Mali™-450 MP graphics processor (GPU) doubles the OpenGL® ES 2.0 performance of the successful family of graphics products. The Mali-450 MP GPU expands the range of performance points by supporting scalability of up to 8 cores, while also doubling vertex processing throughput. The Mali-400 MP GPU’s market-leading performance density has proved that it is possible to deliver both performance and power efficiency. The Mali-450 MP takes those capabilities to new levels and, with additional architectural optimization, is designed to maximize reuse of the available resources during the graphics operation. This minimizes the power and bandwidth requirements of the Mali-450 MP GPU.
The Mali-450 MP takes performance density to new levels and is designed to maximize reuse of the available resources during the graphics operation therefore minimizing the power and bandwidth requirements. The Mali-450 MP has been optimized with a focus on energy and bandwidth savings. Compact size, improved ultra-low power architecture as well as reduced memory bandwidth together with the increased latency tolerance results in lower costs of manufacturing and minimal energy consumption. Linear performance scaling and focus on maximizing processing efficiency, even when rendering at four times Full Scene Anti-Aliasing (4xFSAA), ensures superior image quality and provides a perfect solution for 2K and 4K screen resolutions.
These significant improvements in conjunction with software compatibility across the family of Mali-300, Mali-400 MP and now Mali-450 MP enable current Mali users to expand their range of graphics performance points whilst reusing their existing software investments. Mali-450 MP also lowers the barrier of entry into the market as new entrants can benefit from the vibrant dynamic ecosystem, wide API support; high quality optimized mature drivers all already established on Mali-400 MP.
270MHz
650MHz
in 40nm LP
in 28nm HPM
59Mtri/s, 1.1Gpix/s
142Mtri/s, 2.6Gpix/s
59Mtri/s, 2.2Gpix/s
142Mtri/s, 5.2Gpix/s
4xAA
16xAA
4x Multi-Sampling with virtually no performance drop
16xAA outperforming all implementations of comparable quality