NVIDIA RTX PRO 6000 Blackwell Server vs PowerColor Red Dragon RX 560 14CU 4 GB

Comparison of NVIDIA RTX PRO 6000 Blackwell Server with 96 GB GDDR7 and 24,064 cores vs PowerColor Red Dragon RX 560 14CU 4 GB with 4 GB GDDR5 and 896 cores.

Loading...

Performance Rating

A100 A100
H200 H200
MI325X MI325X

NVIDIA RTX PRO 6000 Blackwell Server

NVIDIA RTX PRO 6000 Blackwell Server

RX 7900 XTX RX 7900 XTX
MI250 MI250
Instinct MI300X Instinct MI300X

PowerColor Red Dragon RX 560 14CU 4 GB

1.9

PowerColor Red Dragon RX 560 14CU 4 GB

1.9

Contents:

Memory ML Performance Compute Power Architecture & Compatibility ML Software Support Clocks & Performance Power Consumption Rendering Benchmarks Additional

Memory

Memory Size

96 GB 4 ГБ

Memory Type

GDDR7 GDDR5

Memory Bandwidth

1.79 TB/s 112.0 GB/s

Memory Bus Width

512 бит 128 бит

ML Performance

FP16 (Half Precision)

126.0 TFLOPS 2.107 TFLOPS

BF16 (Brain Float)

No No

TF32 (TensorFloat)

No No

Compute Power

FP32 (Single Precision)

126.0 TFLOPS 2.107 TFLOPS

FP64 (Double Precision)

1.968 TFLOPS 0.13169999999999998 TFLOPS

CUDA Cores

24,064 896

RT Cores

188 No

Architecture & Compatibility

GPU Architecture

Blackwell 2.0 GCN 4.0

SM (Streaming Multiprocessor)

188 No

PCIe Version

PCIe 5.0 x16 PCIe 3.0 x8

ML Software Support

CUDA Version

12.0 No

Clocks & Performance

Base Clock

1,590 1,090

Boost Clock

2,617 1,176

Memory Clock

1,750 1,750

Power Consumption

TDP/TGP

600 W
🔥 -89% 65 W

Recommended PSU

1000 W
🔥 -75% 250 W

Power Connector

1x 16-pin 1x 6-pin

Rendering

Texture Units (TMU)

752 56

ROP

188 No

L2 Cache

128 MB 1024 KB

Benchmarks

MLPerf, llama2-70b-99.9 (fp4)

3 250 tokens/s

MLPerf, llama3.1-8b (fp4)

5 758 tokens/s

Geekbench AI, FP16

53 322 points

Geekbench AI, INT8

28 264 points

Geekbench AI, FP32

37 299 points

MLPerf, mixtral-8x7b (fp8)

3 767 tokens/s

Additional

Slots

Dual-slot Dual-slot

Release Date

March 18, 2025 July 4, 2017

Display Outputs

4x DisplayPort 2.1b
1x DVI
1x HDMI 2.0b
1x DisplayPort 1.4a

Renting is cheaper than buying