The computer spec is immaterial. The 1% never moves.

The WU finished in 36 minutes, so when I posted earlier, it was 2/3 done, and should have shown 66% complete, and 12 minutes time to completion.

The displayed status in BONIC Manager is totally bogus. Broken. Completely worthless, better to not display anything.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)

You sure do whine alot.

Like, you don't just point out problems that need to be fixed, like everyone else does.

You seem to actually CRY while typing, like bugs in the system are a personal insult to you.

You sure do whine alot.

Like, you don't just point out problems that need to be fixed, like everyone else does.

You seem to actually CRY while typing, like bugs in the system are a personal insult to you.

I *do* get offended when obvious deficiencies are pointed out, over and over, and nothing is done about them.

There are 3 errors here -

* 1. Progress % done indicator - has no relation to reality - stuck at 1 %

* 2. Time to Completion - again, totally non-realistic for two reasons:
A. Counts up instead of down - DOH!
B. Has no relation to crunch time preference set at project - there is no way a 1 hour crunch time preference should EVER display 3:45 time to completion.

* 3. Estimated time for WU - set the same for every WU downloaded. When I attached to the project tonight, I got 50 WUs downloaded, of at least 5 or 6 different types (based on WU names). The first completed in 36 minutes. The next took 3 hours. Leaving aside why it spent 3 hours on a one hour crunch preference, both these first two WUs had the same Estimated Time to Completion, as did the other 48 WUs. This tells me it's obvious that these times are NOT being set realistically by the project.


Edit: 3rd WU finished in almost exactly 1 hour - admirable. However, it stayed at 1% the entire time. I watched steps counting up, to somewhere around 365,000 or so. Seems like it would be easy to figure out early on in the WU how many steps per minute (roughly 10,000 in this case) it was doing, and adjust the % done quickly, then decide how many models would get done in the user's set crunch time preference. That would allow the time remaining to be displayed accurately as well.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)

Seems like it would be easy to figure out early on in the WU how many steps per minute (roughly 10,000 in this case) it was doing, and adjust the % done quickly, then decide how many models would get done in the user's set crunch time preference.

...now you are assuming that the steps all take the same time, and that there is a fixed number of steps per model. If you sit and watch a DOC WU, you can clearly see that time per step varies too. And if you review some of the posts about how Rosetta works, you will learn it plans it way through each model as it goes and so the number of steps is not predetermined.

...but it would seem to get you to a much closer estimate of % completed.

---

Rosetta Moderator: Mod.Sense

You sure do whine alot.

Like, you don't just point out problems that need to be fixed, like everyone else does.

You seem to actually CRY while typing, like bugs in the system are a personal insult to you.

Well as we can see here, the IT people would rather scrape with a person than take on the job of fixing the problem. Must all be University type IT people. This is the way IT people come to the work place. Then in business we have to retrain them so they have just a little customer service in them.

Oh, by the way my jobs are doing the same thing. After running for 4 - 5 hours for a job estimated to be 2:08:56, I just cancell them. O yes, World Community Grid was having the same problem yesterday. They just stopped sending out jobs until the problem could be determined. HMMMM what a solution. No fighting with the people providing the computing power.

Seems like it would be easy to figure out early on in the WU how many steps per minute (roughly 10,000 in this case) it was doing, and adjust the % done quickly, then decide how many models would get done in the user's set crunch time preference.

...now you are assuming that the steps all take the same time, and that there is a fixed number of steps per model. If you sit and watch a DOC WU, you can clearly see that time per step varies too. And if you review some of the posts about how Rosetta works, you will learn it plans it way through each model as it goes and so the number of steps is not predetermined.

...but it would seem to get you to a much closer estimate of % completed.

I believe I've read a couple of places that each WU type has roughly the same number of steps, some as small as 30,000 and some up to 1,5000,000. Are these not being run on "Ralph" before being sent out on the public project? At that time the approximate number of steps per model should be available, and used to determine the % complete, and from there the Time Remaining.
Time to Completion obviously needs to be adjusted and set at DL time based on the client Run Time preference. Using the number of steps, and the run time preference, the number of models that will be crunched can be determined. The fine tuning by DCF will make the necessary adjustment for computer speed.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)

You sure do whine alot.

Like, you don't just point out problems that need to be fixed, like everyone else does.

You seem to actually CRY while typing, like bugs in the system are a personal insult to you.

Well as we can see here, the IT people would rather scrape with a person than take on the job of fixing the problem. Must all be University type IT people. This is the way IT people come to the work place. Then in business we have to retrain them so they have just a little customer service in them.

Oh, by the way my jobs are doing the same thing. After running for 4 - 5 hours for a job estimated to be 2:08:56, I just cancell them. O yes, World Community Grid was having the same problem yesterday. They just stopped sending out jobs until the problem could be determined. HMMMM what a solution. No fighting with the people providing the computing power.

Yes, in the real world, we don't tolerate that. I don't expect it from my developers, and I don't accept it here.

It makes no difference if the science code works magnificently, if the user interface is junk. In the real world, the users just start using something else, and here they vote with their feet. Plenty of other projects that don't have these problems.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)

Well as we can see here, the IT people would rather scrape with a person than take on the job of fixing the problem.

I don't work for the project, genius.

Time to Completion obviously needs to be adjusted and set at DL time based on the client Run Time preference.

Now you are missing the fact that the runtime preference can be changed AFTER the download occurs.

I point this out not to "scrape with" you. I'm simply pointing out that this issue is not as trivial as you paint it to be.

Please take some time, define how the user interface SHOULD look when a reasonable human estimate is 3 hrs for 1 model, and your preferred run time is 1hr. And what should be done when your preference is 8hrs, you crunch for 5hrs and then change your preference to 3hrs. Then post your ideas to this thread, as the entire topic is not specific to 5.54 and that is the topic of this thread.

---

Rosetta Moderator: Mod.Sense

Time to Completion obviously needs to be adjusted and set at DL time based on the client Run Time preference.

Now you are missing the fact that the runtime preference can be changed AFTER the download occurs.

Then the Time to Completion should be adjusted in BOINC Manager when the user changes it after download (How does this happen? It's not an option in MY version of the Manager.)

I point this out not to "scrape with" you. I'm simply pointing out that this issue is not as trivial as you paint it to be.

Well you're neither the "IT People" nor am I the one who used "scrape with", so how is this pertinent?

Please take some time, define how the user interface SHOULD look when a reasonable human estimate is 3 hrs for 1 model, and your preferred run time is 1hr. And what should be done when your preference is 8hrs, you crunch for 5hrs and then change your preference to 3hrs. Then post your ideas to this thread, as the entire topic is not specific to 5.54 and that is the topic of this thread.

If the real runtime estimate (as tested on Ralph) is 3 hours for one model, and the user has a Run Time Preference of 1 hour, then that WU should not be downloaded to that client, otherwise the user will see a WU go significantly beyond the desired run time preference.

In your second case, the WU should end as soon as the current model finishes.

It does no good to design this type of model/run time preference scenario into the client and then not support it in the user interface.

---

Nemesis n. A righteous infliction of retribution manifested by an appropriate agent.

Well as we can see here, the IT people would rather scrape with a person than take on the job of fixing the problem.

I don't work for the project, genius.

No one would imagine that you did work for the project.

If you're not the "IT people", then he obviously wasn't talking to or about you.

---

Nemesis n. A righteous infliction of retribution manifested by an appropriate agent.

There was a suggestion by Angus or Nemesis to send the run time preference and have the server decide which WUs to send to the client, picking ones that will use less than the set run time.

When I suggested sending the system stats to the server and getting 4 hours of work, 8 hours of work, etc. it was explained that the Boinc client at the time didn't support sending information back. Does the latest Boinc client support such 2 way data transfer? And if so, how many people are still using older Boinc clients on this project?

---

There was a suggestion by Angus or Nemesis to send the run time preference and have the server decide which WUs to send to the client, picking ones that will use less than the set run time.

When I suggested sending the system stats to the server and getting 4 hours of work, 8 hours of work, etc. it was explained that the Boinc client at the time didn't support sending information back. Does the latest Boinc client support such 2 way data transfer? And if so, how many people are still using older Boinc clients on this project?

Isn't the Run Time Preference stored on the server, attached to the user account? Why couldn't the server code look up this setting when responding to work requests? It's already getting other info from the user profile, such as daily download limit.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)

Well as we can see here, the IT people would rather scrape with a person than take on the job of fixing the problem.

I don't work for the project, genius.

I'm sure GEM doesn't care for your rude comments....

No one would imagine that you did work for the project.

If you're not the "IT people", then he obviously wasn't talking to or about you.

So he quotes me, and then says that comment is inappropriate for IT people.... BUT THE TWIST IS THAT HE DIDN'T THINK I WAS THE IT PEOPLE!! Ha ha! A true master of deception, he was just quoting me, and then referring to OTHER PEOPLE making rude comments!!!!




I think your dog is smarter than you are.

No one would imagine that you did work for the project.

If you're not the "IT people", then he obviously wasn't talking to or about you.

So he quotes me, and then says that comment is inappropriate for IT people.... BUT THE TWIST IS THAT HE DIDN'T THINK I WAS THE IT PEOPLE!! Ha ha! A true master of deception, he was just quoting me, and then referring to OTHER PEOPLE making rude comments!!!!

So who exactly were you addressing in message 38095?

At this point, I don't think anyone can figure what you're saying, and to whom.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)

Back to the topic.. If seeing 1.xxx% until the first decoy is done is a problem for users that don't start the project by first reading through the FAQ, is it possible to display a counter of the steps involved in the WU? (Working on Step 2453 of an estimated 287,400 for first decoy, minute 12 of 10,800 Run Time preference)

Only more concise?

---

Back to the topic.. If seeing 1.xxx% until the first decoy is done is a problem for users that don't start the project by first reading through the FAQ, is it possible to display a counter of the steps involved in the WU? (Working on Step 2453 of an estimated 287,400 for first decoy, minute 12 of 10,800 Run Time preference)

Only more concise?

Same answer - if the current step is known, and the estimated steps/model is known, it is easy to calculate the percent done, and then calculate the number of models that will complete within the run time preference, and then the time to completion. If I remember correctly, the next model will not start if it won't complete before the run time preference expires.

So - your example slightly modified to be a speedier PC:

287400/(24530/(12*60)) = 8306 seconds to complete first model
10,800/8306 = 1.3 models in the run time preference, so drop the .3 since we won't start another one
Total time to completion becomes 8306/60 = 138.4 minutes or 2:18
The % complete after 12 minutes is 8.6% (either 24530/287400 or 12/138.4)

Even if that is a bit off due to startup costs and changes throughout the model, if the avg steps/min is being recalculated it can stay pretty close to reality. I doubt it would take much overhead to do these calcs once a minute.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)

So who exactly were you addressing in message 38095?

At this point, I don't think anyone can figure what you're saying, and to whom.

Are you really that slow? Or are you just trying to confuse the issue with stupidity?

I called you a crybaby. GEM quoted that and said that was no way for a project IT person to act. Then you started gibbering nonsense about how GEM wasn't addressing me.

No wonder you complain about Rosetta - you just don't UNDERSTAND anything 8)

So who exactly were you addressing in message 38095?

At this point, I don't think anyone can figure what you're saying, and to whom.

Are you really that slow? Or are you just trying to confuse the issue with stupidity?

I called you a crybaby. GEM quoted that and said that was no way for a project IT person to act. Then you started gibbering nonsense about how GEM wasn't addressing me.

No wonder you complain about Rosetta - you just don't UNDERSTAND anything 8)

At least I'm providing constructive suggestions, amid the crap coming from you.

---

Proudly Banned from Predictator@Home and now Cosmology@home as well. Added SETI to the list today. Temporary ban only - so need to work harder :)



"You can't fix stupid" (Ron White)