KU

Kunlunxin R200

XPU-R PCIe 7nm
FP32
32.0
TFLOPS
VRAM
16 GB
GDDR6
TDP
150 W
213.3 TFLOPS/kW
Bandwidth
512 GB/s
memory

Overview

The Kunlunxin R200 is the 16 GB member of the company's second-generation XPU-R accelerator line, a full-height full-length dual-slot PCIe Gen4 x16 card built on 7nm. It delivers 256 TOPS of INT8, 128 TOPS of INT16, 128 TFLOPS of FP16 and 32 TFLOPS of FP32, with 16 GB of GDDR6 at 512 GB/s, ECC, passive cooling and a 150 W board power. Kunlunxin positions it for both inference and training, and the card carries substantial video capability alongside its tensor work: 108 streams of 1080p30 decode and 27 streams of 1080p30 encode. Kunlunxin labels none of these figures dense or sparse and publishes no sparse rate. Note that Kunlunxin's separate second-generation chip page lists INT32 at 128 TOPS but omits the floating-point rates entirely, while this card page omits INT32; only the card page's own table is recorded here.

Performance Metrics

Peak theoretical throughput by precision type

PrecisionBitsPeak TFLOPSEfficiency
INT16 -- 128.0 0.853 TFLOPS/W
INT8 8 256.0 1.707 TFLOPS/W
FP16 16 128.0 0.853 TFLOPS/W
FP32 32 32.0 0.213 TFLOPS/W
FP16 Efficiency
0.853 TFLOPS/W
128.0 TFLOPS / 150W
FP32 Efficiency
0.213 TFLOPS/W
32.0 TFLOPS / 150W

Power Specifications

TDP

150 W

Max Power

173 W

Power Connector

PCIe Slot

Cooling

Passive

Memory Specifications

Capacity

16 GB

Type

GDDR6

Bandwidth

512 GB/s

Interface

--

Hardware & Design

Form Factor

PCIe

Architecture

XPU-R

Process Node

7nm

Launch Year

--

Variant

Standard

Market Segment

Professional

Full Specifications

Memory
VRAM 16 GB
Memory Type GDDR6
Bandwidth 512 GB/s
Interconnect & I/O
PCIe 4.0 x16
Power & Thermal
TDP 150 W
Cooling Passive
Enterprise Features
ECC Memory Yes
General
Form Factor PCIe
Architecture XPU-R

Documentation & Resources

Common Use Cases

General Compute AI/ML Workloads Data Processing

The Kunlunxin R200 is optimized for high-performance computing tasks with XPU-R architecture delivering 32 TFLOPS of compute power.

Where to Rent

Compare cloud providers offering on-demand GPU instances for AI training, inference, and HPC workloads.

Browse GPU Cloud Providers

Similar GPUs

Kunlunxin R200-8F

PCIe ·
FP32: 32.0 TFLOPS
Compare vs R200-8F

Stay Updated on GPU Releases

Get notified when new GPUs are added or specifications are updated.

Loading verification...

No spam, unsubscribe anytime.

Back to GPUs