Thinkmate now offers a wide range of server configurations using the Ampere® Altra® Cloud Native Processor family.
Purpose-built for efficient performance, Thinkmate servers deliver superior performance with superior size, weight, power, and cost (SWaP-C).
Using Ampere® Altra® processors family with 32 to 128 cores and running at a consistent and predictable frequency of up to 3.0 GHz, these servers are ideal for AI inference across a range of frameworks such as PyTorch.
Configure and Buy Below.
The Ampere® Altra® processors family, built with Arm® architecture, competes with the x86 platform by having more cores to tackle compute-bound workloads. Thinkmate servers with Ampere® CPUs offer improvements in total cost of ownership (TCO) by running more efficiently per core and at a lower price per core. Our 1U and 2U servers feature a single socket Ampere® Altra® processor with 128 PCIe (Gen4) lanes, and support for all high-end RAM configurations.
Every Ampere® Altra® processor is single-threaded, runs at consistently high frequencies, and is built with large low-latency private caches. This results in predictable high performance that scales in a linear fashion for all cloud native applications. The architecture also lends itself to high utilization and delivers consistent performance under maximum load conditions.
In addition to being the leader in power efficiency, every Ampere Altra® high-performance Cloud Native Processor contains 128 cores, by far the most in the industry. These processors were architected and designed from the ground up to deliver on the metrics that matter for modern cloud native applications.
Cloud service providers’ relentless drive to achieve carbon neutrality requires power efficient server processors that deliver exceptional performance. Legacy processors fall short of meeting modern cloud compute requirements.
Ampere’s processors, along with Ampere AI® frameworks for Pytorch, Tensorflow, and ONNX, are optimized for AI CPU inference workloads, delivering higher performance than x86 for Vision, NLP, Recommender Engines, Machine learning, and any other AI applications. Ampere AI natively supports FP16 conversion, allowing for half the latency and twice the throughput of FP32 with no accuracy loss.
Ampere AI requires no model conversion from the developer, making it a simple and easy out of the box solution. For heavier AI workloads, such as generative AI and model training combining Ampere CPU with Nvidia GPU delivers the best system efficiency of any AI server on the market.
Unsure what to get? Have technical questions? Contact us and we'll help you design a custom system which will meet your needs.
Thinkmate offers discounts to academic institutions and students on purchases of Thinkmate Systems. Contact us for details.
We offer rapid GSA scheduling for custom configurations. If you have a specific hardware requirement, we can have your configuration posted on the GSA Schedule within 2-4 weeks.
Call 1-800-371-1212 
E-mail us