From eBay · Info as of 6:48 AM ET
Intel/Habana Goya HL-100 ES 8GB Inference Card - Tested, w/ Working Sftwre Stack
$100.00 USD
Free shipping
Buy It Now
- Condition
- Open box
Open box, checked on a server to make sure they are functional.
View listing
Opens on eBay. You buy directly from the seller on eBay.
Listings from eBay. Yoinkly isn't affiliated with, sponsored by or endorsed by eBay.
Price and availability can change; check the listing on eBay before you buy.
Team, league and player names are used only to describe items; no affiliation or endorsement is implied.
Item specifics
- Brand
- Intel
- Chipset Manufacturer
- Habana Labs
- Memory Size
- 8 GB
- Compatible Slot
- PCI Express 4.0 x16
- Unit Quantity
- 1
- Memory Type
- HBM2
- Country of Origin
- United States
- Chipset/GPU Model
- HL-100-08 Goya
Description
Intel/Habana Labs Goya Engineering Sample - 8gb HBM2 Memory
Intel Habana Labs Goya HL-100 engineering sample (ES) — the industry's first purpose-built deep learning inference processor. PCIe x16, 8GB memory, dual-slot passive cooling, full height/full length, 8-pin aux power.
These aren't untested shelf pulls — each card comes from an active development fleet currently running full models and live inference workloads. Verified functional end to end: image classification, object detection, NLP, and recommender workloads. Passive cooling requires chassis airflow (server or workstation with a shroud/fan).
Buyers get a link to a GitHub repo with the most current working software stack for these cards — drivers, runtime, and setup notes to go from bare card to running inference. No hunting down abandoned SDK versions; the hard part is already done.
Engineering sample hardware — sold as pictured, fully working. Multiple available
Full description on eBayIntel Habana Labs Goya HL-100 engineering sample (ES) — the industry's first purpose-built deep learning inference processor. PCIe x16, 8GB memory, dual-slot passive cooling, full height/full length, 8-pin aux power.
These aren't untested shelf pulls — each card comes from an active development fleet currently running full models and live inference workloads. Verified functional end to end: image classification, object detection, NLP, and recommender workloads. Passive cooling requires chassis airflow (server or workstation with a shroud/fan).
Buyers get a link to a GitHub repo with the most current working software stack for these cards — drivers, runtime, and setup notes to go from bare card to running inference. No hunting down abandoned SDK versions; the hard part is already done.
Engineering sample hardware — sold as pictured, fully working. Multiple available
Sold prices
Sold prices coming soon.
We'll only show sold prices from a licensed source, with the source and method next to every number.