New Notfred diskless version out

Come join the... uh... er... fold.

Moderators: just brew it!, farmpuma

Re: New Notfred diskless version out

Postposted on Wed Feb 18, 2009 3:59 pm

Probably isn't as hard as I had it... Seems thats how it goes for me.
Intel Q6600 g0 @ 3.2Ghz Turniq 120 Heatsink BFG 260 OC MaxCore
2x2gb OCZ Platinums @ 1:1 XFX 680i motherboard Corsair 520hx Antec 900


Image
Patonb
Gerbil
 
Posts: 37
Joined: Sat Jan 24, 2009 12:20 pm
Location: Ottawa, Canada there eh

Re: New Notfred diskless version out

Postposted on Thu Feb 19, 2009 3:37 pm

Hey notfred,
Just as an FYI, I replaced all the diskless files from the diskless.zip on my tftp server and rebooted one comp, came back up and continued the WU no problem. So I rebooted them all, 6 out of my 8 computers have to download new WU's for some reason so they all started back up at 0%. Not sure why that happened, or even if its anything to do with your new version, just thought I would relate my experience in case someone has some clients at %90 done or something (like I did) and could potentially loose their work. I should have waited till each one downloaded a new WU to reboot them hehe.

Looks like its going to be a light day for me again. :)
Image
Shinare
Gerbil XP
 
Posts: 352
Joined: Wed Jul 06, 2005 12:48 pm

Re: New Notfred diskless version out

Postposted on Thu Feb 19, 2009 3:56 pm

I know Stanford just came out with a new version of the a2 core for SMP folding because the old version could corrupt the checkpoints. If you'd not rebooted those recently, it could have had the old one and corrupted checkpoints.
notfred
Grand Gerbil Poohbah
 
Posts: 3731
Joined: Tue Aug 10, 2004 10:10 am
Location: Ottawa, Canada

Re: New Notfred diskless version out

Postposted on Thu Feb 19, 2009 4:51 pm

Thats probably it then, I dont think they had been rebooted in a month.
Image
Shinare
Gerbil XP
 
Posts: 352
Joined: Wed Jul 06, 2005 12:48 pm

Re: New Notfred diskless version out

Postposted on Thu Feb 19, 2009 7:59 pm

Support -smp 8 option.

I'm having some trouble with this on the USB stick: If I boot from the USB stick on an 8-core system, it tries to run two 4-core instances. I have to boot from the CD-ROM in order to get all 8 cores to work on one WU.
petcherd
Gerbil
 
Posts: 32
Joined: Tue Jun 17, 2008 12:08 pm

Re: New Notfred diskless version out

Postposted on Fri Feb 20, 2009 9:50 am

Hey notfred,
its not saving the periodic backups like it was with the older version. I just checked the tftp log and theres no longer file transfer activity between it and the nodes, just the initial kernal stuff.
Image
Shinare
Gerbil XP
 
Posts: 352
Joined: Wed Jul 06, 2005 12:48 pm

Re: New Notfred diskless version out

Postposted on Fri Feb 20, 2009 10:07 pm

Mines backing up fine

It's vmwarre though
Intel Q6600 g0 @ 3.2Ghz Turniq 120 Heatsink BFG 260 OC MaxCore
2x2gb OCZ Platinums @ 1:1 XFX 680i motherboard Corsair 520hx Antec 900


Image
Patonb
Gerbil
 
Posts: 37
Joined: Sat Jan 24, 2009 12:20 pm
Location: Ottawa, Canada there eh

Re: New Notfred diskless version out

Postposted on Sat Feb 21, 2009 9:23 pm

petcherd wrote:If I boot from the USB stick on an 8-core system, it tries to run two 4-core instances.
I just checked this and it works fine for me. Remember to edit the syslinux.cfg file on the USB stick and set SMPCPUS=8 on all 4 lines (the one that really counts is in the LABEL fold64 section).

Shinare wrote:its not saving the periodic backups like it was with the older version. I just checked the tftp log and theres no longer file transfer activity between it and the nodes, just the initial kernal stuff.
I checked this as well and it works fine for me again. Make sure you have the files created and writeable by everyone in your tftp directory to allow it to do uploads.
notfred
Grand Gerbil Poohbah
 
Posts: 3731
Joined: Tue Aug 10, 2004 10:10 am
Location: Ottawa, Canada

Re: New Notfred diskless version out

Postposted on Mon Feb 23, 2009 12:51 am

notfred wrote:
Shinare wrote:its not saving the periodic backups like it was with the older version. I just checked the tftp log and theres no longer file transfer activity between it and the nodes, just the initial kernal stuff.
I checked this as well and it works fine for me again. Make sure you have the files created and writeable by everyone in your tftp directory to allow it to do uploads.



Hey sorry i didnt get back here in time to keep you from unnecessary work. It was tftp server being wonky for some reason. There was an IP in the DHCP tab that shouldn't have been there since I have static DHCP IPs via MAC. Anyway, after I noticed it in there, I deleted the IP listing and at least 8 transfer boxes appeared at that same instant. It was probably tftp server locked up or something.
Image
Shinare
Gerbil XP
 
Posts: 352
Joined: Wed Jul 06, 2005 12:48 pm

Re: New Notfred diskless version out

Postposted on Mon Feb 23, 2009 9:23 am

That's OK, it didn't take too long to check, glad to hear you got it working. Fold on!
notfred
Grand Gerbil Poohbah
 
Posts: 3731
Joined: Tue Aug 10, 2004 10:10 am
Location: Ottawa, Canada

Re: New Notfred diskless version out

Postposted on Mon Feb 23, 2009 12:53 pm

Seems to be working fine now on the default 640mb memory setting with A_2 core work units. With the previous version I had to increase the memory to 1024mb. One thing I noticed is when I need to reboot my PC I get some error messages in the VMWare Player window even though the work unit has been restored and is folding fine.

Image
slugbug
Gerbil
 
Posts: 15
Joined: Wed Aug 20, 2008 12:29 pm

Re: New Notfred diskless version out

Postposted on Wed Feb 25, 2009 7:35 pm

^^ I reboot the when i get that, and it'll take care of it.
Intel Q6600 g0 @ 3.2Ghz Turniq 120 Heatsink BFG 260 OC MaxCore
2x2gb OCZ Platinums @ 1:1 XFX 680i motherboard Corsair 520hx Antec 900


Image
Patonb
Gerbil
 
Posts: 37
Joined: Sat Jan 24, 2009 12:20 pm
Location: Ottawa, Canada there eh

Re: New Notfred diskless version out

Postposted on Thu Feb 26, 2009 2:26 pm

notfred,

How do I delete a WU in process??? And the FAH download a new WU.

UPDATED: I delete the WU in process with commands below:

# rm -f /etc/folding/1/work/*_01*
# mount -n -t vfat /dev/hda1 /hda
# rm -f /hda/folding/1/work/*_01*
# umount /hda
# reboot -f

The VM restart, FAH not found work files and donwload a new WU.

Could have an automated way to do this with the Diskless (in the Webpage).

Thank you.
I'm brasilian and my English is bad, very bad.
ethomaz
Gerbil
 
Posts: 25
Joined: Wed Aug 20, 2008 9:00 am

Re: New Notfred diskless version out

Postposted on Fri Feb 27, 2009 11:06 am

Sorry if this has been covered before, but how do I monitor the CPU temps while running the USB flash drive version? Is there some sort of add-on utility that I can install on the USB flash drive? I am a Linux newbie. Thanks.
Edward2
Gerbil In Training
 
Posts: 3
Joined: Fri Feb 27, 2009 11:00 am

Re: New Notfred diskless version out

Postposted on Fri Feb 27, 2009 1:14 pm

Edward2 wrote:how do I monitor the CPU temps while running the USB flash drive version?
With your finger :D

Seriously though, there isn't any software to do voltage / temperature measurements as they are different on every board. You may have a /proc/acpi/thermal_zone/THM0/temperature file that you could periodically check - I think I included the thermal zone in the ACPI options when I built it.
notfred
Grand Gerbil Poohbah
 
Posts: 3731
Joined: Tue Aug 10, 2004 10:10 am
Location: Ottawa, Canada

Re: New Notfred diskless version out

Postposted on Fri Feb 27, 2009 1:15 pm

notfred wrote:
Edward2 wrote:how do I monitor the CPU temps while running the USB flash drive version?
With your finger :D


DOH You beat me to it notfred!
Image
Shinare
Gerbil XP
 
Posts: 352
Joined: Wed Jul 06, 2005 12:48 pm

Re: New Notfred diskless version out

Postposted on Fri Feb 27, 2009 2:10 pm

I checked the /proc/acpi/thermal_zone folder and it is empty. I don't know exactly how the Windows based programs work (such as CoreTemp and RealTemp), but I was hoping that there would be a Linux equivalent. Thanks anyway. Great program as is.

Edit - I see there is a program called lm-sensors for Linux which allows you to read temperature/voltage information from your motherboard via the SMBus. I realize that the Folding USB flash drive is a highly compressed version, but you might consider adding this feature in the future if anyone else is interested. Maybe it's just me.
Edward2
Gerbil In Training
 
Posts: 3
Joined: Fri Feb 27, 2009 11:00 am

Re: New Notfred diskless version out

Postposted on Fri Feb 27, 2009 4:16 pm

lm-sensors and the SMBus is the problem I was talking about of everyone having a different setup.

CoreTemp / RealTemp use the DTS which is supported by the coretemp module under Linux. It normally dumps output in to /sys/devices/platform/coretemp.0/temp1_input but I don't think I have it setup. I can probably get it going and put a little "sensors" command in there that will start it running.
notfred
Grand Gerbil Poohbah
 
Posts: 3731
Joined: Tue Aug 10, 2004 10:10 am
Location: Ottawa, Canada

Re: New Notfred diskless version out

Postposted on Fri Feb 27, 2009 4:26 pm

Another weird problem surfaced. I have two instances of this running on one of my Quad cores. #1 shows an out of memory error but is still folding fine and completing work units. #2 appears to have crapped out but task manager shows the cpu load is still at 94-100%. I can't even access the setup page on #2 though. Shutting down the clients virtual player and reloading it has no affect. Any ideas?
slugbug
Gerbil
 
Posts: 15
Joined: Wed Aug 20, 2008 12:29 pm

Re: New Notfred diskless version out

Postposted on Fri Feb 27, 2009 4:50 pm

Edward2 wrote:... Maybe it's just me.
Nope, it's not just you. The first application I launch at start-up is some form of temperature monitoring. It does no good to be folding at 100% if your cpu is throttling to under 50%. Dust bunnies never sleep.

You could go mechanical with a fan control and/or temp monitoring panel. And some cases, Aspire/Apevia come to mind, have a built-in temp monitoring panel.
Image Image
.* * M-51 * *. .The Whirlpool Galaxy.
farmpuma
Minister of Gerbil Affairs
Silver subscriber
 
 
Posts: 2315
Joined: Mon Mar 22, 2004 12:33 am
Location: Soybean field, IN, USA, Earth .. just a bit south of John .. err .... Fart Wayne, Indiana

Re: New Notfred diskless version out

Postposted on Sun Mar 01, 2009 4:26 pm

OK, for those of you without a thermally calibrated index finger, a new version is out. It will report temps on Intel Core processors and AMD K8 (N.B. NOT K10 i.e. Phenom - that's too new, try back in a couple of months once k10temp is in the Linux kernel mainline).
notfred
Grand Gerbil Poohbah
 
Posts: 3731
Joined: Tue Aug 10, 2004 10:10 am
Location: Ottawa, Canada

Re: New Notfred diskless version out

Postposted on Sun Mar 01, 2009 4:27 pm

slugbug wrote:Another weird problem surfaced. I have two instances of this running on one of my Quad cores. #1 shows an out of memory error but is still folding fine and completing work units. #2 appears to have crapped out but task manager shows the cpu load is still at 94-100%. I can't even access the setup page on #2 though. Shutting down the clients virtual player and reloading it has no affect. Any ideas?
That's odd. I'd try giving it a reboot at this point (leave the VM up, reboot from within the VM via the link on the webpage).
notfred
Grand Gerbil Poohbah
 
Posts: 3731
Joined: Tue Aug 10, 2004 10:10 am
Location: Ottawa, Canada

Re: New Notfred diskless version out

Postposted on Sun Mar 01, 2009 8:50 pm

notfred wrote:OK, for those of you without a thermally calibrated index finger, a new version is out. It will report temps on Intel Core processors and AMD K8 (N.B. NOT K10 i.e. Phenom - that's too new, try back in a couple of months once k10temp is in the Linux kernel mainline).


Thank you, thank you, thank you. :D I'm running this on a spare Dell computer at work, and I really want to see what the temps are.
Edward2
Gerbil In Training
 
Posts: 3
Joined: Fri Feb 27, 2009 11:00 am

Re: New Notfred diskless version out

Postposted on Mon Mar 02, 2009 2:16 pm

I copied over the new diskless folder and see the temperature link now. Love it!

Very cool. My hottest folder is: Core 0 70 Core 1 70. These are all SFF PC's stacked on one another, by the way. Is that too high?
Image
Shinare
Gerbil XP
 
Posts: 352
Joined: Wed Jul 06, 2005 12:48 pm

Re: New Notfred diskless version out

Postposted on Mon Mar 02, 2009 3:56 pm

Welp, all of them have been rebooted with the new client and it looks like all of them are reporting in the low 70's after running for a while. That sounds really hot, is this a problem? All of them are in the 70's except one which is reporting 54-56 for each core. Thats odd too since all the others are so high. *shrug*

Perhapse knowing the temperature is TMI, now I'm going to be worried about it, heh.
Image
Shinare
Gerbil XP
 
Posts: 352
Joined: Wed Jul 06, 2005 12:48 pm

Re: New Notfred diskless version out

Postposted on Mon Mar 02, 2009 4:34 pm

70C is pretty toasty, although most folding GPUs will run that high unless their fan is cranked up manually and C2D CPUs don't start to throttle until they hit 90C. It's also possible that linux is reading core temp instead of the chip temp which most motherboard monitoring software reads and which always reports a lower temp.

Is the system which is reading in the mid 50s running the same work unit as the others? Is it on the bottom of the stack and sitting on the floor?
Image Image
.* * M-51 * *. .The Whirlpool Galaxy.
farmpuma
Minister of Gerbil Affairs
Silver subscriber
 
 
Posts: 2315
Joined: Mon Mar 22, 2004 12:33 am
Location: Soybean field, IN, USA, Earth .. just a bit south of John .. err .... Fart Wayne, Indiana

Re: New Notfred diskless version out

Postposted on Mon Mar 02, 2009 4:43 pm

farmpuma wrote:70C is pretty toasty, although most folding GPUs will run that high unless their fan is cranked up manually and C2D CPUs don't start to throttle until they hit 90C. It's also possible that linux is reading core temp instead of the chip temp which most motherboard monitoring software reads and which always reports a lower temp.

Is the system which is reading in the mid 50s running the same work unit as the others? Is it on the bottom of the stack and sitting on the floor?


I've got two stacks of computers, one of 6 and the other of three (the bottom of the three is the "server"), all sitting on a shelf about halfway up the wall. The one running in the 50's is the top one on the stack of three. All the others are the 70's. The WU is 5102 which none else are working on.

I would imagine that they probably would run a little warm being small thin, SFF cases. Was just worried that 70's might be too warm.
Image
Shinare
Gerbil XP
 
Posts: 352
Joined: Wed Jul 06, 2005 12:48 pm

Re: New Notfred diskless version out

Postposted on Mon Mar 02, 2009 5:30 pm

It's core temp, straight from the processor rather than via any motherboard monitoring stuff.

70C is pretty toasty, it's not going to cause any actual problems but you are pretty close to the limit and have very little room if anything else happens. You can see the actual limits for that sensor by looking in /sys/devices/platform/coretemp.0 temp1_max is when it starts thermal throttling, temp1_crit is when it is cooked (both in milli-centigrade) and temp1_crit_alarm is a flag of whether it has been cooked. My Q6600 is running around 60C on each core with a max of 82C and crit of 100C, but that's in an Antec NSK case.

In other news:
2 March 09: Add support for passkey.
notfred
Grand Gerbil Poohbah
 
Posts: 3731
Joined: Tue Aug 10, 2004 10:10 am
Location: Ottawa, Canada

Re: New Notfred diskless version out

Postposted on Tue Mar 03, 2009 5:37 pm

Welp, it just turned in it's Wu and started another one and its different type, but its still runnin at 57C. Very odd, identical computers in every regard, but one is 57C and the rest are 70! Weird.
Image
Shinare
Gerbil XP
 
Posts: 352
Joined: Wed Jul 06, 2005 12:48 pm

Re: New Notfred diskless version out

Postposted on Tue Mar 03, 2009 10:39 pm

Gotta be the position of the box.
Intel Q6600 g0 @ 3.2Ghz Turniq 120 Heatsink BFG 260 OC MaxCore
2x2gb OCZ Platinums @ 1:1 XFX 680i motherboard Corsair 520hx Antec 900


Image
Patonb
Gerbil
 
Posts: 37
Joined: Sat Jan 24, 2009 12:20 pm
Location: Ottawa, Canada there eh

PreviousNext

Return to TR Distributed Computing Effort

Who is online

Users browsing this forum: No registered users and 2 guests