GeekRadar / The comparison desk

COMPARE.

GPU

NVIDIA

GeForce RTX 5070 SUPER

Graphics processor

Video memory: 18 GB^*
Memory bus: 192 bit^*
Rated memory bandwidth: 672 GB/s^*

No matched benchmark results are available for this entry yet.

Processing & platform

Architecture: Blackwell 2.0^*
Process node: 5 nm^*
GPU cores: 6,400^*
GPU base clock: 2,325 MHz^*
Graphics interface: PCIe 5.0 x16^*
Reported chip die area: 263 mm²^*
Transistors: 31,100 million^*
GPU boost clock: 2,512 MHz^*
GPU L2 cache: 48 MB^*
Graphics processor: GB205^*
Texture mapping units: 200^*
Raster output units: 80^*
Streaming multiprocessors: 50^*
Tensor cores: 200^*
Ray tracing cores: 50^*
Chip foundry: TSMC^*
Card or module format: Dual-slot^*
Declared GPU TDP: 275 W^*
Suggested system PSU: 600 W^*
External power connectors: 1x 16-pin^*
DirectX / feature level: 12 Ultimate (12_2)^*
OpenGL: 4.6^*
OpenCL: 3.0^*
Vulkan: 1.4^*
CUDA compute capability: 12.0^*
Shader model: 6.9^*

Memory & storage

Memory options: 18^*
Video memory: 18 GB^*
Video memory type: GDDR7^*
Memory bus: 192 bit^*
Rated memory bandwidth: 672 GB/s^*

Power specifications

Thermal design power: 275 W^*

System configuration

Detailed specifications

Specifications 1 · 31 fields

proc transistors: 31,100 million^*
proc die size: 263^* mm²
cores: 6,400^*
tensor cores: 200^*
rt cores: 50^*
sm count: 50^*
base clock: 2,325^*
boost clock: 2,512^*
memory clock: 1,750^*
ram: 18^*
name: NVIDIA GeForce RTX 5070 SUPER^*
ram type: GDDR7^*
bus width: 192^*
bus interface: PCIe 5.0 x16^*
ram bandwidth: 672^* GB/s
fp16 performance: 32.15^*
fp32 performance: 32.15^*
fp64 performance: 0.5024^*
tdp: 275^* W
suggested psu: 600^* W
slot width: Dual-slot^*
power connectors: 1x 16-pin^*
api cuda: 12^*
is retail board: No^*
gpi value: 15.72^*
vendor: nvd^*
manufacturer: NVIDIA^*
arch name: Blackwell 2.0^*
card release date: 2026-03-11^*
proc foundry: TSMC^*
proc process size: 5^* nm

Specifications 2 · 39 fields

CUDA compute capability

12.0^*

DirectX / feature level

12 Ultimate (12_2)^*

OpenCL

3.0^*

OpenGL

4.6^*

Shader model

6.9^*

Vulkan

1.4^*

Graphics interface

PCIe 5.0 x16^*

Memory bus

192^* bit

Reported shader processors

6,400^*

Reported theoretical FP16 peak

31.96^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Reported theoretical FP32 peak

31.96^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Reported theoretical FP64 peak

0.4994^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Declared L1 cache configuration

128 KB (per SM)^*

Declared GPU L2 cache

48 MB^*

Reported display outputs

1x HDMI 2.1b, 3x DisplayPort 2.1b^*

Theoretical pixel fill rate

199.8 GPixel/s^*

External power connectors

1x 16-pin^*

Reported chip die area

263 mm²^*

Chip foundry

TSMC^*

Process node declaration

5 nm^*

Transistors

31,100 million^*

Graphics processor

GB205^*

Reported video memory

18^* GB

Rated memory bandwidth

672.0 GB/s^*

Video memory type

GDDR7^*

Raster output units

80^*

Ray tracing cores

50^*

Reported shading units

6,400^*

Card or module format

Dual-slot^*

Streaming multiprocessors

50^*

Suggested system PSU

600 W^*

2:4 tensor sparsity

Yes^*

Declared GPU TDP

275 W^*

Tensor cores

200^*

Theoretical texture fill rate

499.4 GTexel/s^*

Texture mapping units

200^*

GPU cores

6,400^*

Video memory

18^* GB

Video memory type

GDDR7^*

Specifications 3 · 60 fields

Brand: NVIDIA^*
Top BUS WIDTH: 192^* bit
Graphics Processor Process Type: 4N FinFET^*
Graphics Processor GPU Name: GB205^*
Graphics Processor Architecture: Blackwell 2.0^*
Render Config SM Count: 50^*
Graphics Features CUDA: 12^*
Render Config Tensor Cores: 200^*
Graphics Processor Process Size: 5^* nm
Graphics Processor Transistors: 31,100 million^*
Graphics Processor Die Size: 263^* mm²
Graphics Card Release Date: 2026^*
Graphics Card Generation: GeForce 50^*
Graphics Card Production: Unreleased^*
Name: GeForce RTX 5070 SUPER^*
Graphics Card Bus Interface: PCIe 5.0 x16^*
Clock Speeds Memory Clock: 1750 MHz 28 Gbps effective^*
Memory Memory Size: 18^* GB
Memory Memory Type: GDDR7^*
Memory Memory Bus: 192^* bit
Memory Bandwidth: 672^* GB/s
Render Config TMUs: 200^*
Top GRAPHICS PROCESSOR: GB205^*
Render Config ROPs: 80^*
Theoretical Performance Pixel Rate: 201.0 GPixel/s
Theoretical Performance Texture Rate: 502.4 GTexel/s
Board Design Slot Width: Dual-slot^*
Board Design TDP: 275^* W
Board Design Suggested PSU: 600^* W
Board Design Outputs: 1x HDMI 2.1b 3x DisplayPort 2.1b^*
Board Design Board Number: PG147 SKU 65^*
Graphics Features DirectX: 12 Ultimate (12_2)^*
Graphics Features OpenGL: 4.6^*
Graphics Features OpenCL: 3^*
Graphics Features Vulkan: 1.4^*
source file: nvidia2026.json^*
Graphics Processor Foundry: TSMC^*
Graphics Card Predecessor: GeForce 40^*
Recommended Resolutions: Good:1920x1080 | Good:2560x1440 | Good:3840x2160^*
Board Design Length: 245 mm 9.6 inches^*
Graphics Processor GPU Variant: GB205-400-A1^*
Board Design Power Connectors: 1x 16-pin^*
Graphics Processor Density: 118.3M / mm²^*
Board Design Height: 115 mm 4.5 inches^*
Board Design Width: 40 mm 1.6 inches^*
Top TMUS: 200^*
Top ROPS: 80^*
Clock Speeds Base Clock: 2,325^* MHz
Clock Speeds Boost Clock: 2,512^* MHz
Graphics Features Shader Model: 6.8^*
Top CORES: 6,400^*
Render Config Shading Units: 6,400^*
Render Config L2 Cache: 48^* MB
Top MEMORY SIZE: 18^* GB
Theoretical Performance FP32 (float): 32.15^* TFLOPS
Theoretical Performance FP64 (double): 502.4 GFLOPS (1:64)
Render Config L1 Cache: 128 KB (per SM)^*
Theoretical Performance FP16 (half): 32.15 TFLOPS (1:1)
Top MEMORY TYPE: GDDR7^*
Render Config RT Cores: 50^*

* Provisional data