Posts by ivan

log in
21) Message boards : Number crunching : Anybody running SETI on a switch? (Message 1711668)
Posted 20 days ago by Profile ivan
Only 15TB? Try 32TB!

Well, OK, it's only a Mac Pro connected to a NetApp SAN, but I have to beat you in something, right? :-)

Well, it's an old server (20 or so 1 TB drives) that I rescued when it was retired for being out of warranty -- it's our group data server/backup (still 13 TB free). I'm thinking of slotting 4x 8TB Helium drives in the spare bays of one of my 20-core servers if I can convince The Professor to find the dosh, our sysadmin wants to get some idea of their reliability before we switch to them in ~2 years' time (he tells me we'll have something like 1.5 PB in our data centre when the current tender is met and delivered).
22) Message boards : Number crunching : Anybody running SETI on a switch? (Message 1711407)
Posted 21 days ago by Profile ivan
Anybody running SETI on a switch?

No, but I'm running it on a 15 TB disk server.
23) Message boards : Number crunching : extremely slow machine (Message 1711394)
Posted 21 days ago by Profile ivan
I also have a 24 core xeon machine. Comparing your APR of 9.78 GFLOPS to my hosts APR of 11.76 GFLOPS does seem that far off when you compare the CPU specs.

well i was hoping for a lil more oomph from these processors but i guess that is all i'm gonna get out of them unless i shove 7 video cards in it which brings me back to linux and gpu crunching which i think is a problem to get working if you even can get it working

if i install windows on it i doubt i can use all the processor cores in the machine

What are you using to monitor your CPUs? boincmgr (if you can get it working on your particular Linux...) will show how many jobs are running; top shows how many jobs are running and what the instantaneous CPU usage is; gkrellm will show graphs of CPU usage (I usually set mine to 300 secs history) and if you set up lm_sensors you can monitor temperatures and fan speeds, etc, as well (there's a plugin to show CPU speeds, too). gnome-system-monitor also shows individual CPU usage, but overlaid on one graph, so not ideal for a 24-core machine.
24) Message boards : Number crunching : Panic Mode On (99) Server Problems? (Message 1710971)
Posted 21 days ago by Profile ivan
I may have missed this, as I only read up 20 or so posts, but I have 348 tasks in progress, and normally I would only have 300. 100 for the CPU, and 100 each for my GPU's. Did something change?

Yes, you added a computer... 7715048 has 300 tasks, 5483835 has 48.
25) Message boards : Number crunching : extremely slow machine (Message 1710103)
Posted 23 days ago by Profile ivan

This machine here is an extremely slow cruncher

Anyone any suggestions to make this thing hum ?


I presume you have HyperThreading turned on. Your completed tasks for 100 credits take roughly twice as long as my 2x6-core Xeon with HT off so I don't think it's doing too badly. You need to wait a lot longer than two days to see what your RAC finally asymptotes to.
26) Message boards : Number crunching : Panic Mode On (99) Server Problems? (Message 1709846)
Posted 24 days ago by Profile ivan
"...everything is one inch longer than its length..."

Can I apply that recursively?
(Like I do with [Theodore] Sturgeon's Revelation: "90% of everything is crud!")
27) Message boards : Number crunching : SETI/BOINC Milestones [ v2.0 ] - XXIX (Message 1709036)
Posted 26 days ago by Profile ivan
Congratz on the Milestones guys, hit 5 million myself a while ago.

Hmm, just slipped up another point in the overall credit but WB5TUF - Glenn is coming up fast on my six. :-(
tottoatnorway is still leading me, but is his RAC slipping?
Tune in next week for a new exciting adventure of "Pigs in Spaaace!".

Moriarty: "You silly, twisted boy!"
28) Message boards : Number crunching : I seem to have found a Windows 10 / BIONC and/or S@H problem (Message 1707825)
Posted 29 days ago by Profile ivan
i can honestly say i have no issue with the intel igpu driver it works flawless on my i7 nuc and the i5 nuc

My little Celeron J1900 box (£130 at eBuyer) runs as a Win10 test-bed. AFAIK it's up to date with everything Microsoft throws at it. It frequently hangs on GPU jobs, and I have to abort them. Pity, because when they do run to completion they have about 60% of the performance of one of the four CPU cores...
29) Message boards : Number crunching : SETI/BOINC Milestones [ v2.0 ] - XXIX (Message 1707773)
Posted 29 days ago by Profile ivan
My little Tegra TK1 "Jetson" is just coming up to its first birthday (next Wednesday), and its accumulated credit is a shade over 500k! (~107,000th computer overall.)
Hippo, birdie, two ewes...
30) Message boards : Number crunching : New Machine i just bought to crunch (Message 1707458)
Posted 1 Aug 2015 by Profile ivan

What you guys think will that make a good cruncher ? Maybe stick a couple nvidia cards in it like 750 ti's or 960's?

Well, the rear view shows Dell's ID number (under the right-side fan) so you should be able to look up its delivery spec on Dell's support website (assuming it's a shot of the one they'll actually ship...). 16 Xeon cores should be OK, I figure on ~1000 RAC/core, depending on age (I don't run HyperThreading on our work machines, we need maximum single-thread performance as we don't multithread our applications yet).

[Edit] Checked out the ID. The unit pictured shipped with only two CPUs as far as I can see. The manual says that it has 3x PCIE_X4 and 4 PCIE_X8 slots, but it's not clear if the sketch is adequately representing the sockets. You may have trouble putting a PCIe_X16 card in them if they are short sockets. [/Edit]
31) Message boards : Number crunching : Current motherboards that can support 6+ GPUs? (Message 1707109)
Posted 31 Jul 2015 by Profile ivan
Hi all, I've done some searching but everything I've found is several years out-of-date, I was looking for motherboard recommendations that can support at least 6 GPUs. I've found many that have 6 or 7 PCI slots, but all I've seen only appear to support 4 GPUs. Any advice appreciated, thanks.
32) Message boards : Number crunching : Windows 10 - Yea or Nay? (Message 1706242)
Posted 29 Jul 2015 by Profile ivan
If 10 is the F'up I suspect, I may abandon Windows totally, and go for some Linux version.

I hear Linux Mint is nice.

Works OK for me. Put it on my XP machine at home when it lost its HDD; the AthlonX2 that handled 32-bit XP OK wasn't up to 64-bit Linux, so I swapped in an i7 motherboard & CPU. I think part of the problem getting used to it was the Cinnamon display manager, it's a bit top-heavy. I put Mint on an old Atom netbook, using Mate rather than Cinnamon and that seems to work well, though I haven't tried to do much with it.
33) Message boards : Number crunching : Panic Mode On (99) Server Problems? (Message 1706064)
Posted 28 Jul 2015 by Profile ivan
/me runs away, before someone finds the Straight Jacket.

Just a word of advice from a pedant -- it's a straitjacket, from the word strait meaning a hindrance, a hardship, a narrow part or passage; as in dire straits, or Bass Strait, Magellan Strait or Bering Strait, or strait-laced. :-)
34) Message boards : News : Big boost for SETI@home from Yuri Milner's Breakthrough Listen Initiative (Message 1703567)
Posted 21 Jul 2015 by Profile ivan
Wonderful news!!!!!

And all of you who have been crunching for 16 years were REALLY trend-setters!!!

I hope that doesn't make us hipsters then. 0.o
35) Message boards : Number crunching : GPU Curiosity Question (Message 1702634)
Posted 17 Jul 2015 by Profile ivan
I have a dedicated Nvidia 750 TI (no monitor attached) running one task and the temperature of the GPU is 50C. According to the SETI Task Page a typical AP Run Time of 2100 seconds results in a CPU (GPU) Time of around 80 seconds. There is one virtual CPU core free to support the GPU and "top" always shows the load on it at less than 15%. The question is, why would the GPU only show that it is working less than 5% of the time?

I have compared the server times using “ps ax” and the server logs to the times reported by the task page and they seem to agree. I suspect it is what it is, but it would be nice to know why it is. Perhaps the Linux algorithm for reporting the CPU time is flawed, but you would think the GPU temperature would be higher if that were the case. The reported temperature of 50C would certainly indicate that the GPU is not being overtaxed

Perhaps you are confusing GPU with CPU -- "top" knows nothing about GPUs, only CPUs. In fact, I know nothing that will give you an estimate of GPU usage under Linux -- nvidia_smi will tell you the GPU temperature, the percentage of the fan speed and memory usage. But it won't tell you the things that GPUz (purports to) tell(s) you about the GPU on Windows systems.
50 C is a doddle for a 750 Ti -- it will easily run two instances, where the one or the other takes over computing in the "slack" periods that the other job is transferring data to the host PC. Set it to run two instances and watch your RAC climb a bit -- and your GPU run 10 or so (Celsius) degrees warmer. (I've got a 660 Ti in this Linux machine, it's showing 63 C running two Astropulse WUs with an ambient of about 25 C; I also have a 750 Ti in a Windows machine at work, it handles two SETI jobs with ease, though I do notice a bit of mouse lag when one of its jobs is an Astropulse WU).
36) Message boards : Number crunching : Show and tell your machine. Here's mine. (Message 1702299)
Posted 16 Jul 2015 by Profile ivan
Loving that Outlander machine. How well does your Titan run Seti tasks? I was looking into used Titans, and Titan Blacks for the double precision for other projects.

Originally I had 2, but burned one out and was getting a RAC of about 55k. Now with the Titan and 2 750ti I'm getting a RAC of about 46k

They're definitely worth it, but a 780ti works better I believe and would be cheaper than even a second hand Titan.

And then when you really get down to it a 980ti spanks the both of them, so...

Having said that, if you can get a Titan for a good price (I got mine effectively for free) then they're certainly built for it!

Hmm, we've got two 980Tis on order for my cow-orker's N-body simulations. I'd hoped to snaffle his S2070 for one of my 20-core Xeons, but he has other ideas...
37) Message boards : Number crunching : Not that it matters one tiny little bit, but... (Message 1702075)
Posted 16 Jul 2015 by Profile ivan
In terms of Credit I am 99th. In terms of RAC I'm 34th. I only run 6 machines, and 2 of those are helping others.
Well done to you though, but you do have 37 machines :-)))

Just had a look at where I am in RAC. According to BOINCStats I'm second in the UK (and 34th worldwide, how meta!)

At least LHC is up and running again, so my friends at IC high energy physics have moved their farm back to CMS simulation and analysis now. A few months ago their RAC was over a million!
38) Message boards : Number crunching : Not that it matters one tiny little bit, but... (Message 1701646)
Posted 15 Jul 2015 by Profile ivan
(well, for some reason it matters to me)

Helped by the kind loan of 2 more Mac Pros, They may be older but they're cores for the cause.

Cores for the cause. You heard it here first folks! (if it's original, which I doubt, greatly)

Ivan, I'm coming your way :)

Yeah, you gots a couple of ciphers that you'll quickly dispatch, and then Richard next month. Gets a bit harder after that...
Anyway, congrats. I guess there has to be something that Macs are good for!
39) Message boards : Number crunching : Cant merge hosts. But they're the SAME host... (Message 1700772)
Posted 12 Jul 2015 by Profile ivan
So yesterday I had to rebuild Pluto as I managed to break something (un BOINC-related). I thought the backend would either - as in the past when I've had to do this - pick up that it was the same machine or allow me to merge the two instances of the host.

Not today.

What I'm being told is:

Processing pluto.local Can't merge host 7423311 into 7587350 - they're incompatible

Incompatible? They're literally the same computer with the same version of the OS...

The LAN IP address may have changed due to DHCP but that's about it.


Are you sure you're picking the right machines (you gots so many to choose from! :-)? Neither accounts were created yesterday: 7587350 is dated 29 May 2015, 7423311 stems from 2 Nov 2014. AFAICT none of your machines were inaugurated yesterday, so it must have already picked up an earlier incarnation (like my SNAFU with rm -rf * the other day).
40) Message boards : Number crunching : Bad News on BOINC funding (Message 1699110)
Posted 6 Jul 2015 by Profile ivan
Hmm, I wonder if all this is behind the lethargy that's overtaken CMS@Home-dev lately? I'm starting to seriously reconsider my position as its public face.
Interesting times, I'd always felt bending our workflows to fit with the BOINC model was a bit Procrustean, be interesting to see if this prompts a re-think.

Previous 20 · Next 20

Copyright © 2015 University of California