Silly me. I was hoping that with the advent of V7 that users would spend time looking over there rigs and correct the problems that their wingmen have been seeing for months, even years. Yes I'm talking about the dreaded Cuda overflow. You know these users whose rig(s) are generating a near constant flow of "validation inconclusive" that when you look turns out to be a -9 overflow where you found a handful of signals. Sure it gets assigned to another user and you will end up with credit and the overflow guy gets "invalid" but it pains me to see that.

It's like seeing a guy in an expensive suit driving a two year old BMW with it belching smoke and engine making a "bad" sound. Does he not care? I'm mean it's an expensive car, it looks like he could afford to take it in to have it looked at but there he is, every morning at the coffee shop with his nice suit and smokey, noisy car.

So crunchers, take a look of the tasks assigned to each if your rigs. If "validation inconclusive" and "invalid" overwhelms the number of "valid" WUs, something isn't right with your system, have it looked at. Please, just like the car guys at the coffee shop who dig their finger nails into their palms when Mister BMW pulls up for his morning cup of joe, to us computer hardware guys, it's like fingernails on a chalkboard.

---

"Life is just nature's way of keeping meat fresh." - The Doctor

Keith, there is a major difference between "validation inconclusive" and "invalid". I am far more worried about the crunchers that have a pile of "invalid" and "computational error" results than those that are just "inconclusive". The nature of the data we are working on will lead to "inconclusive" results periodically, as there are differences in the precision and accuracy of different processors and applications, and make the final data to be generally of a far higher quality than if would be with only one cruncher being used to determine the result.

The crunchers to be concerned about are those that are very consistently high in their "invalid" and "error while computing" counts.

---

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

I've been keeping an eye on mine...
I have 19 invalids and 119 errors.

The invalids are all when on the new v7 apps, before I re-optimized with the installer on Monday.

Most of the errors are from when I was switching apps.

I have 3,084 valid, 388 inconclusive, and 7,835 pending.

So, I think things are under control at the moment on the kitty crunching farm.
(Other than the RAC bit, LOL.)

---

"Time is simply the mechanism that keeps everything from happening all at once."

@rob smith - Yes but those who have cuda MB throwing off -9 overflows left and right first appear in the "valid inconclusive" pile waiting till they are declared "invalid" when the third arrives. These folks have a few valid and hundreds if not thousands of "invalid" (Completed, marked as invalid) and "valid inconclusive".

@msatter - Yes but you do care about your kitties. You don't let them get all matted down and tick/flea infested. You had all of 2 rigs with a total of 17 Invalid Vs 470 valid Vs 72 inconclusive. I'm talking about users (in my case one user) with one rig that has 7 valid cuda with over 1000+ invalid and 2400+ inconclusive. I am talking just V7 MB here. In what, a week at most.

Yes every once in a while your system may burp. Overheated video card, PSU on it's last legs, whatever. I'm not talking about those users. I'm talking about users who set the system up once and forgot about it. It's sitting in the guest bedroom closet or home office gumming up the works, never checking it's status as if it was some long forgotten zombie you trapped in the unused root cellar.

It just bugs me when I come across these guys.

---

"Life is just nature's way of keeping meat fresh." - The Doctor

It just bugs me when I come across these guys.

Well Keith next time you run across one of these guys PM them and offer to help debug their rig.
Before you do that you may wish to check if these guys have any posts to fora.
Probably less than 1% of people running seti@home ever even visit the message boards.
If errant crunchers were a real problem for the project I'm sure Dr A. would slap them down hard. lol.

It just bugs me when I come across these guys.

Well Keith next time you run across one of these guys PM them and offer to help debug their rig.
Before you do that you may wish to check if these guys have any posts to fora.
Probably less than 1% of people running seti@home ever even visit the message boards.
If errant crunchers were a real problem for the project I'm sure Dr A. would slap them down hard. lol.

I would if I could but I'm not running with an nVidia card so I have no idea where to start in trying to help them. Plus the majority of the ones I come across have never posted on the forums either. And those I've seen posting on the forums and were called out on their faulty crunchers turn around and blame seti@home for putting out faulty apps because their rig is a unique and pretty snowflake that can do no wrong.

Yes in the grand scheme they infrequently mess things up (unless their wingman is wrongly tossing off 30 spike -9 overflows as well). They just frustrate me more than the users who sign up, crunch for a month and then vanish with a queue full of WUs since these guys actually believe that they are helping the project.

---

"Life is just nature's way of keeping meat fresh." - The Doctor

Volunteering Ain't supposed To Be Like Work.

Volunteering Ain't supposed To Cost Any Money.

Volunteering Ain't A Job when Set It and Forget It has been Touted Ad Nauseum.

Volunteering Ain't About Bitchin' and Complainin' 'bout Volunteers.

Bound For IT IT.

---

May we All have a METAMORPHOSIS. REASON. GOoD JUDGEMENT and LOVE and ORDER!!!!!

Well I have done a bit of charity "volunteering" and I in my experience:

Volunteering Ain't About Bitchin' and Complainin' 'bout Volunteers.

Is exactly what it is about :-)

---

Well I have done a bit of charity "volunteering" and I in my experience:

Volunteering Ain't About Bitchin' and Complainin' 'bout Volunteers.

Is exactly what it is about :-)

Yep, Dat's The Best Part of Volunteering.

Carry On RACmeisters. Bitch IT Real Good.

Bound For IT IT and There There.

---

May we All have a METAMORPHOSIS. REASON. GOoD JUDGEMENT and LOVE and ORDER!!!!!

Fine I get it. Don't complain about wingmen whose rigs are hurting more than helping.

---

"Life is just nature's way of keeping meat fresh." - The Doctor

Fine I get it. Don't complain about wingmen whose rigs are hurting more than helping.

Bitch about them wingpeople as much as you want to.

Cheers.

http://setiathome.berkeley.edu/results.php?hostid=5217740

Well this person is being sent CUDA 42 tasks to his 8600 GTS by Boinc.
They are almost all turning into inconclusives.
I thought CUDA 42 was for Fermi cards and Boinc should be sending CUDA23 or CUDA 32 WUs to cards like this.
Then again it could be the amount of VRAM in this older card which is causing the problem . Either way, it doesn't look like its this USERS fault that his WUs are erroring.

john3760

---

http://setiathome.berkeley.edu/results.php?hostid=5217740

Well this person is being sent CUDA 42 tasks to his 8600 GTS by Boinc.
They are almost all turning into inconclusives.
I thought CUDA 42 was for Fermi cards and Boinc should be sending CUDA23 or CUDA 32 WUs to cards like this.
Then again it could be the amount of VRAM in this older card which is causing the problem . Either way, it doesn't look like its this USERS fault that his WUs are erroring.

john3760

False overflows usually point to heat problems or such.

The variants are all viable - 23 would be best but 42 is just slower.

The problem in this case is that the quota system either isn't working or he's getting enough good results in between to keep quota up.

He currently has one v6 task - same problem.

IOW he/she needs to look at his/her card.

---

A person who won't read has no advantage over one who can't read. (Mark Twain)

Given how few crunchers actually come into the forum or look at what their PC is actually doing, and the fact that email addresses are not accessible to even the mods the only way to communicate with a user is a PM not a bitch about them on the forum. Even then, if they have S@H emails on ignore, or are using an redundant email address that message won't get to their eyes.

---

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