Intel® Core™ Processor (Series 3)
Formerly known as Wildcat Lake, Datasheet, Volume 1 of 2
| ID | Date | Version | Classification |
|---|---|---|---|
| 913965 | 05/19/2026 | Public |
A newer version of this document is available. Customers should click here to go to the newest version.
AI Acceleration Extensions in Intel® AVX2
Vector Neural Network Instructions also known as VNNI or Intel® Deep Learning Boost are an extension that can help accelerate Deep learning workloads. The processor supports an AVX2 version of Vector Neural Network Instructions (AVX2 VNNI) which provides similar functionality as the AVX-512 VNNI instruction set but limited to AVX 2.Some platforms introduce support for all signed/unsigned combinations of operands.
For cases where the data input is in FP16 or BF16 data types, Some platforms add fast upconverts to FP32 so that the data can read from memory in FP16/BF16, computed in FP32 and down-converted back to FP16/BF16 for storage in memory.