כתבה
arXiv cs.LG ·
RED-PIM: הפחתת תנועת נתונים ב-Transformers
RED-PIM: Reducing Data Movement for Transformers using Processing-in-Memory
RED-PIM הוא אלגוריתם-ארכיטקטורה שמפחית את זמן האינפרנס של מודלי Transformers. הוא עושה זאת על ידי הפחתת תנועת הנתונים בין יחידות עיבוד וזיכרון. RED-PIM משיג הפחתה של 16.05% עד 99.99% בזמן אינפרנס.
תקציר מקורי באנגליתarXiv:2607.21731v1 Announce Type: new Abstract: Transformers are widely used across many domains, including natural language processing, computer vision, web search, and DNA sequence analysis. Given their broad applicability, improving the performance of transformer models is critical. However, the high volume of data movement between processing units and memory during attention operations significantly limits their efficiency. Processing-In-Memory (PIM) mitigates this issue by performing computations directly inside memory. While prior work has proposed PIM-based transformer implementations, they suffer from costly inter-bank communication, and struggle to scale due to the limited capacity of memory banks. As a result, attention-related data must be split across banks, diminishing the pot
קרא במקור המקורי
arxiv.org
פתח כתבה מקורית