GeekRadar / The comparison desk

COMPARE.

GPU

NVIDIA

NVIDIA DGX Spark

Graphics processor

Video memory: 128 GB^*
Memory bus: 256 bit^*
Rated memory bandwidth: 273.20 GB/s^*

No matched benchmark results are available for this entry yet.

Processing & platform

Architecture: Blackwell 2.0^*
Process node: 5 nm^*
GPU cores: 6,144^*
Graphics interface: PCIe 5.0 x16^*
Reported chip die area: 382 mm²^*
GPU L2 cache: 50 MB^*
Graphics processor: GB20B^*
Texture mapping units: 384^*
Raster output units: 48^*
Streaming multiprocessors: 48^*
Tensor cores: 384^*
Ray tracing cores: 48^*
Chip foundry: TSMC^*
Card or module format: IGP^*
Declared GPU TDP: 140 W^*
Suggested system PSU: 300 W^*
External power connectors: None^*
OpenCL: 3.0^*
CUDA compute capability: 12.1^*

Memory & storage

Memory options: 128^*
Video memory: 128 GB^*
Video memory type: LPDDR5X^*
Memory bus: 256 bit^*
Rated memory bandwidth: 273.20 GB/s^*

Power specifications

Thermal design power: 140 W^*

System configuration

Detailed specifications

Specifications 1 · 28 fields

cores: 6,144^*
tensor cores: 384^*
rt cores: 48^*
sm count: 48^*
base clock: 1,665^*
boost clock: 2,418^*
memory clock: 1,067^*
ram: 128^*
name: NVIDIA DGX Spark^*
ram type: LPDDR5X^*
bus width: 256^*
bus interface: PCIe 5.0 x16^*
ram bandwidth: 273.2^* GB/s
fp16 performance: 29.71^*
fp32 performance: 29.71^*
fp64 performance: 0.4643^*
tdp: 140^* W
suggested psu: 300^* W
slot width: IGP^*
api cuda: 12.1^*
is retail board: Yes^*
gpi value: 17.37^*
vendor: nvd^*
manufacturer: NVIDIA^*
arch name: Blackwell 2.0^*
card release date: 2025-10-15^*
proc foundry: TSMC^*
proc process size: 5^* nm

Specifications 2 · 34 fields

CUDA compute capability

12.1^*

OpenCL

3.0^*

Graphics interface

PCIe 5.0 x16^*

Memory bus

256^* bit

Reported shader processors

6,144^*

Reported theoretical FP16 peak

29.71^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Reported theoretical FP32 peak

29.71^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Reported theoretical FP64 peak

0.4643^* TFLOPS

Scope and conditions

Arithmetic density: Not stated^*

Declared L1 cache configuration

128 KB (per SM)^*

Declared GPU L2 cache

50 MB^*

Reported display outputs

1x HDMI^*

Theoretical pixel fill rate

116.1 GPixel/s^*

External power connectors

None^*

Reported chip die area

382 mm²^*

Chip foundry

TSMC^*

Process node declaration

5 nm^*

Graphics processor

GB20B^*

Reported video memory

128^* GB

Rated memory bandwidth

273.2 GB/s^*

Video memory type

LPDDR5X^*

Raster output units

48^*

Ray tracing cores

48^*

Reported shading units

6,144^*

Card or module format

IGP^*

Streaming multiprocessors

48^*

Suggested system PSU

300 W^*

2:4 tensor sparsity

Yes^*

Declared GPU TDP

140 W^*

Tensor cores

384^*

Theoretical texture fill rate

928.5 GTexel/s^*

Texture mapping units

384^*

GPU cores

6,144^*

Video memory

128^* GB

Video memory type

LPDDR5X^*

* Provisional data