AI Accelerators vs. CPUs: Why They Think Differently
Learn more about AI Accelerators here → https://ibm.biz/~ioJcnGCZ4 The same math can behave very differently across AI hardware. Swagath Venkataramani compares CPUs, GPUs, and AI accelerators through scheduling, parallelism, and data movement. Learn why control flow, dataflow, latency, and throughput shape performance. 00:00 – Compare How AI Hardware Decides What Runs Next 00:23 – Understand CPUs, GPUs & AI Accelerators 01:13 – Compare CPU Latency with GPU Throughput 02:44 – See How CPU Control Flow Works 04:21 – Understand GPU Parallelism & Warp Scheduling 06:47 – Follow Data Through the Memory Hierarchy 07:29 – Explore Dataflow AI Accelerators 08:27 – Schedule Compute and Data Movement Explicitly 09:17 – Understand Systolic Arrays for AI 10:20 – Match AI Hardware to the Workload Continue your education in the course Dataflow Execution Models on Coursera here → https://ibm.biz/~wcnFvkLlm AI news moves fast. Sign up for a monthly newsletter for AI updates from IBM → https://ibm.biz/~cWRl81Lsd AI was used in the creation of the transcript and metadata for this video. #aiaccelerator #cpus #aihardware #gpu --------------------------------------------------------------------------------------------------------- Find us on YouTube: 🔵 IBM Technology: https://www.youtube.com/@IBMTechnology 🔵 IBM: https://www.youtube.com/@IBM 🔵 IBM Developer: https://youtube.com/@IBMDeveloperAdvocates




Join the discussion
Have something to add?
Sign in to reply, follow this thread, and earn XP when your answer helps someone.