Optimised for performance efficiency
The PowerVR AX2185 is a high performing neural network accelerator per mm2. Featuring eight full-width compute engines the AX2185 delivers up to 4.1 Tera Operations Per Second (TOPS) offering high performance per mm2.
In fact, this level of performance is 3.5 times the performance of recent desktop GPUs. This makes it a much more enticing prospect for those companies that are currently using power-hungry full-fat desktop GPUs for neural network inferencing, particularly in the automotive space. Compared to competitors offering competing hardware solutions, using a network based on 4-bits of data, the AX2185 will require just 50% of the bandwidth. When compared to DSP-based competitors the bandwidth requirement is 75% lower.
- Virtual Memory
- 8-bit: 4.1 TOPS
- 16-bit: 2.0 TOPS
MACs per clock rate
- 8-bit: 2048 MACs/clk
- 16-bit: 1024 MACs/clk