AMD's new EPYC 'Venice' CPUs are the first to feature the Zen 6 microarchitecture. This architecture is built on TSMC’s N2P node and represents an evolution from the Zen 5 core. Key architectural changes include a redesigned vector engine, split integer schedulers, and a dual IOD design.
The Zen 6 architecture expands many out-of-order structures and incorporates iterative improvements throughout the core. Notable changes include a re-worked integer scheduler, reduced branch mispredict latency, and the ability for a single thread to utilize both decode clusters.
Preliminary testing of the Zen 6 EPYC 9996 processor demonstrated a 93% performance increase in a 2P configuration compared to previous generations. These benchmarks covered various workloads, including Python, memory, compression, rendering, Java, database, HPC, and compilation tasks.
The performance gains are attributed to the significant architectural changes within the Zen 6 core, building upon the foundation established by the Zen 5 redesign.
AMD executives, including chief Zen architect Mike Clark and corporate VP Madhu Rangarajan, clarified that the Zen 6 family, codenamed Venice, was not explicitly designed for agentic AI. They stated that the development cycle for Zen 6 began before the recent focus on agentic AI, refuting claims made by other companies like Nvidia and Arm regarding their data center CPUs.
Despite not being specifically designed for agentic AI, the Zen 6 CPUs are still capable of handling such workloads, according to AMD.
The Zen 6 architecture includes detailed breakdowns of the Zen 6 and Zen 6c CCD (Core Complex Die) and platform. It also features technologies like UPP, UBPS, and FAST. Preliminary MEXT support is also noted, with claims of achieving 95% performance at 50% DRAM capacity.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's briefNew every morning, and the back catalogue is archived by date.
AMD's Mike Clark, original Zen architect, discussed the upcoming Zen 6 core family, detailing the architectural differences between Zen 6 and Zen 6c. Zen 6c is positioned as the top-of-the-line performance core with higher IPC and full L3 cache, while Zen 6 is the high-frequency design.
AMD executives stated that their Zen 6 CPUs, codenamed Venice, were not explicitly designed for agentic AI, refuting claims made by Nvidia and Arm regarding their own data center CPUs. AMD's chief Zen architect and corporate VP explained that the development cycle for Zen 6 began before the recent focus on agentic AI, but the CPUs are still capable of handling such workloads.
Details about AMD's Zen 6 CPU core family indicate an evolution from Zen 5, with expanded out-of-order structures and iterative improvements. Key changes include a re-worked integer scheduler, reduced branch mispredict latency, and the ability for a single thread to utilize both decode clusters. These updates aim to improve performance and efficiency in the next generation of AMD processors.
AMD has launched its EPYC 'Venice' CPUs, featuring the new Zen 6 architecture built on TSMC's N2P node. This architecture includes a redesigned vector engine, split integer schedulers, and a dual IOD design, leading to significant performance improvements over Zen 5. The Zen 6 EPYC 9996 processor demonstrated a 93% performance increase in a 2P configuration compared to previous generations.