AMD Time Machine

[Steve Lux]

So I was using my time machine next weekend and was looking over some of our old posts from five years from now. We were reminiscing about the old days when officially approved clients were not yet automatically optimised for the manufacturer and model of processors in our systems. We even had to go to each project's web site and set up parameters seperately. Ah, the primitive old days when we crunched on a wing and a prayer.

We also had a lively topic about whether we should invest in one of several available math coporocessor boards or just use one of the new Boinc clients optimized for using the power of our video card processors for crunching. Back then four core dual processor systems will become the standard for crunching.

Yes, we had some interesting times back then five years from now.

[Lagu]

Hi Steve Lux

The good old days is a period that I and other lacks. The time before Boinc when many project had their own client. We never got confused because our earned credits newer fail or as in Boinc when they implemented RAC as can go up and down. The RAC was unknown during this period and there was no other as should verify our works. We got points immediately.
On this time we had slower computers as Intel PI, PII, PIII, and P4 and I think e few AMD K6, Durons and XP 1600.
Lather AMD become more popular and more powerful and today we know what we have and and it isn´t so bad and in the future we can buy computers as is more powerful and faster as todays computers we I guess.

Without doubt, the happened has a lot of on only some few years.

Lagu

[gamer007]

I don't like the whole deadline thing. No one at SETI@Home classic complained about deadlines. Neither did any other project.

[Steve Lux]

Apparently my point was missed. If English isn't your primary language I can't blame you as I was mutilating the past, present and future tenses on purpose.

Conceptually I was exploring the future as if it were the past as a way to illustrate how present issues are likely to fade into the distant past even as future choices will eventually follow. Can you imagine the realistic capabilities of our near future?

I'm thinking that eventually Boinc client software will automatically detect and optimise based on our system's processors, ram, hdd space, other system parameters for the application/s being run. I also imagine that while video cards are making tremendous progress and are likely pushing the envelope with their ability to process math they do seem far more capable in terms of raw math crunching power than motherboard-located processors. Because of this I see two divergent paths, the first is where video processors could be used to crunch math (possibly only a niche market though) and a second path where math co-processors return as an option to once again to handle math intensive processes, initially to support video systems. Either path could be available to us in the next 5 to 10 years.

The thing is that the issues we struggle with now will eventually be part of our easily forgotten past, and the future is likely to offer us some interesting options.

[PcManiac]

I thought it was really funny! good job! I was going to reply, but I couldn't think of anything to say in that same tense, lol. it was one of many highlights of my day! "Crunching from the power of our video cards" haha!

[Nflight]

Looking at posts from 5 years and ago and thinking about, what about the future what will we think about posts from 2011 compared to 2006 and then all the way back to 2001 and before. The future looks very promising for the newbie as there effort to have to adjust every detail we have had to learn and sort out with some frustration will be so low in the future everyone will be able to accomplish so many great things we can't at this present time.

Looking at our future from our present point of veiw and then looking at our past and our accomplishments from present day was a very inquisitive point to promote, My hands down as that was brilliant Steve.

[Steve Lux]

Another thing that will push these changes in technology is the number of crunchers. Life abhores a vacuum and will not let it exist for long. As the number of crunchers increases (Boinc now lists 684,007 chrunchers. There were 540,000 when I joined in February this year) its market base gets larger. Every one of us active crunchers would like to increase our capacity without significant expense or risk to our systems.

Not that such a product is available today, but would you pay $250 for a safe product (software or hardware) that increases from 2x to 10x your system's capacity to crunch numbers for your favorite project? I'd buy two today - if the product were available and proven reliable. I am willing to bet that I'm not alone in this. Just in Distributed Computing there exists a multi-billion $$$ market in this customer base we belong to. In the not too distant future there will be vendors scrambling for their share of that potential wealth.

[Strongbow]

Apple IBoinc


Actually, using the GPU as a coprocessor is a great idea and from searching the web there seems to have been quite a lot of work in this area already so we may get a Nvidia/ATI BOINC sooner than we think!

Have a read here on using a GPU for scientific calculations if you're interested like I was http://www.csit.fsu.edu/~blanco/gpusc/gpusc_project.htm