Home GPU Comparison NVIDIA Tesla V100 FHHL vs AMD Radeon PRO W7800

NVIDIA Tesla V100 FHHL vs AMD Radeon PRO W7800

We compared two Professional market GPUs: 16GB VRAM Tesla V100 FHHL and 32GB VRAM Radeon PRO W7800 to see which GPU has better performance in key specifications, benchmark tests, power consumption, etc.

Main Differences

NVIDIA Tesla V100 FHHL 's Advantages
Larger VRAM bandwidth (829.4GB/s vs 576.0GB/s)
640 additional rendering cores
Lower TDP (250W vs 260W)
AMD Radeon PRO W7800 's Advantages
Released 5 years and 1 months late
Boost Clock has increased by 94% (2499MHz vs 1290MHz)
More VRAM (32GB vs 16GB)

Score

Benchmark

FP32 (float)
Tesla V100 FHHL
13.21 TFLOPS
Radeon PRO W7800 +238%
44.78 TFLOPS
VS

Graphics Card

Mar 2018
Release Date
Apr 2023
Tesla
Generation
Radeon Pro Navi
Professional
Type
Professional
PCIe 3.0 x16
Bus Interface
PCIe 4.0 x16

Clock Speeds

937 MHz
Base Clock
1855 MHz
1290 MHz
Boost Clock
2499 MHz
810 MHz
Memory Clock
2250 MHz

Memory

16GB
Memory Size
32GB
HBM2
Memory Type
GDDR6
4096bit
Memory Bus
256bit
829.4GB/s
Bandwidth
576.0GB/s

Render Config

80
SM Count
-
-
Compute Units
70
5120
Shading Units
4480
320
TMUs
280
128
ROPs
128
640
Tensor Cores
-
-
RT Cores
70
128 KB (per SM)
L1 Cache
256 KB per Array
6 MB
L2 Cache
6 MB

Theoretical Performance

165.1 GPixel/s
Pixel Rate
319.9 GPixel/s
412.8 GTexel/s
Texture Rate
699.7 GTexel/s
26.42 TFLOPS
FP16 (half)
89.56 TFLOPS
13.21 TFLOPS
FP32 (float)
44.78 TFLOPS
6.605 TFLOPS
FP64 (double)
1399 GFLOPS

Graphics Processor

GV100
GPU Name
Navi 31
-
GPU Variant
Navi 31
Volta
Architecture
RDNA 3.0
TSMC
Foundry
TSMC
12 nm
Process Size
5 nm
21.1 billion
Transistors
57.7 billion
815 mm²
Die Size
529 mm²

Board Design

250W
TDP
260W
600 W
Suggested PSU
600 W
No outputs
Outputs
3x DisplayPort 2.1 1x mini-DisplayPort 2.1
1x 8-pin
Power Connectors
2x 8-pin

Graphics Features

12 (12_1)
DirectX
12 Ultimate (12_2)
4.6
OpenGL
4.6
3.0
OpenCL
2.2
1.3
Vulkan
1.3
7.0
CUDA
-
6.6
Shader Model
6.7

Related GPU Comparisons

Related News

© 2024 - TopCPU.net   Contact Us Privacy Policy