Inference Server Throughput Analysis
An engineer is testing two configurations for a language model inference server to determine which one can handle more user requests over time. Analyze the data below and determine which configuration offers higher throughput. Justify your conclusion with a calculation.
0
1
Tags
Ch.5 Inference - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Analysis in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
An inference server needs to process 12 independent user requests. The server's hardware has two processing options:
- Sequential Processing: Handle one request at a time, with each request taking 2 seconds to complete.
- Batched Processing: Group 4 requests into a single batch and process them in parallel, with the entire batch taking 3 seconds to complete.
Based on this information, which statement correctly analyzes the total time required and the resulting efficiency of each approach?
Optimizing Inference Server Performance
Inference Server Throughput Analysis