TPU's WCG/BOINC Team

[agent00skid]

I'm starting to get occasional computation error on my 2p system. Now the errors are becoming more frequent. I know you're all gonna say that it's related to the OC but hear me out first. Before the challenge, I set everything in the BIOS to the recommended defaults for a few weeks, to try and eliminate the issue. During that time I still experienced the errors.

I'm open to suggestions, though I think that I know what the culprit is... and that is the system drive which happens to be an SSD. Not long before the WCG Computation Errors started to appear, Windows started giving me errors stating that "Windows was temporarily unable to access the system drive" or something very similar. So my thinking is that either the sata cable or connection is bad or the drive is failing. What do you guys think?

It appears that its failing most every OET unit but not the others?

I had for some time had OET failing on my J1900 system, so I ended up just disabling OET from that machine.
Never found out why. Only system that I have neither overclocked nor undervolted at the time.

 

[T-Bob]

It's more than just OET units failing. All units seem to fail at times, very sporadic and not too often though and before today it had no problem with OETs. Now it seems that about 9 out of every 10 OET unit fails within the first 10-15 minutes.

 

[Norton]

I'm starting to get occasional computation error on my 2p system. Now the errors are becoming more frequent. I know you're all gonna say that it's related to the OC but hear me out first. Before the challenge, I set everything in the BIOS to the recommended defaults for a few weeks, to try and eliminate the issue. During that time I still experienced the errors.

I'm open to suggestions, though I think that I know what the culprit is... and that is the system drive which happens to be an SSD. Not long before the WCG Computation Errors started to appear, Windows started giving me errors stating that "Windows was temporarily unable to access the system drive" or something very similar. So my thinking is that either the sata cable or connection is bad or the drive is failing. What do you guys think?

It appears that its failing most every OET unit but not the others?

It could be the drive- crunching is capable of hammering drives into an early grave once in awhile. As an example, I lost a couple of HDD's myself and I think [Ion] lost an SSD a while back

 

[T-Bob]

It could be the drive- crunching is capable of hammering drives into an early grave once in awhile. As an example, I lost a couple of HDD's myself and I think [Ion] lost an SSD a while back

That's what I'm thinking especially because of the drive related windows errors. After the challenge is over I'll swap the system drive, do some maintenance and hopefully resolve this issue.

 

[manofthem]

It could be the drive- crunching is capable of hammering drives into an early grave once in awhile. As an example, I lost a couple of HDD's myself and I think [Ion] lost an SSD a while back

I think that's happening my wife's PC, hdd been acting weird I think. Plans are, post Challenge, to get my new ssd in my rig and transfer my current ssd to that PC, so that'll at least help out.

 

[[Ion]]

It could be the drive- crunching is capable of hammering drives into an early grave once in awhile. As an example, I lost a couple of HDD's myself and I think [Ion] lost an SSD a while back

Indeed I did--but I've also lost my fair share of HDDs too. Running OLD drives tends to do this. And it was a terrible very early gen 40GB drive, so I'm surprised it made it as long as it did.

I'd imagine a new drive and a memtest would get things sorted out

 

[t_ski]

If you want to have an SSD in a rig that crunches, I'd recommend putting the BOINC install on an HDD. I have not crunched on my main rig because of the lack of HDD's in the system for this very reason.

 

[T-Bob]

If you want to have an SSD in a rig that crunches, I'd recommend putting the BOINC install on an HDD. I have not crunched on my main rig because of the lack of HDD's in the system for this very reason.

That is actually part of the system maintenance I have planned for after the challenge. I'm going to move WCG to a hdd on my main rig.

 

[hat]

Wow, is WCG really that hard on SSD? How fragile are they to be ruined by WCG? I wonder what I should do about my pagefile then...

 

[[Ion]]

If you want to have an SSD in a rig that crunches, I'd recommend putting the BOINC install on an HDD. I have not crunched on my main rig because of the lack of HDD's in the system for this very reason.

I think you're being overly paranoid--as long as you exclude CEP2 from the mix (since it has by far the highest IO requirements) there's nothing to fear.

Wow, is WCG really that hard on SSD? How fragile are they to be ruined by WCG? I wonder what I should do about my pagefile then...

No--they definitely aren't. I have a Samsung 830 (the cheap non-Pro) version in my 3930k system. It's been running WCG off the SSD pretty much continually since the drive was released.

SSDs will take a lot of beating: http://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead

 

[manofthem]

Wow, is WCG really that hard on SSD? How fragile are they to be ruined by WCG? I wonder what I should do about my pagefile then...

I've been crunching on my Sammy 830 since like November of 2012, and everything has been very very smooth and strong. Soon it's going to transferred to my other pc and continue crunching. My second 830 has been crunching for over a year straight in another pc and going strong, too.

I think you're being overly paranoid--as long as you exclude CEP2 from the mix (since it has by far the highest IO requirements) there's nothing to fear.

I have a Samsung 830 (the cheap non-Pro) version in my 3930k system. It's been running WCG off the SSD pretty much continually since the drive was released.

I think you mean 840. The 830 was only a single version, the one that I have, and it's a beast. When the 840 came out, there was the regular 840 and the 840 Pro, then a little bit later the 840 Evo came out. 850 only has the Pro and the Evo.

 

[[Ion]]

Hmmm....could swear I got it before the 840 series came out. Maybe I'm confusing it with the 840 non-Pro I put in my sister's laptop

 

[manofthem]

Hmmm....could swear I got it before the 840 series came out. Maybe I'm confusing it with the 840 non-Pro I put in my sister's laptop

The 840 is still a good drive. I built my brother a pc with an 840 in it early 2013, and it's still doing well. Granted he doesn't use his pc quite like we do

 

[T-Bob]

The 840 is still a good drive. I built my brother a pc with an 840 in it early 2013, and it's still doing well. Granted he doesn't use his pc quite like we do

I agree both models are great drives.

Part of my problem might have to do with the 840 Evo firmware issues and that I have yet to update the firmware on mine. This is one more item on my post challenge system maintenance list.

Despite this ongoing problem it's still putting up good numbers.

 

[manofthem]

Part of my problem might have to do with the 840 Evo firmware issues and that I have yet to update the firmware on mine. This is one more item on my post challenge system maintenance list.

It's funny that we all have these "post-challenge" maintenance lists that need to be taken care of

But yes, definitely update the firmware. Iirc the Sammy software makes it pretty easy.

 

[t_ski]

The idea is just the write cycles go up as the disk is being utilized more. More writes means reaching the write-limit of an SSD faster. Add on top of that the idea that WCG (or folding, mining, etc. - anything that stresses the system at/near 100% for extended periods of time) is just plain harder on hardware. We as a team have seen a lot of failures for one reason or another.

 

[t_ski]

GRRRRRR... fix one problem and cause another

One of my 2P systems has had a ton of "base system device" items listed in the unknown devices section of the control panel. I had a few minutes to look into it and saw that the fix was pretty easy: just update the Intel Chipset drivers and I was good to go. Just sat down a little while ago and checked FreeDC to see that this rig has very little output for the day. Go back to the rig and move the mouse - nothing. Move it again - still nothing. Bash a key on the KB and I'm prompted to "log in" which I've never had to do before. OK, so I log in, and I see it's still crunching away. Weird. Go to look at the power management setting to make sure the "prompt for login" is turned off and I see it's set to "balanced" instead of "maximum performance."

Long story short: I guess the "base system devices" had a little bit to do with power management settings never being set up right in the first place...

 

[FordGT90Concept]

IME (Intel Management Engine) I think has a lot to do with power saving features. It was probably one of the drivers you installed.

http://www.slideshare.net/codeblue_jp/igor-skochinsky-enpub

It works independently from the main CPU, can be active even when the rest of the system is powered off, and has a dedicated connection to the network interface for out-of-band networking which bypasses the main CPU and the installed OS. It not only performs the management tasks for which it was originally designed, but also implements features such as Intel Identity Protection Technology (IPT), Protected Audio-Video Path, Intel Anti-Theft, Intel TPM, NFC communication and more.

Maybe better off...not using it at all...but I don't think you can even disable it...WTFcakes!?!

 

[[Ion]]

I'm heading out of town for the weekend, so all of my systems at home have been turned off until Sunday night. Fortunately I have most of my output offsite, but still that's at least 25k PPD down

 

[manofthem]

I'm heading out of town for the weekend, so all of my systems at home have been turned off until Sunday night. Fortunately I have most of my output offsite, but still that's at least 25k PPD down

Enjoy the weekend! (perfect timing with the Challenge ending today)

 

[[Ion]]

Enjoy the weekend! (perfect timing with the Challenge ending today)

Thanks! Going camping with some friends, so that should be a nice relaxing time

 

[FordGT90Concept]

Running on all cylinders (i7 920 ~4 hours/day, i7 6700K ~16 hours/day, and Xeon 1230 v3 24 hours/day) again for the first time since July 20. I'm happy and sad about that; that was several months of 24 hours/day crunching missed.

 

[[Ion]]

I'm home and have my systems running again. MIGHT have found a host for the 48c 4P setup. Even if not, I have two more of the T5500s to upgrade to six-core CPUs tomorrow or Tuesday.

 

[ThE_MaD_ShOt]

I think either later tonight or tommorow I am going to blow off the dust on one of the rigs I have sitting on the side line.

 

[Steevo]

Its getting cold at night, so my main rig is back to folding,