NVIDIA GeForce RTX 5070 Ti vs NVIDIA GeForce GTX 680
We compared two Desktop platform GPUs: 16GB VRAM GeForce RTX 5070 Ti and 2GB VRAM GeForce GTX 680 to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA GeForce RTX 5070 Ti 's Advantages
Released 12 years and 11 months late
Boost Clock has increased by 132% (2452MHz vs 1058MHz)
More VRAM (16GB vs 2GB)
Larger VRAM bandwidth (896.0GB/s vs 192.3GB/s)
7424 additional rendering cores
NVIDIA GeForce GTX 680 's Advantages
Lower TDP (195W vs 300W)
Score
Benchmark
FP32 (float)
GeForce RTX 5070 Ti
+1252%
43.94 TFLOPS
GeForce GTX 680
3.25 TFLOPS
3DMark Time Spy
GeForce RTX 5070 Ti
+1277%
27607
GeForce GTX 680
2004
Graphics Card
Feb 2025
Release Date
Mar 2012
GeForce 50
Generation
GeForce 600
Desktop
Type
Desktop
PCIe 5.0 x16
Bus Interface
PCIe 3.0 x16
Clock Speeds
2295 MHz
Base Clock
1006 MHz
2452 MHz
Boost Clock
1058 MHz
1750 MHz
Memory Clock
1502 MHz
Memory
16GB
Memory Size
2GB
GDDR7
Memory Type
GDDR5
256bit
Memory Bus
256bit
896.0GB/s
Bandwidth
192.3GB/s
Render Config
-
-
-
70
SM Count
-
8960
Shading Units
1536
280
TMUs
128
96
ROPs
32
280
Tensor Cores
-
70
RT Cores
-
128 KB (per SM)
L1 Cache
16 KB (per SMX)
48 MB
L2 Cache
512 KB
-
-
-
Theoretical Performance
235.4 GPixel/s
Pixel Rate
33.86 GPixel/s
686.6 GTexel/s
Texture Rate
135.4 GTexel/s
43.94 TFLOPS
FP16 (half)
-
43.94 TFLOPS
FP32 (float)
3.250 TFLOPS
686.6 GFLOPS
FP64 (double)
135.4 GFLOPS
Board Design
300W
TDP
195W
700 W
Suggested PSU
450 W
1x HDMI 2.1b
3x DisplayPort 2.1b
Outputs
2x DVI
1x HDMI 1.4a
1x DisplayPort 1.2
1x 16-pin
Power Connectors
2x 6-pin
Graphics Processor
GB203
GPU Name
GK104
GB203-300-A1
GPU Variant
GK104-400-A2
Blackwell 2.0
Architecture
Kepler
TSMC
Foundry
TSMC
5 nm
Process Size
28 nm
45.6 billion
Transistors
3.54 billion
378 mm²
Die Size
294 mm²
Graphics Features
12 Ultimate (12_2)
DirectX
12 (11_0)
4.6
OpenGL
4.6
3.0
OpenCL
3.0
1.4
Vulkan
1.1
12.0
CUDA
3.0
6.8
Shader Model
5.1
