Posts by NJMHoffmann

1) Message boards : Number crunching : Problems with Rosetta version 5.41 (Message 32211)
Posted 7 Dec 2006 by NJMHoffmann
Post:
I had the app not responding error too, while looking at the graphics (no screensaver). Single core AMD XP 2200+ with Nvidia GeForce4 MX 440.

Suspending the task was not possible (i.e. the task showed as suspended but still used cpu). Aborting the graphics later aborted the WU (http://boinc.bakerlab.org/rosetta/workunit.php?wuid=44752875).

Norbert
2) Message boards : Number crunching : Archieve of newbie Q&A (Message 28219)
Posted 22 Sep 2006 by NJMHoffmann
Post:
HP_Owner: Your client told the server, that it needs at least 584 seconds of new work to fill its queue. So the server did send you one WU (and not more). The size of the next WU to be sent is not taken into account when sending one WU. It is only used to compute if additional WUs are needed.

Norbert
3) Message boards : Number crunching : How to get a moderator to delete your post (Message 27189)
Posted 17 Sep 2006 by NJMHoffmann
Post:
Can everyone agree that \'optomizers\' are not, and were not \'cheaters\'?

No.
4) Message boards : Number crunching : RAC dropping (Message 26431)
Posted 9 Sep 2006 by NJMHoffmann
Post:
This new credit system does\'nt reflect the actual work done either. We have a situation now where lower end machines are getting higher credits than they would otherwise have got and higher end machines are getting lower credits than they would otherwise have got. In other words, the higher end machines are dragging the lower end machines up and the lower end machines are dragging the higher end ones down.

Do you have some facts to back up this? It doesn\'t get true by repetition alone.
5) Message boards : Number crunching : Discussion of the new credit systen (2) (Message 26250)
Posted 7 Sep 2006 by NJMHoffmann
Post:
This new credit system seems fair to all participants in that it places everyone on the same playing field.

It does not. See David Kims comments/dialogue about Power Mac users.

It may not be fair, that nobody compiles a Rosetta build for Power Macs, that does use all of its power. But the credits given (same credits for same work done) are fair.

Norbert
6) Message boards : Number crunching : Discussion of the new credit systen (2) (Message 25469)
Posted 29 Aug 2006 by NJMHoffmann
Post:
An unoptimized benchmark is incapable of correctly measuring the potential of a CPU. IF you are going to measure a cpu, you need to measure its entire potential to be fair, not its castrated potential. This is at the BOINC level, as it it not in their control whether a project optimizes their code. What use is taking a measurement you know isn\'t fully accurate?

What use is taking a measurement you know is wrong? What credits are given for is not potential but work done. The new system is on the right track, not optimal but much better then the old system. Thinking loud: Perhaps we need a per-project-benchmark which per host uses those (and only those!) accelerators the science application will use? But no, I don\'t want a different amount of credits per hour given from the projects. Credits should not be a cause to choose a project.

Norbert
7) Message boards : Number crunching : Discussion of the new credit systen (2) (Message 25419)
Posted 29 Aug 2006 by NJMHoffmann
Post:

The old Crunch3r BOINC clients were fine, as are the Trux optimised clients. They do take into account the increase in throughput that can be attributed to extra instructions as found in SSE etc. But 5.5.0 just seems to make numbers up. They are physically impossible.


Physically impossible as throughput, as stated above, yet not as a comparison. If the numbers were made up, I doubt the rosy staff would have allowed its use.


I wouldn\'t call \"allowed\" what more was \"didn\'t care\". In the meantime things have changed and the Rosetta team had to take actions against so called \"optimizations\", that made credits almost worth- and senseless. What the projects display on their websites as \"TeraFLOPs\" as a measure for their throughput is nonsense now - because of clients that benchmark something, but not the used instructions.

Norbert
8) Message boards : Number crunching : Discussion of the new credit systen (2) (Message 25249)
Posted 28 Aug 2006 by NJMHoffmann
Post:
mmciastro:
Try being a dial up user. I was, the best you can do is set to display \"most recent post first\" and then hit the stop button on the browser to avoid waiting 3 minutes for all the old (already read) posts to load.
There is the \"if a thread contains more than ... display only the first and the last ... postings\" setting.
[edit] Some projects, with a very old forum code, still don\'t have it.[/edit]
9) Message boards : Number crunching : Discussion of the new credit systen (2) (Message 25244)
Posted 28 Aug 2006 by NJMHoffmann
Post:
This thread is a continuation of the dicussion of the new credit system. All prior posts can be found here.

@Mod.DE (OT):Why do you think, you must split a thread because of \"modem users\"? They will have their display settings in the \"message board preferences\" set to fit their needs.
10) Message boards : Number crunching : Discussion of the new credit system (Message 25236)
Posted 28 Aug 2006 by NJMHoffmann
Post:
The only usefull info would be in actual work done, or FLOPS returned. Unless the FLOPS were measured from the benchmarks.. then they mean as much as the credits.
They are. So only the project can tell, if there really is a decrease in \"science\" returned. Everything else is wild speculation.
11) Message boards : Number crunching : Discussion of the new credit system (Message 25226)
Posted 28 Aug 2006 by NJMHoffmann
Post:
Well, there it is afterall. A drop of about 13% of credits so far.

What was expected. So what? Losing inflated credit does not mean losing anything worthful.

Norbert
PS: And if somebody now tells me, that the the FLOPs for Rosetta too dropped by 13%, he doesn\'t understand anything.
12) Message boards : Number crunching : New Forum Software (Message 25225)
Posted 28 Aug 2006 by NJMHoffmann
Post:
SIMAP is a boinc project, it uses phpBB

- and lacks integration of accounts,
- will never have \"forum preferences\" set centraly (ok. - Boinc now doesn\'t replicate them, but you could wish),
- fights with spammers (because not being able to grant post permission only to active crunchers)
There\'s always two sides.

Norbert
13) Message boards : Number crunching : Points/Credits on DC in General (Message 24761)
Posted 24 Aug 2006 by NJMHoffmann
Post:
Saenger & Trog Dog: Thanks for the correction. My memory slipped.
Norbert
14) Message boards : Number crunching : Points/Credits on DC in General (Message 24729)
Posted 24 Aug 2006 by NJMHoffmann
Post:
I`m at WCG which like most operates a quorum (how many I know not). My points are all over the place, here maybe a reason.
If WCG hasn\'t changed latetely, they use Boinc only to distribute work. Their credits are calculaded by points their own SW would have given times a fixed factor.

Norbert
PS: Their factor used to be a bit to the high side (a share of 75 for WGC and 100 for other projects gave equal RAC for me) but seems to be adjusted lately.
15) Message boards : Number crunching : Multiple project processing... (Message 24426)
Posted 23 Aug 2006 by NJMHoffmann
Post:
1) There is no attempt made to wait for a task to finish other than waiting for the next checkpoint. So, if things are running normally, and there is no checkpoint between the re-schedule and the end of the task, it does wait.
At Rosetta this leads to WUs stopping at 100% and waiting for the next timeslice for Rosetta. This is not the fault of the scheduler but a consequence of the Rosetta code, that seems at the end to:
- tell Boinc \"I\'m at 100%\"
- write a checkpoint (why?)
- want to clean up and exit, but is replaced by another app before this.
At the next start it loads the checkpoint, cleans up and exits :-)

Norbert
16) Message boards : Number crunching : Report Problems with Rosetta Version 5.25 (Message 24151)
Posted 21 Aug 2006 by NJMHoffmann
Post:
So I think the labelling should be changed, as it\'s also possible that a result is really invalid, for example when the hardware is faulty and delivers no useful results.
It would be difficult to decide: Is the result invalid, because the computer failed? Or is the result invalid, because the used \"routines / parameter combination\" doesn\'t work? The second is a very useful result for Rosetta.

Norbert
17) Message boards : Number crunching : Report Problems with Rosetta Version 5.25 (Message 24146)
Posted 21 Aug 2006 by NJMHoffmann
Post:
Now that ralph is here, the incidents of failed wus is very low, and I hope they stop that practice.
No. (see my answer to Saenger for my argument).

Norbert
18) Message boards : Number crunching : Report Problems with Rosetta Version 5.25 (Message 24142)
Posted 21 Aug 2006 by NJMHoffmann
Post:
Do I understand this right: In the old system you even got credit for invalid results? Why should this be?

Because here at Rosetta@home the software is tested. Part of the data, sent to you with a new WU, is code to test. So bugs in the software should not effect credit.

Norbert

PS: It\'s not new. IIRC Seti does this for years, when aborted WUs get credit. At Seti it\'s corrupt data (or useless data), that causes the aborts (Error 9??).
19) Message boards : Number crunching : Report Problems with Rosetta Version 5.25 (Message 22576)
Posted 16 Aug 2006 by NJMHoffmann
Post:
Feet1st: I have BOINC set to run 24/7 (which it does, my PC is up 24/7), the unit is set to be removed from memory when it switches to another project. Interval is 1 hour, then it moves to the next project and so on.

Rosetta still :-( has WUs with checkpoints more then 1 hour apart. So you should
- leave in memory or
- use a bigger interval or
- use a boinc version, that waits for a checkpoint before switching (e.g. 5.5.13)

Norbert
20) Message boards : Number crunching : Report Problems with Rosetta Version 5.25 (Message 22533)
Posted 16 Aug 2006 by NJMHoffmann
Post:
I\'m having the same problems with Rosetta@Home \"hanging\" (it shows \"running\" but the CPU is at 0%).

You don\'t happen to have Boinc alpha 5.5.10 running?

Norbert


Next 20



©2020 University of Washington
http://www.bakerlab.org