ARM has just announced their new Cortex A72 ARMv8 core with 3.5 times the performance of Cortex A15 ARMv7 cores, together with Mali-T880 GPU delivering 1.8 times the performance of Mali-T760, and CoreLink CCI-500 Cache Coherent Interconnect to link the new CPU, GPU and I/Os together.
- Architecture – ARMv8-A
- Multicore – 1-4x SMP within a single processor cluster, and multiple coherent SMP processor clusters through AMBA 5 CHI or AMBA 4 ACE technology
- ISA Support
- AArch32 for full backward compatibility with ARMv7
- AArch64 for 64-bit support and new architectural features
- TrustZone security technology
- NEON Advanced SIMD
- DSP & SIMD extensions
- VFPv4 Floating point
- Hardware virtualization support
- Cache – 48KB I-cache, 32KB D-cache, and 512 KB to 2MB L2 cache with ECC
- Debug & Trace – CoreSight DK-A57
- Process – 16nm FinFET
A cluster can support up to 4 Cortex A72 cores clocked up to 2.5 GHz in mobile devices and higher frequencies for larger form factor devices such as servers. Cortex A72 cores may also be combined with low power Cortex A53 cores in big.LITTLE configuration for power efficient SoCs.
ARM also claims 75% power usage reduction in typical mobile workloads thanks to the new 16nm FinFET manufacturing processor. Expected applications include premium smartphones, larger screen mobile devices, enterprise networking, servers, wireless infrastructure, digital TV, and automotive ADAS/IVAI.
- Anti-Aliasing – 4x MSAA, 8x MSAA, 16x MSAA
- API Support
- OpenGL ES 1.1, 1.2, 2.0, 3.0, 3.1
- OpenCL 1.1, 1.2
- DirectX 11 FL11_2
- Bus Interface – AMBA4, ACE-LITE
- L2 Cache – 256 to 2048KB (256 to 512KB for every 4 shader cores
- Memory System – Virtual Memory with built-in Memory Management Unit (MMU)
- Multi-Core Scaling – 1 to 16 cores
- Adaptive Scalable Texture Compression (ATSC) – Low dynamic range (LDR) and high dynamic range (HDR) for 2D and 3D images
- ARM Frame Buffer Compression (AFBC) – 4×4 pixel block size (lossless image compression format)
- Transaction Elimination – 16×16 pixel block size (Only performs a partial update to the frame buffer with the changed pixel blocks)
- Smart Composition – 16×16 pixel block size (Identical pixel blocks of input surfaces are not read, not processed for composition and not written to final frame buffer)
The GPU is also manufactured with 16nm FinFET process. Mali-T880MP16 can be clocked up to 850 MHz, and outputs up to 1700 million triangles per second, and 13.6 gigapixels per second. That’s 1.8 times better performance than Mali-T760, and ARM also claims 40% more energy efficiency.
ARM Cortex A72 processor, Mali-T880 GPU, CoreLink-C500 cache coherent interconnect, Mali-DP550 display processor unit, and Mali-V550 video processor can be combined to create SoCs support 4K video output, and decoding up to 120 fps.
Ten partners have already become licensees including Rockchip, Mediatek, and HiSilicon, and products are expected in (early) 2016. Further details may be found on ARM’s Cortex A72 and Mali-T880 product pages.