Computer vision is a computational expensive tasks. Most of existing commercial product in the market requires GPU in deployment. The limitation of using GPU is product proprietary and hardware expensive, especially in the age of GPU demanding for crypto currency mining.
To overcome this limitation, our underlying engine is well designed to supports variety of platform including GPU, CPU and NPU. The technique used is called 'parallel computing' which distribute processing load into all processing unit cores and gathering back for the result. The difficulty is task splitting, assignment and combine processed part of result into single one output.
We also apply our technology into data mining and big data processing. The project nick name is called 'No Mercy Computation: NMC'. The NMC is set of data mining algorithm that redesigned to support multicore processing. Its functions including, classification, clustering and association mining. Below is an example demonstration of how fast can boost data mining processing on the same machine.
To test the performance, we use the dataset from one of world's leading database software testing dataset of market basket analysis. We run both software and capture time used in minutes as well as compare the final result. We found that the final result is exactly the same but with NMC, 33 times faster than traditional processing.
![entry image](http://crestkernel.com//include/images/blog/standard/nmc_chart.jpg)
NMC is a well designed engine that currently support data mining on multicore processing. It shares the same technology with DeepEyes which specifically build for computer vision. Please consider contact us if there are suitable use case of NMC in any computation intensive area you might have. We provide on the shelf software, on premises and customize project to fits your business requirements.