Folding@home flops per client

[TalRW]

I found these images someone made on the FAH wikipedia article. I think they are great illustrations of how efficient each client is.

http://img206.imageshack.us/img206/6384/800pxf40hflopsbyplatforck0.png
http://img206.imageshack.us/img206/5827/569pxf40hflopsperclientiv1.png

[Adak]

This doesn't mention that the SMP client will undoubtedly be refined further, and can be used on all kinds of simulation projects - whereas the GPU client can not.

Adak

[WarriorII]

Holly Cow !!!

I guess we now know what the best bang for the buck is.

[TalRW]

Holly Cow !!!

I guess we now know what the best bang for the buck is.

Yeah, but you gotta remember this was made in terms of actual flops per client not points. So if you are purely after points GPU is severely docked in terms of actual folding power but if you are in it for the science it quickly becomes apparent which folds the most.

[dz_jad]

ehh....well the GPU isnt compared on points like the CPU....they have the "benchie" rig that they consider for the points....100 PPD on a P4 or something like that. Now the rig has a x1900 or some variant in it...I think they clocked it for around 600 ppd.....I think....Its in the GPU faq.

thanks for the graphs though! Its cool to check 'em out.

[jws2346]

To coin an old phrase I heard on this forum "holly cow" and not being one of the "sharpest knives in the drawer" would someone please explain what all this "flops" stuff is about? Is it something for the average folder to learn? :confused: I just got the SMP client straightened out :bang head and wonder if I didn't miss out on this "flops" deal :eek: . I'm trying to get a "handle" on this folding stuff and I want to learn everything that would make me a better folder. :shrug: Thanx for any answers. :D

[dwschoon]

To coin an old phrase I heard on this forum "holly cow" and not being one of the "sharpest knives in the drawer" would someone please explain what all this "flops" stuff is about? Is it something for the average folder to learn? :confused: I just got the SMP client straightened out :bang head and wonder if I didn't miss out on this "flops" deal :eek: . I'm trying to get a "handle" on this folding stuff and I want to learn everything that would make me a better folder. :shrug: Thanx for any answers. :D

http://en.wikipedia.org/wiki/FLOPS

[dz_jad]

to quote the first line:
In computing, FLOPS (or flops) is an acronym meaning FLoating point Operations Per Second. This is used as a measure of a computer's performance, especially in fields of scientific calculations that make heavy use of floating point calculations.

[Sleepy_Steve]

Generaly, flops are used more when looking at computational power than the ocilating frequencey of the chip (hertz)

[jws2346]

http://en.wikipedia.org/wiki/FLOPS

Now, I am throughly confused. In that article it states a cpu " By comparison, a common AMD A64 or Intel Pentium 4 general-purpose PC would have a FLOPS rating of around ten GFLOPS" :eek: but "the Xbox 360 has been announced as having total floating point performance of around one TFLOPS, while the PS3 has been announced as having a theoretical 2.18 TFLOPS":eek: . Eekers, with all these GFLOPS and TFLOPS what I'm wondering is if this apply's more to "state of the art" video cards or game rigs. (1950xt, Xbox 360, PS3, etc), then to a simple cpu? The way I read that article, remember "I'm not the sharpest knife in the drawer", is this probably pertains more to a GPU client than a CPU client? :confused:

Edit: I just thought of something a "cheap flop" is someplace cheap to rent, but this probably doesn't apply. :attn:
Yeah, I got it "FLoating point Operations Per Second".

[dz_jad]

yah....well...the GPU power is why stanford came out with a GPU (beta) client. It does their calculations really well. The reason why your normal CPU doesnt have the power that a GPU does, is because of specific applications. It took stanford much time to get a client that worked properly on the GPU; and one specific card at that (then they were able to expand). You have to know just a little bit more about hardware to understand this in full.
To put it simply, because your CPU is 'general-purpose', that's why its 'slower'. Its made a certain way that you can run anything on it, and it wont be that hard to 'get the software to run on the hardware'. GPUs are more specific, they have more specific drivers, so the software has to be modified even more to fit that specific application.
so yeah, thar ya go!

** I was talkin with a guy here at work the other day, and he says there is a company out in Utah, who for years has been trying to develop an effecient hardware that can be flashed in the blink of an eye to change the specific functionality-to change it from multi-purpose, to optimized/effecient performance of one task. the current CPUs are made to run everything, and optimized for certain tasks. Thats the limit of my knowledge, I think.