Heterogeneous Computing in the Edge
Software infrastructure optimized for conventional CISC / RISC processors is evolving to meet the specific needs of the edge . Examples of edge SW platforms include Microsoft Azure Edge 6 , Amazon Greengrass 7 , VMware Edge 8 and open source edge software projects from the Eclipse Foundation 9 , EdgeX Foundry 10 , and the Linux Foundation 11 ( among others ).
These software packages manage the operating system infrastructure , configuration , security , orchestration , management , etc . of CISC / RISC processors in edge nodes . Once one of these software infrastructure packages is up and running on the processor chip of an edge node , algorithms , protocol stacks , and application software can be loaded on top to complete the functionality of the edge system .
These systems serve the requirements well for systems whose computational workloads primarily use single thread architectures and run-to-completion models . However , they start to break down for applications that use massive parallelism , or do not map efficiently to standard CISC / RISC CPU data paths . Different types of processors , such as GPUs , TPUs , and FPGAs can supplement the system ’ s CISC / RISC processors to optimize its key performance attributes . That is where heterogenous processors come in .
KEY PERFORMANCE ATTRIBUTES OF EDGE COMPUTING
There are several key performance attributes that can be used to judge the suitability of a certain processing architecture to a specific set of applications . These attributes relate to performance , efficiency , scalability , density , cost , and many similar areas .
Throughput is probably the highest priority attribute . It is about how much of a specific model of processing can be accomplished by a given edge node . Measuring this is very application dependent . Throughput could be quantified using measures like sessions / users / devices supported , link bandwidth processed , latency , model complexity evaluated , transactions / inferences / operations per second , and similar measures . This is often the numerator of a performance ratio in the form of :
6
7
8
9
10
11
- 46 - June 2021