Message boards : News : [dna] credit totals fixed
Author | Message |
---|---|
I've fixed the issue with credit totals. In one place credit was getting multiplied however with the new multi-app code it wasn't getting multiplied where actually needed. This resulted in the credit totals going down a bit for the DNA@Home credit, so I apologize for this and any confusion it might have caused. On another note, now that the DNA@Home credit is actually being displayed accurately everywhere -- what are your thoughts on the amount of credit being awarded (always a great way to stir up a can of worms)? If it is too low I'll look back into adding in a multiplier, or trying to get things working with fixed credit. | |
ID: 5350 · Rating: 0 · rate:
![]() ![]() ![]() | |
I think the general consensus is to avoid Credit New at all costs. It really is a terrible system for scoring. If you could get DNA a consistent run time, then I would say go with fixed credit. But right now there appears to be different duration needs. | |
ID: 5354 · Rating: 0 · rate:
![]() ![]() ![]() | |
Travis, there is a CSG challenge going on right now and a couple of hours ago I lost 150,000 credits in that challenge. I'm assuming it was your fix. I'm not really concerned about losing the challenge credits, but I do note that my credits didn't change at the CSG site. Had they gone down a like amount I'd have no credit for the work I've done since I joined. Something doesn't seem quite right yet. | |
ID: 5355 · Rating: 0 · rate:
![]() ![]() ![]() | |
Mine dropped 500K as I was over 1.2 Million. However, my ~700k is more accurate to where it should have been. My scores updated here accordingly. Just as others had mentioned, the sub project scores here did not reflect the DNA point inflation. So, perhaps that is where the confusion is? | |
ID: 5356 · Rating: 0 · rate:
![]() ![]() ![]() | |
Going to credit new is a negative step for the project. Many will be discouraged by this. DNA credit now appears to be too low compared to subset and other projects that encourage participation by a more positive attitude towards credit offered. | |
ID: 5357 · Rating: 0 · rate:
![]() ![]() ![]() | |
What's going on?? I just started with BOINC. My 9500pts for CSG just dropped to 2500. That's crazy. | |
ID: 5359 · Rating: 0 · rate:
![]() ![]() ![]() | |
What's going on?? I just started with BOINC. My 9500pts for CSG just dropped to 2500. That's crazy. Only a few moments of your time reading through this forum will answer most questions. For Example: read this thread> http://csgrid.org/csg/forum_thread.php?id=2069#5360 Or perhaps also read the one you just posted to. The short answer is: This project just moved servers and, as always, there are small things that surface that need to be found and fixed. A problem with the 3 project total being too large because of a mistake in DNA program credit was corrected. If you don't have time to read, That's crazy. ____________ ![]() | |
ID: 5361 · Rating: 0 · rate:
![]() ![]() ![]() | |
Going to credit new is a negative step for the project. Many will be discouraged by this. DNA credit now appears to be too low compared to subset and other projects that encourage participation by a more positive attitude towards credit offered. Could not agree more with those points. Maybe there could be a separate formula to use for DNA (avoiding "credit new"). Since there are different sizes of WUs you can't have X credits for all. How about using: Peak working set size times .25 = DNA credits? eg. 880 x .25 = 220 credits (same WU is about 88 credits now (10%) and takes my cpu about 2 hours 45 minutes each. Maybe that is too high but using the work file size seems fair, do-able and it could be used with a variety of WU sizes and still account for the processor power we are donating. I think a bump in the credits structure would equal a bump in future participation. | |
ID: 5362 · Rating: 0 · rate:
![]() ![]() ![]() | |
Going to credit new is a negative step for the project. Many will be discouraged by this. DNA credit now appears to be too low compared to subset and other projects that encourage participation by a more positive attitude towards credit offered. It has been credit new the whole time, it's just that for awhile there was a multiplier being applied to the DNA@Home credit that was being added to the combined total (however the credit for the DNA@Home total was not being multiplied). The amount of time spent for a WU is pretty much deterministic on it's parameters however, so I can do fixed credit for the WUs. It just won't get done until I'm back from Reykjavik (so probably this weekend). | |
ID: 5367 · Rating: 0 · rate:
![]() ![]() ![]() | |
Compared to other projects I run, credit is too high. | |
ID: 5370 · Rating: 0 · rate:
![]() ![]() ![]() | |
Compared to other projects I run, credit is too high. You might be the first person I've found saying credit new is too high. :) On another note, right now there are basically WUs with a few different runtimes for DNA@Home. If anyone would want to offer some suggestions as to credit for the different runtimes I should be able to plug things into the FPOPS calculation to come up with a fixed credit value. From longest running to shortest running: largest size -- for these, snail should run longer than slug: snail_hg19_1000fa_3motifs_6width_norepeats_1 slug_hg19_1000fa_3motifs_6width_norepeats_1 snail_hg19_1000fa_2motifs_6width_norepeats_1 slug_hg19_1000fa_2motifs_6width_norepeats_1 snail_hg19_1000fa_1motifs_6width_norepeats_1 slug_hg19_1000fa_1motifs_6width_norepeats_1 middle size -- snail and slug should run almost identically: slug/snail_hg19_100fa_3motifs_6width_norepeats_1 slug/snail_hg19_100fa_2motifs_6width_norepeats_1 slug/snail_hg19_100fa_1motifs_6width_norepeats_1 small size -- snail and slug should run almost identically: slug/snail_hg19_10fa_3motifs_6width_norepeats_1 slug/snail_hg19_10fa_2motifs_6width_norepeats_1 slug/snail_hg19_10fa_1motifs_6width_norepeats_1 Basically, the larger the value in front of fa, the longer the runtime, likewise the more motifs, the longer the runtime. | |
ID: 5371 · Rating: 0 · rate:
![]() ![]() ![]() | |
Going to credit new is a negative step for the project. Many will be discouraged by this. DNA credit now appears to be too low compared to subset and other projects that encourage participation by a more positive attitude towards credit offered. It was credit new with a multiplier being applied :) Given the memory requirements for DNA running it on multicore systems is not just a connect and forget application unless your systems have a lot of memory. On 8 core systems, less than 16 GB available and you will start to run into system overload conditions. Simply applying credit new with no extra reward doesn't address this issue alone. A fixed credit system would be better with an appropriate reward for the extra work required to run DNA. The next best solution would be to continue to add a multiplier to the base credit new points system. | |
ID: 5376 · Rating: 0 · rate:
![]() ![]() ![]() | |
Given the memory requirements for DNA running it on multicore systems is not just a connect and forget application unless your systems have a lot of memory. On 8 core systems, less than 16 GB available and you will start to run into system overload conditions. I definitely agree. I want to move to fixed credit as it gives me a bit more flexibility, and I do think there should be a bonus due to the high memory requirements and the fact that the project is still fairly alpha and has things to debug. Hopefully I'll get some feedback on credit for the different WU sizes and we can move over to that soon. | |
ID: 5377 · Rating: 0 · rate:
![]() ![]() ![]() | |
I guess the best way to determine the initial credit would be to look at the average run times for the different apps based on an average system and then propose the credit based on that. | |
ID: 5378 · Rating: 0 · rate:
![]() ![]() ![]() | |
No doubt :P but there was around a 3x increase in my daily credit when CSG had work available, so clearly it was too much generous compared to my other projects :P | |
ID: 5390 · Rating: 0 · rate:
![]() ![]() ![]() | |
Fixed credit has to be the way to go, equated to the amount of value to the project. Whether that be determined by file size or results. That means a true level playing field as faster computers produce more work and rightly earn more credit. | |
ID: 5393 · Rating: 0 · rate:
![]() ![]() ![]() | |
You need to be careful with Credit New system it uses a faulty benchmark system to determine the base credit. For instance in Numberfields the AMD rigs I have running (a 64 core 6276 running at 2900Mhz and a 48 core 63xx rig running at 3800Mhz) are faster than the 64 core 4650 rigs I have running at 3140Mhz. On an average the AMD rigs complete the WU's quite a bit faster but receive substantially less credit per hr of run time than the Intel rigs do. It was confirmed by Eric here at this link. http://numberfields.asu.edu/NumberFields/forum_thread.php?id=209#1097 | |
ID: 5396 · Rating: 0 · rate:
![]() ![]() ![]() | |
Message boards : News : [dna] credit totals fixed