The Thing

[dfonda]

When you install choose the option that install's the updates as it installs the OS I think its a little faster. If that's even an option on 10.10? I went with 11.10, chose that option and it was about 25Min.

I always have a hard time wading through those advanced settings on the Ubuntu install's...I finally gave up and went with the auto, since I knew it wasn't going to matter in my case. That would be a nice project to take screenshots of the steps for the advanced HD install on Ubuntu.^Hint

 

[Norcalsteve]

Ok, After I accidently JUNKED a 100% p6903 this morning (dont ask, I didnt have my coffee yet!), I reinstalled my entire set up on EXT3. I actually followed [H]'s install guides (here) to a T. No screen shots nessiscary... just make sure you dont look over the "EDITS" and "UPDATES"... i.e. DO NOT even bother to run/install updates, even when installing on EXT3, as the updated OS can still cause a file writing issue, i.e. after compresion of a W/U, the "done," and the "SHUTTING DOWN CORE"

Any who... got me a p6901 this time... Not familiar with the TPF on 6174's crunching p6901's, but does 6:30 sound ok? PPD is sitting at about 300K (i like the 6903's better! ~450K)... the total turn in will be smaller as I couldnt start the W/U right away, as I had to download it, and a A5 Core to install the Kraken, make some tweaks, then start the client.

 

[dfonda]

Oh man sorry about the Wu. I thought my night/morning went bad! Glad your back at it. Did you leave Numa enabled even though we are in Linux/AMD? I left it enabled. I remember Torin, I think saying he needed it disabled. he was on SR2 though. I picked up a 6904@215K(No kraken, see thread )

 

[Norcalsteve]

I didnt see a setting for numa, but I did ensure PowerNow (for the temp monitoring client I am using) and Node Interleaving (under the memory settings) was disabled. I think Numa is for the dual+ intel boards. unless I am wrong and someone would like to point me to the correct bios setting (EVERY sub menu was searched, and I did not see "numa" on my AMD SuperMicro board)

 

[dfonda]

I didnt see a setting for numa, but I did ensure PowerNow (for the temp monitoring client I am using) and Node Interleaving (under the memory settings) was disabled. I think Numa is for the dual+ intel boards. unless I am wrong and someone would like to point me to the correct bios setting (EVERY sub menu was searched, and I did not see "numa" on my AMD SuperMicro board)

I searched every where and didn't see the "Node Interleaving" I should have taken a Pic of the Numa spot. It had to be in advanced near the bottom. I only wanted to shut the rig down once since every second counts. I'll get a Pic ASAP. Its enabled by default.

We do have the same Mobo right? H8QGL-if+ ?Lets take some Pics of our screens the next time we are in there.

If Numa is for Intel I wonder why I even have it in my BIOS?

as a side note of trivia Numa was the name of Lion, in the Ape/English dictionary from The Tarzan books written by Edgar Rice Burrughs. The new John Carter movie coming out in march, was also written by Burroughs. Around 1912 or so. Numa is also the name of the organization in the Clive Cussler books. Sorry I digress...but everytime I see that name I recall my child hood.

 

[Norcalsteve]

Looks as if the default settings dont need to be changed then... if you say Numa is enabled by default, and I seen the Node Interleaving is Disabled by default, and we have the same board... we are both good to go then. Just the PowerNow needs to be turned off if you installed Turion Power Control for monitoring temps...

 

[Adak]

If you turn off PowerNow, and choose the Best Optimization setting, then you should be the set with the best settings for your board and BIOS. I tested several BIOS options, and none of them were better than using the BIOS's own "Best" settings.

For Tanker, NUMA was on by default, and in my tests, NUMA allowed shorter tpf.

Numa is not just for Intel - it's used even in Super Computers, depending on the architecture of the system.

 

[dfonda]

CD? lol, yeah, I actually wanted to ask you guys about that... I seen that everything was working fine, so I didnt bother with the CD. What is this "sound" that you speak of? Actually, I dont have a spot on the board that my front pannel audio can hook into, and I dont see anything on the back. Not that I need the sound, but in the future, If I wanted to test that "dual" mode I was talking about, It would be nice to have sound for the VM Windows install.

I think the drivers for sound and vid are Windows only, when I get sound working. I'll post up. I am going to try a speaker on the "speaker" pins, see if it transmits more then beeps.

Sort of useless if I so much as look at the screen the fah slows down. We can't have that.

 

[dfonda]

Looks as if the default settings dont need to be changed then... if you say Numa is enabled by default, and I seen the Node Interleaving is Disabled by default, and we have the same board... we are both good to go then. Just the PowerNow needs to be turned off if you installed Turion Power Control for monitoring temps...

No I just used lm sensors, 2 lines of code and it was good. The SM temp monitor on the CD has 10 times more commands then Kraken and 4 times as complicated. So that was a no go.

Thanks Adak I have read up on Numa and node interleaving this morning, its very interesting and built just for what we are doing, multi core multi threading.

 

[harlam357]

Guess I need to go look at my bios settings... all I can recall is loading optimized defaults and running with it.

 

[dfonda]

Guess I need to go look at my bios settings... all I can recall is loading optimized defaults and running with it.

and don't forget the CD Quite a few things on there none of us would ever need.

 

[Norcalsteve]

I'd probably install the display driver if i got bored enough... What kinda GPUs are on these boards anyways? never heard of Matrox...

 

[dfonda]

I'd probably install the display driver if i got bored enough... What kinda GPUs are on these boards anyways? never heard of Matrox...

Drivers are for Windows Steve, Ubuntu installed the vid, and sound drivers on install. Other then the Manual,which is kinda sketchy, I have not found anything useful.

I was just giving H the business since he was busting my chops for looking at it.

 

[Norcalsteve]

I kinda gatherd after i hit submit... Linux is good about popping up the "display driver install" thing when It doesnt have the driver... I just like to see myself type

After seeing the thread Leo just posted, I am stoked to see the OC guide for our board soon... I just may! I am glad I didnt go with the Tyan board and had the patience to wait for the SM board.

 

[dfonda]

Yeah, I owe you one for getting us both on the Supermicro, looks like it may pay dividends for our anxious waiting time last month.

 

[Norcalsteve]

WTF (and not 'way to fold') happened here?!?! anyone seen this before? I also noticed the time until i CTRL+C'd the client, the time exceeded where I should be for a step to complete...

[01:54:06] Project: 6903 (Run 6, Clone 13, Gen 57)
[01:54:06] 
[01:54:06] Assembly optimizations on if available.
[01:54:06] Entering M.D.
[01:54:15] Mapping NT from 48 to 48 
[01:54:20] Completed 0 out of 250000 steps  (0%)
[02:11:16] ng M.D.
[02:11:22] Using Gromacs checkpoints
[02:11:30] Mapping NT from 48 to 48 
[02:11:48] fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.Resuming from checkpoint
[02:11:48] fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.mdrun returned 3
[02:11:48] Gromacs detected an invalid checkpoint.  RestartingfcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.fcSaveRestoreState: I/O failed dir=0, var=0000000001E107F0, varsize=21120
[02:11:48] Can't restore state.
[02:11:55] Folding@home Core Shutdown: UNKNOWN_ERROR
[02:11:55] CoreStatus = 62 (98)
[02:11:55] + Restarting core (settings changed) 
[02:11:55] 
[02:11:55] + Processing work unit
[02:11:55] Core required: FahCore_a5.exe
[02:11:55] Core found.
[02:11:55] Working on queue slot 02 [March 4 02:11:55 UTC]
[02:11:55] + Working ...
[02:11:55] - Calling './FahCore_a5.exe -dir work/ -nice 19 -suffix 02 -np 48 -checkpoint 15 -notermcheck -verbose -lifeline 5095 -version 634'

[02:11:55] 
[02:11:55] *------------------------------*
[02:11:55] Folding@Home Gromacs SMP Core
[02:11:55] Version 2.27 (Thu Feb 10 09:46:40 PST 2011)
[02:11:55] 
[02:11:55] Preparing to commence simulation
[02:11:55] - Looking at optimizations...
[02:11:55] - Not checking prior termination.
[02:12:02] - Expanded 57247929 -> 71846524 (decompressed 50.4 percent)
[02:12:02] Called DecompressByteArray: compressed_data_size=57247929 data_size=71846524, decompressed_data_size=71846524 diff=0
[02:12:03] - Digital signature verified
[02:12:03] 
[02:12:03] Project: 6903 (Run 6, Clone 13, Gen 57)
[02:12:03] 
[02:12:03] Assembly optimizations on if available.
[02:12:03] Entering M.D.
[02:12:12] Mapping NT from 48 to 48 
[02:12:17] Completed 0 out of 250000 steps  (0%)

any thoughts? You think Kracken/Langoust or something more could cause this?

 

[Adak]

Yes.

Tear mentions it in his post - I was testing different settings (NUMA on, NUMA off, etc.), in the BIOS, and in the starting and stopping, Tear's "baby's", got stuck in the wrong client (I am guessing), anyway, I was borked big time.

After fussing around, to no good effect, I did a clean install of Ubuntu again, and did NOT re-install Tear's two programs. I just don't have the Linux-mojo to fix these programs, when they go sideways.

The way I look at it is, during any testing phase, the Kraken and Langouste programs are not to be used. Use them when you have finished your other testing, and just want to fold with the rig.

 

[Norcalsteve]

seems to be working fine now... I decided to restart and check BIOS settings, and load "optimal" defaults like you did. I finally found the NUMA setting too... enabled as default...

I remember you saying something about Server boards take a long time to "post" but there have been restarts that sit there for 15 minutes (running, but no screen or beeps), but when I hit the reset button during that phase, it posts with in a 15 seconds... Is that normal? I can always tell that it will post normal from a restart by the network cable... there is a green and amber light... the amber is always on... when I restart, the amber light blinks, then will go out, and come back on with the green light, and post... But, when it hangs there, it is just the amber light either steady or flashing and just sits there...... just dont seem right. I did a restart a few days ago prior to dinner, and when we were done, it was still sitting there!

as for for the power lights, i see the "heart beat" flashing as it should, and the steady green over by the 24 and 8 pin power connectors.

 

[Adak]

By a long time to POST (POST = Power On Self Test), I meant a *relatively* long time, not longer than 5 minutes.

Little anecdote - when I first fired up Tanker, I got no display at all (default was apparently no display for boot up messages in the BIOS).

So after about 30 seconds, I'm thinking there had been a failure in POST, somewhere. The psu was also quiet as a tomb. Hmmm... I reset it.

Only to have it repeat with no display! This went on until the third time, I was more patient, and saw the first Supermicr* splash screen!

Oh!

This is the beep code post errors, that our mobo's have. (BIOS's are altered at times to fit certain boards, so it may be slightly different, but should be very close to this.)



After the POST, if your board has a remote monitoring feature (mine does), then that will take a good 1-3 minutes to detect and set up. This only has to be done on cold boot - reboots (warm boots), will not cause the remote monitoring function to be set up anew.

You can set an option in your BIOS to have it display the POST & BIOS messages, as they are generated, during boot up.

So what's causing your boot up problem, I'm not sure. More than 5 minutes should not be needed to reboot from a cold start up, however. Can you set up your BIOS so it prints out what it's doing, as it boots? That could possibly pinpoint the problem, right there.

Your lights are right, according to my mobo document.

 

[dfonda]

Yes.

Tear mentions it in his post - I was testing different settings (NUMA on, NUMA off, etc.), in the BIOS, and in the starting and stopping, Tear's "baby's", got stuck in the wrong client (I am guessing), anyway, I was borked big time.

After fussing around, to no good effect, I did a clean install of Ubuntu again, and did NOT re-install Tear's two programs. I just don't have the Linux-mojo to fix these programs, when they go sideways.

The way I look at it is, during any testing phase, the Kraken and Langouste programs are not to be used. Use them when you have finished your other testing, and just want to fold with the rig.

I have not been using Langouste at all, and I have quietly not been using Kraken, I only need to get burned once. (It was at 0% of the wu) I hate dumping wu's and I really hate bigadv going south. I'll take my medicine (and make my measly 215K) and I'll start using Kraken again when I understand it better.

I think it sets up the same on smp, I may be alot better off experimenting with my 2600K rig where an hour or 2 of slow down while I work on it, isn't going to matter as much.