Multi-core support
NoiseMap can, as an option, fully utilise computers that have any number of cores or processors. The extra cores are used during noise contour or receiver calculations. Typically, a quad-core processor can obtain a 5-times speedup over a single-core machine, and this can do away with the need for distributed computing.
Multiple instances
NoiseMap can, as a further option, allow multiple instances of NoiseMap to be run simultaneously in the same machine. This lets you compare two models or scenarios side-by-side, or to transfer models between databases. It is not intended to speed up calculations as the resources of the computer would simply be shared between the instances, which could cause input/output bottlenecks and might slow down the throughput.
Multi-core computers
Most Intel and AMD central processor chips used in the current (2019) generation PCs have multiple cores – in essence this means they are like several independent computers built into one chip. This is useful if you want to multi-task several independent programs – such as running a word processor, spreadsheet and web-browser simultaneously, as each one can be allocated to its own core. This is done automatically by the Windows operating system.
But for a single program to take advantage of a multi-core processor, the computing task needs to be divided into separate tasks that can be run in parallel. In other words, the tasks must be run at the same time without interfering with each other, or indeed without causing new bottlenecks which could slow things down and lose any advantage of the multi-tasking. There are many complex issues in doing this successfully, but these have been resolved by NoiseMap, to the extent that a quad-core processor can bring about a five times speedup in processing of calculations. This is because some of the necessary preliminary analysis performed by NoiseMap used by all the processors – thus reducing the overhead.
In simple terms, the latest (9th) generation of Intel Core i3 processors for desktop use have four cores, i5and i7 processors have six cores, and i9 processors offer up to eighteen cores. Clock speed is also important and the newer processors have a turbo-boost mode to 4.7 GHz.
What is the best option? A fast clock speed, memory and disk are the first requirement. Heat generation is a limitation on clock speed, meaning that multi-core chips can usually run at maximum when utilising one core, but may have to use less boost when using many cores. Thus, when choosing a multi-core computer, ensure that it has an efficient cooling system to maintain high clock speeds over long time periods.
Much will depend on your model size and the amount of design work that you need to do. For intensive projects, the choice could be between having several cheaper computers collaborating on a task, or having one multi-processor computer.
At the current time, it makes sense to buy preferably a six-core machine. You will need to buy a NoiseMap licence that lets you run the six-core version. If you have to calculate contours for very large areas, then you may prefer to have the ‘remote server’ version of NoiseMap which allows many separate computers to share the processing load. The extra cost of setting up the server is certainly justified for a medium to large organisation, but a small organisation may prefer to keep things simple by using the ‘standalone database’ version of NoiseMap with a six-core processor such as the Intel Core-i5-8500 running at 3.0 GHz, with 4.1 GHz turbo. A quality machine can be had with this processor for around GBP 700 plus VAT (not including the peripherals.)
A machine with the specification could easily host the database server as well, in which case you could take advantage of the ability to add queue calculations to be done in the background. By additionally running within this machine a parallel ‘instance’ of NoiseMap that was permanently listening to the calculation queue, calculations could take place in the background without any disruption to your main design work.