TPU's Rosetta Milestones and Daily Pie Thread

[phill]

Well after messing that up the first time, lets try again shall we

Rosetta Pie Stats for the 7th March 2022!!.....

And now for the milestones and stoners for the day....

Sadly no milestones today but some very very nice contributions and more than one person contributing!! Great work there!! I hope everyone is doing well and is ok!!

 

[phill]

And here we go for the 8th March 2022 for Rosetta Pie Stats .....

And now for the milestones and stoners for the day.....

Sadly no one for the milestones today and sadly only our @mstenholm contributing for our team as well which is no bad thing just wish others could get some work units, including myself!! Here's hoping tomorrow will bring us some better results on all accounts!!

Great work everyone and apologies for the mess up

 

[mstenholm]

And here we go for the 8th March 2022 for Rosetta Pie Stats .....

View attachment 239282

View attachment 239283

And now for the milestones and stoners for the day.....

View attachment 239284

Sadly no one for the milestones today and sadly only our @mstenholm contributing for our team as well which is no bad thing just wish others could get some work units, including myself!! Here's hoping tomorrow will bring us some better results on all accounts!!

Great work everyone and apologies for the mess up

Please, please download work when it’s there, like NOW!

 

[MachineLearning]

Please, please download work when it’s there, like NOW!

Thanks for the heads-up, just got some WU's

 

[phill]

Please, please download work when it’s there, like NOW!

I'd love to mate but I was actually asleep for a change!! Can't keep machines on over night generally either

 

[phill]

Well here we are again, the Rosetta Pie Stats for the 9th March 2022......

And now for the milestones and stoners for the day.......

Sadly no one today but I hope and think that a few shall be making an appearance shortly I hope that everyone is doing well and are ok.. Amazing efforts everyone, proud bodgit captain here today!! Just like most days to be honest.... lol How and why you guys and gals put up with me is anyone's guess!!

Take care all, hope to see you again tomorrow !!

 

[mstenholm]

I'd love to mate but I was actually asleep for a change!! Can't keep machines on over night generally either

If both you and your machines sleep then you have to rely on the team for points
Today has been more or less dry in the sense of new work but since I got +400 when I wrote the message yesterday I’m good for 72 hours. I’m afraid that that is 99% of the total queue for the team. Better luck tomorrow.

 

[Zachary85]

I have about 3 days of work ready to run. I never have any luck clicking on the update button. My machine seems to get work on it's own (when work is available)

 

[mstenholm]

I have about 3 days of work ready to run. I never have any luck clicking on the update button. My machine seems to get work on it's own (when work is available)

If you have about three days work (3x3x8 = 72 for a 8 thead CPU with normal 8 hour runtime) then you are golden. Remember that they only have three day deadline and and being late can result in zero points. In the past, 3-4 months ago, we had longer periods with steady supply and there were no need to use the update button. Now you have to be lucky that you machine ask when they send out the big batches.

I have noticed that all the re-sends I get is from Windows machines that errors out, never from Linux machines. Just a hint of where to optimize a dedicated cruncher. An other way to get more out of the work you have received is the change the length of the run time for each (Preferences for this Project, Target CPU run time). The setting has to be applied before the unit starts. Again don’t be late, if the re-send is returned before you complete your job then you get zero points. If you abort a job before it has completed then you will get close to normal points for the part of the job that is completed.

 

[MachineLearning]

Just got some WU's, everyone update if needed

 

[mstenholm]

Just got some WU's, everyone update if needed

Small detail….you got ARM jobs and you are the only one on the team that have ARM equipment in the arsenal . Good luck, I will for sure follow the progress.

Edit: it can very well be, that the jobs could have been assigned to other platforms. It’s new to me. The floating point value for your ARM equipment is similar to a mid-range CPU so why not. The points you get will be interesting to see. How much power does this 8 thread ARM-thing use?

 

[phill]

Well here we are with a Rosetta Pie Stats update for the 10th March 2022.....

And now for the milestones and stoners for the day......

Sadly no one today but here's hoping someone will be here soon!.....
Hope everyone is doing well and ok?
I see we managed to have a load of people contributing today which is amazing, where did you all find work units?? I've had machines on but sadly still nothing has come down, ah what it must be like for having cheap electric lol I hope we can continue with this many members getting work, its a lovely welcomed change

 

[phill]

Well here goes for the last update for today, the Rosetta Pie Stats for the 11th March 2022....

And now for the milestones and stoners of the day.......?

@agent00skid !!
@xrobwx71 !!
@MachineLearning !!

Well, there's a turn up for the books... We get some work units and members hit milestones........ Its as if the two are connected...... Weird don't you think!?!?! Well, now I've finished taking the michael.... ..... Great work everyone for grabbing those work units and making some points happen!! I suppose this might just show that it's not @mstenholm's Rosetta Pie Stats thread after all.......

Great stuff everyone, time for me to hit the sack and hopefully tomorrow we'll share the same positives of more members and some milestones!! Take care everyone!! Speak soon!!

 

[mechtech]

I remember folding from 15+ years ago, was a big thing, seti, protiens, etc. etc.

Then it seemed to disappear like a fad.

Is it still big and doing good things???

 

[MachineLearning]

I remember folding from 15+ years ago, was a big thing, seti, protiens, etc. etc.

Then it seemed to disappear like a fad.

Is it still big and doing good things???

Absolutely, so big that at the beginning of covid F@H ran out of WU's And Rosetta has basically none right now, only in bursts, unless you use ARM such as an Android phone / ChromeOS / etc. It never fully went away, just disappeared from the public's radar temporarily.

For Rosetta, use BOINC, using CPU is more efficient than GPU for it. (Though you won't find any jobs except random spurts, if you keep it on constantly)
For Folding@Home, use its built-in client and disable CPU folding (or leave at least 1 core remaining to feed GPU work, if not doing Rosetta CPU work).
Make sure you set a passkey for F@H for bonus points, sign up for TPU's team on Rosetta and F@H, and you're back in!

@mstenholm is extremely experienced and knowledgable, if you wanted to get back into things.
~
@mstenholm :

Edit: it can very well be, that the jobs could have been assigned to other platforms. It’s new to me. The floating point value for your ARM equipment is similar to a mid-range CPU so why not. The points you get will be interesting to see. How much power does this 8 thread ARM-thing use?

Just saw this edit: I'll clarify things so you don't need to guess the hardware I'm not sure how much you can see, haven't looked for any stats on the team for Rosetta yet. It's my Samsung Galaxy S20, Snapdragon 865. It seems that should only be about 5 watts TDP - I'm impressed it was anywhere near a mid-range CPU in any category!

Also, here is the job if this gives you any answers:

Annoyingly even if I retry transfers, these are stuck on Uploading.

No reason for them to be suspended in the parameters I set either.

 

[mstenholm]

Just saw this edit: I'll clarify things so you don't need to guess the hardware I'm not sure how much you can see, haven't looked for any stats on the team for Rosetta yet. It's my Samsung Galaxy S20, Snapdragon 865. It seems that should only be about 5 watts TDP - I'm impressed it was anywhere near a mid-range CPU in any category!

Also, here is the job if this gives you any answers:

Annoyingly even if I retry transfers, these are stuck on Uploading.

No reason for them to be suspended in the parameters I set either.

This is what I see:
ARM
AArch64 Processor rev 14 (aarch64) [Impl 0x51 Arch 8 Variant 0xd Part 0x805 Rev 14]
(8 processors)

So that translate into a Snapdragon. I guess I could have found out if I spent some time in the company with Google.

Do you still have upload problems? As far as I can see no jobs have been returned yet from this device.

As for the performance I went by the reported

Measured floating point speed	3923.48 million ops/sec
Measured integer speed	17115.89 million ops/sec

My old W3670 (i7 970) is

Measured floating point speed	4118.53 million ops/sec
Measured integer speed	11184.66 million ops/sec

That's why I compared it to a mid-range CPU. Well it was a good CPU 12 years ago, calling it a mid-range by today standard is not entirely correct but since Rosetta uses the floating point speed as a basis for points you could get something like 25 points/hour per thread, and for 5 W that's fantastic , but if the jobs doesn't upload.....

 

[MachineLearning]

This is what I see:
ARM
AArch64 Processor rev 14 (aarch64) [Impl 0x51 Arch 8 Variant 0xd Part 0x805 Rev 14]
(8 processors)

So that translate into a Snapdragon. I guess I could have found out if I spent some time in the company with Google.

Do you still have upload problems? As far as I can see no jobs have been returned yet from this device.

As for the performance I went by the reported

Measured floating point speed	3923.48 million ops/sec
Measured integer speed	17115.89 million ops/sec

My old W3670 (i7 970) is

Measured floating point speed	4118.53 million ops/sec
Measured integer speed	11184.66 million ops/sec

That's why I compared it to a mid-range CPU. Well it was a good CPU 12 years ago, calling it a mid-range by today standard is not entirely correct but since Rosetta uses the floating point speed as a basis for points you could get something like 25 points/hour per thread, and for 5 W that's fantastic , but if the jobs doesn't upload.....

In such a context, that seriously is impressive. I have quite a few of those hexa-core 1366 Xeons, they're still wicked fast if you OC even today.

It seems that yes, uploading is still a problem. I'll give it until the job expires, hopefully this issue is temporary. Otherwise, I'll take it offline. You may also notice a 4c Snapdragon chip, that's an S4 (Snapdragon 800) which I don't think will ever find work but it's worth leaving up anyway.

 

[phill]

Well everyone in Rosetta Pie Stats land, how are we today?? Here's a Stats update for the 12th March 2022....

And now for the milestones and stoners for the day......

@mstenholm !!
@Zachary85 !!

Well there's the big man @mstenholm getting the next milestone in his path and well done to @Zachary85 for hitting his next milestone too!! I'm jealous of both of you as I can't seem to grab any work units when my crunchers are online!! So frustrating lol Still, very glad and happy to see members hitting milestones and more so, even sending in points for the team!! Amazing stuff guys and gals, please do keep the great work up!!

Well best go grab some more data before bed time and then get myself to bed! Take it easy everyone, I hope you are doing well and ok!!

 

[mstenholm]

So I finally got some work. Preet units but they errored out on 3 systems after 30 seconds, Linux. It seems like I’m not the only one with a problem.

Update: Windows has a similar problem.

 

[MachineLearning]

So I finally got some work. Preet units but they errored out on 3 systems after 30 seconds, Linux. It seems like I’m not the only one with a problem.

Update: Windows has a similar problem.

Same exact situation here, can Update for new tasks but those error out too (W10)

 

[mstenholm]

Same exact situation here, can Update for new tasks but those error out too (W10)

Last time it happened, 10 days ago, at least Linux could handle the jobs. Now we have to wait for volunteers to error out the 100,000 or so jobs. I opted out on two of my machines and turned them of for the first time since I build them 3 years ago. Strange with the silence . My last machine was working on 12 rb jobs so I just paused it getting new jobs. I would hate to be asked to wait to get new jobs due to too many errors.

 

[phill]

Well here we go for a quick update to Rosetta Pie Stats for the 13th March 2022....

And now for the milestones and stoners for the day....

Sadly no one today but there's hopefully another around the corner!! Everyone doing well and ok I hope!!

 

[phill]

And here we go for the Rosetta Pie Stats for the 14th March 2022.....

And now for the milestones and stoners for the day....

Well sadly no one today, but here's hoping, there's another around the corner!! I hope everyone is doing well and ok?? Hopefully catch up with you all tomorrow.. Bit late in the day to try and do it now! Take care everyone!!

 

[phill]

Well here we are for another quick update for Rosetta Pie Stats for the 15th March 2022......

And now for the milestones and stoners for the day.....

@Zachary85 !!

Great work from @Zachary85 for hitting their next milestone but I do wonder how you've managed it since you've got any points by your name today!?! lol I'm guessing it must have happened whilst you where validating the work units I guess
It seems again no one but @mstenholm has had some spare work for todays contribution.. I wonder if many of the other teams are having these lack of work unit issues with their team members... Seems such a shame they can't push out enough work units for the the demands. Still, never mind

I hope that everyone is doing well and is ok?? Take care everyone, hope to see you back again tomorrow for another update

 

[Zachary85]

It's not a lack of work ... there have been plenty available, but all the units the past day or two error out in less than a minute. I stopped downloading new work last night.