Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
![]() |
World Community Grid Forums
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
No member browsing this thread |
Thread Status: Active Total posts in this thread: 109
|
![]() |
Author |
|
roundup
Veteran Cruncher Switzerland Joined: Jul 25, 2006 Post Count: 838 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
... here are some examples of what i personaly consider less meaningfull projects than science projects --- ... and for me RC5-72 on DNETC is by far the most meanigless project in the BOINC world. DNETC is a wrapper between BOINC and distributed.net. Presently DNETC is working on RC5, which was launched years ago by RSA lab, which is a EMC² company. The subject of RC5 is to crack a certain code using the brute force method (which is the most stupid method for decryption). RSA withdrew from this project, so there is nobody left who is interested in the results of this DNETC project - but is still running and running and running .... |
||
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
I think DNETC give a very high points ratio for their pointless project to attract crunchers who are just interested in getting as many points as possible.
So much for Boinc vetting projects! |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The cross project comparison is only a lesson in futility. I have seen a computer with a fast quad core (it is a fast CPU) that is credited with 45X the daily credits that my AMD Phenom 3 core is earning. If that computer were on WCG it would be earning no more than 3X my credits.
You just cannot try to compare. My rank in WCG has improved by about 50 places over the last 5 months but has worsened by 1100 places in BOINC over the same time. |
||
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
If Boinc declared a point's scheme and said all projects must use it, or they cannot use Boinc, then such inequalities would not exist.
I think there may still be some points cheats about, but some projects do offer far too many points to begin with. These tend to be the worthless projects. GPU's should be separated from CPU stats; there should be separate CPU and GPU Boinc points tables. I would say most of the people that past you would be using GPUs; they get so much more credit. I have 2 systems that get 60K Boinc Credits per day; one has a GTX470 and the other has four GT240's. Just two GT240 cards could do more work than all the 50+ cores I have attached to Boinc projects. Even with GPUs there is uneven credit among projects. Some of this is due to the nature of the projects and cards, with ATI being better at Colatz and MW, and NVidia being so much better at GPUGrid that they did not even release their ATI app (though that might happen in the near future). MW requires double precision, and GPUGrid does not and uses CUDA. However the points awarded for Colatz and MW are very high. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
If only there are gpu apps here preferably for ati (I only have 1 PC with nvidia), I will gladly switch them over to WCG. Its been a while since gpu computing was made available to other projects but the scientists and admins here seems to have no interest or just cant find time developing gpu apps. Seriously, technology is constantly moving why are we still stuck with only 32 bit cpu apps. There are a lot of WCG crunchers with high end gpus that are only used doing solitaire or pinball. Even the new onboard gpus from nvidia and ati are way faster than any single core cpu.
----------------------------------------MilkyWay as mentioned above need double precision cards so It is only run on my 4850 and 3850, while the single precision cards, the 5670 and 4650, are running Collatz. Now I dont mind that I have been running WCG for almost 4 years now with only 897K BOINC credits and will continue to run WCG as long as I am able. I don't care even though WCG have a very low payout in terms of BOINC credits. My WCG credits is the cake while the credits from other projects is the icing. Still 2 days short for bronze badge in beta testing. :( I've tried the cpu app of Collatz and it is also very highly optimized but still take more than 16 hrs to finish a task on my athlon x4 620 (OC to 3.2ghz) on 32-bit. On the other hand, using 64-bit app shaves about 3-4 hrs off from the 32-bit time. Doing that on an ati 760-790 onboard gpu only takes about 5-6 hrs. (Note: cpu and gpu wus in Collatz and MW are the same thats why you can also see wus done by gpus validated against wus done by cpus or vice versa) But still, even within WCG projects, performance is still biased in favor of Intel machines. For comparison sake, my athlon x4 620 @ stock (2.8ghz quad) gets about 1100 something boinc rac while my celeron e3300 @ stock (2.5ghz dual) gets almost 900 rac. [Edit 2 times, last edit by Former Member at Jun 16, 2010 6:51:24 AM] |
||
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
What is your NVidia card?
Would be useful if Boinc or other projects told us in simple terms which CPUs were the best for which project. I think Intel are generally better, due to the instruction set, but there might be instances in which AMD are better. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
An 8500gt bought 3 yrs ago I think.
----------------------------------------Yes, Intel is better but I prefer AMD procs because of the onboard gpus of the AMD based motherboards. AMD IGPs perform on par sometimes even better than the cheapest/low-end nvidia discrete card. While using the IGP as primary display and use a discrete card for crunching, I can watch movies or play games when bored without problems. Can be done with nvidia IGPs too but I can only pair an nvidia discrete card with it (If paired with an ATI card, I get BSODs when the card is crunching and I do other things on the IGP, heck even GPU-Z triggers a BSOD). It would really be great when WCG have gpu apps, I can discard the power hungry cpus and get ion atoms and pair it with fast discrete cards. We can save a lot on the hardware and also save a lot on power consumption. And of course WCG can attract some of the crunchers with tens of PCs with 3-4 GPUs per system that only crunch for credit race, more science done in a short time. Distributed computing has come a long way. Before, people participate because they are interested on the project. Now, its become a hobby to most of rich people out there. Well, its better to spend to race with boinc credits than spending thousands of dollars on cars or motorcycles and get into accidents. :) [Edit 1 times, last edit by Former Member at Jun 17, 2010 8:03:41 AM] |
||
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
With only 16 stream processors your 3 year old 8500GT is not up to much, an entry level GPU. I dont think it could be used anywhere for crunching, possibly Folding@home, but possibly not fast enough there or anywhere else. Hang onto it just in case you have a good Motherboard without a GPU, and you just want to use the CPU.
You have some descent ATI GPUs, but at the minute there are only a few ATI projects that use Boinc, and none of them worth the effort/electric (IMO). It's only points grabbers that are interested in these PiET'r projects. ATI are good for double precision, so if WCG ever introduces such a project it would be attractive to many people. I only have one ATI card, in a box of junk somewhere. Its a 4850, but they are power hungry, loud and did not exactly fly through Folding (a non Boinc project, but a good one). I just get GPUs for GPUGrid, so I have to use NVidia, which tend to be more expensive and more power hungry, and I am not rich! It uses CUDA and does not need double precision, so the NVidia cards are a good match there. It's really down to the code, the type of calculations and the number and speed of the shaders, but ATI should be the default choice, and only change if you need CUDA. |
||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I stopped buying nvidia since they started renaming their cards and was disappointed at Fermi when it came out. Now that nvidia cannot manufacture boards for new Intel cpus, the future of nvidia is not good IMO.
Yeah, ATI support would sure attract lots of people considering there are only 3 projects offering them. MW (which I consider meaningful of the 3 projects) and Collatz ati apps are stable and fast while DNETC is a bit problematic especially for people with mixed gpu models. Right now, the only thing that will attract more people to WCG is to increase the BOINC credit per WCG credit. I've been looking for health science boinc projects that support gpus but with no luck. Lattice seems good but I think its cuda app is still in development because I didnt get any work when I attached to the project. I wish the WCG admins and the project scientists can collaborate with other projects that already have gpu apps. Cluster Physik (responsible for the ati gpu apps of MW and Collatz) seems nice and approachable but I think he's still busy looking for a post doc position and some projects at his university. |
||
|
sk..
Master Cruncher http://s17.rimg.info/ccb5d62bd3e856cc0d1df9b0ee2f7f6a.gif Joined: Mar 22, 2007 Post Count: 2324 Status: Offline Project Badges: ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Pity GPUGrid dont have huge financial clout. They could do with somone to help them add ATI support. Not sure he would fancy the move to Barca though?
|
||
|
|
![]() |