Today's Numbers: Get 'Em While They're Hot!

[Plan.B]

Great day for a great team

Indeed!! Congrats everybody!!

[Sparky]

More....!!!

[trn]

^^^1000th post?!?

says the guy with 29,000+

Cool milestone, I almost missed that , I've been here for about a year and crunching for less, you have to remember that Dave has been here on XS for about 29 years and most of his post are spam anyways

[Jaco]

^^ yup and his postcount was reset several times already

[Havis]

18mil is nice
Keep on crunchers I suspect close to 20mil tomorrow

[croto5000]

This. Is. Awesome.

[Movieman]

Cool milestone, I almost missed that , I've been here for about a year and crunching for less, you have to remember that Dave has been here on XS for about 29 years and most of his post are spam anyways

Nah, just 5 years this month..Where has the timegone.
And what do you mean by spam?
I resemble that remark!

^^ yup and his postcount was reset several times already

Just once at 3000+..

[lahn]

Congratulations guys! I joined just a day prior to the week starting, have my old machine crunching this week too, so two running 24/7. Wish I could squeeze out more and seeing as there is clients for the PS3, why aren't there any for my Xbox360?! :P

[zalbard]

Awesome!



And this is just the beginning!

[D_A]

Congratulations guys! I joined just a day prior to the week starting, have my old machine crunching this week too, so two running 24/7. Wish I could squeeze out more and seeing as there is clients for the PS3, why aren't there any for my Xbox360?! :P

Thanks for joining!

But there is no WCG client for the PS3, sorry. That client is specific to PS3Grid. No other project supports the architecture.

[artemm]

Thanks for joining!

But there is no WCG client for the PS3, sorry. That client is specific to PS3Grid. No other project supports the architecture.

The day they develop an efficient GPU client for WCG is the day XS will rise head and shoulders over 2CH and all other teams. Unfortunately, I don't think it's in the cards... not for a looooong time.

[shoota]

The day they develop an efficient GPU client for WCG is the day XS will rise head and shoulders over 2CH and all other teams. Unfortunately, I don't think it's in the cards... not for a looooong time.

why not?

[artemm]

why not?

Because developing something like that takes a lot of time.

They're only considering HCC using OpenCL. Maybe they'll have it done in a year, maybe 2... maybe never, nobody knows.

[D_A]

GPU programming is very different than CPU programming. Add to that the issues of constantly changing language support (CUDA v OpenCl(?) ) and hardware and you have a very complicated development environment. The notion of programming and running arbitrary code on video cards is still quite new, so getting your head around the the massively parallel GPU is a real challenge.

[Sam_oslo]

Because developing something like that takes a lot of time.

They're only considering HCC using OpenCL. Maybe they'll have it done in a year, maybe 2... maybe never, nobody knows.

It shouldn't take longer time than this one, I guess the problem is the lack of common standers and disagreement between AMD/ATi and nVidia. Besides, Intel doesn't want this thing at all, they don't like competition for CPUs. All this (and some other issues too) makes the future of GPGPU a bit shaky and unsecure.

Nobody wants to spend time on developing a application that can become obsolete in near future. Otherwise, GPGPU is the future of HPC and supercomputers, only if these guys could agree and move on.

[shoota]

well other projects didn't let it stop them from using gpu's. it just seems like such a waste to let that much power go unused

[Bobsama]

well other projects didn't let it stop them from using gpu's. it just seems like such a waste to let that much power go unused

I agree fully. Even if they can only get 10% of the GPU's potential, it'll be well-worth their time and energy. My HD3850 does a theoretical 427 GFLOPs; even 10% of that (42.7 GFLOPS) would be more performance than all 6 cores I have online right now.

[Sam_oslo]

Yeah, They should start working on GPGPU, maybe they are doing so already. A high-end GPU can eat 10 high-end CPUs for lunch in these kind of tasks (parallel computing).

I don't know why these guys are waiting, exactly, but I know others having concerns about changing standards and such. CUDA and ATI Stream are "closed" and both been evolving/changing too. OpenCL is a common and open standard, but it is very young one, and it's still a bit shaky. On top of this, Intel have been trying to sabotage/delay all these, and will do the same in the future too, I guess.

Some people are insecure what kind of platform, SW, standards, etc they should choose, because this a very young field, and all standards are changing/evolving rapidly. And all major plays are fighting each other too.

[DimmuKlokSC2]

Excellent job guys, my prediction is 20+ for the next update.

[Bobsama]

Yeah, They should start working on GPGPU, maybe they are doing so already. A high-end GPU can eat 10 high-end CPUs for lunch in these kind of tasks (parallel computing).

I don't know why these guys are waiting, exactly, but I know others having concerns about changing standards and such. CUDA and ATI Stream are "closed" and both been evolving/changing too. OpenCL is a common and open standard, but it is very young one, and it's still a bit shaky. On top of this, Intel have been trying to sabotage/delay all these, and will do the same in the future too, I guess.

Some people are insecure what kind of platform, SW, standards, etc they should choose, because this a very young field, and all standards are changing/evolving rapidly. And all major plays are fighting each other too.

They should hire students and ask them to get something running for GPU clients. I don't think CUDA/Stream are going anywhere fast, but they have the potential to make the field a lot more lucrative by supporting it themselves.

[lkiller123]

They should hire students and ask them to get something running for GPU clients. I don't think CUDA/Stream are going anywhere fast, but they have the potential to make the field a lot more lucrative by supporting it themselves.

I've realized that it will be very hard since each sub-project from WCG are on its own.

For example, C4CW is from Tsinghua University in China while HFCC is from Chiba University in Japan.

[slaveondope]

I can feel it....

[shoota]

all it should really take is one big player to pick a standard and then everyone would go for that. problem is, i don't wcg is a big enough player to pull that off..

[trn]

all it should really take is one big player to pick a standard and then everyone would go for that. problem is, i don't wcg is a big enough player to pull that off..

A big player would be... maybe Nvidia devoting some of their software engineers to work with a project and develop a Cuda app with them. If Nvidia really wants to drive GGPU like they keep saying it might be worth of them to start look at more aggressively helping to create Apps that can use their GPU's. (But if we really are going to look at it like that then they would be working with Adobe or Apple for Final Cut Pro or something where businesses could suddenly have a much wider use for GGPU's besides drawing a windows desktop .)

I wonder if GGPU could actually process a widely parallel server task such as a database server or web-server. It kinda makes a little sense, both of those server task take only a little processing power, are independent. A large database or webserver can get a large amount of these small request all at once. (Good luck getting Intel to give up their grip on that cash cow though )

[WFO]

A big player would be... maybe Nvidia devoting some of their software engineers to work with a project and develop a Cuda app with them. If Nvidia really wants to drive GGPU like they keep saying it might be worth of them to start look at more aggressively helping to create Apps that can use their GPU's. (But if we really are going to look at it like that then they would be working with Adobe or Apple for Final Cut Pro or something where businesses could suddenly have a much wider use for GGPU's besides drawing a windows desktop .)

I wonder if GGPU could actually process a widely parallel server task such as a database server or web-server. It kinda makes a little sense, both of those server task take only a little processing power, are independent. A large database or webserver can get a large amount of these small request all at once. (Good luck getting Intel to give up their grip on that cash cow though )

I've only been here 3 days but why isn't gpugrid considerd WCG? It's an option on install.