AMD Launches On-Premises AI Coding Platform Instinct Coder 355 with Dell and Spectro Cloud

·2 views

AMD announced on November 9, 2026 Instinct Coder 355, a turnkey on-premises AI coding platform built with Dell and Spectro Cloud. Each node is a Dell PowerEdge server with AMD EPYC CPUs and eight Instinct MI355X GPUs, and AMD says it can support up to 85 developers, with 50 working concurrently. The system runs a local GLM-5.3 model and can route harder tasks to frontier models from Anthropic, OpenAI, or Google, integrating with tools like Claude Code, Cursor, and Visual Studio Code. AMD claims up to 75% lower total cost of ownership than cloud frontier models, a figure drawn from its own announcement and a Spectro Cloud benchmark run on the previous-generation MI325X hardware. The platform's roughly $450,000 hardware cost makes it best suited to heavy-use teams and organisations that cannot send code to cloud APIs.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Benchmarking Node.js setTimeout at Scale: Memory, Lateness and Restart Risks

A DEV Community article argues that the question of how many setTimeout calls Node.js can handle depends on the specific machine, so it offers a benchmark script rather than quoting figures. The author deliberately withholds personal results, since they would not transfer to other servers, Node versions or workloads. The benchmark measures per-timer memory cost, lateness relative to scheduled times, and the effect of many pending timers on event-loop delay. It recommends running the script at several sizes, adding background CPU load and using realistic payloads for more representative results. The article also flags that pending timers are lost when a process restarts, a risk that cannot be measured by the script itself.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

CloudSlasher Aims to Turn Cloud Cost Visibility into Actionable Optimization

The author, a developer who built their own technology products, found that cloud spending was hard to trace and wanted to make cost optimization more actionable. That experience led to the creation of CloudSlasher, a platform intended to go beyond cost dashboards. The piece argues that knowing a resource is expensive does not show teams whether it can be changed safely, since operational context such as production dependencies and performance needs matters. Because AWS, Azure, Oracle Cloud Infrastructure, Google Cloud, and Kubernetes each use different resource models, pricing, and APIs, building a multi-cloud tool poses engineering challenges. The article outlines these considerations, including resource discovery, utilization analysis, and preserving provider-specific context.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Germany's public energy register lists 9.7 million installations with operator contacts

Germany's Federal Network Agency maintains the Marktstammdatenregister, a public register that legally must record every energy installation in the country. The records cover solar parks, wind turbines, home batteries and biogas plants, each with GPS coordinates, capacity, commissioning dates and voltage level. Operating companies' business contact details are also included, and the data is published under Germany's open data license. The author describes a JSON endpoint behind the website that can be queried without authentication, and shows how filters can isolate utility-scale solar parks. The article also promotes a paid scraping tool built on this endpoint and notes that private individuals are anonymized while GDPR still applies to downstream use.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How to Calculate Topsoil Volume in Cubic Yards With a Simple Formula

A short guide explains that a topsoil calculator needs only one multiplication and one division, with most errors coming from measurement rather than arithmetic. Volume in cubic yards is found by multiplying length, width, and depth in feet, with depth in inches divided by 12, and then dividing by 27. Worked examples include a 4 by 8 foot raised bed at 6 inches deep, needing about 0.59 cubic yards, and a larger 10 by 12 foot bed at 4 inches, needing about 1.48 cubic yards. Round beds are handled by computing circular area from the diameter, and irregular shapes should be split into rectangles and summed. The author recommends averaging several depth readings and ordering about 10 percent extra soil to allow for settling.

0 comments Read more at DEV Community