| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Community
Forum: Chat Room
Thread: What does the cloud computing model mean for the future of distributed computing projects? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 7
|
|
| Author |
|
|
ngmwcg
Cruncher Joined: Jan 2, 2010 Post Count: 18 Status: Offline Project Badges: 
|
Ten to twenty years ago, when a consumer purchased a new PC, they would generally purchase one with the latest and greatest cpu (although they would have a few clock speeds to choose from). Sure there would be plenty of computers in people's homes without the latest and greatest cpu, but that would only be because they weren't brand new and the user had decided it wasn't time for them to upgrade yet. Users wouldn't buy a new computer with a two year old cpu to save money, such a practice was uncommon then. Things have changed dramatically over the past few years. Take Intel's lineup of x86 processors for instance, it's difficult to count how many active CPU's they offer but it looks as though there are almost a hundred active cpu's to choose from. From Atoms and Celerons to i7 extremes and Xeons.
----------------------------------------Given the shift to cloud computing, consumers are increasingly opting for low end processors to support thin clients like tablets and lightweight laptops. Meanwhile the higher performing chips are going to data centers where the business is all about efficiency and maximizing utilization. If a data center invests in a $4000+ multi-core cpu, or more likely thousands of them, they're going to do everything in their power to ensure that cpu is used to it's full capability 24/7. This is a major driver of the rise for virtualization (i.e. a server that handles a lot of web traffic may shift to handling e-mail when the web traffic starts to drop off). And since many users are sharing the same cpu, data centers have been consistently decreasing the amount of idle time. So with consumers opting for lower performing hardware and data centers doing everything in their power to reduce idle time; it seems like this could have quite a negative impact on projects like World Community Grid. World Community Grid and other projects are dependent on idle or wasted cpu's hours, something the efficiency seeking information technology business community has gone to war with. I'd like to hear thoughts from others in the community. [Edit 1 times, last edit by ngmwcg at Apr 10, 2011 10:44:48 PM] |
||
|
|
KWSN - A Shrubbery
Master Cruncher Joined: Jan 8, 2006 Post Count: 1585 Status: Offline |
On the flip side of that, the sheer volume of work that is accomplished today dwarfs what processors were capable of even a couple years ago.
----------------------------------------It's true that the power of individual computers may level off some, but there will always be a certain number of home computers crunching away and as the power of processors continually increase, the same amount of work (or more) can be accomplished on far fewer machines.  Distributed computing volunteer since September 27, 2000 |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Are there companies out there that run projects like WCGrid on srevers?
The company I work for doesn't run WCGrid at all, but it does run a similar grid software for its actuaries. It gets put on some end-user workstations, and certain number of CPUs specifically dedicated to it. It doesn't get put on servers, at least not servers intended for other uses. Reliability and responsiveness outweigh idle time concerns. |
||
|
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges: 
|
We have entered a time when small, less powerful systems are being increasingly used; netbooks are now fairly common, entry level laptops and basic desktop systems with CPU's such as the Atom are also common place. This low performance range of computerized devices is set to extend into the touchscreen slates, palmtops, mobiles, TV's... However these devices are not desktop replacements and not everything can be done over the Internet and in the cloud. I see this as a development of a new market/extension of existing processing device ranges, rather than a move away from mainstream computing.
Even with slates and mobiles there is performance competition; slates will be competing against other slates and laptops. The need for speed will remain. With such systems the architecture is moving towards low frequency processors with many cores, but in time the performances will rise. Perhaps the challenge for research projects will be to include such devices and extend the architectures that can be used. The cloud should be seen as an extension of online services. Yes it may overlap/compete with on-system apps, but it will not replace desktop computing any time soon. I expect overall CPU performances will continue to increase in all areas of computing, probably faster than research projects; we seem to be seeing increasing outages across all the Boinc projects. While the average CPU core frequencies of work and home computers may not continue to rise as sharply, frequency is only one performance factor and is not an accurate way of comparing performance. Again, the trend is moving towards many cores/threads for better multi-tasking performance. This is what the immediate challenge is; to adapt research science projects to better utilize low frequency multicore CPU's. This might mean some projects will eventually move to applications that can use multiple cores for the one task. |
||
|
|
RedMenace
Cruncher Canada, eh! Joined: Nov 19, 2007 Post Count: 28 Status: Offline Project Badges: 
|
So far gaming is still alive and well. I still prefer a PC over a console.
----------------------------------------
"If you want to go fast, go alone. If you want to go far, go together!" -- African Proverb
|
||
|
|
ngmwcg
Cruncher Joined: Jan 2, 2010 Post Count: 18 Status: Offline Project Badges:
|
Don't get me wrong. The top of the line chips available for purchase are continuing to get more powerful, particularly those gaining more cores.
But, if you look at market trends you see that low end cpu purchase are cannibalizing some of the high end cpu purchases. So 20 years ago 80-90% of computer users would purchase a top of the line cpu. Now, maybe only 5% of users opt for the most powerful chips. I'm making up the percentages but the trend is somewhere along those lines. With that said those high end cpu's that make up 5% of purchases continue to get more powerful. And yes the overall performance of the grid has continued upward, but the point of my discussion is that I feel there is the possibility that it could plateau or even decline depending on which ways things go going forward. In addition to my concern about the number of chips to choose from, I think there is some degree of a digital divide that is expanding. That is the gap between low end and top of the line cpu's continues to grow with respect to overall performance. And while there are certainly multiple metrics to look at with regard to cpu's such as overall performance vs. performance / watt the latest chips are capable of throttling these tradeoffs, making consolidation more practical. I think is an interesting discussion and I would have to agree with most of your comments. |
||
|
|
nasher
Veteran Cruncher USA Joined: Dec 2, 2005 Post Count: 1423 Status: Offline Project Badges: 
|
realize that computers are getting faster (even the low end ones) so yes more people will probably go for lower end computers in the future if they can use the cloud more BUT i do not see that slowing the Distributed computing.. it may slow its increase but people rarely downgrade there systems that they currently are running.
---------------------------------------- |
||
|
|
|