4851...

Have a beer. It will help you not worry about it.

Good idea my friend

6960!! Some AP's might have gotten validated:)))

---

rOZZ
Music
Pictures

Gee, the way I remember it SETIv7 was Much Slower than SETIv6, causing Much Longer Run-Times for the Same Credit resulting in Lower RAC. Yes, I'm Positive about it.

The part that slowed things down were the auto correlations, which shouldn't have resulted in a 50% reduction in RAC as the processing time didn't double.

You might want to check on that. There is proof the old nvidia cards, such as mine, ran a v6 shorty in 4 minutes and produced around 40 credits. I still have one of those cards running. Check out how long it takes to produce 40 credits today; http://setiathome.berkeley.edu/results.php?hostid=6796475&offset=20&show_names=0&state=4&appid=11
Granted...it's a little slower in Win 8, but, 1000secs is much slower than 240secs in most everyone's book.

I can't remember the actual run times, only that when v7 came out my credit dropped by half, and while the run times did increase they didn't come close to doubling.

---

Grant
Darwin NT

Gee, the way I remember it SETIv7 was Much Slower than SETIv6, causing Much Longer Run-Times for the Same Credit resulting in Lower RAC. Yes, I'm Positive about it.

The part that slowed things down were the auto correlations, which shouldn't have resulted in a 50% reduction in RAC as the processing time didn't double.

You might want to check on that. There is proof the old nvidia cards, such as mine, ran a v6 shorty in 4 minutes and produced around 40 credits. I still have one of those cards running. Check out how long it takes to produce 40 credits today; http://setiathome.berkeley.edu/results.php?hostid=6796475&offset=20&show_names=0&state=4&appid=11
Granted...it's a little slower in Win 8, but, 1000secs is much slower than 240secs in most everyone's book.

I can't remember the actual run times, only that when v7 came out my credit dropped by half, and while the run times did increase they didn't come close to doubling.

Jesus;
Just some early observations on V7 CUDA work.

Shorties that use to take 4-5mins are now taking 18-20mins.

Workunits that took 10-12mins now take about 25mins.

Workunits that took about 20mins now take around 30mins.

And those odd heavy 1's that took around 25mins are now taking around 35mins.

*Rolls Eyes*

Gee, the way I remember it SETIv7 was Much Slower than SETIv6, causing Much Longer Run-Times for the Same Credit resulting in Lower RAC. Yes, I'm Positive about it.

The part that slowed things down were the auto correlations, which shouldn't have resulted in a 50% reduction in RAC as the processing time didn't double.

You might want to check on that. There is proof the old nvidia cards, such as mine, ran a v6 shorty in 4 minutes and produced around 40 credits. I still have one of those cards running. Check out how long it takes to produce 40 credits today; http://setiathome.berkeley.edu/results.php?hostid=6796475&offset=20&show_names=0&state=4&appid=11
Granted...it's a little slower in Win 8, but, 1000secs is much slower than 240secs in most everyone's book.

I can't remember the actual run times, only that when v7 came out my credit dropped by half, and while the run times did increase they didn't come close to doubling.

Jesus;
Just some early observations on V7 CUDA work.

Shorties that use to take 4-5mins are now taking 18-20mins.

Workunits that took 10-12mins now take about 25mins.

Workunits that took about 20mins now take around 30mins.

And those odd heavy 1's that took around 25mins are now taking around 35mins.

*Rolls Eyes*

The time increase was for NVidia (CUDA) processing.

The credit baseline is fixed by CPU tasks, and the time increase was much smaller for them.

Killjoy.

Well, you have to admit the large number of GPU tasks tends to lower the overall average resulting in most of the MBv7 loss being attributed to a much slower GPU App. The discrepancy in Shorty times is atrocious. Again, APv7 is NOT slower than APv6 and in a perfect world with a functioning CreditFew there shouldn't be any change in AP scores. Just as there wasn't any change at Beta.

There are fewer people using their CPUs for crunching all the time. Which is partly why I just suspended my remaining APv6 CPU tasks to run APv7 CPU tasks on my last machine still holding APv6 CPU tasks. I fear it won't make a dent though...

The discrepancy in Shorty times is atrocious.

Sadly, that will remain the case until Jason can find a technique (or a CUDA version) for more efficient autocorrelation processing on a CUDA GPU. That doesn't look like an easy problem to solve.

The discrepancy in Shorty times is atrocious.

Sadly, that will remain the case until Jason can find a technique (or a CUDA version) for more efficient autocorrelation processing on a CUDA GPU. That doesn't look like an easy problem to solve.

Part of the autocorrelation problem was that there are lots of small inefficient transfers over the PCIe Bus, if that can be worked around then there should be an improvement.

I never asked Raistmer if OpenCL MBv7 was similarly affected, it has ZeroCopy so shouldn't be.

Claggy

It may still be too early to make this comment on the V7 apps, with 2 GPU's, 20 +/- out of 70 +/- AP jobs validated, it looks to me like the long awaited credit correlation between AP and MB has arrived.

---

It may still be too early to make this comment on the V7 apps, with 2 GPU's, 20 +/- out of 70 +/- AP jobs validated, it looks to me like the long awaited credit correlation between AP and MB has arrived.

I agree, and it is not for the best. From what I have seen on my own validated WUs, and on those of some heavy hitter PCs with a lot more results than I have, we are getting 1/2 to 1/3 the credit of APV6. I used to see a lot of 500 results plus or minus and now they are 200 plus or minus - mostly minus.

Hopefully things will improve as some have indicated might happen when there are more results, but you will have to forgive me if I am doubtful.

I guess the good news is that it might stop the AP feeding frenzies of the past. It seems like a high price to pay though.

I hope I am wrong, but it appears the party is over.

It may still be too early to make this comment on the V7 apps, with 2 GPU's, 20 +/- out of 70 +/- AP jobs validated, it looks to me like the long awaited credit correlation between AP and MB has arrived.

I agree, and it is not for the best. From what I have seen on my own validated WUs, and on those of some heavy hitter PCs with a lot more results than I have, we are getting 1/2 to 1/3 the credit of APV6. I used to see a lot of 500 results plus or minus and now they are 200 plus or minus - mostly minus.

Hopefully things will improve as some have indicated might happen when there are more results, but you will have to forgive me if I am doubtful.

I guess the good news is that it might stop the AP feeding frenzies of the past. It seems like a high price to pay though.

I hope I am wrong, but it appears the party is over.

There was a similar discussion over on Beta when AP_v7 testing first began. It too began with low scores. This is how it turned out;
Valid tasks for computer 72274

So, I wouldn't head to the bank just yet.

This being the first time I have been through an upgrade I'm confused. I received a Boinc notice that my "app_info.xml file doesn't have a usable version of AstroPulse v7". Question I have is will the usable version be automatically added to my S@H app_info.xml file? Or do I need to manually download the necessary update? I am running Lunatic software and assume that will need an update to correctly process v7.

This is probably pretty basic to those of you who have been in the project for many years but we 'newbies' get lost as those basics are unknown to us.

Any helpful comments would be greatly appreciated.

---

"Sour Grapes make a bitter Whine." <(0)>

This being the first time I have been through an upgrade I'm confused. I received a Boinc notice that my "app_info.xml file doesn't have a usable version of AstroPulse v7". Question I have is will the usable version be automatically added to my S@H app_info.xml file? Or do I need to manually download the necessary update? I am running Lunatic software and assume that will need an update to correctly process v7.

This is probably pretty basic to those of you who have been in the project for many years but we 'newbies' get lost as those basics are unknown to us.

Any helpful comments would be greatly appreciated.

You will want to use the new lunatics installer once it is released. Unless you feel up to modifying your app_info.xml yourself.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

You will want to use the new lunatics installer once it is released. Unless you feel up to modifying your app_info.xml yourself.[/quote]

I previously edited the file to increase GPU performance, can you give me specific changes to make that will make the v7 portion process correctly? After seeing the changes needed I can decide if I want to tackle that chore or wait for the new installer.

Thanks for the input.

You will want to use the new lunatics installer once it is released. Unless you feel up to modifying your app_info.xml yourself.

I previously edited the file to increase GPU performance, can you give me specific changes to make that will make the v7 portion process correctly? After seeing the changes needed I can decide if I want to tackle that chore or wait for the new installer.

Thanks for the input.

You won't have an AP v7 portion, you'll need a new app as well, the old AP v6 app isn't compatible,
you can either wait for a new Installer, revert to Stock, or download the apps from Crunchers Anonymous, decompress them, copy them to the project directory, and run aimerge or aimergeg to combine the aistub's into an app_info

Claggy

As I recall when MB v7 came out it was said, lets wait and see, the credits will hopefully get better over time and match MB v6 credits, that was said week after week, we all know how that ended. Seems the same may happen now, so my hope isn't that high ;)

does anyone ever complain about too much RAC?

Before Aqua shutdown things went a little crazy over there:



I didn't complain though, LOL :)

---

Member of the People Encouraging Niceness In Society club.

download the apps from Crunchers Anonymous, decompress them, copy them to the project directory, and run aimerge or aimergeg to combine the aistub's into an app_info
Claggy

Uh........I'll wait for the new Lunatics installer, being true to their name I'm sure it's being worked on now :). I may try the download and copy if I run out of weekend things to do.

Thanks again, everyone.

download the apps from Crunchers Anonymous, decompress them, copy them to the project directory, and run aimerge or aimergeg to combine the aistub's into an app_info
Claggy

Uh........I'll wait for the new Lunatics installer, being true to their name I'm sure it's being worked on now :). I may try the download and copy if I run out of weekend things to do.

Thanks again, everyone.

Richard stated in the AP7 problems thread that the new installer may be released in the next 48 hours or so.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Richard stated in the AP7 problems thread that the new installer may be released in the next 48 hours or so.

That is very good news, I'm not very happy crunching MB.

---

Richard stated that the new installer may be released in the next 48 hours or so.

Indeed good news, I will work on getting my old workstation back on line and happily crunch S@H 7 until the release. I really didn't want to edit code, it gives me a headache.