UnitedGerbilNation's drive to the Top

[flybywire]

Which f@h client version are you running? You should be running 6.29 found here. Also, have you entered your passkey into the config?

[Gerbil Jedidiah]

Which f@h client version are you running? You should be running 6.29 found here. Also, have you entered your passkey into the config?

I'm getting bonuses on the 6XXX WUs, as you can see in my stats. Shouldn't that mean that I am running 6.29 and my passkey is being recognized?

To answer your question, though, I just donwnloaded and installed 6.29 last week. And I got a passkey. I will double check when I get home to make sure nothing squirrely is going on.

[Ragnar Dan]

My fastest machine has had 3 a3 WU's fail to upload on time or at all. One took 45 extra hours to upload, the other 2 are still trying after I forget how long.

And now the count is 3 failed uploads, and after extended waits for upload attempts, the 6.29 client downloads an a1 WU.

While the server status appears normal and accepting uploads for both of the a3 servers (130.237.232.140 and 171.64.65.54), it also appeared normal during the time period that server 171.67.108.21 was off in the weeds. I imagine the a3 servers are both using the new and 'improved' software, but they may not have yet received the band-aid which was recently applied to the 171.67.108.21 server.

Still, it is perplexing that flybywire is seeing normal a3 activity. Have you both checked your log files to make certain the a3 servers are not giving you the same work unit over and over?

I'm not getting repetition, except for a WU that my machine puked on which I got again about 3 WU's later.

They're definitely screwed up on more than a few of their servers. There was also a period when they were updating user statistics on the hour and doing it, according to them, minutes before the end of the hour. That lasted a few days, and now they're not completing them until almost 15 minutes past the top of the hour. Not that it means much in itself, but it's a pattern of claims and failures that is long established.

[Ragnar Dan]

Changing gears... Am I maxing out the points on the a3 cores now? I'm only getting around 10k PPD. I thought it was supposed to be MUCH more than that?

If you're doing units 6011 - 6025 then the ppd sounds about right. Units 2681 - 2683 are the big bonus units, where you should be netting 20k - 25kppd.

Arrrggghhh, I got another a1 today. Here are my current flags, "-verbosity 9 -bigadv -smp 8".

I don't think I'm getting the 2681 to 2683. IIRC I've been getting 6XXXs, and a mix of a1s. Currently chewing on a 2653.

Try it without -verbosity 9 and see what happens.

[Gerbil Jedidiah]

If you're doing units 6011 - 6025 then the ppd sounds about right. Units 2681 - 2683 are the big bonus units, where you should be netting 20k - 25kppd.

Arrrggghhh, I got another a1 today. Here are my current flags, "-verbosity 9 -bigadv -smp 8".

I don't think I'm getting the 2681 to 2683. IIRC I've been getting 6XXXs, and a mix of a1s. Currently chewing on a 2653.

Try it without -verbosity 9 and see what happens.

OK now I'm down to "-bigadv -smp 8" flags. I'm at 80% on this 2653 WU, so will be a couple of hours before I get another chance to pull an a3. Update will follow.

[MaxTheLimit]

It's nice to see our numbers start to climb again.
Seeing them so low was depressing.
Sorry to hear of the annoyances you fellows have been coming across.

[Gerbil Jedidiah]

It's nice to see our numbers start to climb again.
Seeing them so low was depressing.
Sorry to hear of the annoyances you fellows have been coming across.

Hey MTL! Yes our numbers recently have been pretty sad. There's a new SMP client out now though that makes the annoyances worth the hassle. People are reporting 20K PPD or more from one quad core i7 CPU. So far the best I've done with it is 13k, but I continue to tweak. How are your rigs running? Can you add the new SMP client?

[Ragnar Dan]

Just for fun, I notice all 3 of my GPU's are waiting for new WU's.

Code: Select all

[19:14:23] Connecting to http://171.64.65.71:8080/
[19:14:24] - Couldn't send HTTP request to server
[19:14:24] + Could not connect to Work Server
[19:14:24] - Attempt #7  to get work failed, and no other work to do.
Waiting before retry.


That's the one which has only been waiting for ~10 minutes. Another is > 53 minutes, and the longest one has been waiting ~60 minutes.

[flybywire]

Try it without -verbosity 9 and see what happens.

That shouldn't have any effect on the performance as all that switch does is adjust the level of output in the log file.

Looking at the list of systems runing the 'bigadv' switch here, they're either running under Linux or Mac OS; I don't see see the Windows client listed at all. So it may be that the Windows client cannot run the big bonus units at the moment.

[Ragnar Dan]

I just wanted to make sure that flag wasn't screwing up anything. You can never tell with those guys. As for Windows and -bigadv, I can't remember enough about it, but if it won't run with that, then it should have a different version number.

Edit: And BTW, the GPU's now have WU's after restarting 2 of them, after the assignment server connected them to a working server (171.67.108.11).

[flybywire]

I think the very purpose of version 6.28 and above is to allow the usage of the A3 cores, and not just allow the big bonus units.

[Gerbil Jedidiah]

OK now I'm down to "-bigadv -smp 8" flags. I'm at 80% on this 2653 WU, so will be a couple of hours before I get another chance to pull an a3. Update will follow.

Just downloaded a 6021. *sigh* So based on what FBW stated, I need to install a UNIX VM in order to get the big bonus WUs.

I haven't done that in a couple of years. What VM software is out there that supports 8 threads? Is there an image of UNIX out there that works well for this purpose?

[Flying Fox]

I think the very purpose of version 6.28 and above is to allow the usage of the A3 cores, and not just allow the big bonus units.

I thought the bigadv big bonus units are still running on A2's? Eventually it will be moved to A3's plus the not-so-big ones with the bonus system (like those 602x's).

Just downloaded a 6021. *sigh* So based on what FBW stated, I need to install a UNIX VM in order to get the big bonus WUs.

I haven't done that in a couple of years. What VM software is out there that supports 8 threads? Is there an image of UNIX out there that works well for this purpose?

VMware Player 3. On foldingforum I remember reading someone said they managed to modify notfred's image to do the 8 thread thing but there's also someone who made an image too. The relevant threads may have been linked here so you can try search for it. Failing that, a Google/Bing should get you where you need to go.

[Ragnar Dan]

I just hacked my versions of notfred's ISO to run the new Linux 6.29 client by downloading it using wget, renaming the original fah6 and mpiexec files to something else (though it appears the mpiexec files are identical), and then unpacking and running the contents of the .tgz file. You could do the same, but it's a pain in the rear to keep doing that manually.

Right here is a thread about an apparently modified version of notfred's VM that is supposed to work with -bigadv. They call it the FredNotFred VM. You can be our guinea pig. It says you need VMWare Player 3 to run it, and to make sure the "checkbox for advmethods is turned OFF". And that it uses 5 GB.

[Gerbil Jedidiah]

And that it uses 5 GB.

Uh, that's a lot of GB. I'm not sure I want to dedicate 5GB to a VM.

[flybywire]

I thought the bigadv big bonus units are still running on A2's? Eventually it will be moved to A3's plus the not-so-big ones with the bonus system (like those 602x's).

Yes, that is true, the big bonus units are A2's. The point I was trying to make is that even though the version numbers are the same across the Windows and Linux clients, it doesn't mean that both are able to fold the big bonus units. At least that is what I have surmised taking into account David's experience with the Windows client and the list of systems folding the big bonus units in the link that I had provided.

David, here's a link to show you have to setup the vm for the big bonus. VMware Player only supports 2 processors out of the box, so to speak, and what it boils down to is that you have to edit its configuration file to enable support for more than 2 processors. This not only comes in handy with big bonus vm setup, but also with the 60xx bonus units; I have a q9450 system running a vm with 3 cores folding the 60xx units while the other core is dedicated to the gpu.

[flybywire]

And that it uses 5 GB.

Uh, that's a lot of GB. I'm not sure I want to dedicate 5GB to a VM.

Oh yeah, and the big bonus stuff will eat it all up.

[Flying Fox]

And that it uses 5 GB.

Uh, that's a lot of GB. I'm not sure I want to dedicate 5GB to a VM.

Out of your 1TB storage subsystem? It is ok IMO.

[Ragnar Dan]

And that it uses 5 GB.

Uh, that's a lot of GB. I'm not sure I want to dedicate 5GB to a VM.

Out of your 1TB storage subsystem? It is ok IMO.

That 5 GB is really 5 GiB of RAM, for those who prefer overprecision to a conversational tone.

If it were me, I'd experiment with the amount of memory after a few successful BIGADV WU's are submitted. Maybe you can do reasonably well with less memory dedicated. I haven't looked at it to see what others are saying about it, though there may be discussion about such matters on the folding forum.

[MaxTheLimit]

Hey MTL! Yes our numbers recently have been pretty sad. There's a new SMP client out now though that makes the annoyances worth the hassle. People are reporting 20K PPD or more from one quad core i7 CPU. So far the best I've done with it is 13k, but I continue to tweak. How are your rigs running? Can you add the new SMP client?

Everything seems to be running real well here.
My 8800GT system is still a bit unstable, but the 8800GTSs-s-es....the three 8800GTS cards, are working fine.
I may also load that SMP client on my two fore i5 750 machines. They are gaming rigs though, so I don't promise steady points from them.

[Gerbil Jedidiah]

Decisions, decisions... I just grabbed a Powercolor 5970 on Newegg for $530 after taxes and shipping (open box).

Not sure if I want it or not, lol, but wanted to buy it before it went out of stock. Now I have till in the morning to figure out if I want to keep it.

What do you guys think? It would put a hurt on my folding numbers I imagine, but it would be nice to have the features of the new cards.

[Gerbil Jedidiah]

Decisions, decisions... I just grabbed a Powercolor 5970 on Newegg for $530 after taxes and shipping (open box).

Not sure if I want it or not, lol, but wanted to buy it before it went out of stock. Now I have till in the morning to figure out if I want to keep it.

What do you guys think? It would put a hurt on my folding numbers I imagine, but it would be nice to have the features of the new cards.

I canceled it. If anyone wants a cheap 5970 it should be back on the egg soon. I thought about it, and after putting up with all the noise from a 9800GX2 and a GTX295, then seeing in TR's review that the 5970 also gets pretty noisy, I decided not to bother with it. Truth be told I think I'll get a 5850, 2 if I need it to drive the candy.

[JPinTO]

Here is a VMWare folding guest which works fine for BIGADV and non BIGADV folding. All the instructions are nicely provided: http://www.evga.com/FORUMS/tm.aspx?m=4464

The A1 units are random, and it sucks to have a i7 processing loafing on a low point A1 unit. They are saying that they are just wrapping up the old units.

If you use -verbosity 9 switch, put it at the end of your switches, as in: -smp 4 -advmethods ... -verbosity 9

- JP

[Gerbil Jedidiah]

Bastards tried to hit me with anothe a1 core. SO, I copied my SMP client, changed the maching ID on one, then kicked them both off with "-smp 4 -bigadv - advmethods". Now they are both running a3 cores, although a 6 min per frame instead of 3. At least they are a3. Not sure if I got the a3s based on anything that I did.

Also just launched the SMP client on my wife's Q9450. I wonder if I should kick it off on the Athlon X2? Meh

[Flying Fox]

My 10th WU should be submitted in the afternoon so I will see how this bonus business works.

One other thing I have noticed about the A3's. It seems that the 45nm CPUs (Nehalem family excluded) from both Intel and AMD are neck and neck. In the old days my PDC E6300 @2.8GHz runs circles around the Athlon II X2 250 @3.0GHz. Now HFM.net is predicting that the X2 250 has a slight edge of points (~100ppd) over the E6300. So things are back to "normal" that their clock speeds seem to be comparable. No more fancy compiler optimizations and stuff? I am sure the Nehalem's have skewed the comparison again, but it's nice to see that a 3.0GHz processor is slightly faster than a 2.8GHz one again.

[Gerbil Jedidiah]

My 10th WU should be submitted in the afternoon so I will see how this bonus business works.

One other thing I have noticed about the A3's. It seems that the 45nm CPUs (Nehalem family excluded) from both Intel and AMD are neck and neck. In the old days my PDC E6300 @2.8GHz runs circles around the Athlon II X2 250 @3.0GHz. Now HFM.net is predicting that the X2 250 has a slight edge of points (~100ppd) over the E6300. So things are back to "normal" that their clock speeds seem to be comparable. No more fancy compiler optimizations and stuff? I am sure the Nehalem's have skewed the comparison again, but it's nice to see that a 3.0GHz processor is slightly faster than a 2.8GHz one again.

I have a 780G mobo that supports 140W CPUs. AM2+ that is. Should I replace the athlon x2 with some sort of quad core folding goodness?

[Flying Fox]

One other thing I have noticed about the A3's. It seems that the 45nm CPUs (Nehalem family excluded) from both Intel and AMD are neck and neck. In the old days my PDC E6300 @2.8GHz runs circles around the Athlon II X2 250 @3.0GHz. Now HFM.net is predicting that the X2 250 has a slight edge of points (~100ppd) over the E6300. So things are back to "normal" that their clock speeds seem to be comparable. No more fancy compiler optimizations and stuff? I am sure the Nehalem's have skewed the comparison again, but it's nice to see that a 3.0GHz processor is slightly faster than a 2.8GHz one again.

I have a 780G mobo that supports 140W CPUs. AM2+ that is. Should I replace the athlon x2 with some sort of quad core folding goodness?

This just in: E6300 - 1694.9ppd (min) / 1689ppd (avg); X2 250 - 1921.2ppd (min) / 1871.3ppd (avg), according to HFM.net on project 6023 running Xubuntu/Ubuntu VMs. I am not sure how much a quad can give you, but for me, a 9600GT/GT240 doinng 3000ppd will own the SMP client still. So if you have a GTX 260 or something that may still beat an X4. Not sure about the Phenom II X4 but an i5/i7 should still be in the lead, at a higher price premium for the platform altogether. Of course, ppd per watt may be a different story. That's why I would like to know the stats I was asking for a few posts up. If getting ~1000ppd more means double the power consumption, it may not be that worth it for some.

Once they move everything over to the thread-based A3 and get rid of all that mpich/deino nonsense, may be they will make WinSMP runnable as a service and all would be well again.

[Gerbil Jedidiah]

That's why I would like to know the stats I was asking for a few posts up. If getting ~1000ppd more means double the power consumption, it may not be that worth it for some.

Once they move everything over to the thread-based A3 and get rid of all that mpich/deino nonsense, may be they will make WinSMP runnable as a service and all would be well again.

My GT240 should come in today. I'll be putting it in the 780G rig and will let you know how much it bumps the wattage. I forget if that mobo has two PCI-E slots ( I think no) but if it does 2 GT240s I'm betting would be a better power per watt solution than running the SMP client on my athlon X2 plus one GT240. It'll be interesting to find out.

After moving over to TWO SMP clients in win 7 last night with "-smp 4 -bigadv -advmethods -verbosity 9" as my flags I've gotten 4 a3s in a row. Maybe this is the magic setting, I dunno, but I did this so that if I do get a couple of a1s I'll still be maxing out all 8 threads.

[Gerbil Jedidiah]

I'm not sure if this will be taken overall well, or not so well... but... I'm thinking about folding under my own username. Since UGN's numbers have been so erratic lately, I would like to be able to track my own results. I switched my machines over last night. I'm not completely sold on the idea yet, but I am mulling it over and will fold under Gerbil_Jedidiah for a while irregardless just to see what kind of numbers I can put up.

No matter what, I'll stay part of the UGN crowd, and will most likely offer up one rig or card to support UGN even if I do decide to keep most of my stuff under my own name.

[jeffry55]

I'm not sure if this will be taken overall well, or not so well... but... I'm thinking about folding under my own username. Since UGN's numbers have been so erratic lately, I would like to be able to track my own results. I switched my machines over last night. I'm not completely sold on the idea yet, but I am mulling it over and will fold under Gerbil_Jedidiah for a while irregardless just to see what kind of numbers I can put up.

No matter what, I'll stay part of the UGN crowd, and will most likely offer up one rig or card to support UGN even if I do decide to keep most of my stuff under my own name.

UGN 2nd fiddle? I need a hug.