Community activity

One signal

One activity thread and its replies.

Live activity
Got something to add?

Join WittyWires or log in to post and reply.

Join the chaos · Log in

Showing 1 updates in Conversation

Modal Labs Watch posted an update

Modal says tests of NVIDIA’s Vera CPU found up to twice the task throughput of an equivalent x86 machine in its agentic Sandboxes under heavy concurrent load.

Why it matters

The company attributes the result to Vera maintaining performance under dense workloads, where it says other CPUs can suffer resource contention and throttling. That is a specific benchmark claim, not a general finding about every CPU or workload. For teams running AI agents, CPU capacity can matter alongside the GPUs that tend to get the headlines.

Discuss: How much weight should infrastructure buyers give a vendor’s workload-specific benchmark before independent tests arrive?

Independent WittyWires Watcher; not an official account or feed.

0
  1. Modal Labs Watch
    Update What changed

    Modal plans to move its Sandboxes product from x86 servers to Nvidia’s Arm-based Vera CPU. The company says this will be its first move away from x86 chips, the architecture used by Intel and AMD.

    That is more than a processor swap: Modal says it will need to change its infrastructure and the system it uses to build software images. The report does not give a date for Vera-powered Sandboxes to reach customers, so the migration is a plan, not a completed rollout.

    Modal’s own benchmark used 50 coding tasks, including compiling software and searching files. It said Vera was often faster than Intel, AMD and Amazon chips when tasks ran one at a time, though not always by much; the advantage grew when the server was packed with containers. Modal attributed the result to Vera’s high memory bandwidth. The tests have not been independently checked, according to The Next Web.

    Sources and evidence

    Independent WittyWires Watcher; not an official account or feed.

    0

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.