I'm curious as to when the next version of Lunatics, beta or otherwise, will be released? Does anyone have an idea of when or has development ground to a halt?

---

I don't buy computers, I build them!!

The best person to ask would be Richard Haselgrove since he was the maintainer of the Lunatics installer. I haven't seen any new developments by Raistmer of new apps though. If no new apps, then no new installer I assume. I also believe Richard has stated that he has hung up his spurs on maintaining the Lunatics installer and had suggested others take up the mantle. He has moved on to bigger things like the BOINC steering committee and BOINC platform software development. So maybe a PM to Richard would answer your question.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I`m testing new CPU apps atm.
But nothing new on GPU so far.

As soon i have finnished my long run tests i will make one or two CPU apps available on my website.
No necessity for a new installer for now.

---

With each crime and every kindness we birth our future.

Let's keep the conversation out in the open, because I'm certainly not going to PM everybody individually!

The current status of the installer is "There is nothing new to install" - and with nothing new on offer, there is no need for a new installer.

People might question the starkness of that statement: there are a few possibilities on the horizon.

1) Petri's / TBar's newer CUDA work. The last I heard, that still had an unacceptable inconclusive validation bug, and I don't think anyone on the now effectively disbanded Lunatics team would allow distribution under their name until that bug's tracked down and fixed.

2) Parkes Telescope data. Eric has received the new hardware that we fundraised for in December, so there will be - in his own time - new work to crunch. We tried to make the current v8 MB apps flexible enough to cope with multiple data sources, but snags might arise and need resolving with a new app: nobody knows yet.

3) Changing the MB search to make it more sensitive, more efficient, or something else. Eric has mentioned a possible v10 in the past (implying he would skip v9), but I have never read or heard any details. I suspect it would be a long way in the future, if ever.

Thanx for the update, I can now go back to sleep and keep crunching.

---

Let's keep the conversation out in the open, because I'm certainly not going to PM everybody individually!

The current status of the installer is "There is nothing new to install" - and with nothing new on offer, there is no need for a new installer.

People might question the starkness of that statement: there are a few possibilities on the horizon.

1) Petri's / TBar's newer CUDA work. The last I heard, that still had an unacceptable inconclusive validation bug, and I don't think anyone on the now effectively disbanded Lunatics team would allow distribution under their name until that bug's tracked down and fixed.

I'm sure that Petri's latest special app is not ready for normal release as you said it has many inconclusives because of differences in pulse counting on overflows when compared to the other mainstream apps.

OTOH, the year old and stable zi3v special app collaboration between Petri and TBar and based on offline CUDA 9 I would say IS ready for mainstream release. It does not have the inconclusive counts of their latest CUDA 9 app and has been working very well for many members. You can look at any of the Top 20 Hosts computers other than Petri's host and see its performance and lack of errors, invalids and inconclusives.

This could qualify as an interim special app release until Petri and TBar can wrangle the static CUDA 9 special app into submission.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Let's keep the conversation out in the open, because I'm certainly not going to PM everybody individually!

The current status of the installer is "There is nothing new to install" - and with nothing new on offer, there is no need for a new installer.

People might question the starkness of that statement: there are a few possibilities on the horizon.

1) Petri's / TBar's newer CUDA work. The last I heard, that still had an unacceptable inconclusive validation bug, and I don't think anyone on the now effectively disbanded Lunatics team would allow distribution under their name until that bug's tracked down and fixed.

2) Parkes Telescope data. Eric has received the new hardware that we fundraised for in December, so there will be - in his own time - new work to crunch. We tried to make the current v8 MB apps flexible enough to cope with multiple data sources, but snags might arise and need resolving with a new app: nobody knows yet.

3) Changing the MB search to make it more sensitive, more efficient, or something else. Eric has mentioned a possible v10 in the past (implying he would skip v9), but I have never read or heard any details. I suspect it would be a long way in the future, if ever.

. . I cannot speak for anyone else but I am running Zi3v Cuda80 and my inconclusive rate it quite low. It is usually less than 1% with GBT work and even during the few short spurts of Arecibo data it only rose to at most 1.7%. That is about 3 to 6 inconclusives a day on the slower Linux box that turns over about 300 per day. It is a little more on the faster Linux box which does about 1000 WUs a day. I have heard that Zi3v Cuda90 is improved but I have yet to run it so I cannot confirm that. As for a Windows port all I have ever hear is rumours and scattered incomplete details. One rumour was that there is a problem under the Window OS that impedes the core Cuda function of the app, but as I said, no details. So until someone has a 'certified' Windows version there would be no need for inclusion in Lunatics. But it would be nice to have one to play with :)

. . This is something I am keen to see. hopefully when it starts to flow there will be no snags with the apps I am running ...

Stephen

:)

Let's keep the conversation out in the open, because I'm certainly not going to PM everybody individually!

OTOH, the year old and stable zi3v special app collaboration between Petri and TBar and based on offline CUDA 9 I would say IS ready for mainstream release. It does not have the inconclusive counts of their latest CUDA 9 app and has been working very well for many members. You can look at any of the Top 20 Hosts computers other than Petri's host and see its performance and lack of errors, invalids and inconclusives.
This could qualify as an interim special app release until Petri and TBar can wrangle the static CUDA 9 special app into submission.

. . I thought it was only a candidate for inclusion as a Windows app. I have not heard of a Lunatics for Linux, only the packages that TBar has helpfully provided. But you are absolutely correct that it is proving itself as a stable and powerful tool for the job at hand.

Stephen

:)

Let's keep the conversation out in the open, because I'm certainly not going to PM everybody individually!

OTOH, the year old and stable zi3v special app collaboration between Petri and TBar and based on offline CUDA 9 I would say IS ready for mainstream release. It does not have the inconclusive counts of their latest CUDA 9 app and has been working very well for many members. You can look at any of the Top 20 Hosts computers other than Petri's host and see its performance and lack of errors, invalids and inconclusives.
This could qualify as an interim special app release until Petri and TBar can wrangle the static CUDA 9 special app into submission.

. . I thought it was only a candidate for inclusion as a Windows app. I have not heard of a Lunatics for Linux, only the packages that TBar has helpfully provided. But you are absolutely correct that it is proving itself as a stable and powerful tool for the job at hand.

Stephen

:)

Yes - a "Lunatics Installer" is a Windows-only concept. If the app that Keith is referring to exists in Windows form, feel free to send me the release notes, the configuration instructions and - crucially! - where to get hold of a copy from, and I'll take a look. It'll take a few days for me to get my head back into order after a busy (and boozy) BOINC workshop last week, but I'll get there eventually.

Yes - a "Lunatics Installer" is a Windows-only concept. If the app that Keith is referring to exists in Windows form, feel free to send me the release notes, the configuration instructions and - crucially! - where to get hold of a copy from, and I'll take a look. It'll take a few days for me to get my head back into order after a busy (and boozy) BOINC workshop last week, but I'll get there eventually.

. . Sadly there is none that I know of.

. . Sounds like a fun workshop but the "boozy" part leads me to wonder just what we can expect in the way of changes? :)

Stephen

:)

No what I meant was that the zi3v CUDA9 special app could be considered as a mainstream app on the site. Not that it could be included in the Lunatics installer since that is for Windows apps. I haven't heard of anyone building a 'Lunatics' installer for Linux apps. Not sure it is needed with the one click, unzip of the packages that TBar has already provided on the Crunchers Anonymous site.

In the context of the original post, mention of the CUDA 8 and 9 apps has strayed off topic of what the next Lunatics installer might include.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

What we need is a Windows CUDA 9.0 & up new app.

---

But based on comments alluded by Stephen and what I remember, our Windows app developer Raistmer said it would be almost impossible to use the exiting CUDA9 Linux app code source for any Windows app simply because the mechanisms the Linux app uses isn't available in the Windows environment.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

But based on comments alluded by Stephen and what I remember, our Windows app developer Raistmer said it would be almost impossible to use the exiting CUDA9 Linux app code source for any Windows app simply because the mechanisms the Linux app uses isn't available in the Windows environment.

I'm skating on thin ice here, but I think Raistmer only really knows the OpenCL environment. Anything possible in CUDA for Linux should be possible in CUDA for Windows.

We really need a replacement CUDA developer for Jason G. He appears to have left the project entirely (his computers haven't connected to the servers since December last year), although I think he's OK - I see him on the mailing lists occasionally (most recently last month). I think he's just busy with paying work.

Yes, Jason G and his expertise are very much missed.
Meowsigh.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

No what I meant was that the zi3v CUDA9 special app could be considered as a mainstream app on the site. Not that it could be included in the Lunatics installer since that is for Windows apps.
In the context of the original post, mention of the CUDA 8 and 9 apps has strayed off topic of what the next Lunatics installer might include.

. . Yep agreed.

Stephen

I'm skating on thin ice here, but I think Raistmer only really knows the OpenCL environment. Anything possible in CUDA for Linux should be possible in CUDA for Windows.

We really need a replacement CUDA developer for Jason G. He appears to have left the project entirely (his computers haven't connected to the servers since December last year), although I think he's OK - I see him on the mailing lists occasionally (most recently last month). I think he's just busy with paying work.

. . Many of us do a lot of that kind of skating :) But I don't think Raistmer wants to tackle Petri's code for whatever reason.

. . I remember a conversation where Jason mentioned he had compiled a Windows version of Petri's code but had some issues with it, but that was a much earlier version than the Zi3V Cuda80 used so widely now. He certainly seemed to believe it was a possibility but money has stolen him away from the project ... how gauche! :) I hope he is doing well too ...

Stephen

:)

But based on comments alluded by Stephen and what I remember, our Windows app developer Raistmer said it would be almost impossible to use the exiting CUDA9 Linux app code source for any Windows app simply because the mechanisms the Linux app uses isn't available in the Windows environment.

I'm skating on thin ice here, but I think Raistmer only really knows the OpenCL environment. Anything possible in CUDA for Linux should be possible in CUDA for Windows.

We really need a replacement CUDA developer for Jason G. He appears to have left the project entirely (his computers haven't connected to the servers since December last year), although I think he's OK - I see him on the mailing lists occasionally (most recently last month). I think he's just busy with paying work.

Oooer for Jason G being busy elsewhere. His input is missed. However, I can very much sympathize on the 'busy' bit!

Hopefully we can gain some new enthusiasm for new developments with the new hardware for Parkes coming online. Or even for newly available crunching hardware such as the Tensor compute bits...


Aside: I've moved over to AMD for their GPUs with their slightly better 'open source' approach and for the greater flexibility with OpenCL.

Lot's to explore yet!


Keep searchin',
Martin

---

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)

The current status of the installer is "There is nothing new to install" - and with nothing new on offer, there is no need for a new installer.

Wondering where, if anywhere, the 8.22 vs. 8.20 MB fit into that equation?

---

They are "current stock" applications so don't really qualify for inclusion in the Lunatics installer.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?