I have seen CUDA tasks that have suffered from an incorrect -9 result.
The validator is still giving them credit!
Example Work Unit

There is a Memory Leak in the cuda app.
See: Work unit
This was on 32bit windows xp sp3 after an error..

It has errors with VHAR work.

This app had NOWHERE near enough testing.
It is producing Invalid work.

The servers were already having trouble keeping up with the current CPU demand. Now they release this crazy fast CUDA app full of errors to make no work available for CPU's which are creating valid work.

Why was all this done? More donations? More Science?
Whatever the reason it is WRONG.

I would suggest you suspend all cuda work until these issues are Fixed.

~Bob

P.S. In case you are wondering yes I am upset with this.

---

Do you Good Search for Seti@Home? http://www.goodsearch.com/?charityid=888957
Or Good Shop? http://www.goodshop.com/?charityid=888957

I have seen CUDA tasks that have suffered from an incorrect -9 result.
The validator is still giving them credit!
Example Work Unit

There is a Memory Leak in the cuda app.
See: Work unit
This was on 32bit windows xp sp3 after an error..

It has errors with VHAR work.

This app had NOWHERE near enough testing.
It is producing Invalid work.

The servers were already having trouble keeping up with the current CPU demand. Now they release this crazy fast CUDA app full of errors to make no work available for CPU's which are creating valid work.

Why was all this done? More donations? More Science?
Whatever the reason it is WRONG.

I would suggest you suspend all cuda work until these issues are Fixed.

~Bob

P.S. In case you are wondering yes I am upset with this.

I have posted publicly and privately......

WTF were they thinking releasing this????

Where are the friggin' fish?????

Dumb move.......enough said.


How much did nVidea donate to push this through???? The kitties really wanna know!

Yeah, admins.......I am not stupid.....

---

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

Completely 100% agree. Pull it out of SETI Main, put it back in Beta. Iron out the bugs, release back here in a few months.

The servers can't keep up with the demand...
It's producing bad results...
Ordinary CPU users can't get WU's as easy as you once could....
And if Arecibo is going to shutdown in 2011, This is only salt in the wound.

And it's alienated long time crunchers of the SETI@home project. What happened? CUDA is what happened.

I'm not going to follow what Nvidia want us to do - I have a CUDA GPU, but I'm not going to cruch with it. I'm boycotting CUDA & Nvidia. Who knows Nvidia, my next graphics card might just be ATI!

---

- Luke.

Completely 100% agree. Pull it out of SETI Main, put it back in Beta. Iron out the bugs, release back here in a few months.

The servers can't keep up with the demand...
It's producing bad results...
Ordinary CPU users can't get WU's as easy as you once could....
And if Arecibo is going to shutdown in 2011, This is only salt in the wound.

And it's alienated long time crunchers of the SETI@home project. What happened? CUDA is what happened.

Well.....I, for one....am not gonna rush out to plunk down big bucks for a freakin' Cuda card......

And if my rigs don't get work on Seti due to the demand from Cudas......I am outta here..........

This was reeeeeeeeeeely a dumb move.....

Upping the ante when you can't even cover your bets......sheesh.

And if you want to maintain your street cred with me,

Explain it to me, Eric.

---

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

Completely 100% agree. Pull it out of SETI Main, put it back in Beta. Iron out the bugs, release back here in a few months.

The servers can't keep up with the demand...
It's producing bad results...
Ordinary CPU users can't get WU's as easy as you once could....
And if Arecibo is going to shutdown in 2011, This is only salt in the wound.

And it's alienated long time crunchers of the SETI@home project. What happened? CUDA is what happened.

Well.....I, for one....am not gonna rush out to plunk down big bucks for a freakin' Cuda card......

And if my rigs don't get work on Seti due to the demand from Cudas......I am outta here..........

This was reeeeeeeeeeely a dumb move.....

Upping the ante when you can't even cover your bets......sheesh.

Join the Boycott with me, Mark....
I had trouble getting Work this morning for my laptop.
And the Number Crunching forum is a swamp...

---

- Luke.

Completely 100% agree. Pull it out of SETI Main, put it back in Beta. Iron out the bugs, release back here in a few months.

The servers can't keep up with the demand...
It's producing bad results...
Ordinary CPU users can't get WU's as easy as you once could....
And if Arecibo is going to shutdown in 2011, This is only salt in the wound.

And it's alienated long time crunchers of the SETI@home project. What happened? CUDA is what happened.

Well.....I, for one....am not gonna rush out to plunk down big bucks for a freakin' Cuda card......

And if my rigs don't get work on Seti due to the demand from Cudas......I am outta here..........

This was reeeeeeeeeeely a dumb move.....

Upping the ante when you can't even cover your bets......sheesh.

Join the Boycott with me Mark....

I am not about to boycott anything.........yet.

This needs to get ironed out.......and fast.

I am not pulling out of the project......as long as I can still get work reliably for my rigs.......

But if that falls through......the kitties be looking for retribution......

---

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

Completely 100% agree. Pull it out of SETI Main, put it back in Beta. Iron out the bugs, release back here in a few months.

The servers can't keep up with the demand...
It's producing bad results...
Ordinary CPU users can't get WU's as easy as you once could....
And if Arecibo is going to shutdown in 2011, This is only salt in the wound.

And it's alienated long time crunchers of the SETI@home project. What happened? CUDA is what happened.

I'm not going to follow what Nvidia want us to do - I have a CUDA GPU, but I'm not going to cruch with it. I'm boycotting CUDA & Nvidia. Who knows Nvidia, my next graphics card might just be ATI!

Well, it certainly sounded like a good idea when I read of what it does- or should do. But, as I found when I was working for EDS before it was an HP Company, sometimes things are released to have final testing in the real world before they are ready. Sometimes it is buisness pushing things too fast because there is too little money to finish the testing, or things fall behind scehdule. Other times, the testing just does not come up against enough of the situations that are encounterd in the real world.

---

Well, it certainly sounded like a good idea when I read of what it does- or should do. But, as I found when I was working for EDS before it was an HP Company, sometimes things are released to have final testing in the real world before they are ready. Sometimes it is buisness pushing things too fast because there is too little money to finish the testing, or things fall behind scehdule. Other times, the testing just does not come up against enough of the situations that are encounterd in the real world.

I am waiting for Eric to repond.......it is his project.......not nVidea's.....

---

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

Completely 100% agree. Pull it out of SETI Main, put it back in Beta. Iron out the bugs, release back here in a few months.

The servers can't keep up with the demand...
It's producing bad results...
Ordinary CPU users can't get WU's as easy as you once could....
And if Arecibo is going to shutdown in 2011, This is only salt in the wound.

And it's alienated long time crunchers of the SETI@home project. What happened? CUDA is what happened.

Well.....I, for one....am not gonna rush out to plunk down big bucks for a freakin' Cuda card......

And if my rigs don't get work on Seti due to the demand from Cudas......I am outta here..........

This was reeeeeeeeeeely a dumb move.....

Upping the ante when you can't even cover your bets......sheesh.

Join the Boycott with me Mark....

I am not about to boycott anything.........yet.

This needs to get ironed out.......and fast.

I am not pulling out of the project......as long as I can still get work reliably for my rigs.......

But if that falls through......the kitties be looking for retribution......

I stand beside what I said - The "boycott" is simply not crunching CUDA. Join me at anytime.
Oh - and anyone who's going to argue with me - I'm not even going to listen.

---

- Luke.

I have seen CUDA tasks that have suffered from an incorrect -9 result.
The validator is still giving them credit!
Example Work Unit

There is a Memory Leak in the cuda app.
See: Work unit
This was on 32bit windows xp sp3 after an error..

It has errors with VHAR work.

This app had NOWHERE near enough testing.
It is producing Invalid work.

Agreed. It's clearly beta state, not production state.
But it all began not with CUDA, it began with AstroPulse on beta.
Alpha AP version was released on beta in beginning. It's just logical continuation. If beta-alpha, then main-beta, it's simple.
Moreover:
How you suppose to test CUDA app on beta now? With adaptive replication ON ? It's just impossible! Yes, we can see computational errors still, but how we can detect if -9 pulse overflow is legal or not for this particular task if we even don't know if this task has wingman or it will go directly into master database cause owner of this host has good reputation. I stress, owner (host), not newly released application under testing !
It's just crazy approach IMHO. I created thread on beta with demand to stop doing 2 uncompatible things - testing newly unproved app and testing adaptive replication... no answer, no reaction....
So, beta become not even alpha, beta become useless. Lets' do testing on main, what we can do else ?....

I have seen CUDA tasks that have suffered from an incorrect -9 result.
The validator is still giving them credit!
Example Work Unit

There is a Memory Leak in the cuda app.
See: Work unit
This was on 32bit windows xp sp3 after an error..

It has errors with VHAR work.

This app had NOWHERE near enough testing.
It is producing Invalid work.

Agreed. It's clearly beta state, not production state.
But it all began not with CUDA, it began with AstroPulse on beta.
Alpha AP version was released on beta in beginning. It's just logical continuation. If beta-alpha, then main-beta, it's simple.
Moreover:
How you suppose to test CUDA app on beta now? With adaptive replication ON ? It's just impossible! Yes, we can see computational errors still, but how we can detect if -9 pulse overflow is legal or not for this particular task if we even don't know if this task has wingman or it will go directly into master database cause owner of this host has good reputation. I stress, owner (host), not newly released application under testing !
It's just crazy approach IMHO. I created thread on beta with demand to stop doing 2 uncompatible things - testing newly unproved app and testing adaprive replication... no answer, no reaction....
So, beta become not even alpha, beta become useless. Lets' do testing on main, what we can do else ?....

You are sounding like MSoft.......

Can't quite figure it out......let's put it out in the wild, and see what happens..........

---

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

The question is what was the rush for?
To little money? Nope.. Sure they are short on money but how much does it cost to test an app anyway? Oh wait... They still are testing it in BETA!
Testing just does not come up against enough of the situations that are encountered in the real world. Was that an issue? Nope. A few error reports were listed but none were fixed before it went live.

The limited Beta test it did have (2 days?) was albeit limited due to the adaptive replication test that was going on... The initial results returned have STILL not validated... So there was ZERO validation tests before it came to main.

---

Do you Good Search for Seti@Home? http://www.goodsearch.com/?charityid=888957
Or Good Shop? http://www.goodshop.com/?charityid=888957

I stand beside what I said - The "boycott" is simply not crunching CUDA. Join me at anytime.
Oh - and anyone who's going to argue with me - I'm not even going to listen.

So I will be not argue with you (don't like to speak in space) I will argue to another participants.
CUDA boycott is wrong decision IMHO.
CUDA still has very big perspectives. Moreover, it already has big advantage. My host from 4-core become 6-core with CUDA, really (and I have not too fast GPU). So we NEED CUDA.
All we don't need - is such ugly project and beta management like we have now.
But again, we need testing CUDA app to see what bugs it has and how them can be removed. As I said already it's almost impossible to get such info on beta now, so, please, keep crunching with CUDA on main and report any new bug you encounter here or on beta side. Project still needs that reports!

You are sounding like MSoft.......

Can't quite figure it out......let's put it out in the wild, and see what happens..........

Not me, I just describe current situation, not I created it.

I stand beside what I said - The "boycott" is simply not crunching CUDA. Join me at anytime.
Oh - and anyone who's going to argue with me - I'm not even going to listen.

So I will be not argue with you (don't like to speak in space) I will argue to another participants.
CUDA boycott is wrong decision IMHO.
CUDA still has very big perspectives. Moreover, it already has big advantage. My host from 4-core become 6-core with CUDA, really (and I have not too fast GPU). So we NEED CUDA.
All we don't need - is such ugly project and beta management like we have now.
But again, we need testing CUDA app to see what bugs it has and how them can be removed. As I said already it's almost impossible to get such info on beta now, so, please, keep crunching with CUDA on main and report any new bug you encounter here or on beta side. Project still needs that reports!

When I say boycott - it's simply not crunching CUDA until everything is "fixed" and it's ready for primetime...
Really easy to do. Just don't crunch until an optimum is reached.

---

- Luke.

The question is what was the rush for?
To little money? Nope.. Sure they are short on money but how much does it cost to test an app anyway? Oh wait... They still are testing it in BETA!
Testing just does not come up against enough of the situations that are encountered in the real world. Was that an issue? Nope. A few error reports were listed but none were fixed before it went live.

The limited Beta test it did have (2 days?) was albeit limited due to the adaptive replication test that was going on... The initial results returned have STILL not validated... So there was ZERO validation tests before it came to main.

Yes... Again, completely agreed.
IMHO rush was from nVidia, it's my opinion.
They were need beauty press-release to increase GPU sale before christmas... Damn market economy, it again hurts science!

I stand beside what I said - The "boycott" is simply not crunching CUDA. Join me at anytime.
Oh - and anyone who's going to argue with me - I'm not even going to listen.

So I will be not argue with you (don't like to speak in space) I will argue to another participants.
CUDA boycott is wrong decision IMHO.
CUDA still has very big perspectives. Moreover, it already has big advantage. My host from 4-core become 6-core with CUDA, really (and I have not too fast GPU). So we NEED CUDA.
All we don't need - is such ugly project and beta management like we have now.
But again, we need testing CUDA app to see what bugs it has and how them can be removed. As I said already it's almost impossible to get such info on beta now, so, please, keep crunching with CUDA on main and report any new bug you encounter here or on beta side. Project still needs that reports!

WE don't NEED Cuda........

The project has had enough trouble keeping up with the demands for work from 'normal' rigs.......much less OCd ones like mine...
If they had a problem getting the work crunched....I might agree......but computing power has kept up just fine, I think.........

There were some bigger things in play here that have not been relayed to the rest of us, I think.

---

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

When I say boycott - it's simply not crunching CUDA until everything is "fixed" and it's ready for primetime...
Really easy to do. Just don't crunch until an optimum is reached.

Very nice, and how we will know what needs to be corrected ?
BTW, it's relative small % of errors now. IT don't exceed gain from speed increase. And invalid results here can't harm project - each result has wingman (it will harm only for CUDA agains CUDA validation though), but on beta it has much more chances to come into master database unchecked now.

I join the boycott. I have a 88GYT - WTF? should have read 8800GT - on the way as Xmas present form the Mrs, but i'll stick to crunching in OSX.

This was no where near ready.

---

I join the boycott. I have a 88GYT on the way as Xmas present form the Mrs, but i'll stick to crunching in OSX.

This was no where near ready.

Thank you Satan...

---

- Luke.