Tenstorrent Wormhole: Scalable AI Processing with Tensix Cores

Tenstorrent's Wormhole™ PCIe boards, featuring Tensix Cores™, provide powerful and scalable AI processing. Designed for superior performance at a competitive cost, Wormhole™ offers flexible solutions for a wide range of AI workloads.

Company website

Wormhole Product Lineup:
Wormhole™n150d

The Wormhole™ n150d:
PCIe add-in board, designed for AI processing

- It's powered by Tenstorrent's Wormhole™ Tensix Processor, a specialized AI processor.
-It operates at up to 160W, aiming to deliver superior performance per dollar compared to traditional GPUs.
- It supports a wide range of data precision formats, which is crucial for various AI workloads.

  • Cooling:
    The board includes a 2.5-slot active cooling solution to manage thermal performance.

  • Target use:
    Workstations and servers..

  • Memory:
    12GB GDDR6 Memory & 72 Tensix Cores

    In essence, the n150d is designed to provide a flexible and cost-effective solution for AI processing, with a focus on scalability and developer accessibility..

Wormhole™ n300d

Wormhole™ n300d: The higher-performance option, designed for users with more demanding AI workloads.

- The most significant difference from the n150d is that the n300d features
-It operates at up to 300W, reflecting the increased processing capacity.
- Like the n150d, superior performance for cost compared to traditional GPUs and supports broad data precision formats. The dual processors naturally lead to greater overall performance.
-It retains the ability to network into a multichip mesh, allowing for scalable performance in workstations and servers, just like the n150d.

  • Cooling:
    It includes a 2.5-slot active cooling solution.

  • Target use:
    Workstations and servers.

  • Memory:
    24GB of GDDR6 memory, 128 tensix Cores(64 per ASIC)

    Key Differences from n150d:

    - The n300d has double the processing power due to its dual Tensix Processors.
    - The n300d consumes significantly more power (300W vs. 160W).
    - Doubled memory and core count.

Contact us

Neurometric provides engineering consulting services for AI hardware. We can work with CPUs, GPUs, and have unique and rare expertise in many of the new and novel AI hardware chips. If you are looking to benchmark your model against various types of compute, need help designing a new AI chip into your device, or want help implementing AI hardware in a new project, please fill out this form, or give us a call.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Contact us

Neurometric provides engineering consulting services for AI hardware. We can work with CPUs, GPUs, and have unique and rare expertise in many of the new and novel AI hardware chips. If you are looking to benchmark your model against various types of compute, need help designing a new AI chip into your device, or want help implementing AI hardware in a new project, please fill out this form, or give us a call.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.