NVIDIA GeForce RTX 4080 Super

The Nvidia GeForce RTX 4080 Super is a high-end desktop graphics card based on the Ada Lovelace architecture. It uses the AD103 chip in full configuration with 10,240 cores and 16 GB GDDR6X graphics memory (256-bit bus with 23 Gbits).

The ray tracing performance is outstanding thanks to 80 dedicated cores and the 320 Tensor cores can be used for DLSS 3.

The gaming performance without DLSS is ideal for 4k with around 60fps and 2k with higher frame rates in demanding games without ray tracing effects (e.g. F1 22 or Cyberpunk 2077). With ray tracing, the use of DLSS is recommended. Compared to the competition, the performance without DLSS and ray tracing is slightly better than the significantly cheaper AMD Radeon RX 7900 XT.

The currently available RTX 4080 cards are very large (2.5 or 3 slots) and use the new 12VHPWR connector (which has caused some problems with adapters).

The AD103 chip is produced at TSMC in 5nm (4N with EUV) and is specified with a TGP (Total Graphics Power) of 320 watts (Nvidia recommends at least a 750 watt power supply).

GeForce RTX 4000 Series NVIDIA GeForce RTX 4090 16384 @ 2.24 - 2.52 GHz 384 Bit @ 21000 MHz NVIDIA GeForce RTX 4080 Super 10240 @ 2.3 - 2.55 GHz 256 Bit @ 23000 MHz NVIDIA GeForce RTX 4080 9728 @ 2.21 - 2.51 GHz 256 Bit @ 22400 MHz NVIDIA GeForce RTX 4070 Ti SUPER 8448 @ 2.34 - 2.61 GHz 256 Bit @ 21000 MHz NVIDIA GeForce RTX 4070 Ti 7680 @ 2.31 - 2.61 GHz 192 Bit @ 21000 MHz NVIDIA GeForce RTX 4070 SUPER 7168 @ 1.98 - 2.48 GHz 192 Bit @ 21000 MHz NVIDIA GeForce RTX 4070 5888 @ 1.92 - 2.48 GHz 192 Bit @ 21000 MHz NVIDIA GeForce RTX 4060 Ti 16G 4352 @ 2.31 - 2.54 GHz 128 Bit @ 18000 MHz NVIDIA GeForce RTX 4060 Ti 8G 4352 @ 2.31 - 2.54 GHz 128 Bit @ 18000 MHz NVIDIA GeForce RTX 4060 3072 @ 1.83 - 2.46 GHz 128 Bit @ 17000 MHz
Codename	AD103-400-A1
Architecture	Ada Lovelace
Pipelines	10240 - unified
TMUs	320
ROPs	112
Raytracing Cores	80
Tensor / AI Cores	320
Core Speed	2295 - 2550 (Boost) MHz
Theoretical Performance	52.22 TFLOPS FP16, 52.22 TFLOPS FP32, 816 TFLOPS FP64
Cache	L2: 64 MB
Memory Speed	23000 effective = 1438 MHz
Memory Bus Width	256 Bit
Memory Type	GDDR6X
Max. Amount of Memory	16 GB
Shared Memory	no
Memory Bandwidth	736.3 GB/s
API	DirectX 12 Ultimate (12_2), Shader 6.7, OpenGL 4.6, OpenCL 3.0, Vulkan 1.3
Power Consumption	320 Watt
Transistor Count	45.900 Million
Die Size	379 mm²
technology	5 nm
PCIe	PCIe 4.0 x16
Displays	4 Displays (max.), HDMI 2.1, DisplayPort 1.4a
Date of Announcement	31.01.2024

Benchmarks