NVIDIA DGX Spark
Updated 2 h ago · first seen 11 Sept 2026
hw_01M293SPNPQ432V0871EHP07HA
- Kind
- system
- T2 · 4 h ago
- Memory
- 128 GB LPDDR5x
- T2 · 4 h ago
- Bandwidth
- 273 GB/s
- T2 · 4 h ago
- TDP
- 140 W
- T1 · 2 h ago
- Released
- 2025
- T2 · 4 h ago
Specification
- Release date
- 2025
Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 4 h agomedium
- Architecture
- NVIDIA Grace Blackwell
Source:NVIDIA — data center product pagesT1observed 2 h agohigh
- Compute fp4 tflops
- 1,000 TFLOPS
Source:NVIDIA — data center product pagesT1observed 2 h agohigh
- Cpu
- 20-core Arm, 10 Cortex-X925 + 10 Cortex-A725 Arm
Source:NVIDIA — data center product pagesT1observed 2 h agohigh
- Gpu architecture
- Blackwell Architecture
Source:NVIDIA — data center product pagesT1observed 2 h agohigh
- Kind
- system
Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 4 h agomedium
- Manufacturer
- NVIDIA
Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 4 h agomedium
- Memory bandwidth
- 273 GB/s
Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 4 h agomedium
- Memory
- 128 GB
Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 4 h agomedium
- Memory type
- LPDDR5x
Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 4 h agomedium
- Power supply watts
- 240
Source:NVIDIA — data center product pagesT1observed 2 h agohigh
- Runtimes
- cuda, tensorrt-llm, vllm, llama.cpp
Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 4 h agomedium
- Spec sheet
Source:AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2observed 4 h agomedium
- Storage
- 4 TB NVME.M2 with self-encryption
Source:NVIDIA — data center product pagesT1observed 2 h agohigh
- TDP
- 140 W
Source:NVIDIA — data center product pagesT1observed 2 h agohigh
Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →
Provenance
Attributed facts
15
Source tiers
T1T27 / 8
Freshest observation
2 h ago
Conflicts
None
Models whose estimated memory footprint fits this device — see estimate method.
No models recorded
- Manufactured by
- NVIDIA
As of
Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.
Claim history · Architecture
Architecturearchitecture2
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| NVIDIA Grace Blackwell | → current | current | NVIDIA — data center product pagesT1 | high | deterministic |
| Grace Blackwell (GB10) | → | superseded | AI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2 | medium | curated |
Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →
NVIDIA DGX Spark: architecture changed from Grace Blackwell (GB10) to NVIDIA Grace Blackwell
ArchitectureGrace Blackwell (GB10)→NVIDIA Grace Blackwellnvidia_specs
| Source | Document | Type | Tier | Last observed | Snapshots |
|---|---|---|---|---|---|
| NVIDIA — data center product pages | nvidia.com/en-us/products/workstations/dgx-spark | spec_page | T1· Official | 2 h ago | 1 |
Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.