I think they are all going to "Invalid" now as I haven't had any credit show up on my personal user stats webpage I have for 6 hrs (it updates every hour).
Hope they notice this issue and fix soon. I'm shocked it has gone on this long.
I think they are all going to "Invalid" now as I haven't had any credit show up on my personal user stats webpage I have for 6 hrs (it updates every hour).
Hope they notice this issue and fix soon. I'm shocked it has gone on this long.
I'm not quite sure that's correct. If you look at the file names, they've been ending in a 2, 3, or 4 (at least), indicating that they are being treated as a 2nd, 3rd, or 4th file for testing. This indicates to me that they may be resent out to everybody for the 2nd, 3rd, or 4th time to get the file finally validated. Normally I've had only "0" or "1", occasionally "2" shows up, but all files need to be validated by 2 individuals (or "computers").
.....[EDIT].....
Plus, in my "task" section, I have only a few tasks marked as "Validate Error" which is much less than what I was having before.
Same here. But I still think there a way too many Invalids. I have never had anything even close to this many Invalids on this project before...even with a buggy AMD driver for my Vega 64 lol.
Thanks, all. All my "validate errors" are gone. One task received "Completed, too late to validate", which is also something I don't think I've seen before.
But, more importantly, the invalid rate went up substantially, though, so I'm going to keep an eye on that. Sounds like @Bluestang saw the same thing. I want this to be a random thing, but it doesn't feel like it.
While the majority of tasks seem to have been resolved (and that's great, I appreciate that) I do have a handful of invalid tasks from work-units such as this one. In case that's no longer accessible, there are six tasks sent from that work-unit, and only the last two returned are marked as completed and validated. The other four (including mine), are all marked 'too late to validate'. Probably too few of them to bother doing anything about, but it is a shame there are still a few not resolved properly.
Not for me. I have an informal standard: I look at the task list on my account for an individual machine and look at the ratio of the number reported invalid to the number reported valid. If that number is not too far from 1% I think it pretty normal for me on Einstein on GPU gamma-ray pulsar work in recent years. Rarely below 0.4%, nor much above 1.5%.
Right now my three machines cluster near 0.8%, so well inside what I think is my normal range. Among the three machines, the range for the currently visible valid task is about 1,300 to 2,500. Machines processing far less work per day will likely see a higher variation in this percentage from time to time for ordinary sampling noise.
I think they are all going to
)
I think they are all going to "Invalid" now as I haven't had any credit show up on my personal user stats webpage I have for 6 hrs (it updates every hour).
Hope they notice this issue and fix soon. I'm shocked it has gone on this long.
bluestang wrote:I think
)
I'm not quite sure that's correct. If you look at the file names, they've been ending in a 2, 3, or 4 (at least), indicating that they are being treated as a 2nd, 3rd, or 4th file for testing. This indicates to me that they may be resent out to everybody for the 2nd, 3rd, or 4th time to get the file finally validated. Normally I've had only "0" or "1", occasionally "2" shows up, but all files need to be validated by 2 individuals (or "computers").
.....[EDIT].....
Plus, in my "task" section, I have only a few tasks marked as "Validate Error" which is much less than what I was having before.
Proud member of the Old Farts Association
No, I am seeing a massive
)
No, I am seeing a massive increase in invalid count also.
Doesn't matter that the tasks are going out to more wingmen before it is validated.
Once you've been proclaimed invalid, you are on the losing end no matter when the task eventually gets validated.
Please see this quick
)
Please see this quick update...
Cheers,
Oliver
Einstein@Home Project
It seems fixed now. Most of
)
It seems fixed now. Most of my invalids are now validated. It also means that for some there are 6 users that receive credit...
S@NL - John van Gorsel
)
Same here. But I still think there a way too many Invalids. I have never had anything even close to this many Invalids on this project before...even with a buggy AMD driver for my Vega 64 lol.
Thanks, all. All my
)
Thanks, all. All my "validate errors" are gone. One task received "Completed, too late to validate", which is also something I don't think I've seen before.
But, more importantly, the invalid rate went up substantially, though, so I'm going to keep an eye on that. Sounds like @Bluestang saw the same thing. I want this to be a random thing, but it doesn't feel like it.
Thankyouuuu! if the
)
Thankyouuuu!
if the picture did not show (above) ...
Here is a link to my recent avg screenshot: My avg!
While the majority of tasks
)
While the majority of tasks seem to have been resolved (and that's great, I appreciate that) I do have a handful of invalid tasks from work-units such as this one. In case that's no longer accessible, there are six tasks sent from that work-unit, and only the last two returned are marked as completed and validated. The other four (including mine), are all marked 'too late to validate'. Probably too few of them to bother doing anything about, but it is a shame there are still a few not resolved properly.
Soli Deo Gloria
bluestang wrote:I still think
)
Not for me. I have an informal standard: I look at the task list on my account for an individual machine and look at the ratio of the number reported invalid to the number reported valid. If that number is not too far from 1% I think it pretty normal for me on Einstein on GPU gamma-ray pulsar work in recent years. Rarely below 0.4%, nor much above 1.5%.
Right now my three machines cluster near 0.8%, so well inside what I think is my normal range. Among the three machines, the range for the currently visible valid task is about 1,300 to 2,500. Machines processing far less work per day will likely see a higher variation in this percentage from time to time for ordinary sampling noise.