YouChao

This name is used as an internal codename only.

Glenfly HMUA 1.0 Architecture

Compact Power, Exceptional Performance

IMR rendering pipeline

Geometry, Rasterization, Pixel Processing, Framebuffer

SIMD + VLIW parallel computing

SIMD Execution Units, VLIW Execution Units

For portable PCs · laptop integrated GPU

Lower Power
Higher Integration
Codec Acceleration

  • IMR rendering pipeline

    Efficient, scalable graphics-processing pipeline

  • SIMD + VLIW general-purpose compute array

    Data parallelism + Instruction parallelism for high performance and efficiency

  • Multi-stream video codec

    Multi-format encode/decode and transcoding acceleration

  • Shared-memory control system

    High-bandwidth, low-latency shared-memory architecture

Chip Specifications

Domestic process technology

28nm

IMR

Immediate-mode rendering pipeline architecture

SIMT cores

256-2048

Scalable massively-parallel
compute array

Glenfly Arrow

1st Generation Glenfly HMUA Architecture

IMR · SIMD · Codec · Convolutional AI

Display interfaces

DP · eDP · HDMI · VGA

Codec

Full-function encode/decode large core


High-efficiency decode small core

Mem Fabric

0.25M~4M

High-efficiency memory-access fabric

Matrix Core

INT8 · FP8 · FP16 · BF16

FP32ALU : FP8ALU

1:4 ~ 16:64

Contact Us