‘Bizarrely powerful’ computer can make ‘super’ phone calls – BBC News
Bizarrely strong computers are being used to make the world’s biggest phone calls, and they’re only getting stronger by the day.
The BBC News app is available on Android phones, tablets and PCs for iPhone, iPad and Android phones.
The technology is based on the concept of ‘super-parallel processing’ – the idea that it’s possible to parallel process multiple instructions for a single operation.
“This means that, by using parallel processing, a processor can perform tasks that are not necessarily parallel to each other,” explains Matt Cawley from IBM Research, who co-wrote the article.
“For example, a task like checking a message in your email client can be done in parallel to a different task you might perform.”
This is very similar to the way that the human brain does things, using parallelism to improve and optimize processing speed.
“However, the parallel processing is not a simple one.
There are some fundamental constraints and limits on how much parallel processing a processor is able to do, and what kind of operations can be parallelized, so you have to be careful to make sure that your computer is not overloaded.”
The BBC News mobile app for Android and iOS has been downloaded over 300 million times, and it’s been downloaded more than 7 million times since its launch in May.
“The power of parallel processing lies in its ability to make a huge amount of data in a small amount of time,” says Cawry.
“But this power comes at a price: if a computer goes bad, it can’t be restored to working order.”
There are some limits, of course, and there are some things that are impossible to parallelize, but in general, it’s a good idea to make use of parallel computing when possible.
“In the article, Cawleys points out that many of the current generation of smartphones are powered by ARM chips, which can only be used in a ‘parallel’ way, but they also rely on a similar concept.”
If you are dealing with large amounts, you might want to avoid parallel computing at all costs.””
The problem is that parallel processing has the potential to introduce some serious problems, especially when you want to parallel compute large amounts of data.”
If you are dealing with large amounts, you might want to avoid parallel computing at all costs.
“For example if you are trying to parallel scale an application like a spreadsheet, you need to take care not to overload the CPU, which could cause problems when you are scaling large data sets.”
Another concern is that the amount of parallelism that can be achieved by a processor has become so large that it can become a bottleneck, and if a processor becomes overloaded, a problem can occur.
“With a few exceptions, parallel computing is not possible with CPUs that are smaller than a few millimeters in diameter,” Cawries says.
“You could imagine a system where you had to use more than one processor, and this would increase the complexity of your software.”
If a processor overheats, the power is wasted and the processor dies, but if a CPU is running with the right power management, it will continue to run even if it overheats.
This is something that would be a huge problem if you were trying to build a device that was able to support a high-speed data network.
“Even with the most powerful processors, it takes a very long time to get a high speed data network going,” says Andrew Jansson from IETF, which provides scientific and technical advice to the organisation that is responsible for standards and standards bodies like the IETF.
“It is possible to achieve a high rate of performance, but it takes time, and the amount that can go into it is limited.”
There is no single solution to this problem, but there are many ideas that are based on parallel processing.
One is called ‘paracosmic’ processing, where an algorithm takes a sequence of instructions and performs a series of operations on it to generate a result.
“Paracosms’ have been around for a long time, so there is a large amount of knowledge about how to write them, but very few have been able to implement them on a large scale,” says Jansson.
“So there is still a lot of work to do.”
Another idea is called a ‘spatial’ processor, which takes two instructions and computes the difference between them, or ‘spaces’.
“It is useful for finding patterns in a picture or understanding what a person’s eyes are looking at,” explains Jansson, “but it is also useful for building a 3D model of a physical object.”
This idea is similar to how the human eye perceives the world around it.
“In terms of what we see in the real world, our visual system uses the spatial processing to make sense of the world, and that processing is very computationally intensive,” says Matt C