Message boards :
Number crunching :
AP v7 Credit Impact
Message board moderation
Previous · 1 · 2
Author | Message |
---|---|
Lionel Send message Joined: 25 Mar 00 Posts: 680 Credit: 563,640,304 RAC: 597 |
There is a down side to this that I have not mention in the above, but have spoken of in PMs. I think it is worth mentioning. HAL9000 said "It has been mentioned numerous times now of how the credit did the same thing at Beta. In fact it started out much lower at Beta than it here here on Main. Over time it ramped up to a average near the same as AP v6" {and please don't think I am picking on HAL9000 because I mentioned him first, I'm not}. Richard H. said: "If the blanked tasks take less time than using v6 (as they will), then you would expect less credit per task, but more tasks per day, resulting in the same RAC. If the credit system worked as documented, that is. ;)". TBar said: "I think part of the problem on main is the absurdly High Time Estimates. On my machine the stock estimate is 14 hours for an ATI task. Actual Time is 33 minutes, which it is now estimating since achieving 11 completions. From experience with 2 dissimilar GPUs, the GPU finishing under the estimate receives a Lower 'score'. 33 minutes is Much Lower than 14 hours, hence a Lower 'score'. Unfortunately it took around 50 validations to receive 11 'Completions' so now all the Bad history must be overcome. Unlike SETIv7, APv7 is producing a Higher APR than APv6 and SHOULD produce HIGHER Scores than APv6. That assumes CreditFew actually follows basic math principles. It would have been Nice if it wouldn't have recorded 50 valid tasks at an Estimate of 14 Hours...". The above are some of the comments and inferences that imply that APv7 credits will rise to match or be greater than APv6 credits. If this is truly the case, then the disparity between APv7 and MBv7 should be greater than the disparity between APv6 and MBv7. This will stimulate a higher level of aberrant behaviour. If this is not the case, then aberrant behaviour may be less, but dissatisfaction will likely increase. I think it is time to start at the beginning and go back to the first premise. Should a WU that is crunched by any processor using the standard application get the same credit regardless of processor? That is the first question that needs to be answered. All else flows from there. Lastly, from an outsiders perspective looking in, for example, you don't ask a machine to do more work (for example auto correlation) which then takes that machine more time to complete (a work unit) and then penalize that machine for doing the extra work that you asked it to do. The credit system appears to be structurally flawed and needs a fresh new look at. Those in the "privy council" need to act. cheers |
Bernie Vine Send message Joined: 26 May 99 Posts: 9954 Credit: 103,452,613 RAC: 328 |
The credit system appears to be structurally flawed and needs a fresh new look at. I was under the impression that Jason and others were doing exactly that. |
Lionel Send message Joined: 25 Mar 00 Posts: 680 Credit: 563,640,304 RAC: 597 |
I am not being sensitive here to the comment above, it has been almost 18 months since the flaws in the credit system were exposed with MBv7. Nothing has been truly done to correct the philosophical structure of the code. Bernie, I would ask you this: If you think this is a reasonable time to get where they are, then what time do you think would be reasonable to finish the job and implement a working solution ?? I don't see any movement forward, nor do I see discussion towards fixing the issue. In the past I have seen comments from others about Dr A. not believing that there is a problem ... maybe he is the blocker that needs to be removed so that this thing can go forward and rectified. cheers |
qbit Send message Joined: 19 Sep 04 Posts: 630 Credit: 6,868,528 RAC: 0 |
Couldn't they use the data from Beta to get credit right here on main from the start? |
Donald L. Johnson Send message Joined: 5 Aug 02 Posts: 8240 Credit: 14,654,533 RAC: 20 |
Just finished an AP v6 task where all three of us validated, but only got 260 credits, when just a few weeks ago they were yielding 650-750. That is about what many AP v7 tasks are paying. I don't really care about the credit, but it just looked wierd. Donald Infernal Optimist / Submariner, retired |
James Sotherden Send message Joined: 16 May 99 Posts: 10436 Credit: 110,373,059 RAC: 54 |
Its called parity. I remenber when nobody wanted to run an AP work unit. Then when MBv7 came out and RAC dropped everyone said, Screw MB I want AP. Well initial reports say APv7 pays less then MBv7. If that turns out to be true guess what will happen? Dont get on lionel's case because he reports what he is seeing. And yes I know that blanking and angle range has a lot to do with credit granted, But so does DR. A. And NO, Im not leaving the project over stinking credit. [/quote] Old James |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
The credit system appears to be structurally flawed and needs a fresh new look at. A group of us were recently , tongue firmly in cheek, referred to as the 'CreditNew Vigilantes', and it's that group that's variously assessing code, system behaviour, and refreshing on control systems engineering practices. My own contributions cover the last part of the system, headed toward full scale modelling of the control systems as should have been done before (but clearly weren't, you find as you get deeper) As things stand, the overall approaches and extensive study of the existing mechanism have been discussed and poked at in various ways, and we get closer to wholesale patching and systemic recommendations, though there are still quite a few finer points to be looked at in more detail as a refined model is constructed. Not least of those is comparing the complexity of maintaining/patching the existing hodgepodge over wholesale reengineering. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
Julie Send message Joined: 28 Oct 09 Posts: 34053 Credit: 18,883,157 RAC: 18 |
|
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13732 Credit: 208,696,464 RAC: 304 |
... Not least of those is comparing the complexity of maintaining/patching the existing hodgepodge over wholesale reengineering. As painful as constructing something from scratch is, often it is the best option to take as trying to patch something that is badly broken often only leads to further problems. Cosmetic problems are easily repaired. Some structural problems can be repaired, with considerable effort. However when the core structure is significantly damaged, demolition & reconstruction is the best option. Grant Darwin NT |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
... Not least of those is comparing the complexity of maintaining/patching the existing hodgepodge over wholesale reengineering. Yep, pretty much sums up the point we're at in many ways. There's a hiatus while I refamiliarise with some modern tools (such as Matlab and estimate localisation techniques), and I know Richard's been busily gathering data and symtpoms for some time. That will yield an idealised model for direct comparison against the existing core, giving us at least some scope for how much needs replacement, along with pros & cons. Frustratingly what isn't feasible at the moment is any kindof timeline, though there have been benefits to extended observation & research time nonetheless. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
There are many possible points on view. Depends of what particular human values more. For example: No, they should not pay the same. Older CPU takes much more energy to complete that task. So credits should punish it as being ineffective and pay more credits if task completed on power-efficient GPU. I think such approach would be very popular between "Green party" members ;) |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
Couldn't they use the data from Beta to get credit right here on main from the start? Unfortunately, current BOINC server code state and quality did not allowed this even between version changes on beta itself, no speaking about reliable transfer between projects. |
©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.