About the Series8XT

The Series8XT family of GPU cores is the first based on the Furian architecture. Targeting high-performance SoCs, they are designed to deliver high-resolution, immersive graphics content and data computation within strict power budgets for premium mobile devices and automotive.

Key features:

  • Class-leading performance per watt with up to 80% increase in fill-rate density over previous generation products
  • Full hardware virtualization with near zero-context switching overhead
  • Support for all major compute and vision APIs
  • Fully coherent for flexible deployment in heterogeneous compute system

Key benefits:

  • Multi high-resolution display support for automotive
  • Enables secure deployment of automotive services
  • Delivers the graphics & compute performance required for automotive use cases
  • Optimised for power efficiency in premium mobile devices
  • Class-leading embedded performance and energy efficiency for next-gen applications such as AR and VR

Specifications

Series8XT Description
API Support OpenGL®ES 1.1, 2.0, 3.0, 3.1, 3.2*
Vulkan1.0®*
OpenCL™ 1.1, 1.2, 2.0*
Android NN HAL
RenderScript™
Full support for latest Khronos and Android API based 2D/3D graphics and compute/vision applications.
Bus Interface AMBA®4 ACE & ACE-LITE Compatible with a wide range of bus interconnect and peripheral IP.
Security TrustZone® and multidomain security Hardware security support which exceeds the current industry requirements for protection of Digital Right Management (DRM) content. Offers baseline TrustZone concept of two isolated domains, one secure and one non-secure with enhanced support for multiple protected domains (up to seven) also available for futuristic applications.
Hardware virtualization Up to eight operating systems Full hardware virtualization support enabling the hosting of up to eight operating systems. Enables true isolation offering system robustness while maintaining maximum hardware utilisation.
Memory System Virtual Memory Integrated two-stage Memory Management Unit (MMU) to support virtual memory in combination with virtualization.
Texture Compression ETC, ETC2, ASTC and PVRTC Third-party and proprietary native compressed texture formats support for reduction in memory footprint, bandwidth and system power.
PVRIC Frame Buffer Compression 8×8, 16×4, 32×2 pixel block size Integrated within the GPU and also available as a standalone for system wide deployment is PowerVR’s lossless image compression engine. It supports compression of ARGB and YUV surfaces and a range of block sizes for optimal integration with other IP blocks. It is used to reduce memory bandwidth and therefore system power consumption.
SPU range 1 and 2 The Scalable Processing unit (SPU) consists of the geometry pipeline, fragment pipeline, shader and texture units. Highly optimised for maximum power efficiency to address the premium mobile, tablet and automotive market requirements.
Pixel rate per clock Up to 8 pixels per SPU per core The number of pixels that can be read or written per clock per SPU. Optimised for highest fillrate per mm2 for maximum UI performance per mm2.
Floating point operations per clock 192 per SPU per core High number of floating point operations per clock per SPU, for high compute/shader performance per second. Optimised for highest utilisation efficiency for maximum performance efficiency per watt.