MemVerge in the News
Cadence Collaborates with MemVerge to Increase Resiliency and Cost-Optimization of Long-Running High-Memory EDA Jobs on AWS Spot Instances
In a move that promises significant cost savings and enhanced efficiency for design engineers, the Cadence and MemVerge collaboration solves this challenge by implementing a transparent, low-overhead incremental checkpoint/restore solution that makes these EDA jobs resilient (hot restart) to Spot pre-emptions or without needing to change the underlying EDA application.
Memcon 2024 Key Takeaways
Memory Tiering is hard. Like, really hard. It’s one of the biggest gaps in the market in terms of solutions, but Memverge and Kove are both taking different approaches to solving it.
What role does CXL play in AI? Depends on who you ask
At GTC, MemVerge, Micron and Supermicro demonstrated how CXL can increase large language model GPU utilization without the addition of more processing units. CXL does this by expanding the GPU memory pool to increase high-bandwidth memory usage at a lower cost than scaling out the infrastructure with more GPUs or more HBM. The tradeoff, however, is performance.
Startup claims to boost LLM performance using standard memory instead of GPU HBM — but experts remain unconvinced by the numbers despite promising CXL technology
Between Flexgen’s memory offloading capabilities and Memory Machine X’s memory tiering capabilities, the solution is managing the entire memory hierarchy that includes GPU, CPU and CXL memory modules.
MemVerge uses CXL to drive Nvidia GPU utilization higher
CXL v3.0 can pool external memory and share it with processors via CXL switches and coherent caches. MemVerge’s software virtually combines a CPU’s memory with a CXL-accessed external memory tier and is being demonstrated at Nvidia’s GTC event. Micron and MemVerge report they are “boosting the performance of large language models (LLMs) by offloading from GPU HBM to CXL memory.”
The promise of CXL still hangs in the balance
“… servers operate as one of three pillars — compute — with the other two being networking and storage. AI is changing this,” Fan said. “While the three [pillars] will continue to exist … a new center of gravity has emerged where AI workloads take place.”
Introducing Memory Fabric Forum (formerly known as CXL Forum)
The focus of the Memory Fabric Forum in 2024 is to engage the IT community. We’re looking to IT pros to deploy PoCs this year but survey data shows their awareness of CXL is very low. In response, we’re rolling out new digital channels to reach a community that will soon appreciate the availability of abundant, composable fabric-attached memory for their memory-hungry apps.
Samsung, MemVerge, and H3 Build 2TB CXL Memory Pool
Samsung, MemVerge, H3 Platform, and XConn have jointly unveiled their 2 TB Pooled CXL Memory System at the Flash Memory Summit. The device can be connected to up to eight hosts, allowing them to use its memory when needed.
Embedded Executive: CXL, the Latest High-Speed Interconnect
Podcast | If you are not familiar with this spec, you need to know it, understand it, and potentially design to it. To get a better understanding of this relatively new specification, I spoke to Mark Orthodoxou, the Vice President of Strategic Marketing, SoCs, at Rambus on this week’s Embedded Executives podcast.
Rest In Pieces: Servers and CXL
…talk inevitably turns to increasing bandwidth and capacity for CPUs, GPUs, and FPGAs and lowering the overall cost of main memory that gives the systems of the world a place to think. And that leads straight to the CXL protocol, of course.