--- title: "AMD Learns to Save Millions on Server Memory: MEXT Technology Offloads Data from DRAM to SSDs Using AI" description: "AMD introduced MEXT, a technology transferring infrequently used data from expensive DRAM to cheap SSDs using AI forecasting. Across 145 servers, the company estimated savings at about $5 million." date: 2026-11-11T21:36:56.623Z lang: en url: https://xab.info/en/posts/amd-mext-server-memory-savings-ssd-dram tags: [amd, mext, server-memory, dram, ssd, epyc, machine-learning] author: "Roman V." publisher: "XAB.info" --- # AMD Learns to Save Millions on Server Memory: MEXT Technology Offloads Data from DRAM to SSDs Using AI ![A virtual digital RAM stick glows with neon blue light inside an abstract technological cyberspace.](https://xab.info/media/2026/11/11/amd-mext-server-memory-savings-ssd-dram/amd-mext-server-memory-savings-ssd-dram-1.webp) ## Key Points - MEXT technology offloads inactive data pages from DRAM to cheaper SSDs while preserving memory capacity for applications. - Predicting subsequent memory accesses is performed by a model based on the Transformer architecture without requiring a GPU. - In internal tests across 86 servers, MEXT provided about 58 TB of extra memory with estimated savings of $2.78 million (82%). - Across 145 server systems, AMD estimated total savings at roughly $5 million; commercial license distribution is planned for the next stage. AMD has introduced MEXT, a technology that allows large server systems to significantly reduce operating memory costs. The mechanism moves infrequently used data from expensive DRAM to significantly cheaper SSD storage, while machine learning algorithms ensure this does not lead to a noticeable drop in performance. Across the first 145 server systems, the company estimated the achieved savings at approximately $5 million. ### Memory as the Primary Server Expense According to AMD Vice President of Business Development Gary Smerdon, who previously led the MEXT project, RAM modules in large servers can account for 60 to 90% of the cost of the entire hardware. Replacing part of DRAM with cheaper flash memory makes it possible to reduce costs without necessarily decreasing the amount of memory available to applications. It was precisely this price disparity that determined the direction of development. ### The Performance Gap Challenge The key complexity lies in the performance difference between memory types. According to Smerdon's estimation, flash memory can be 50 times cheaper than DRAM per bit and consume 30 times less energy, yet operate roughly 500 times slower. Ordinary data offloading to SSDs can severely slow down applications, especially latency-sensitive ones, making simple data "shuffling" without intelligent management unacceptable. ### How MEXT Works MEXT utilizes a machine learning model that tracks memory accesses and attempts to predict which data pages the processor will need next. Inactive pages are moved to storage, and returned to DRAM before a repeated access. For the operating system and applications, this process must remain transparent, which ensures the seamless operation of the mechanism. ### Prediction Based on Transformer Architecture At the core of the prediction mechanism lies a model derived from the Transformer architecture. Instead of the next word, as in language models, it predicts the next demanded memory page. According to Smerdon, initial training takes less than a second on a single EPYC processor core, after which the model continues adapting to the workload pattern. A separate graphics accelerator is not required for its operation. ### Results Depend on the Application Performance figures depend heavily on the specific application. In AMD tests, a system running Apache Spark maintained 99% of its original performance when half of the DRAM was replaced with flash memory. In the Synopsys PrimeTime suite, a similar configuration provided 95% performance, and when physical DRAM was reduced to a quarter of its initial volume, the figure dropped to 80%. ### Internal Deployment and Estimated Savings AMD also published the results of MEXT's internal deployment. Across 86 company servers, the software provided about 58 TB of additional available memory. Purchasing a comparable volume of DRAM would have cost roughly $3.3 million, whereas the deployment costs of MEXT amounted to about $601,000. Thus, the estimated savings reached $2.78 million, or 82% of the cost of traditional memory expansion. ### Availability, Partners, and Context The software is intended primarily for servers running AMD EPYC processors. On Linux systems, a kernel module is provided for its operation, and the RAM View utility is offered for workload analysis. AMD is collaborating with server manufacturers, including HPE, to prepare hardware for MEXT usage, and commercial licensing distribution through partners is scheduled for the next deployment phase. AMD itself acquired the MEXT developer in June 2026, and the technology's capabilities were showcased alongside EPYC Venice server processors based on the Zen 6 architecture. ## Fact-Check & Sources - [AMD придумала, как сэкономить миллионы на серверной памяти — SSD без потерь «заменят» половину дорогой DRAM](https://3dnews.ru/1151172) - Primary source: description of MEXT, Gary Smerdon quotes, savings figures ($5 million across 145 servers, $2.78 million/82% across 86 servers), Apache Spark and Synopsys PrimeTime results, acquisition of developer in June 2026, EPYC Venice Zen 6. - [AMD придумала, как сэкономить миллионы на серверной памяти — SSD без потерь «заменят» половину ...](https://3dnews.ru/1151172/amd-pridumala-kak-sekonomit-millioni-na-servernoy-pamyati-ssd-bez-poter-zamenyat-polovinu-dorogoy-dram) - Duplicate link to the same material; short description matches the core of the article. ## FAQ ### What is MEXT and how does it work? MEXT is an AMD technology that moves infrequently used data pages from DRAM to cheaper SSDs. A machine learning model tracks memory accesses and predicts which pages will be needed next, returning them to DRAM prior to re-access so the process remains transparent to the OS and applications. ### How much cheaper and slower is flash memory compared to DRAM? According to Gary Smerdon, flash memory can be 50 times cheaper than DRAM per bit and consume 30 times less energy, but operate roughly 500 times slower. ### What performance results did testing show? With Apache Spark, the system retained 99% of original performance when replacing half of the DRAM with flash memory. In Synopsys PrimeTime, a similar configuration yielded 95%, and reducing physical DRAM to a quarter of the volume lowered the figure to 80%. ### How much did AMD save in internal deployment? Across 86 servers, MEXT provided about 58 TB of additional memory. Purchasing a comparable volume of DRAM would have cost roughly $3.3 million, while implementing MEXT cost about $601,000, yielding an estimated savings of $2.78 million (82%). ### When will MEXT become commercially available? Commercial license distribution through partners is scheduled for the next deployment phase. AMD is already collaborating with server vendors including HPE, having acquired the MEXT developer in June 2026.