Moving towards Mobile: Catching up with the decline in PC sales

Message boards : Promotion : Moving towards Mobile: Catching up with the decline in PC sales
Message board moderation

To post messages, you must log in.

AuthorMessage
noderaser
Avatar

Send message
Joined: 2 Jan 14
Posts: 276
United States
Message 52078 - Posted: 22 Jan 2014, 6:49:42 UTC

I'm sure that everyone has heard by now that sales of traditional laptop and desktop PCs have started to slide (one estimate I saw was 10% decrease in sales in 2014 vs. 2012) in favor of mobile devices (phones and tablets) and appliances such as set-top boxes and gaming consoles. While BOINC for Android might be able to recuperate some of the potential losses due to declining PC sales, mobile devices have nowhere near the same computing power of a traditional desktop with (or without) a GPU. Grid computing for profit ventures (bitcoin mining, etc) are even moving away from GPUs, toward purpose-built ASIC and FPGA devices. While developments in mobile computing in the future will reduce the overall loss of computing power, I doubt that mobile devices will ever be on par with or superior to their stationary cousins.

Some might see this as a death knell for distributed computing ventures like BOINC and their need for traditional systems with open programming interfaces and general-purpose computing hardware. While it's true that the number of potential users and their traditional hosts may be declining, it will increase the value of the contributions from those who are still purchasing new PCs. The potential contribution of corporations, government, schools and other organizations with pooled resources will become more important, even though returns from such pools can be difficult to acquire and administer.

I'm not saying that this is all bad; the development of mobile computing devices has improved energy efficiency and increased access to technology and information through cheaper devices that allow people to connect with each other in ways that were the realm of science fiction a few years ago. However, it is a trend that BOINC and its reliant projects will need to keep up with, to ensure continued access to volunteer computing resources. If BOINC is no longer a viable and productive platform for them, they will be forced to find other options--such as discontinuing research, finding other areas of study or requiring lumps of cash to farm the work out.

Now, more than ever, promotion of BOINC and the projects it supports will be essential to keeping the movement alive and moving forward. I think there are two major areas that need to be addressed; first, continuing to develop the core BOINC platform and making sure that it gets released to as many new platforms as possible, to follow the trend of more but less powerful devices. Second, BOINC itself and the projects need to become a little more attuned with the general populace. I've tried to recruit other people to BOINC, but I think it sounds a little too tech-y and intimidating to them; installing software, finding projects and signing up, then managing those projects and preferences so as not to interfere with user of their computer, etc. We need some PR and marketing gurus to come up with a way to make BOINC appeal to the mainstream computing user; comparing benchmarks and stats isn't going to do it. And, the projects need to communicate with their volunteers better, whether it be by ensuring rapid response to user inquiries, or simply showing that the volunteers' contributions actually mean something and are being used to achieve a goal. In over 10 years of being a grid computing volunteer, the proof of computational worth has remained the most elusive, and is something I struggle with all the time when considering my project load.

In all of this, it's important to remember that BOINC needs us. It needs us to volunteer computing resources. It needs us to spread the word and encourage others to participate. It needs us to start projects that use the framework, and contribute changes so that others may benefit. It needs us to help out with financial, funding and grant concerns. So far, BOINC has been a great platform for both the researchers and the select few of us who are crazy enough to crunch endlessly in the pursuit of an unglamorous virtual trophy--and I'm sure that BOINC will continue to be great with community support and dedication towards using what we have to make something else better.
My Detailed BOINC Stats
ID: 52078 · Report as offensive
noderaser
Avatar

Send message
Joined: 2 Jan 14
Posts: 276
United States
Message 52117 - Posted: 24 Jan 2014, 3:38:59 UTC

I read today that Lenovo has the leading bid to acquire part of IBM's server business, which they are looking to part with due to declining sales... Yet, they are still third in the server sales ranking. Not a development I would have expected, since servers are the backbone of everything online--especially a lot of mobile and "cloud" apps.

I suppose some of the decline in sales could be due to market saturation; simply, everyone already has a computing device--whether it be a traditional PC or mobile device.
My Detailed BOINC Stats
ID: 52117 · Report as offensive
Coleslaw
Avatar

Send message
Joined: 23 Feb 12
Posts: 186
United States
Message 52238 - Posted: 31 Jan 2014, 18:45:33 UTC

Virtual Machines are probably a large part of this too. Companies don't need dedicated boxes any more for each individual purpose. One large server to control them all isn't that expensive compared to multiple middle grade ones. I know the company I work for went from several down to mainly two in corperate. The one handles most needs and the second is just a backup. Then there is a server in each town for the local needs. Their server purchasing drastically changed and cut a ton of their costs.
ID: 52238 · Report as offensive
mretome1960

Send message
Joined: 30 Apr 14
Posts: 2
United States
Message 53845 - Posted: 30 Apr 2014, 1:06:13 UTC

I am not so sure that mobile devices won't be able to pick up the slack. I just got a Samsung ATIV SE phone with a quad-core 2.3 GHz cpu. It runs Windows Phone 8 so there isn't a Boinc port that I can run but the S5 has a quad-core 2.5 GHz and it runs Android.
ID: 53845 · Report as offensive
noderaser
Avatar

Send message
Joined: 2 Jan 14
Posts: 276
United States
Message 53849 - Posted: 30 Apr 2014, 3:15:44 UTC
Last modified: 30 Apr 2014, 3:17:59 UTC

By the two measurements BOINC uses for performance, floating point and integer speed, mobile devices really aren't anything special. Even though they don't have the raw power, what makes them attractive is the performance versus power required. That is where they are superior to traditional architectures, and why we're seeing a lot of developments in ARM devices for servers and other applications where the low power consumption is desirable over raw computational ability.

My example is a bit outdated (Galaxy S3, 1.5 GHz dual-core) but is still useful for comparison. According to a project, it is capable of 525 million floating point ops/sec and 1,560 million integer ops/sec per core.

A comparable desktop that produces 511 million floating point ops/sec is my trusty Pentium III 550 MHz.

A comparable desktop that produces 1,479 million integer ops/sec is a Pentium III 1 GHz.

My Core i5 (2400S, 2.5 GHz) desktop runs 3,090 million floating point ops/sec and 8,390million integer ops/sec, for each of its four cores.

However, all of those desktop machines (if I still had them today) would draw around 100 watts or more. My phone, charging and running BOINC draws about 4 watts.

Naturally, any additional hosts are beneficial, but it will take a lot of mobile devices to replace the raw computational power of desktops and laptops.
My Detailed BOINC Stats
ID: 53849 · Report as offensive

Message boards : Promotion : Moving towards Mobile: Catching up with the decline in PC sales

Copyright © 2018 University of California. Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.