gbps
- Memory Bandwidth: 736 GB/s
- Base Clock: 2295 MHz
- Boost Clock: 2550 MHz
- Total Graphics Power (TGP): 320W
- Display Outputs: 1x HDMI 2.1, 3x DisplayPort 1.4a
the RTX 4080 SUPER goes all-in, unleashing the full potential of the “AD103” silicon. It utilises all 80 streaming multiprocessors (SMs), resulting in a monstrous 10,240 CUDA cores, 320 Tensor cores, and 80 RT cores.
The RTX 4080 SUPER boasts 23 Gbps GDDR6X across a hefty 256-bit bus, translating to 736 GB/s of bandwidth. RTX 4080 Super also get a boost in the clock speeds by 100MHz compared to the RTX 4080.
RTX 4080 Super demands a healthy 320W TGP (total graphics power). Custom designs ditch the single 8-pin PCIe connector for the beefier 12VHPWR standard, ensuring ample juice flowing to unleash the SUPER’s full potential.
Stay tuned with Yantragyan for benchmarks and performance.
Leave a Reply
You must be logged in to post a comment.