Random Musings About the Value of CPUs vs CUDA

Message boards : Number crunching : Random Musings About the Value of CPUs vs CUDA
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 . . . 11 · Next

AuthorMessage
Fred W
Volunteer tester

Send message
Joined: 13 Jun 99
Posts: 2524
Credit: 11,954,210
RAC: 0
United Kingdom
Message 848436 - Posted: 3 Jan 2009, 0:17:24 UTC - in response to Message 848429.  

CUDA is Fast on the right GPU..
My GTX 260 core 216 would be top 10 IF the app was stable... (current clocks are 700/1510/1070)

60 credit wu takes 10 min 24*60/10 = 144 * 60 = 8640
14 credit wu takes 3.25 min 24*60/3.25 = 443 * 14 = 6203

So Depending on the work assigned I would be from top 10 to top 50...
Again this is IF the app were stable.

~Bob

Then add on another 6000 from 4 CPU cores each crunching 2 AP WU's per day and where does that put you?

F.
ID: 848436 · Report as offensive
Profile gregh

Send message
Joined: 10 Jun 99
Posts: 220
Credit: 4,292,549
RAC: 0
Australia
Message 848445 - Posted: 3 Jan 2009, 1:01:09 UTC - in response to Message 844044.  

Its good to see Seti able to do more workunits and get the results faster from less computers.....but I have to ask where does this leave those of us with older (non-CUDA capable) PCs.......

I for one dont need to nor want to upgrade my PCs...not worth the effort for just Seti....plus even if I wanted to the $$ arent there (for 4 PCs) to do so......

So does this leave those of us using only CPU's for crunching to the fate of a 486 PC ? Also-rans in the world of overclockers and new GPUs ??? At 1/10 the work done it hardly seems worth the price of the electricity....

I'll keep crunching away (for now) and see what comes of this situation. But from what I have been reading I'm not optimistic.


For my part, it's 6 of one and half a dozen of the other. I started crunching simply because I wanted to do it. I started doing it, in 1999 on a P2 233mhz (since junked) and continued, later that year, on a P3 600 that I built. I still have that same machine working and doing ONLY Seti along with 2 x P4 3.0Ghz machines. They are definitely not state of the art, any of them, but they DO actually do the work and that is all that is really needed.

Will people finish their WUs faster than me? Sure! They always have no matter WHAT machinery I have had. Will I ever make the top of the list? Sure! If I win the lottery I will buy an array of machinery JUST to do Seti stuff in and build a big room for the computers and I only. That will work, too! Until then, me and my 2 x P4 3.oGhz and 1 x P3 600Mhz machines will happily continue doing crunching on what we have.

Sure, I could also go buy a new video card that is CUDA capable but I dont feel the need for it right now. Eventually I will. Until then, as the great Bill and Ted said "Party on, Dude!"

ID: 848445 · Report as offensive
Profile James Sotherden
Avatar

Send message
Joined: 16 May 99
Posts: 10436
Credit: 110,373,059
RAC: 54
United States
Message 848453 - Posted: 3 Jan 2009, 1:31:15 UTC

On May 16 09 I will have ten years of doing seti work. Its the only project ive wanted to do. How ever that being said I dont know how long I will do seti. For many years i had a pentium 1 taking 24 hours to get a WU done then when i got a p4 it was like wow this is fast. Ive let my computer run 24/7 frpm the start. In April of this year I bought an Imac dual core intel and this baby is faster still. Now both computers run 24/7. The occaisinol outage dosent get me down as my old p4 is running an AP and its happy for 5 days. But ever since cuda started it seems theres nothing but outages. Now technology will move on but i dont think i have the cash to keep being current with it. So when the day comes that my CPU isnt welcome at seti, I will find another project or just start turning off my computers when im done with them.


[/quote]

Old James
ID: 848453 · Report as offensive
OzzFan Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Apr 02
Posts: 15691
Credit: 84,761,841
RAC: 28
United States
Message 848577 - Posted: 3 Jan 2009, 8:16:36 UTC - in response to Message 848453.  

On May 16 09 I will have ten years of doing seti work. Its the only project ive wanted to do. How ever that being said I dont know how long I will do seti. For many years i had a pentium 1 taking 24 hours to get a WU done then when i got a p4 it was like wow this is fast. Ive let my computer run 24/7 frpm the start. In April of this year I bought an Imac dual core intel and this baby is faster still. Now both computers run 24/7. The occaisinol outage dosent get me down as my old p4 is running an AP and its happy for 5 days. But ever since cuda started it seems theres nothing but outages. Now technology will move on but i dont think i have the cash to keep being current with it. So when the day comes that my CPU isnt welcome at seti, I will find another project or just start turning off my computers when im done with them.


It may be a very long time before your CPU isn't "welcome" at SETI, considering even an old Pentium 60MHz CPU was still able to turn in SETI@Home MultiBeam workunits on time (using an optimized application). I'd imagine that even old Pentium 200's are still plenty useful. Besides, if you add up all the power of the "low power" computers, they contribute far more to the science than all the CUDA cards, and all the power cruncher farms put together.
ID: 848577 · Report as offensive
Profile Francois Piednoel
Avatar

Send message
Joined: 14 Jun 00
Posts: 898
Credit: 5,969,361
RAC: 0
United States
Message 848580 - Posted: 3 Jan 2009, 8:28:52 UTC

For the moment,I see no evidence of CUDA version being fast, nowhere in the top 100 ... I guess, there is something NV does not tell you ...

I got a surprise coming ... i am p.ss off ... Last time I did, I toke the Top 1, just to make a point ...

Come and beat me if you can mister NV ... lol
lol
ID: 848580 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 848586 - Posted: 3 Jan 2009, 8:49:13 UTC - in response to Message 848580.  
Last modified: 3 Jan 2009, 9:07:35 UTC

For the moment,I see no evidence of CUDA version being fast, nowhere in the top 100 ... I guess, there is something NV does not tell you ...

I got a surprise coming ... i am p.ss off ... Last time I did, I toke the Top 1, just to make a point ...

Come and beat me if you can mister NV ... lol
lol

RAC: 345 - about what are you talking at all ??
CUDA here not for silly top 100 competition, it's about speeding up project computations.
ID: 848586 · Report as offensive
Profile Francois Piednoel
Avatar

Send message
Joined: 14 Jun 00
Posts: 898
Credit: 5,969,361
RAC: 0
United States
Message 848591 - Posted: 3 Jan 2009, 9:07:45 UTC - in response to Message 848586.  
Last modified: 3 Jan 2009, 9:28:23 UTC

For the moment,I see no evidence of CUDA version being fast, nowhere in the top 100 ... I guess, there is something NV does not tell you ...

I got a surprise coming ... i am p.ss off ... Last time I did, I toke the Top 1, just to make a point ...

Come and beat me if you can mister NV ... lol
lol

RAC: 345 - about what are you saying at all ??
CUDA here not for silly top 100 competition, it's about speeding up project computations.


It you look closely, you ll figure out that my RAC history can make you look like a beginner, with all the modesty that goes with it.
I even got to the top 20 participant 2 years ago ... In about the same periode as you, i did compute more than double your credits ... So ,I invite you to speed up a little my friend, it is not with 7000 that you will catch up ... lol

look at my RAC tomorrow, and monday! have fun !

LOL x 100000000000000000
ID: 848591 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13855
Credit: 208,696,464
RAC: 304
Australia
Message 848599 - Posted: 3 Jan 2009, 9:31:41 UTC - in response to Message 846613.  

Is it dumb not to use cuda? I see gpus for less money than a memory upgrade ,$59 with what sounds like more improvment.

Don't know where you are, but here in Australia it's not possible to get a reasonably fast CUDA capable card for any less than about $200.

Grant
Darwin NT
ID: 848599 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13855
Credit: 208,696,464
RAC: 304
Australia
Message 848601 - Posted: 3 Jan 2009, 9:32:40 UTC - in response to Message 846704.  
Last modified: 3 Jan 2009, 9:33:05 UTC

But I do strongly believe that the administration thought that this new app would quickly attract lots of new users (which seems to work) to squeeze money out of them...

How would it squeeze money out of them?
Grant
Darwin NT
ID: 848601 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13855
Credit: 208,696,464
RAC: 304
Australia
Message 848602 - Posted: 3 Jan 2009, 9:36:47 UTC - in response to Message 846882.  

From my observations so far CUDA seems to be about level with my 1.86GHz Core2Duo time wise....nothing like the up to 10x claimed...crunching with an factory OC'd 8800GT

Like most claims i expect they were comparing the top of the line video card at stock speed running the CUDA application with a top of the line CPU at stock speed running the offcial app; not an overclocked to the absolute limit rig running the optimised app.

Grant
Darwin NT
ID: 848602 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 848603 - Posted: 3 Jan 2009, 9:45:02 UTC - in response to Message 848591.  

[offtopic]
For the moment,I see no evidence of CUDA version being fast, nowhere in the top 100 ... I guess, there is something NV does not tell you ...

I got a surprise coming ... i am p.ss off ... Last time I did, I toke the Top 1, just to make a point ...

Come and beat me if you can mister NV ... lol
lol

RAC: 345 - about what are you saying at all ??
CUDA here not for silly top 100 competition, it's about speeding up project computations.


It you look closely, you ll figure out that my RAC history can make you look like a beginner, with all the modesty that goes with it.
I even got to the top 20 participant 2 years ago ... In about the same periode as you, i did compute more than double your credits ... So ,I invite you to speed up a little my friend, it is not with 7000 that you will catch up ... lol

look at my RAC tomorrow, and monday! have fun !

LOL x 100000000000000000


ROFL... if you could be more stable in your achievements you could bring much more processing power to project than you do now. Try too keep this "fantastic" RAC you speak about not 1-2 days or weeks but constantly - then we will talk again ;) To process twice more than my hosts did - it's not big achievement at all. There is only one quad in my hosts and even not from high end type of quads. And it crunches only few months already. So try to be more constant in fast crunching. Your blowed RAC records do very little help here...

[/offtopic]

And about CUDA and top 100 - when one of top CPU will be combined with top GPU - when we return to this question - will this CPU+GPU pair outperform matched CPU or not. Now you compare eggs with bananas - top CPU w/o GPU and weak CPU with weak GPU....
ID: 848603 · Report as offensive
Fred W
Volunteer tester

Send message
Joined: 13 Jun 99
Posts: 2524
Credit: 11,954,210
RAC: 0
United Kingdom
Message 848604 - Posted: 3 Jan 2009, 9:53:46 UTC - in response to Message 848591.  


<snip>
CUDA here not for silly top 100 competition, it's about speeding up project computations.


It you look closely, you ll figure out that my RAC history can make you look like a beginner, with all the modesty that goes with it.
I even got to the top 20 participant 2 years ago ... In about the same periode as you, i did compute more than double your credits ... So ,I invite you to speed up a little my friend, it is not with 7000 that you will catch up ... lol

look at my RAC tomorrow, and monday! have fun !

LOL x 100000000000000000

Sorry, Who?, the record is getting a bit worn so the message isn't getting through. Are you saying that your employers are planning to launch something new in a few months and we have to suffer your interminable sniping at everyone who is trying to contribute to the project until then?

F.
ID: 848604 · Report as offensive
Profile Francois Piednoel
Avatar

Send message
Joined: 14 Jun 00
Posts: 898
Credit: 5,969,361
RAC: 0
United States
Message 848605 - Posted: 3 Jan 2009, 9:57:35 UTC - in response to Message 848603.  

[offtopic]
For the moment,I see no evidence of CUDA version being fast, nowhere in the top 100 ... I guess, there is something NV does not tell you ...

I got a surprise coming ... i am p.ss off ... Last time I did, I toke the Top 1, just to make a point ...

Come and beat me if you can mister NV ... lol
lol

RAC: 345 - about what are you saying at all ??
CUDA here not for silly top 100 competition, it's about speeding up project computations.


It you look closely, you ll figure out that my RAC history can make you look like a beginner, with all the modesty that goes with it.
I even got to the top 20 participant 2 years ago ... In about the same periode as you, i did compute more than double your credits ... So ,I invite you to speed up a little my friend, it is not with 7000 that you will catch up ... lol

look at my RAC tomorrow, and monday! have fun !

LOL x 100000000000000000


ROFL... if you could be more stable in your achievements you could bring much more processing power to project than you do now. Try too keep this "fantastic" RAC you speak about not 1-2 days or weeks but constantly - then we will talk again ;) To process twice more than my hosts did - it's not big achievement at all. There is only one quad in my hosts and even not from high end type of quads. And it crunches only few months already. So try to be more constant in fast crunching. Your blowed RAC records do very little help here...

[/offtopic]

And about CUDA and top 100 - when one of top CPU will be combined with top GPU - when we return to this question - will this CPU+GPU pair outperform matched CPU or not. Now you compare eggs with bananas - top CPU w/o GPU and weak CPU with weak GPU....


Right now, the CUDA stuff is buggy like hell ... That's it ...
By the time they get it fixed, I ll have my own toys to make them look slow.

Again, you can say what ever you want, they can't beat the MACs, neither skulltrail

I propose they return in the drawing board, i look at the code, they will never get fast this way, they touch too many times the data. I know one product that will terminate it ... please keep fixing the bugs! lol.

I am really annoyed by shameless marketing claims, because they are not supported by anything.
ID: 848605 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 848609 - Posted: 3 Jan 2009, 10:16:11 UTC - in response to Message 848605.  


I propose they return in the drawing board, i look at the code, they will never get fast this way, they touch too many times the data. I know one product that will terminate it ... please keep fixing the bugs! lol.

I am really annoyed by shameless marketing claims, because they are not supported by anything.


Your shameless marketing claims not supported by anything also ;)
Even your computers are hidden. About what "toys" and what achievements you talk ?? You looked on code - so - any suggestions of improving? Or just general words again as it was with other opt apps ?
Yes, current CUDA app buggy - can you fix it? No ? Just pointing finger on first real cool effort (yes, it still buggy, but it works most of the time now - I haven't seen any working code from you, even buggy one) and giving void promises ?
ID: 848609 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 848611 - Posted: 3 Jan 2009, 10:22:07 UTC - in response to Message 848605.  
Last modified: 3 Jan 2009, 10:22:35 UTC


Again, you can say what ever you want, they can't beat the MACs, neither skulltrail

Is it really SO hard idea to comprehend ? GPU is co-processor. Combine this skulltrail with GPU - and you will beat this skulltrail alone.
Pretty simple, don't you think so ?
ID: 848611 · Report as offensive
Profile Francois Piednoel
Avatar

Send message
Joined: 14 Jun 00
Posts: 898
Credit: 5,969,361
RAC: 0
United States
Message 848612 - Posted: 3 Jan 2009, 10:29:28 UTC - in response to Message 848611.  


Again, you can say what ever you want, they can't beat the MACs, neither skulltrail

Is it really SO hard idea to comprehend ? GPU is co-processor. Combine this skulltrail with GPU - and you will beat this skulltrail alone.
Pretty simple, don't you think so ?


For the moment, your statement is wrong, you can repeat it many many times, it will not change the result, it does NOT increase the speed of a good processor.

Is it so hard to comprenhend, you are repeating again and again something that it is not true.
ID: 848612 · Report as offensive
Profile Francois Piednoel
Avatar

Send message
Joined: 14 Jun 00
Posts: 898
Credit: 5,969,361
RAC: 0
United States
Message 848613 - Posted: 3 Jan 2009, 10:30:47 UTC - in response to Message 848609.  
Last modified: 3 Jan 2009, 10:38:17 UTC


I propose they return in the drawing board, i look at the code, they will never get fast this way, they touch too many times the data. I know one product that will terminate it ... please keep fixing the bugs! lol.

I am really annoyed by shameless marketing claims, because they are not supported by anything.


Your shameless marketing claims not supported by anything also ;)
Even your computers are hidden. About what "toys" and what achievements you talk ?? You looked on code - so - any suggestions of improving? Or just general words again as it was with other opt apps ?
Yes, current CUDA app buggy - can you fix it? No ? Just pointing finger on first real cool effort (yes, it still buggy, but it works most of the time now - I haven't seen any working code from you, even buggy one) and giving void promises ?


really?
You did not see the code I posted here? you are blind?

Just to show how badly you do not follow stuffs ... http://setiathome.berkeley.edu/forum_thread.php?id=38030&nowrap=true#525611 (true, I refuse to provide build code, because the legal departement of my company does not want to agree with the license, and I use SETI only as habby)

http://setiathome.berkeley.edu/top_hosts.php
3666921 | tasks 1 awit 11,210.50 1,370,822 GenuineIntel
Intel(R) Xeon(R) CPU X5482 @ 3.20GHz [x86 Family 6 Model 23 Stepping 6] Darwin
9.5.0
What about your claims? Show me an NV set up in the top 20 ?
ID: 848613 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 848615 - Posted: 3 Jan 2009, 10:35:41 UTC - in response to Message 848612.  


Again, you can say what ever you want, they can't beat the MACs, neither skulltrail

Is it really SO hard idea to comprehend ? GPU is co-processor. Combine this skulltrail with GPU - and you will beat this skulltrail alone.
Pretty simple, don't you think so ?


For the moment, your statement is wrong, you can repeat it many many times, it will not change the result, it does NOT increase the speed of a good processor.

Is it so hard to comprenhend, you are repeating again and again something that it is not true.


Yes, I repeating again and again what I measured by myself. And this measurement in accordance with other results.

Any data from you or just loud words?
Benchmarks ? What support your words ?
ID: 848615 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 848616 - Posted: 3 Jan 2009, 10:37:25 UTC - in response to Message 848613.  


really?
You did not see the code I posted here? you are blind?


No, I'm not blind and yes, I didn't see the code you posted here. Also I didn't see applications that was built based on this code too. Links please ?

ID: 848616 · Report as offensive
Profile Francois Piednoel
Avatar

Send message
Joined: 14 Jun 00
Posts: 898
Credit: 5,969,361
RAC: 0
United States
Message 848617 - Posted: 3 Jan 2009, 10:38:43 UTC - in response to Message 848616.  
Last modified: 3 Jan 2009, 10:40:44 UTC


really?
You did not see the code I posted here? you are blind?


No, I'm not blind and yes, I didn't see the code you posted here. Also I didn't see applications that was built based on this code too. Links please ?



HAHAHHHAHAHA http://setiathome.berkeley.edu/forum_thread.php?id=38030&nowrap=true#525611 ... see. code ... haaaaaaaaaaa

I did post code too, the one with the red tongue in the alien face :-P
I guess you did not pay attention one more time.
ID: 848617 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 . . . 11 · Next

Message boards : Number crunching : Random Musings About the Value of CPUs vs CUDA


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.