• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

DC Vault 2

I won't be sleeping or hibernating my main PC, because SR-2s can be finicky enough already - factor in overclocking and then resuming from sleep right onto load, and it'd be a nightmare. I don't particularly want to subject it to a cold boot every day either, so I may have to settle for leaving it running during the day, but not crunching...it'll probably take some experimenting before I find out what works best for the PC, for the projects, and for me.
I was thinking of those servers I have yet to set up - if they do sleep and resume properly, which I think they will, that could be an option rather than having them shut down and wake on a timer. Especially as the fans go nuts during POST. I only want to run the bulk of my crunching overnight as power is much cheaper than in the day. So yeah, if they sleep and resume properly, they could be a candidate for RNA. VMs wouldn't be ideal because I'd have to remote into each one and start the VM, or maybe schedule it...either way I think sleeping them would be better if it works properly, I'll see when I get there.

Thanks for the tips on POEM, I haven't run a decent card on BOINC at all before, and the last time I did run a GPU was a 5770 which I just ran Collatz on, seemed to do fine with no tweaking. I assume that the GPU utilisation will be <100% if there's still headroom to run more units concurrently?
 
Well..the servers will do fine on BOINC if you leave LAIM turned on. The BOINC client has the option of telling it what times and days to run. So, hybernation might work. You will have to play with it.

GPU utilisation being <100% is exactly why there is still headroom. You just have to tinker with it to find the sweet spot. Also, running different cards in the same system can get troublesome. So, make sure to account for that if you mix. My Q8200 has 3 entirely different cards in it right now, so I have SETI/BETA just using 2 units per card. That is probably worthless on the 8400 (only 128MB VRam) but the other two cards can handle it.
 
Kai said:
That's an incredible amount of Borged systems! I can just imagine you going door to door selling the idea of science to people in your neighborhood... "Hello sir, have you ever wanted to cure diseases, find aliens or map the cosmos?"

Oh, the 7970 is working perfectly in a system with a 7990 as a primary. No issues with it thus far, but I'm just waiting for the lock-up/fire when I come home from work one day.
Click to expand...

I have actually gone as far as emailing every Cancer Research not for profit I could find email addresses for informing them of the contributions their IT depts. could be doing. Most people are not receptive or don't want to spend the time learning about it. I have found that most are willing to let you borg rather then them join the cause actively. It is that whole idea of donating money rather then donating time thing.
 
Gilthanis said:
Well..the servers will do fine on BOINC if you leave LAIM turned on. The BOINC client has the option of telling it what times and days to run. So, hybernation might work. You will have to play with it.
Click to expand...

Yeah, I just don't want to leave them turned on in the day even idle, as they still suck down a fair bit of power.

GPU utilisation being <100% is exactly why there is still headroom. You just have to tinker with it to find the sweet spot. Also, running different cards in the same system can get troublesome. So, make sure to account for that if you mix. My Q8200 has 3 entirely different cards in it right now, so I have SETI/BETA just using 2 units per card. That is probably worthless on the 8400 (only 128MB VRam) but the other two cards can handle it.
Click to expand...

Got you. They're matching cards so shouldn't have a problem there.
 
Last edited:
Well, I'm game for POEM, as it was actually next on my list after I finish my current video card push on Einstein - I'm 165k points away from my 1M goal, and at 11k ppd, I should hit that right about October 1st. After that, my 9800gtx will be going to POEM, and likely joined by a 680...

So I like the idea of a POEM GPU & a Correlizer CPU push. Although I'll have to remember to set myself up a Correlizer account!
 
Gilthanis said:
If you do run POEM GPU work, make sure to use an app_config file so that you can run multiple work units at a time on your card. A 7970 could probably handle around 5 tasks at a time on the card. An example of the app_config would be:
<app_config>
<app>
<name>poemcl</name>
<gpu_versions>
<gpu_usage>0.20</gpu_usage>
<cpu_usage>1.0</cpu_usage>
</gpu_versions>
</app>
</app_config>

Just open notepad and paste that in there. Save it as app_config.xml into the POEM data directory. If you have two cards, that would also allow you to run 10 total for the system. You can change those numbers to experiment with how many you could push your system for.
Click to expand...
I've tried this before with ATI cards, but does it work for Nvidia cards as well?
 
Jathanis said:
I've tried this before with ATI cards, but does it work for Nvidia cards as well?
Click to expand...

I haven't turned on POEM GPU yet since I added that config. I know my SETI/BETA config is working just fine and I'm using nVidia cards there. Give it a shot and let me know. ;) I may only be able to do two on some of my cards, so my numbers will be a bit different.
 
Last edited:
Random BOINC / WCG observation of the day: FAAH Vina 7.06 WUs for Android have gotten extremely long! Last week, they were taking about 8-10 hours on my overclocked 1.2 Ghz Galaxy Tab and 6-8 on my 1.5 Ghz Galaxy S3. I currently have one on the S3 at 60% after 11 hours, and the one on the Tab is at only 56% with 28+ hours run time! They are showing progress, though, so I'll let them run...

We now return you to your regularly scheduled program...
 
The one running on my Motorola Bionic (dual core 1GHz) is currently sitting at 69.5% and has ran 14 hours and 11 minutes. I know WCG is still tweaking the sizes to find a happy medium between Android devices and traditional PC's. If they make them too small, the servers can't handle the load.
 
For those looking for rare points, Pirates@home has work units going out today. They typically run work on 9/19 since it is Talk like a pirate day. :)
 
So the FAAH Android WU on my S3 finished in 17:36:10. Looking at my Tab as I left for work, it was at 99.444% with a run time of 49:58:20... Can't wait to see if it's going to be valid!
 
Happy Birthday to me yesterday...I hit my 1 million point SETI goal. Now for SETI BETA.
 
Android BOINC updates:

- the 38 hour WU on my Galaxy Tab 7 ended up as valid; the 50 hour one is still pending...
- tried to change Android projects today, as WCG/FAAH is starting to take too long and I hit my goal in Primegrid. Found that of the other projects: Yoyo is a joke, refusing to download WUs unless the phone has over 2 GB RAM (my Galaxy S3 only has 1.5); Asteroids' website is completely down; and Oproject has no WU for the only subproject that supports Android.
- So of the 6 I can run, 2 I already do, 3 I currently can't, and 1 (Einstein) I don't want to start until I hit my 1M GPU goal because I'm too lazy to set up separate preferences... :D

Oh, well, one more FAAH for tonight, and hopefully Asteroids will be back up tomorrow!
 
Jathanis... Did YoYo's requirements change? My Bionic only has 1GB of ram and has processed many.
http://www.rechenkraft.net/yoyo/show_host_detail.php?hostid=99421
http://www.rechenkraft.net/yoyo/show_host_detail.php?hostid=84295

Both of those are the same device. One is nativeBOINC and the other is BOINC.
Also, my list of BOINC/nativeBOINC capable projects are;
OProject (which stopped having capable work long before BOINC had a working client)
SETI BETA (which I have yet to run but they are asking for testers)
Milkyway
SubsetSum (which hasn't had work for months due to the Admin focusing on Wildlife)
WUProp
Asteroids
YoYo
Enigma
Albert
theskynet Pogs
PrimeGrid
WCG
Einstein
SETI (which I haven't ran on BOINC but did run on nativeBOINC)
 
I'm not sure if the Yoyo requirements have changed or if my devices are not being properly recognized. I do know both of my devices listed the same 2059 MB requirement to download a WU, and I even tried flushing the databases and redownloading them. I did a lot of googling, and found a thread or two on the Yoyo forums of others with a similar problem, and at least one gave up. Another seemed to fix it by tweaking the configuration file, but that seemed like too much trouble for me... It doesn't help that many of the posts aren't in English...

As for your list of projects - thanks! I was going off of the list on Berkeley's FAQ page, so I didn't realize those others... I may take a look at Albert, as I've already got an account set up there (I think), and I've never been huge on SETI. For some reason, it made perfect sense in my mind to do stuff like Primegrid or Milkyway, but I've always thought looking for ET was a waste of computer power... But maybe after trying Albert or Pogs, I'll give SETI a try!
 
Yeah...SETI brought me to BOINC so I am thankful. However, I think the odds of their science actually producing any results is highly unlikely. At least we use Prime numbers in our daily lives. However, I'm giving them at least 1 Million points as a goal and as thanks for bringing me in. :)

Is it requiring that much ram or is it looking for that much hard drive space? Developers tend to reference memory as both mediums some time. Very confusing to many. It shouldn't matter since I have run YoYo on some pretty old hardware. If that doesn't work out, try nativeBOINC instead and see if you get better results.
 
Well, good news/bad news...

Good news is that thanks to Kai, my GTX680 came in the mail, and was a very simple plug and play procedure! Instantly dropped times on Enigma from 2h 11m per WU to 1h 22m. Once I finish a few single WU to benchmark, I'm going to start firing up 2, then 3, etc...

Bad news is that the card doesn't seem to want to download any work from GPUGrid or DistrRT. It's also showing up in GPU-Z as not having CUDA enabled :mad: I'm wondering if it's because I have it secondary to an ATI card, and the Catalyst drivers are somehow blocking it from working in CUDA mode... Of course, there's a good chance that the drivers are just messing with GPU-Z!
Edit: Primegrid downloaded a 40 hour CUDA Genefer WU, so it's the projects, not the card...

But for now, I'll be at my Einstein 1M goal in just a few days, and will be jumping on POEM for the October push. But for now, I'm too tired to keep troubleshooting...
 
Last edited:
Jathanis said:
Well, good news/bad news...

Good news is that thanks to Kai, my GTX680 came in the mail, and was a very simple plug and play procedure! Instantly dropped times on Enigma from 2h 11m per WU to 1h 22m. Once I finish a few single WU to benchmark, I'm going to start firing up 2, then 3, etc...

Bad news is that the card doesn't seem to want to download any work from GPUGrid or DistrRT. It's also showing up in GPU-Z as not having CUDA enabled :mad: I'm wondering if it's because I have it secondary to an ATI card, and the Catalyst drivers are somehow blocking it from working in CUDA mode... Of course, there's a good chance that the drivers are just messing with GPU-Z!
Edit: Primegrid downloaded a 40 hour CUDA Genefer WU, so it's the projects, not the card...

But for now, I'll be at my Einstein 1M goal in just a few days, and will be jumping on POEM for the October push. But for now, I'm too tired to keep troubleshooting...
Click to expand...


Are Enigma WU's running on GPU's?

You may also want to test it on POEM to make sure it is working too.

When you say it wont download any work from GPUGrid or DistRTgen, are you getting any response why from your Event Log?
 
Last edited:
Gilthanis said:
Are Enigma WU's running on GPU's?

You may also want to test it on POEM to make sure it is working too.

When you say it wont download any work from GPUGrid or DistRTgen, are you getting any response why from your Event Log?
Click to expand...

Sorry, I meant Einstein, not Enigma... And all I'm getting on GPUGrid and DistRTgen is a generic "not requesting new work" in the event log. Let me make sure I have event logging set correctly when I get home tonight, and I'll post a screen capture of the event log...
 
Have you suspended work from all other GPU projects? The client sometimes gets stupid when GPU's are involved. Try removing other projects and see if GPUGrid or DistRTgen picks up work. The manager "should" eventually download work when the resource "debt" kicks in. Has this machine ran work at those projects before the upgrade?

Also, I'm not a fan of mixing cards, but how do you have yours setup? Is the AMD in the first slot and connected with a monitor? Is the nVidia connected to a monitor? Did you just install the Windows detected drivers or the complete package from nVidia?
 
On a separate note, Correlizer may be a bad idea as our CPU project. Looking in their forums, they seem to have had difficulties last month and work units may not be available or consistent. However, if everyone still wants to have that project pushed next month I would recommend changing your settings to also allow CPU work units at POEM so that they don't run dry of work.

http://svahesrv2.bioquant.uni-heidelberg.de/correlizer/forum_thread.php?id=109
http://svahesrv2.bioquant.uni-heidelberg.de/correlizer/forum_thread.php?id=89

Quite frankly, there are a few projects that should be removed from the Vault due to not meeting Vault requirements. However, I can't make an account on their forums due to them blocking my email. If someone else has an account and is willing to post my suggestions on there, let me know. I can shoot you a PM.
 
Gilthanis said:
Have you suspended work from all other GPU projects? The client sometimes gets stupid when GPU's are involved. Try removing other projects and see if GPUGrid or DistRTgen picks up work. The manager "should" eventually download work when the resource "debt" kicks in.
Click to expand...
I only suspended the NVidia projects (Einstein & Primegrid) and ran out the queues, but left Moo! running on the ATI. Was too excited/lazy last night to take time to work things through to this level of detail... :D

Gilthanis said:
Has this machine ran work at those projects before the upgrade?
Click to expand...
No, I've tried, but my old 9800GTX+ would never run it. When I researched that, I think I discovered they had updated and my card didn't support the CUDA level needed - but it's been a while, so I'm not 100% sure! To be honest, the main reason I want to run them is that I'm tired of seeing 0 points in those projects at the bottom of my list! (Only slightly less is the easy points for personal goals...)

Gilthanis said:
Also, I'm not a fan of mixing cards, but how do you have yours setup?
Click to expand...
I'm not either, but since my other system is Win 8, there's no driver support for the ATI. And since that same other system is a shared computer, and BOINC GPU projects won't work on fast user switching, the only way I could use both is to put them in my XP HTPC together...

Gilthanis said:
Is the AMD in the first slot and connected with a monitor?
Click to expand...
The AMD has to be in the primary PCI-E slot or the catalyst drivers won't recognize the card. Nvidia has no such issues.

Gilthanis said:
Is the nVidia connected to a monitor?
Click to expand...
Nope, I just use the Display properties to extend the desktop, which works fine in XP... And since the mouse pointer disappears, it seems to be working...

Gilthanis said:
Did you just install the Windows detected drivers or the complete package from nVidia?
Click to expand...
Nope, was using 310.something for the 9800GTX+, and when those didn't work last night, I uninstalled them, ran driver cleaner, then installed the newest 327.XX drivers.


QUICK UPDATE: I just found out I have to go out of town starting immediately after work tomorrow, so I'll come back to this mess next week or after the POEM push is done, one of the two... Thanks for trying, Gilthanis! And I promise to forward you the entire story as I promised whenever I can find a free moment...
 
Last edited:
Try removing all of nVidia's drivers and run stock MS ones. I can't recall off the top of my head, but I think I read that you can't install their software. This way the card shows up as a coprocessor instead of trying to be a full fledged gaming graphics card. I'm not sure though. I'm kinda interested in your results. I just don't have the time anymore to try it out. Maybe in a few years. :)

I will also forwarn those who may want to run CPU POEM work units. The point payout isn't very good for them. So, you may just want to focus on GPU's for that project.
 
Last edited:
Also, I believe I had to separately install the CUDA toolkit (5 at the time, but 5.5 now I think) in order for it to play nice with AMD. Though I installed the AMD after the nVidia card was already in, so your mileage may vary. Significantly perhaps.
 
If Correlizer aren't likely to be reliable enough and CPU work at POEM isn't great "value", is there something else we could benefit from focusing on or have as a secondary push if/when Correlizer goes down or runs out?

Waiting on delivery of two hex cores for the SR-2, decided to sell off the quads. Stuck with two other hexes that don't seem to play with each other, not sure if it's the chips or what as the quads worked just fine. I have all the other parts, bar a cosmetic item, in place and ready to go, I just need time to build the thing now!

My gf had her MRI scan last night, apparently it was pretty scary stuff. They originally told me I could come into the room with her, then they basically barred me at the door and told me I had to wait in the waiting room, which was weird. Anyhow...they said 2 weeks or so for the MRI results, and up to 3-4 weeks for the EEG results, so we'll see how it goes. They want me to take a video of a seizure to show them at her next neorologist's appointment, but it's pretty difficult as usually my first concern is to make sure she's safe, and by the time I've got her in the recovery position and away from solid objects she could bang her head or something on, the fit is usually just about over.
 
I would suggest possibly running Asteroids. It is ram "light" and we could get a good push in there. That project seems pretty busy too, so we need to get a strong positioning at some point. It also has a nice return on Vault points.
 
Sounds good to me, do you want to push that on its own, CPU-wise, or alongside Correlizer? I don't mind either way. Asteroids is supported on Android too so I can get my two Androids on it for what they're worth.
 
I personally don't care what we crunch because it all goes to advancing the team in some way. Whether it be offensive or defensive. NFS pays well, but we need over a million points (CPU only) before we have any chance of moving up. Asteroids will probably be mostly Defense until the newness wears off or some heavy hitters joins in. *cough get those servers going cough* ;)
 
Let me check my schedule from SWMBO, see if I can get some time off for good behaviour and get some real work done ;)
 
Well, seeing as I just switched both my Android devices over to Asteroids, I'm good with that!

My wife decided she wants a Kindle for her birthday next month, and of course, my first thought was: 2.2 Ghz quad core? More than doubles my Android BOINC output!
 
Jathanis said:
Well, seeing as I just switched both my Android devices over to Asteroids, I'm good with that!

My wife decided she wants a Kindle for her birthday next month, and of course, my first thought was: 2.2 Ghz quad core? More than doubles my Android BOINC output!
Click to expand...

....as long as she leaves it plugged in. LOL
 
Got the tablet all switched over, I'll have my rigs on asteroids here by tomorrow. Huzzah!
 
Gilthanis said:
You may also want to test it on POEM to make sure it is working too.
Click to expand...

Fired up POEM, 3 WU running no problem - but also takes 3 of 4 CPU cores out of service on the computer... Oh, well, I'll see how it's doing when I get home Monday morning, and maybe see if I can reduce the CPU a bit then. See you all next week!
 
Jathanis said:
Well, seeing as I just switched both my Android devices over to Asteroids, I'm good with that!

My wife decided she wants a Kindle for her birthday next month, and of course, my first thought was: 2.2 Ghz quad core? More than doubles my Android BOINC output!
Click to expand...

I had similar thoughts when my gf got a Note II on her contract, haven't managed to borg it yet though sadly.

I'll set Enigma to no new work on my dedicated Android crunchers tonight, and then they can pick up some Asteroids units. Will do the same with the PCs I crunch on, swap them over from SIMAP.
 
Jathanis said:
Fired up POEM, 3 WU running no problem - but also takes 3 of 4 CPU cores out of service on the computer... Oh, well, I'll see how it's doing when I get home Monday morning, and maybe see if I can reduce the CPU a bit then. See you all next week!
Click to expand...

Jathanis, this is pretty common. Many find that freeing up one core per GPU increases production because the work unit doesn't have to wait on the core to free up. Even WCG assigned a full core to their GPU work units when HCC was still active. So, don't put too much time into it because it may be fruitless.
 
theSkynNet is being discussed as a potential Vault project. For those who aren't familiar with it, they actually have 2 clients they support. theSkyNetPogs which is the BOINC client and then they have their own client as well. Their main site merged the two for stats purposes, but the BOINC page does not reflect that and most likely the BOINC stats pages probably wont either. However, I have not run the non-BOINC app and can't confirm that. We do have a few Commandos that have tallied points with the other client. However, as I check that page, I noticed it said that Razor_FX_II and I had left that team. It must have been something to do with merging the two projects. phoenicis is still listed as on that team which is why I'm suspecting it was from the merge. He hasn't joined our team at pogs yet. I would suggest that he do so to help unify the teams standing. http://www.team-ninja.com/vbulletin/showthread.php?t=50874&page=2
 
I ran theSkyNet for a little while elsewhere, just to see what it was like. Haven't run it in recent times but I'll take a look.

Haven't picked up any Asteroids work on my laptop overnight, anyone else having issues? I was also finding that the remaining estimated time was wildly inaccurate - some units, I had done maybe 50% but the remaining time was estimated as 12-24 hours, way more than expected. Not sure if that's related at all, but obviously it was playing hell with my cache settings, because I did have it set to cache 0.5 days of work which in this case meant it was only caching one unit at best, usually none.

Edit: Asteroids seems to be working fine now, for no apparent reason it grabbed a bunch of units and is now crunching them normally, taking about 50 minutes per unit.
 
Last edited:
Was this behavior right after attaching? If so, that is common behavior of the BOINC client doing its learning curve. The numbers should become more accurate as you complete work.
 
You must log in or register to reply here.
Back
Top