AMD EPYC 7003-series Server CPUs to be released in March

AMD's Server Business Unit's senior vice president and general manager, Dan McNamara, stated on Thursday that the firm is on pace to release its highly anticipated EPYC 7003-series CPUs with…

By Nivedita Bangari

March 4, 2022

2 min read

Comments

AMD’s Server Business Unit’s senior vice president and general manager, Dan McNamara, stated on Thursday that the firm is on pace to release its highly anticipated EPYC 7003-series CPUs with 3D V-Cache in March. According to AMD’s server boss, the codenamed Milan-X CPUs are sampling clients.

“We are sampling [Milan-X] products in GA today, and we are going to launch that product actually at the end of this month,” said McNamara at the SIG Annual Tech Conference 2022. “So, we are very, very excited about that.”

AMD’s Milan-X server products were released at the end of March, bringing the PC market closer to the release of the company’s Ryzen 7 5800X3D processor with 3D V-Cache, which is geared at gamers (and which promises to be one of the best CPUs for gaming).

AMD’s EPYC 7003-series ‘Milan-X’ processors include a 64MB SRAM 3D V-Cache chipset atop the computational die, which is connected to it using AMD’s proprietary packaging technique, which provides for TSVs and direct copper-to-copper hybrid bonding.

The 3D V-Cache from AMD operates as an extension of the processor’s L3 cache, boosting its size to 768MB. A Milan-X CPU with 64 cores will have a total of 804MB of cache: 32MB L2, 256MB L3, and 512MB 3D V-Cache.

Applications that require high memory bandwidth and good single-thread performance benefit greatly from large caches. As a result, AMD’s 3D V-Cache will boost the performance of both high-core-count CPUs and low-core-count processors.

AMD anticipates considerable performance gains in high-performance computing (HPC) applications, physics modelling, and electronics design automation with EPYC 7003-series ‘Milan-X’ CPUs (EDA).

“[Milan-X] is going to deliver up to a 50% performance increase across technical computing workloads like computational fluid dynamics, EDA, and any sort of advanced physics modelling is going to see a tremendous uplift here,” said McNamara. “So we are excited about this family. It is going to triple the amount of cache offered from Milan and the uplift we see, as I mentioned, is pretty dramatic.”