Mini is the one in my sig. I don't have a PS3. Are they any good? I've heard there is a program for them tho.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Finally got my 7 digits
- Thread starter twoodcc
- Start date
- Sort by reaction score
Mini is the one in my sig. I don't have a PS3. Are they any good? I've heard there is a program for them tho.
oh sorry, i didn't look at your sig. the PS3 will get you around 1,100 PPD or so. not great, but if you already had one, it might be worth it. i wouldn't go out and buy one (unless you already wanted a PS3).
do you have any other computers laying around?
I was going to buy one but I just ended up buying a BD drive for my Mini instead
.I have an iMac G5, 12" PB and my MBP around
Which I had set up for F@H but haven't been using it.
HA! good moveI was going to buy one but I just ended up buying a BD drive for my Mini instead
seeing as though you cant install Linux on them anymore I have an iMac G5, 12" PB and my MBP around
not that they would help very much
i might give my PS3 a go - i wonder what its power consumption is like.
I did see a PS3 online locally with a busted optical drive. Would that make a great dedicated F@H machine since everything else works? It's only $100.
I have an iMac G5, 12" PB and my MBP around
well those might not be all that great, but if you aren't using them, might as well give them a try.
not that they would help very much
i've heard that they old PS3's used like 200 watts, and new slim ones is closer to 100 watts, but i could be wrong. but i read about it somewhere
it wouldn't be bad for that price. is it a slim one, or an older one?
200watts! seriously!? thats nothing!!
straight from folding site. granted, that's out of date. but it should give you a good idea
no complaints from here - nice memory
thanks for that. gets 2x what my iMac does for probably 1/2 the consumption! does that make it 4x better?
no complaints from here - nice memory
no problem
. haha, yeah you could look at it like that. but really the nicer video cards and fast cpus are better. a nice GPU will get you 5,000 ppd easy. [in windows though]. they can get over 15,000 ppd, but those are $400-$500 cards.
a fast cpu can get you anywhere from 20,000 - 33,000 ppd, and when you have 2 cpus in one system, that can go up to over 80,000 ppd in one system!
15k? wowwww.no problem
a nice GPU will get you 5,000 ppd easy. [in windows though]. they can get over 15,000 ppd, but those are $400-$500 cards.
a fast cpu can get you anywhere from 20,000 - 33,000 ppd, and when you have 2 cpus in one system, that can go up to over 80,000 ppd in one system!
i would LOVE a few teslas
15k? wowwww.
i would LOVE a few teslas
i wouldn't mind a new 12-core mac pro. depending on the unit, anywhere from 55k - 85k ppd!
teslas cost about the same as a MP but would easily quadriple the output of a MP *dreams*
excuse the lame reply - at work.
GTX280 - 290GFLOPS
High end Tesla - 1.2TFLOPS.
these are outdated cards now, but represent the changes you would expect. the new 4xx series Teslas are incredible!
hmm. i'll try to look into it. i'll probably start a new thread about it
EDIT: after looking around, a tesla isn't any better at folding than a regular nvidia gpu. in fact, they appear to be worse than some higher end cards. a mac pro with 12 cores would be much better at folding
most Teslas have quadro GPUs - which will benchmark terribly in games.
a Tesla has 4xquadro cards - each card has 448 cores, thats 1792 cores - compared to a 12 core MacPro? it doesnt stand a chance
and this, straight from F@H
most Teslas have quadro GPUs - which will benchmark terribly in games.
a Tesla has 4xquadro cards - each card has 448 cores, thats 1792 cores - compared to a 12 core MacPro? it doesnt stand a chance
and this, straight from F@H
but for folding, they would just use the regular GPU client. and i've read all over the place that they are just an gtx 280 without the all the connections, and have more RAM.
so you might get 10,000 ppd out of one. the 12 core mac pro you can get close to 90,000 ppd
F@H supports CUDA and Tesla cards - you'd expect 200,000+ ppd even from the most basic Tesla machines.
linky
where are you getting this ppd figure? no where have i read about people getting any better ppd numbers than regular nvidia cards.
that link is old. but still it doesn't give any indication that you would get super numbers from tesla cards
notice that everyone in that thread ignored him.
i look to the hard forum when i need help. they are currently the top folding team right now, with over 3 billion points. i searched for tesla there, and i read about them being actually worse for folding than regular cards
haha.
let's compare the Teslas to the Quadros.. one is a GPGPU, the other is a "workstation/gaming" card.
the Tesla has 448 cores, no video out (so no overhead for that) and 4-6 GB RAM.
the quadro 480 has 480 cores (more then Tesla) however it has HALF the VRAM (1.2GB RAM) than the Tesla counterparts. so whilst there is more cores, then amount of data fed to each core is less overall. the bandwidth on these 400s is smaller also, again resulting in less performance.
all of these small differences combine to make some massive differences in computing power, given the correct software (F@H is such software) - the differences should be night and day. F@H has been written to utilise all the RAM for those GPUs for better computation, the Teslas have TONS more RAM and therefore they have lots of better performance.
*still searching for universal benchmarks that arent biased*
this is a bit hard to find. due to the fact that not many people will purchase the Teslas
haha.
let's compare the Teslas to the Quadros.. one is a GPGPU, the other is a "workstation/gaming" card.
the Tesla has 448 cores, no video out (so no overhead for that) and 4-6 GB RAM.
the quadro 480 has 480 cores (more then Tesla) however it has HALF the VRAM (1.2GB RAM) than the Tesla counterparts. so whilst there is more cores, then amount of data fed to each core is less overall. the bandwidth on these 400s is smaller also, again resulting in less performance.
all of these small differences combine to make some massive differences in computing power, given the correct software (F@H is such software) - the differences should be night and day. F@H has been written to utilise all the RAM for those GPUs for better computation, the Teslas have TONS more RAM and therefore they have lots of better performance.
*still searching for universal benchmarks that arent biased*
well i just don't think there is a folding client that takes advantage of the tesla cards. as far as i know, there is only 2 gpu clients, and they are not designed for tesla cards.
i realize that these cards are very powerful, but i just don't think folding takes advantage
having just talked to somebody who uses 32 PS3s for computational work (and uses a myriad of Teslas for their work also - she says that the Teslas are many many times faster then the consumer models, and that F@H really can take advantage of the bandwidth/memory/frequency increases of the cards. make of that what you will - its extremely hard to find solid benchmarks.