But we must be very very close to the point where the DB server runs out of RAM. I already get occasional "Project has no tasks available" although the RTS queue has plenty of stuff.

I agree. It would be good if boinc tasks or another piece of software could push short tasks to the front of the queue. Does anybody know of any software that does this?

Is there a significant performance hit when SETI uses less than 1 CPU for the GPU? If so, will lowering "Use at most..." a bit more so another CPU is free help? Will the increase in GPU performance offset the loss of a CPU for other projects' tasks?

I use ethyl alcohol myself.

Cheers.

Isn't that the remains of the old Seti V7? We are now on V8.

BOINC will do exactly what you ask it to do, so if you ask it for a two day cache and your main is out, it will make a two day cache of your backup. As those deadlines will be different, it will likely go into EDF mode and crunch the backup until done! Unfortunately we can't set different caches for different projects.

Moral is, if you have a backup project(s) don't get greedy with the cache or it will fill with the backup!

I have two X5675 3.06GHz 6 core cpus and two Asus R3E motherboards, 1 motherboard needs a new cpu socket, but buying this motherboard used is not inexpensive, not at about $450, which is not guaranteed, damaged was about $100 to buy and repairs about $70 or so, the repair place puts up special prices sometimes so the price varies, then I need to get a 24pin cable repaired on an Enermax Revolution+ 1050w psu(quoted $110 at one time, I'd have to spend another $20 to ship the psu to FL I think, the psu was not like this when I bought it at one time) and the cable is attached otherwise I'd replace the cable on this semi modular psu. I'll get these done asap.

It is after all only a pastime and a hobby

Secondly, modern GPU's vastly out perform CPU's of any kind, so upgrading in that direction is more worthwhile.

The above WU is estimated to take "1d 4:43:xx" time to complete. Really? Has anyone else come across anything like this while not running a 486 PC? ;)

It is very difficult to find cpu tasks on the website completed list because the ratio of gpu to cpu task is so high. You either have to whack in a 1000 offset in the URL to stumble upon them or spend minutes paging through the lists. The easier way to find them if you are using BT, is to simply highlight the cpu task in your History list that you are interested in and copy it to a text file. Highlight the task name in the text file and copy it. Now go to your Seti web page and use the search function on your Tasks list and paste the name into it. !Voila! You've found your cpu task.

Results Returned in Last Hour has hit 237,593 and the RTS is practically empty (30K). The system is still up though. It is 7am in California, so hopefully soon someone will notice and throw us an Aricebo dataset , if they have one.

If you are lucky enough to snag some of the work from the RTS buffer, it is all for naught because the tasks currently there are all noise bombs.

But in the BOINC discussion, Jord mentions that benchmarks are also run when you change the BOINC configuration manually - specifically, when you increase the number of CPUs BOINC is allowed to use. Could something like that be happening?

I am still wondering if there is a way to get an E5-2670 to run at "extended turbo-boost" all the time? According to Intel "Extended Turbo Boost" runs all the cores, all the time, in turbo-boost (I think).

I was chastised that claiming 10% was too much. You still should see some benefit from running at stock P0 speeds or with a overclock. I always see a boost.

And bonus, most 1050tis (minus the FTW model from EVGA) don’t require external power. All from the PCIe slot.

But they can cause power problems with some motherboards when using multiple cards and are no good with dual rail PSU's.

Cheers.

I just thought it was interesting you could make the graph line on Host Total steeper by doing fewer tasks.

Hi,

AP on Titan V vs GTX 1080. I run 2 at a time (gpu_usage = 0.5).

http://setiathome.berkeley.edu/workunit.php?wuid=2915879436


Petri

Makes sense with half the chip turned off. Probably will drop cpu temps too since you won't be spinning processes on the HT cores.

More issues from Carolyn. I have had disk index error messages from that server continuously since November. They still haven't fixed it. Database is screwed up. Whether by failing disk or corruption, who knows?

In Australia it's between Oct - April

mb is only a year old.....darn it

Yes. "Being processed" on the SSP means that work has started on that tape. But when the splitters are paused (e.g. for maintenance yesterday), or when new tapes are loaded, the tape selection mechanism sometimes picks a different tape for processing. So then the old tape goes into a 'waiting to run' state, such as we sometimes see on our BOINC Managers - except you can't see it on the SSP. Those are your ghosts: they are very common, even if you haven't noticed them before. And they're nothing to worry about - eventually the tape selection process will pick them out again and give them another go.

Hello,
Everytime then the WU are withdrown, my crunchers encounts the same behavior: the slower & less core one, which has few WU's left to crunch, receives new WUs , before the second cruncher (faster & more cores) which is totally withdrown of CPU WUs , receive it's new batch. Also when it receivs it got first GPUs (which some are already from the previous batch) and no CPU WUs. How schould I configure it for letting more CPU WUs be downloaded in advance instead of GPUs? Every outrage I would need some 100 CPU WUs more to have the work sustained on the faster machine.

Two items of interest:

1) E5-2680 v2 (10 core, 20 thread) are now under $200 on ebay
2) E5-2670 v1 are > $100.

So you can upgrade your dualie cruncher for around $200. The v2s are 100mhz faster than the v1s and will run 8 more threads and at the same wattage.

Just a headsup, guys!

It's great to have plenty of work, it's not great to get even less acknowledgement for the work we're doing

What motherboard do you have to put the CPUs into?

I have an Intel s2600cp2j from a Natex bundle and kind of wondering if this would work for me too. $59 per CPU upgrade seems pretty good.

Hey - it seems the file 13ja07ad has been stuck for a while. Has anyone else noticed this? Or am I crazy?

It's working, but just maybe a little slower than usual as AP's as well as MB's are being split off that file at the same time.

Cheers.

Just four hours to go, folks. We've got our $100K (and then some), but I see 248 donations - 247 if we link both of Juan's into one. Can someone make it up to 250?

They have to cancel flights out of Sky Harbor airport in Phoenix every summer because it is too hot for planes to take off. No lift.

I've thrown an overthrow on bed just in case for the night anyway. ;-)

Cheers.

Ahh, fond memories. I wish that were my case. I've been dealing with +43° C. for the past two weeks. A/C blasting night and day .... still miserable.

How do you pull up the entire list? When I click it I get the host he referenced, but didn't see a way to show the entire list of hosts. Thanks!

That was a major issue with Dell during the days of DDR1/2 in that no more than 6 banks of ram could be used. As I haven't dealt with any Dell machines using DDR3, don't know if that issue has been resolved.

0010 1010 is the answer ...

The project is suspended because of a database problem. It should be available soon.

I like your signature....:-)

Total credit 16,451,341
Recent average credit 50,770.82
The GTX 1080 really crunches and does 28,000 to 30,000 WU on average with a EVGA Classified MB, Corsair Dominator Platinum DDR4 64Gb and a Intel 6950X 10 core cpu.
I have 3 other machines crunching.
ASUS i7 G73JW Laptop with Nvidia 460 gpu
Desktop with GTX 750Ti (Intel 975X 4 core cpu).
Desktop with GTX 1050Ti (Intel 980X 6 core cpu).
All are overclocked.
I think my electric bill went up $100 a month!

Use_sleep with a 1080?? I would drop that and use -hp instead. You have to remember that a 5% increase in GPU times will be more than your entire CPU can do for RAC.

EDIT: And I would update to r3584 as well.

How many hours does it take for a GTX 1080 to go trough the 100WU limit?

Go one 4 watt Android tablet only, and all problems suddenly disappears.
It's so calm and wonderful. I have all the WU's I need.
LOL

I just added a 3rd 750ti but Boinc is only using 2 cards. I recall seeing that you needed to put a load on the third card for it to run. and how to make an adapter with a resistor to supply that load. Is this making sense? How do I get the 3rd card to run short of buying another monitor?

Just curious to know your MB reference .....

Sopigt projekt helt enkelt. Det är aldrig för sent att ge upp.

And empty now since many hours (the last ones were CPU tasks and iGPU tasks, the 980 has been dry for almost 24 hours).
Now the RAC only lives on 402 Pendings, and when they're all validated, the free fall will begin.

We're back from the outage and it looks like they removed that file from the queue. :)

Oh no, not degrading again.....

Save us Obi-Wan Korpela, you're our only hope...

As usual, all 5 boxes here are full-up (1800 total tasks) within 30 minutes after the outage.
What's different?

It is the measure of the feeling we have already sensed that GBTs make us sweat more, but bring less credit.
The possible reasons have been discussed here by Raistmer et alii.
Probably, already people using the Linux Special application may have a different feeling.

Crunch on!

Enjoy the shorties now because you will have BLC files nipping at your butt for weeks after this!

. . But hopefully something other than Blc5s :(

Stephen

.

Stephen may I ask why you are not wanting blc 5s?

. . Because they are the gristliest, nastiest Greenbank WUs that take far longer to crunch than anything else. 100% longer than Arecibo tasks and up to 50% longer than other Guppies. And for some reason they are the ones we get the most of :( ...

. . Does that answer your question :)

Stephen

:)

Well today's 9.5hr long outrage saw my 1060's out of work for 6 minutes this time before the servers came back online, another few more minutes and I would've grabbed some backup work for them (so much for that 3-4hr notice these days). :-(

Cheers.

When were those exactly? It seems like 5-6 has been more typical for the past few years at least.

True, but today's was especially of rare quantity.

100 WUs per CPU, 100 WUs per GPU.

. . The RAC calculations are a strange and mystical thing. But if you are tracking your work output I suspect you will find it has not gone down, well certainly by that amount :)

Are they really at such a disadvantage compared to Intel CPU's?

The only CPU only tasks that I can think of is Arecibo vlar tasks on NVIDEA. That is tasks like 04fe09...vlar SETI does not send them to NVIDEA cards due to lockups (I believe).

My question is,

If you have room to add a third card in between your other two, Why not just move a card and see how it works?

I know the 1070's are a bit less power-hungry but I've got 3 of them sandwiched together in 2 different machines with no problems (not founders edition but still reference blower designs). I do ensure they're fed with cool air from the front of the case, though.

I would ask, are these the EVGA cards? Are they ACX design?X

Do they have backplates on them?

Have you checked to see if that gives the best throughput?

I tried running quad cards with backplates and no gaps and temperatures would just skyrocket. I tried two risers but that wasn't reliable for me. So now I'm running triple with one slot gap in between each of the cards, 91C/72C/84C

Depends what you're pasting it into. Look for 'Paste special - unformatted text', or take it via a plain-text editor.

Here is some data on run times for the E5-2670's at 3.0GHz with the AVX app.
32 threads: VLARs ~120m, Normal AR ~150m
16 threads: VLARs ~75m, Normal AR ~90m

Ideally, you want a very thin layer of compound, because the whole point of the compound is just to fill-in the pits in the porous surface of both the CPU's heat spreader and the base of the heatsink.

2) Process Monitor event log was pretty interesting; from the Registry var access pattern I could see it doing stuff for each card and then for each extra card past the first one there would be a flurry of activity followed by a 1.5 to 2 minute jump in the log. I could tell it was doing NVidia driver stuff because of the reg-key names and nothing else was going on in that interval.

There usually is a fast boot option in the latest BIOS. Also turn off any fullscreen MB logo posting. I just have a 5 second POST notice and that's it. This on Windows 7 64 bit. The Win 10 machine has the faster startup though checking the same amount of memory as the Win 7 machines.

1) Go to Power Options
2) Click “Choose what the power buttons do” on the left side of the window.
3) Click “Change settings that are currently unavailable.”
4) Under “Shutdown settings” uncheck “Turn on fast startup”.

I got the last bits to fully spool up the S2600CP2J with E5-2670's today. After a quick BIOS version update to add the turbo option it's cooking 32 CPU AVX tasks at 3.0GHz. The Noctura coolers seem to be keeping the CPU temps below 50ºC at the moment as well. A few hundred tasks should tell how it is going to settle in long term.

Nice! Let us know how it goes when you get it. Any thoughts on the Procs and memory you are going to be putting into it?

We're back and uploading is painfully slow.

Has anyone seen any good deals on dual LGA2011 motherboards recently for the Xeons E5-2670/E5-2660's? The CPUs are still going for very little, but the MBs seem to be a really hard thing to come by.
With my recent change in job and increase in finds I was hoping to put something along these lines together for my home VMware box.

You can tell it you only want it to use 0.2 CPU...but it is going to do whatever it wants despite what anyone here tells you.

-use_sleep will knock it down to some level but it will never be the 0.2 you want.

Good luck

Whatever it is that's causing memory usage to spike up to 4GB (and we did have that problem with an Astropulse (IIRC) application once, with a certain type of task), I very much doubt that it's the BOINC configuration. It might even be a glitch in the reporting tool, or as others have said the interactions of all the command line switches leading to the SoG application itself to over-commit single instances of resources, rather than allowing the multi-tasking tools in the OS to distribute them normally.

????

There is little benefit in running more than 1WU per GPU with the SoG application.
For best performance 1 CPU core per GPU WU is required.
If you want to run multiple WUs on multiple GPUs it would be worth reserving 1 Core for each WU.
Using app_config.xml is generally the best method; if you make a mistake you won't trash your cache which is what tends to happen if you make a mess of app_info.xml

Mike mentioned earlier in the thread that while you have 8 CPU cores, there are only 4 FPUs. 2 Cores share 1 FPU. If 2 GPU WUs are trying to share the 1 FPU, the work will stall.

Did you reinstall the drivers after the swap?

No

Thanks. I'll keep an eye on computer 8107155 - you seem to have had a high rate of inconclusive validations recently with the older v8.12/r3430, but none yet with v8.19/r3528. Let's hope that's another good sign.

....So why do Lunatics now?

Because you'll get to use the optimised CPU apps instead of using the slower stock CPU apps. ;-)

Cheers.

If you are running Lunatics, you should probably use the latest one if you want SoG

If you let the server send you wantever it wants, you will get a random sampling of the different work units.

I don't know what version of Lunatics you use but v45 beta 4 has the SoG app in it.

you probably should copy over the command lines to the .sah program as well.
Its seems that the server will deliver a cluster of .sahs and a then alternate with a cluster of .SoGs.

Just been posted in news

https://setiathome.berkeley.edu/forum_thread.php?id=80371

If you want to see the run times better, just stop your network activity for awhile and let some tasks built up waiting to be returned.

In this case, the answer is 18 May 2016 - which makes the stock versions (r3430) quite a bit older, and more troublesome, than the latest Betas (r3528). But hopefully all three sources (stock, Beta, Lunatics) will be brought into line in the next few days.

If I select the Performance tab in Windows Task Manager and slide the mouse above the values, you may notice that the mouse gets into edit mode even though the values may not be changed.

This means that the value is what the Operating System (Windows) is seeing.

Not sure about the solution, but curious to see if the substantially slower memory (~40%) has any impact in your crunching.

Windows repair install maybe ? Be (just an idea)

not in my neck of the woods.

The lifespan of a cpu/or PSU isn´t unlimited (even mainboard). What will u try first ?

I have seen that typically the limit is lifted by a factor of 4 as completed WUs are reported. So report 1 you get 4 back. It doesn't take long to rebuild cache.

When not the DRAM, how old is the PSU build in the pc ? And which is it ?

Are you overclocking your CPU?

Did you change the voltage in any way to the CPU?

Was there any unusual wonkieness during the shutdown process of that machine, or did it seem to go normally? Did you suspend/shut down BOINC before shutting the machine off? Are there any unusual messages in your log when it booted up?

*edit* good thought on the memory errors, you may want to blow out/clean the board, and try re-seating the memory, over time there can be issues if it doesn't have perfect contact on the connector, and can cause errors. If I suspect an issue like that, I take a normal pencil eraser and carefully clean all the the contacts, and then try installing it again, sometimes it helps, others it's just a bad DIMM.

Maybe a DRAM failure.

it's a 100% blanked work unit.

If I remember correctly, you need a certain number of validated work units before it establishes a correct credit for your system


AstroPulse v7 (anonymous platform, NVIDIA GPU)
Number of tasks completed 7
Max tasks per day 45
Number of tasks today 0
Consecutive valid tasks 12
Average processing rate 849.79 GFLOPS
Average turnaround time 1.04 days

I'm probably remembering this wrong.

Just tell us when it's finished, please, and works out of the box.

I've got to say it's a tempting build if you don't mind the power bills.

EXCELLENT! I will give that a shot, when looking at it, it looked sort of like a type of rivet or pressed stamping(which I thought was quite weird for something as delicate as a motherboard, but it is HP, so who knows?), as the holes didn't appear symmetrical, at least it seemed that way when I looked at it, so they must be pretty small torx bits. Any idea as to the size they take? I'd love to get that thing torn down and assembled tonite. Have to start digging for my torx set, though from what I remember, I may have to get some smaller bits, I think the smallest I have is T10 if I remember correctly, and I'd guess that that would be something between a T4 to T6? Thanks again for the great news.

Why do people keep saying there is a design flaw.

With a normal AR task running ~3.5 hours on a E5-2670 you would be looking at closer to 192-224 tasks a day for a 16c/32t system.

"Credit" should reflect the real work done, so every machine - regardless of it's speed - should "earn" the same amount (for this unit!). If a machine is slower, it gets less credit just because it can't crunch as much units, as a faster machine - simple as that. Same work - same credit.

Very tempted to get a tower unit instead of a stack of laptops.

Yes purchase price to performance is good, but compared to laptops, running costs are high. My stack of 6 laptops use between 3 & 20 watts so with average of say 15 watts, that should give total usage of around 90 watts. Tasks vary between 30 and 65 per day with credits between 3,600 and 3,900 per day. Two machines run Rosetta as well as SETI@Home and I have just added Enigma to a separate machine as well.

Perhaps some of our resident smart guys can fix it? I'm looking at you, Jason!!!!

I don't think it's fair to single him out. He already has a lot on his plate and donates his time and energy.

Everyone is in the same boat.

Oh I appreciate the 'faith', only warn that the time I have at the moment is pretty patchy/limited. Between all the apps being experimented with at the moment by others, and the knowledge we all have accumulated, I'm sure we'll find the better options. For now just assume it's going to take time, and there may not be simple one-size-fits-all solutions here. My contributions for the short term will probably be limited to testing some theories (with some special test pieces). For one thing it looks like speed/throughput on these is only a part of the problem, while scaling/load-balancing on vastly different machines may be a bigger part of the issues than before.

Unfortunately, I can't seem to find what temperature it is rated for.

see http://stats.free-dc.org/stats.php?page=hostbycpid&cpid=311e83fe46e6c6ae2d92e1d3d037110e
and
http://stats.free-dc.org/stats.php?page=hostbycpid&cpid=4c4c2c2ab5e7217bdfff066e2970f4ef

Scroll both of them down. The titan gets 54000 a day. It is a new machine and its RAC is still climbing.

Hey - I just noticed that I have NO GBT WUs on my GPUs now.

Been that way for about a week or so now, other than the odd re-issue.

When I was able to run only AP V6 only on my 2 meager Seti hosts I was pretty much maintaining a 8K RAC now they are MB V8 boxes, not by my choice and I am enjoying a tremendous drop. It seems like I'll never get that free toaster. I wonder if I continue to crunch 24/7 can I get my RAC down to zero?

To put some figures on the difference in runtimes: all tasks are VLAR, no -9 overflow, on I think my most stable machine.

Arecibo (23 tasks): average 5,325.73 seconds, std. dev. 159.50
Guppi (4 tasks): average 4,441.34 seconds, std. dev. 241.30

Not as big a difference as I thought, but noticeable.

I have 2 blc3 files on Syphilis. Estimated time of 1hr 12min. It'll take a little while to get to them though...

Are the GBT data sets more "channels" per set? Either not all of the loaded sets are being displayed or each one contains over 100 channels.

I tested further and found that if I run 8 GPU tasks, all 12 CPU tasks run at the same time but not with 9 GPU tasks (<cpu_usage>.04). So this doesn't quite fit what described by BilBg. I wonder what I am overlooking here.
If I suspend all GPU tasks, then all 12 CPU tasks run, of course.
Now I suspended all CPU tasks and observed the CPU usage by GPU tasks (Why didn't I think about this earlier?), and I saw 8 threads keep working for GPU tasks and 4 threads get down to 0, average CPU usage being approximately 14%. Meanwhile average GPU usage of 85% (with CPU tasks running) goes up to 95%.
I still don't have a clue how CPU contributes to GPU tasks but for my 3 GPUs running 9 tasks seem to require 15% of total CPU usage on average.
It looks like if I can dedicate 2 threads to GPU tasks (16.7%), I may be able to maximize my GPU usage.
But again, I don't know the internal work of CPU assignment by the software to GPU tasks, this may not be the case...
Will see.

Looking at my system usage, 8 cores + 2 GPU tasks would be 5.5MB, he only has 4MB. That's not enough.

Running 3 on a GTX 980.

In most cases, yes, you need to install the software driver. The onboard RAID function of most chipsets out there are "fake" RAID in that all the calculations are handled via the driver, and hence use up more CPU time.

Not on mine!

One of my machines is a 4790K (with HT OFF) with 2 x GTX980, and currently is doing 43K RAC stock apps vs your 23K RAC with 4790K with HT ON and 1 x GTX 980 running SoG. Seems to me that SoG has essentially NO advantage over the stock v8, then, to a first approximation. (I am running 3 WUs/GPU, but it has almost the same RAC as when I was running 2/GPU, judging by the slope of the STATS tab line in BOINC before and after that change).

Do math. 23*2=46>43.

After having done about 4000 WU's with the SoG app here on main, without any invalids or errors whatsoever, and showing that on my system at least, it's considerably faster than CUDA, I would say that it certainly have proven its usability (at least on my system.)

But until it is released here on main as stock, we will never know, how it reacts in the wild, so to speak.

Do you (or anyone else ) know if this is/will be an option in Lunatics?

Do you mean lunatics installer or what? App available on Lunatics in downloads sections more than week already...