Always Getting Crappy SMP WUs on F@H; Could Something Be Wrong?

[LOL_Wut_Axel]

I'm not sure about the "smaller". Systems with more ram could hold larger chunks and are generally newer, faster systems with more cores that will return WUs quicker. So the projects most important to them will go to these systems and have very short deadlines to get the results back soonest so they can send the next WU out and get it back faster (remember that the next in line WU is created from returned results). If the line is 100 WUs long, quick returns makes a huge difference.
Return in one day versus return in five days is 100 vs. 500 days to find a possible cure. (Uniprocessor client WUs can have deadlines on the order of 30 or 60 days or more?) So as soon as a short deadline WU is returned, a new WU is created and sent out to the next qualified computer needing another WU. If a 16-core system needs another WU and there is no "special" WU to send the server sends something else. It works the same for your 8-core system.
Also, there is priority set for each project so some fast 8-core systems will get other projects. Imagine a researcher begging the priority setter to give his project higher priority than another researcher!?! So the server will send this one for 10 minutes & and the next for 5 minutes & the next for 20 minutes. What you get depends on where in the priority "steps" the server is.


Same for me!
After 7+ years of folding (4+ years doing a daily F@H stats thread) I decided to explore BOINC a little more. Now I am having fun trying to get many projects over different Milestones.

I'm interested. Do you have a link?

I've never run any of the BOINC projects/clients, though they do seem interesting and I've read about them. I've heard they use a lot of internet bandwidth, though, right? My connection right now is less-than-ideal.

 

[petrusbroder]

Depends on project. Some projects use WUs and results in the size of 10 - 50 kBytes (e.g. some PrimeGrid subprojects), other 10 - 15 Mbytes (often protein folding projects).
You decide which project to run on BOINC - depending on your interest, the scientific value as you see it, bandwidth, crunching time ...

 

[LOL_Wut_Axel]

Depends on project. Some projects use WUs and results in the size of 10 - 50 kBytes (e.g. some PrimeGrid subprojects), other 10 - 15 Mbytes (often protein folding projects).
You decide which project to run on BOINC - depending on your interest, the scientific value as you see it, bandwidth, crunching time ...

I'll try BOINC; F@h likes to wildly fluctuate. Right now I just got a 7005 project in smp4 when I was gonna call it quits. Not all of the WUs are absolute crap, so I'll keep going at it. I'll run smp4 on the day and smp8 at night.

Which BOINC projects don't consume a ton of bandwidth? Right now, unfortunately, I'm tied to a 2Mbps connection until I get 4Mbps installed some two weeks later, and even then that's not fast. Unfortunately, 16Mbps here costs $90 a month.

 

[VirtualLarry]

Unfortunately, 16Mbps here costs $90 a month.

That's not completely horrible. My friend pays $70/mo for 16Mbit down with Comcast.

 

[GLeeM]

I'm interested. Do you have a link?

If you were asking about the stats threads - I quit about 2 years ago, 30+ minutes per day to do them was wearing.

As for BOINC and bandwidth - I have satellite internet with 512 Kbps down and 128 Kbps up so I don't think you will have a problem. I also have a total per 30 day limit of 7500 MB down and 2300 MB upload - this makes things difficult!
Bandwidth doesn't really matter because you can have many hours or days worth of WUs qued up.

If you want to stick with med projects there are for CPU, World Community Grid has many choices with most small up/down and Correlizer also has small up/downloads. Roseta and Simap (Simap only has WUs at beginning of month) has lots of TAers, but has larger up/down. For Nvidia GPU there is GPUGRID - large up/down, but like 10 hour WUs.
You could still crunch F@H on GPU and BOINC on CPU.

 

[GFA]

i accidentaly placed a wrong username in FAH6 win Smp
what should i do?

 

[GLeeM]

Hello GFA and welcome to the forum

Just edit the username and restart.

TeAm number here is 198

Oh wait ... is FAH6 the one where you have to configure to change the name? If so then you will have to start the app with the "config" or "configonly" flag. Yikes ... it's been too long since I've done it and can't remember. Someone else will be along to help.

PS: you should have started a new thread

 

[DooKey]

i accidentaly placed a wrong username in FAH6 win Smp
what should i do?

Run the FAH6.exe with the -configonly flag and reset the name.