AI’s next build-out: Inference racks that resize themselves for every new model
SiliconAngle reports that memory pooling is becoming a key tool in AI inference economics, as companies try to get more output from each graphics processing unit. The outlet notes that most AI servers ship with a fixed GPU-to-memory ratio for the life of the machine, which it describes as a poor fit when future model requirements are unknown. According to SiliconAngle, pooling lets companies share GPUs and memory across servers, with the aim of improving tokens per dollar and easing AI supply constraints.
Based on SiliconAngle's report - only one outlet has covered this so far. Read the original.Coverage 1 source · 1 report
Chronological.
-
AI’s next build-out: Inference racks that resize themselves for every new model
Memory pooling is fast becoming one of the sharpest tools in AI inference economics, with companies racing to wring more output from every graphics processing unit. Most AI servers still ship with a fixed GPU-to-memory ratio that holds for the life of the machine. But that is a poor fit when nobody knows what model […] The post AI’s next build-out: Inference racks that resize themselves for every new model appeared first on SiliconANGLE .
Discussion 0 comments