SNIA Developer Conference September 15-17, 2025 | Santa Clara, CA
Pramod Peethambaran
Director of Engineering
Samsung
Pramod Peethambaran is a visionary technology leader with 20+ years of experience building multi-functional teams and world class products catering to cloud infrastructure focused on Storage/memory, System Software, AI/ML platform & Networking. As Director of Engineering in Samsung Semiconductors, he has led memory software platform, Samsung Cognos. His vast experience spans building software products for Semiconductor, Multimedia, Automotive, telecom and networking industry. He has IEEE publication on Cyber Security (computational storage-based deep learning inference functionality to combat ransomware) and patents granted in the AI/ML processor architecture optimizations (transaction-based instruction set architecture) and patents pending in the field of offloading based performant storage solutions and VoIP/networking. He has also served as a webinar presenter on optimizing data architecture for scale and performance.
Related Sessions
Compute Express Link (CXL) as a Scalable and Highly Cost-Effective Memory Architecture for Modern Computing Workloads
This presentation examines the transformative role of Compute Express Link (CXL) technology in addressing the critical challenges of memory scaling within modern computing architectures. As data-intensive applications in graph analytics, machine learning, and recommendation engines continue to demand exponentially increasing memory resources, traditional memory architectures face significant limitations in scalability and efficiency. The presentation provides a comprehensive analysis of current memory architecture challenges, including CPU memory channel limitations and resource utilization inefficiencies, before exploring how CXL's innovative protocol stack offers solutions through its three key components: CXL.IO, CXL.cache, and CXL.mem.
We present some real data points from our experiment using distributed in-memory database as workload running on a clustered platform with CXL used as memory expander, well complemented by Samsung Cognos. The experiment shows 5x capacity increase and meeting the targeted performance/SLA.
These findings suggest that CXL represents a fundamental shift in memory architecture design, offering a viable pathway for meeting the escalating memory demands of next-generation computing applications while maintaining efficiency and cost-effectiveness.