NVIDIA RTX 5000 Ada Generation vs NVIDIA GeForce RTX 3090 Ti
We compared two Desktop platform GPUs: 32GB VRAM RTX 5000 Ada Generation and 24GB VRAM GeForce RTX 3090 Ti to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.
Main Differences
NVIDIA RTX 5000 Ada Generation 's Advantages
Released 1 years and 7 months late
Boost Clock has increased by 37% (2550MHz vs 1860MHz)
More VRAM (32GB vs 24GB)
2048 additional rendering cores
Lower TDP (250W vs 450W)
NVIDIA GeForce RTX 3090 Ti 's Advantages
Larger VRAM bandwidth (1008GB/s vs 576.0GB/s)
Score
Benchmark
FP32 (float)
RTX 5000 Ada Generation
+63%
65.28 TFLOPS
GeForce RTX 3090 Ti
40 TFLOPS
3DMark Time Spy
RTX 5000 Ada Generation
14170
GeForce RTX 3090 Ti
+54%
21832
3DMark Time Spy Extreme
RTX 5000 Ada Generation
+12%
12700
GeForce RTX 3090 Ti
11318
3DMark Speed Way
RTX 5000 Ada Generation
+9%
6473
GeForce RTX 3090 Ti
5903
Blender
RTX 5000 Ada Generation
6013
GeForce RTX 3090 Ti
+3%
6217
OctaneBench
RTX 5000 Ada Generation
+25%
816
GeForce RTX 3090 Ti
649
Graphics Card
Aug 2023
Release Date
Jan 2022
Quadro Ada
Generation
GeForce 30
Desktop
Type
Desktop
PCIe 4.0 x16
Bus Interface
PCIe 4.0 x16
Clock Speeds
1155 MHz
Base Clock
1560 MHz
2550 MHz
Boost Clock
1860 MHz
2250 MHz
Memory Clock
1313 MHz
Memory
32GB
Memory Size
24GB
GDDR6
Memory Type
GDDR6X
256bit
Memory Bus
384bit
576.0GB/s
Bandwidth
1008GB/s
Render Config
-
-
-
100
SM Count
84
12800
Shading Units
10752
400
TMUs
336
176
ROPs
112
400
Tensor Cores
336
100
RT Cores
84
128 KB (per SM)
L1 Cache
128 KB (per SM)
72 MB
L2 Cache
6 MB
-
-
-
Theoretical Performance
448.8 GPixel/s
Pixel Rate
208.3 GPixel/s
1020 GTexel/s
Texture Rate
625.0 GTexel/s
65.28 TFLOPS
FP16 (half)
40.00 TFLOPS
65.28 TFLOPS
FP32 (float)
40.00 TFLOPS
1020 GFLOPS
FP64 (double)
625.0 GFLOPS
Board Design
250W
TDP
450W
600 W
Suggested PSU
850 W
4x DisplayPort 1.4a
Outputs
1x HDMI 2.1
3x DisplayPort 1.4a
1x 16-pin
Power Connectors
1x 16-pin
Graphics Processor
AD102
GPU Name
GA102
AD102
GPU Variant
GA102-350-A1
Ada Lovelace
Architecture
Ampere
TSMC
Foundry
Samsung
5 nm
Process Size
8 nm
76.3 billion
Transistors
28.3 billion
609 mm²
Die Size
628 mm²
Graphics Features
12 Ultimate (12_2)
DirectX
12 Ultimate (12_2)
4.6
OpenGL
4.6
3.0
OpenCL
3.0
1.3
Vulkan
1.3
8.9
CUDA
8.6
6.7
Shader Model
6.6
