Recherche

Afficher tous les résultats pour

Accueil

Personne

Babak Falsafi

À propos
Confidentialité
Mentions légales

Graph Chatbot

Publications associées (89)

Page 3 sur 5

Avant-Garde: Empowering GPUs with Scaled Numeric Formats

Babak Falsafi, Simla Burcu Harma

The escalating computational and memory demands of deep neural networks have outpaced chip density improvements, making arithmetic density a key bottleneck for GPUs. Scaled numeric formats, such as FP8 and Microscaling (MX), improve arithmetic density by a ...

ACM2025

Server Architecture from Enterprise to Post-Moore

Babak Falsafi

Luiz Barroso started his career at Digital Equipment Corporation, investigating workload-optimized multiprocessor server architectures marketed to enterprises in the 1990s. These high-margin, low-volume products lost their market to more cost-effective ent ...

2024

A Case for Specialized Processors for Scale-Out Workloads

Anastasia Ailamaki, Babak Falsafi, Michael Ferdman, Yusuf Onur Koçberber, Stavros Volos, Dorde Jevdic, Ilknur Cansu Kaynak, Adrian Daniel Popescu, Al Mutaz Adileh, Mohammad Alisafaee

Emerging scale-out workloads need extensive amounts of computational resources. However, datacenters using modern server hardware face physical constraints in space and power, limiting further expansion and requiring improvements in the computational densi ...

Ieee Computer Soc2014

Quantifying the Mismatch between Emerging Scale-Out Applications and Modern Processors

Anastasia Ailamaki, Babak Falsafi, Yusuf Onur Koçberber, Stavros Volos, Dorde Jevdic, Ilknur Cansu Kaynak, Michael Ferdman, Adrian Daniel Popescu, Al Mutaz Adileh, Mohammad Alisafaee

Emerging scale-out workloads require extensive amounts of computational resources. However, data centers using modern server hardware face physical constraints in space and power, limiting further expansion and calling for improvements in the computational ...

Association for Computing Machinery2012

The Mondrian Data Engine

Babak Falsafi, Boris Robert Grot, Javier Picorel Obando, Alexandros Daglis, Mario Paulo Drumond Lages De Oliveira, Dmitrii Ustiugov, Nooshin Sadat Mirzadeh

The increasing demand for extracting value out of ever-growing data poses an ongoing challenge to system designers, a task only made trickier by the end of Dennard scaling. As the performance density of traditional CPU-centric architectures stagnates, adva ...

2017

Confluence: unified instruction supply for scale-out servers

Babak Falsafi, Boris Robert Grot, Ilknur Cansu Kaynak

Multi-megabyte instruction working sets of server workloads defy the capacities of latency-critical instruction-supply components of a core; the instruction cache (L1-I) and the branch target buffer (BTB). Recent work has proposed dedicated prefetching tec ...

ACM Press2015

SecureCells: A Secure Compartmentalized Architecture

Andrés Sánchez Marín, Babak Falsafi, Mathias Josef Payer, Siddharth Gupta, Atri Bhattacharyya, Yuanlong Li, Florian Hofhammer

Modern programs are monolithic, combining code of varied provenance without isolation, all the while running on network-connected devices. A vulnerability in any component may compromise code and data of all other components. Compartmentalization separates ...

2023

Scale-out Systolic Arrays

Pascal Frossard, Babak Falsafi, Ahmet Caner Yüzügüler, Canberk Sönmez

Multi-pod systolic arrays are emerging as the architecture of choice in DNN inference accelerators. Despite their potential, designing multi-pod systolic arrays to maximize effective throughput/Watt-i.e., throughput/Watt adjusted when accounting for array ...

ASSOC COMPUTING MACHINERY2023

Thermal Characterization of Cloud Workloads on a Power-Efficient Server-on-Chip

Babak Falsafi, Dorde Jevdic, Pejman Lotfi Kamran

We propose a power-efficient many-core server-on-chip system with 3D-stacked Wide I/O DRAM targeting cloud workloads in datacenters. The integration of 3D-stacked Wide I/O DRAM on top of a logic die increases available memory bandwidth by using dense and f ...

2012

FPGAs versus GPUs in Data centers

Babak Falsafi

This article presents position statements and a question-and-answer session by panelists at the Fourth Workshop on Computer Architecture Research Directions. The subject of the debate was the use of field-programmable gate arrays versus GPUs in datacenters ...

Institute of Electrical and Electronics Engineers2017

An Analysis of Load Imbalance in Scale-out Data Serving

Babak Falsafi, Edouard Bugnion, Boris Robert Grot, Alexandros Daglis, Stanko Novakovic

Despite the natural parallelism across lookups, performance of distributed key-value stores is often limited due to load imbalance induced by heavy skew in the popularity distribution of the dataset. To avoid violating service level objectives expressed in ...

2016

Near-Optimal Cache Block Placement with Reactive Nonuniform Cache Architectures

Anastasia Ailamaki, Babak Falsafi, Michael Ferdman

The growing core counts and caches of modern processors result in data access latency becoming a function of the data's physical location in the cache. Thus, the placement of cache blocks determines the cache's performance. Reactive nonuniform cache archit ...

2010

Near-Memory Data Services

Babak Falsafi

This column introduces seven invited position papers about the challenges and opportunities in near-data processing. "Near-Memory Data Services" by Babak Falsafi, "Automata Processing: The Memory Is the Processor!" by Mircea Stan, Kevin Skadron, "Overcomin ...

Institute of Electrical and Electronics Engineers2016

Near-Memory Address Translation

Babak Falsafi, Javier Picorel Obando, Dorde Jevdic

Memory and logic integration on the same chip is becoming increasingly cost effective, creating the opportunity to offload data-intensive functionality to processing units placed inside memory chips. The introduction of memory-side processing units (MPUs) ...

Ieee2017

Data transformer apparatus

Christoph Koch, Babak Falsafi, Siddharth Gupta, Mark Johnathon Sutherland, Mario Paulo Drumond Lages De Oliveira, Arash Pourhabibi Zarandi, Zilu Tian, Hussein Kassir

Data transformer apparatus comprising at least a dispatcher module (D), a reader module (R), a converter module (C) and a writer module (W). The dispatcher module (D) is configured to: receive a data transformation request (DTR) including: a first informat ...

2021

Optimus Prime: Accelerating Data Transformation in Servers

Christoph Koch, Babak Falsafi, Siddharth Gupta, Mark Johnathon Sutherland, Mario Paulo Drumond Lages De Oliveira, Arash Pourhabibi Zarandi, Zilu Tian, Hussein Kassir

Modern online services are shifting away from monolithic applications to loosely-coupled microservices because of their improved scalability, reliability, programmability and development velocity. Microservices communicating over the datacenter network req ...

ACM2020

Multi-Grain Coherence Directory

Babak Falsafi

Conventional directory coherence operates at the finest granularity possible, that of a cache block. While simple, this organization fails to exploit frequent application behavior: at any given point in time, large, continuous chunks of memory are often ac ...

2013

Temporal streams in commercial server applications

Anastasia Ailamaki, Babak Falsafi, Michael Ferdman

Commercial server applications remain memory bound on modern multiprocessor systems because of their large data footprints, frequent sharing, complex non-strided access patterns, and long chains of dependant misses. To improve memory system performance des ...

2008

AstriFlash: A Flash-Based System for Online Services

Lei Yan, Babak Falsafi, Siddharth Gupta, Mark Johnathon Sutherland, Yunho Oh, Lei Yan

Modern datacenters host datasets in DRAM to offer large-scale online services with tight tail-latency requirements. Unfortunately, as DRAM is expensive and increasingly difficult to scale, datacenter operators are forced to consider denser storage technolo ...

2023

Unlocking Energy

Rachid Guerraoui, Babak Falsafi, Javier Picorel Obando, Vasileios Trigonakis

Locks are a natural place for improving the energy efficiency of software systems. First, concurrent systems are mainstream and when their threads synchronize, they typically do it with locks. Second, locks are well-defined abstractions, hence changing the ...

Usenix Assoc2016