BREAKING
NVIDIA Pushes Local AI Rigs
From Data Center to Desktop
0
B
LLM params
0
M
token context
0
K
AI video
RTX Spark vs DGX Spark
RTX Spark
~1 PFLOP
●
128GB unified memory
●
120B LLM, 1M tokens
●
4K video and 3D
DGX Spark
~20 PFLOPS
●
128GB, ARM CPU
●
Runs 1T-param models
●
Fast setup
Memory Across the Systems
RTX Spark
128
DGX Spark
128
Community rig
336
Local AI Goes Mainstream
AI NEWS BLITZ
NVIDIA is inviting the community to share their home-built AI clusters.