21)
Message boards :
Number crunching :
Response to concerns regarding the new credit system.
(Message 311638)
Posted 20 May 2006 by Xaak Post:
Actually, if you're in the quorum (first 3), it pretty much guarantees you'll get more credit. Here's how. Without trux that box was claiming 7, which was almost always the lowest, so the next lowest, usually around 15 - 20 became the grant. After trux, that same claim was 32 or so, making it one of the highest. So now that 15-20 became the lowest, mine became the middle or highest, giving me anywhere from the same or a little more credit granted to 32 granted. Average increase was around 8 credits per wu. True, there's no guarantee, but the data I collected at the time I swiched proved that increase, and the increase was soley due to the change Trux made in claimed credit. |
22)
Message boards :
Number crunching :
Response to concerns regarding the new credit system.
(Message 311597)
Posted 20 May 2006 by Xaak Post: @Xaak I forgot to mention, both were running Crunch3r's SSE3 optimized seti client for windows. That's where some of the increased numbers came from. Trux added an average of about 8 credits per wu when I was part of the granting quorum vs. no trux. With an average of 85 wu/day from the 630, that worked out to approximately 680 additional credits granted just from the trux calibrations. |
23)
Message boards :
Number crunching :
Seti Enhanced Credit Fair?
(Message 310497)
Posted 19 May 2006 by Xaak Post:
People have free choice as to why they crunch, when they cruch, and if they crunch. I don't advocate anyone being excluded from the project, and I don't see where anyone is being excluded. However, if someone is unhappy with the way credits are granted, it's their choice if they no longer wish to support a project. That's each individual's right, to do with their computer's resources as they see fit. However, it's my right to have the opinion that it's rediculous for a person to support a project soley because it grants credit in a manner that person approves of. |
24)
Message boards :
Number crunching :
Response to concerns regarding the new credit system.
(Message 310475)
Posted 19 May 2006 by Xaak Post:
There's no need to factor in benchmarks in unless the speed of a machine is a factor in how many credits per wu are granted. The Enhanced system grants fixed credit per wu, theoretically at least, and benchmarks play no part in that calculation. |
25)
Message boards :
Number crunching :
Response to concerns regarding the new credit system.
(Message 310469)
Posted 19 May 2006 by Xaak Post:
Wow, I never realized the entire seti project exists just to allow Seti.usa to pass other teams faster. And here I thought I was crunching for a science project ;-). |
26)
Message boards :
Number crunching :
Response to concerns regarding the new credit system.
(Message 310415)
Posted 19 May 2006 by Xaak Post: @Zaak Sorry, I misread your initial post. Slowest boxes: Pentium 630 running at 3.825ghz with an Asus P5ld2-VM board (Intel 945p chipset) 1 gb DDR2-667 (pc2-5300) Regular Seti wus average under 35 minutes 2 at a time Pentium 530 running at 3.3 ghz with a gigabyte 945p chipset motherboard 2 gb DDR2-533 (pc2-4300) @ DDR2-586 speed. Regular seti wus average around 40 minutes 2 at a time Both of those were previously running ATI chipset board with standard DDR (Asus P5-RD* series boards) and crunch times for both were over an hour/wu 2 at a time. Someone previously stated that DDR2 has much higher bandwidth, which is true, and seti appears to take full advantage of that. Plus, the 945/955 chipset also takes advantage of being able to run memory async very well too. |
27)
Message boards :
Number crunching :
Response to concerns regarding the new credit system.
(Message 310358)
Posted 19 May 2006 by Xaak Post:
Agreed, but the entire intended credit system for Seti (before enhanced) wasn't meant to have faster machines claim the same credit per wu as slower machines, it was meant to grant credit more on a time spent crunching approach. Trux attempted to change that to an equal per-wu approach, and therefore did circumvent the intent of the cobblestone system set up by Berkeley for the standard seti app. If you did crunch completely non optimized, a fast computer did claim less credit per wu than a slow computer. If you optimized (leaving trux out of the equation), credit per wu dropped drastically, but not credit per hour. You crunched more per hour so got less credits per wu. With enhanced, they're now granting credit on a per-wu basis. Everyone crunching gets the same credit/wu regardless of machine speed, however, faster machines crunch more in a given time so credits/hour is higher than slower machines. But they based that credit per wu on what an unoptimized setup would generate, which is the lowest common denominator. Keep in mind, even if 100,000 users optimized and used trux, they're still in a pretty small minority, and I really doubt anywhere near that number used both the optimized clients and trux. My guess would be well below 10,000 did, making that a statitstically insignificant group to base anything on. But the graphs seem to tell the story much better than anything else. Trux users are experiencing a big drop in RAC, myself included, while the overall project numbers staying relatively unchanged so far. And since unoptimized systems automatically download the Enhanced client, we know that very large numbers of users are already crunching Enhanced wus. |
28)
Message boards :
Number crunching :
Response to concerns regarding the new credit system.
(Message 310343)
Posted 19 May 2006 by Xaak Post:
Of course I did and we did. Did you bother to read http://setiathome.berkeley.edu/forum_thread.php?id=30918#309580 ? The point is that Berkeley isn't unlikely to distribute something that alters statitstics, even if they allow it's use, and using just Crunch3rs optimized Boinc Core client resulted in low claims, althouth not as low as the non-optimized core client. |
29)
Message boards :
Number crunching :
Response to concerns regarding the new credit system.
(Message 310023)
Posted 19 May 2006 by Xaak Post:
Actually it would. Running the optimized client and the regular optimized boinc core client (not trux) resulted in my pentium D machines claiming around 5 credits per wu. Trux changed that equation, and only by reporting inaccurate benchmarks and crunch times. |
30)
Message boards :
Number crunching :
Response to concerns regarding the new credit system.
(Message 309637)
Posted 18 May 2006 by Xaak Post: @Zaak The three top machines I'm running are (2) dual core Pentium D (Pressler core) 920s, running at or over 4ghz on air cooling and 1 930, liquid cooled, running at 4.06ghz. All 3 user the Asus p5wd2 series motherboards (Intel 955x chipset) The memory is running at the equivilent of DDR2-850 or higher on all machines. All use (used) Crunch3r's optimized SSE3 client v 4.11 and the Trux calibrating client, and crunch the old 4.18 wus between 18 and 19.5 minutes each for mid-ar wus. What I've found is that non-intel chipsets are significantly slower in crunching than the 945x and 955x chipsets. Socket 478 cpus are significantly slower than socket 775 counterparts, especially when using said intel chipsets and DDR2 memory with the socket 775s. For example, I had an 630 (3ghz) running in an ATI based chipset board with DDR-400 memory that took over an hour optimized to crunch a wu. Upgrding to a 945 chipset and DDR2-533, and pushing the overclocking a bit more took the times that processor ran to below 35 minutes/wu. |
31)
Message boards :
Number crunching :
Response to concerns regarding the new credit system.
(Message 309580)
Posted 18 May 2006 by Xaak Post: The whole problem here was really due to the Trux calibrating client. As I understand it, previously, credit was granted more based on time spent cruncing rather than how many wus you did per day. The thing that trux did was artificially inflate the credits claimed so that every wu crunched, no matter how fast or slow, was normalized to credit based on 32.x cobblestones for the standard unit. This allowed fast computers to claim much more credit per hour than non trux or non-optimized computers, and greatly inflated the claimed credit, which in turn raised the granted credit. When I first heard about the Trux Calibrating Client, I though it was basically cheating . However, since it was openly discussed on the boards here, appeard to be in widespread use, was available equally to anyone who chose to use it and Berkeley was silent on the issue, I took that silence as tacit approval so after a while I decided to use Trux myself and recommend it to my teammates. Now, everyone gets the same credit for wus. Faster machines complete wus faster, and with optimized clients even moreso. Everyone is on a completely even playing field, not only within seti@home, but among other projects running on boinc. The credit thing to me is a complete non-issue. I've spent money on fast machines too, and had 3 in the top 20 computers list with a 4th rising, and the opinions of some seti crunchers here just looking for continued inflated rac do not represent me. I'm happy with the new credit system, and it's as fair as it can be made all around. I applaud the project developers for finding a way to put everyone on a fair playing field. Xaak Proud member of BroadbandReports.com Team Starfire |
32)
Message boards :
Number crunching :
Seti Enhanced Credit Fair?
(Message 308482)
Posted 17 May 2006 by Xaak Post: First off, I think the whole issue of credits, and those complaining about them, is pretty funny. I like getting credit, don't get me wrong, as it's both a recognition of the work I've done for the project, and fosters competition, which is fun in most cases. However, credits don't get you anything. So, you're getting less credits than you used to. Big deal, so is everyone else. If the project management suddenly doubled the amount of credit they're giving out, it would have absolutely no meaning whatsoever, because everyone's credits would double at the same time. My rac was over 24,000, and now it's dropping probably to the 10k range. I don't care, as everyone else's will do the same. My fast computers, once things settle down, will still be in the top 20, I'll still be gaining on the same people as I was. What I can't understand even more is people and teams that aren't switching to enhanced until they absolutely have to simply because their credits will drop. Xaak Proud member of BroadbandReports.com Team Starfire |
©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.