Collatz Conjecture Information

[GLeeM]

Last summer I started crunching Collatz Conjecture again and discovered that they created a new type of WU called "solo_collatz".

These WUs validate themselves ... you don't wait for a wingman.

The best part is that they get double points of the other type.

If you are running Collatz, go into your Preferences and unselect the regular and the mini type WUs and select only "solo_collatz" to double your ppd.

You can also "tweak" how these WUs run by adjusting numbers in the file: "solo_collatz_4.07_windows_x86_64__cuda50.config" to increase production even more!!
The following is from my .config file:
verbose=0
items_per_kernel=18
kernels_per_reduction=9
threads=8
sleep=1

My GTX560 Ti 448 core does best running one WU at a time.

 

[petrusbroder]

Thanks for the info - nice, very nice!

 

[GLeeM]

Bump!

Icecold and zzuupp both need to see this information as they are still running the old type WUs.

 

[Icecold]

Thanks for the heads up GLeeM. I saw this post before, but didn't realize it applied to me. I had to take down my Collatz machine for a couple of days, but will check that out when I bring it back online this weekend. A 6950 was already doing 150-200k ppd so it's crazy to think that this would double it.

I may switch it to MilkyWay@home anyways but last time I tried to it was failing on all WU's pretty much immediately. Thanks again for the info.

 

[faxon]

Thanks for the heads up GLeeM. I saw this post before, but didn't realize it applied to me. I had to take down my Collatz machine for a couple of days, but will check that out when I bring it back online this weekend. A 6950 was already doing 150-200k ppd so it's crazy to think that this would double it.

I may switch it to MilkyWay@home anyways but last time I tried to it was failing on all WU's pretty much immediately. Thanks again for the info.

make sure you're running the latest BOINC and try 13.4 catalyst drivers for milkyway@home if you're having issues with your 6950. ran into similar issues with some failing work units but for those having issues apparently 13.4 fixes it for most

 

[GLeeM]

A 6950 was already doing 150-200k ppd so it's crazy to think that this would double it.

A 6870 is doing about 300,000 ppd. Is a 6950 much better?

 

[zzuupp]

Bump!

Icecold and zzuupp both need to see this information as they are still running the old type WUs.

Thanks for thinking about me. I had been running a mix for WUPROP hours.
My .config:

items_per_kernel=19
kernels_per_reduction=9
threads=9
sleep=1

 

[GLeeM]

My .config: items_per_kernel=19 kernels_per_reduction=9 threads=9 sleep=1

What kind of card do this config do best on?

 

[zzuupp]

What kind of card do this config do best on?

It works really well on the gtx670, less on the 560ti (regular version).

I set verbose to 1; I might play around a bit this weekend.