Posts by ivan


log in
1) Message boards : Number crunching : i7 4770K overheating when running seti@home (Message 1505615)
Posted 21 hours ago by Profile ivan
Hello,
I've recently got a new rig and try running seti@home on it, using the latest boinc currently available. I am getting high CPU temperatures where each core is reaching 100 degrees C after a maximum of half an hour (even if I lower max cpu to 70%).

I first suspected my stock fan, but Intel Extreme Tuning Utility stress test for half an hour didn't bring the cpu temps above 80 degrees C. So I'm starting to suspect that there's something wrong with the application itself.

Is there anyone else facing this issue? any ideas?

Specs:
- i7 4770k, stock fan, stock configuration
- MSI Z87 MPOWER MAX AC
- MSI GeForce Gaming N780 TF 3GD5/OC
- ThermalTake VN300M1W2N Chaser MK-I (3 default fans, back, top, front)
- Kingston HyperX Beast 16GB(2x8GB)
- windows 8.1 professional

I swapped in a new motherboard with a 4770 recently and found that running s@h it quickly ran into the "rev limiter" at 100 C, which dropped the CPU clock speed down to 3600 MHz from the normal 3900. I installed an Arctic Cooling Freezer 7 Pro rev 2 cooler and the temperatures dropped to around 70 C whilst still running at 3900 MHz.
Don't believe artificial tests -- always test with the production load.
Short answer: buy a better cooler, the stock fan is only good for testing that you've installed everything OK, not for extreme loads.
2) Message boards : Number crunching : Low-power Power .... New Arm SOCs (Message 1499858)
Posted 14 days ago by Profile ivan
New, cheaper, 64-bit MinnowBoards coming out from Intel too. Including GPUs on die. Whether they'll be any use for s@h, I don't know...


The Bay Trail range of processors is really going to hurt the ARM chips. The spec looks very nice indeed. It is nice to see open source GPU driver so people can built as they please. I think S@H already has Intel GPU support somewhere.

1 Lane PCIe, SATA and gigabit Ethernet, lovely.

I might have to try to get one when they become available -- my Atom/ION Acer nettop died after I'd switched it off for some power-distribution work. Never did get it to run again; I salvaged the HDD, memory, WiFi card and fan from it today and junked the rest, as they are moving us to a smaller office.
3) Message boards : Number crunching : Panic Mode On (87) Server Problems? (Message 1498282)
Posted 17 days ago by Profile ivan
Oh dear, that was just the sound of a GPU fan spinning down. No more AP WUs for my Linux machine at home. One day I'll get around to adding Nvidia MB capability as well; I tend to be lazy (and hayfever-afflicted) lately...
4) Message boards : Number crunching : Low-power Power .... New Arm SOCs (Message 1498279)
Posted 17 days ago by Profile ivan
New, cheaper, 64-bit MinnowBoards coming out from Intel too. Including GPUs on die. Whether they'll be any use for s@h, I don't know...
5) Message boards : Number crunching : I'm falling, I bought a parachute. From 100% AP, to 100% MB. (Message 1495988)
Posted 23 days ago by Profile ivan
Well, I'll continue, partly because I'm interested in where my particular setup levels out, and partly because I may be crazy :-)

Still? After all these years?


Still what? Interested, or crazy?

Ask Paul Simon.
6) Message boards : Number crunching : I'm falling, I bought a parachute. From 100% AP, to 100% MB. (Message 1495985)
Posted 23 days ago by Profile ivan
Well, I'll continue, partly because I'm interested in where my particular setup levels out, and partly because I may be crazy :-)

Still? After all these years?
7) Message boards : Number crunching : New HP Z400 (Message 1493335)
Posted 28 days ago by Profile ivan
The "big issue" issue is the 475W PSU. This is a gen 2 machine and I am trying to find out if it is a proprietary HP PSU or an ATX form factor PSU. If it's a proprietary PSU, I understand I am limited to a 600W HP PSU upgrade, if not then I can just add in a 850W ATX PSU. Anyone knowing anything about the HP Z400 would be a real help.

If it's like the proprietary power supply in my [(hp)] Elite 8200 it won't have any extra PCI 6- or 8-pin 12V leads to power the thirstier GPU cards. I didn't find the service manual for the Z400 on the [(hp)] web-site but it's probably buried there somewhere.
8) Message boards : Number crunching : Adding ssd to speed up projects (Message 1492277)
Posted 29 days ago by Profile ivan
One brand had a reputation of failing after one and a half years no matter how much it was used. The problem was a write to the same spot every hour when the computer was on. A firmware update brought all the dead drives back to life.

Our company purchased a 512GB OCZ SSD for a particular user who was running out of space on his 320GB mechanical. After imaging this drive with our corporate edition of Windows 7 the drive repeatedly failed to boot after POST unless you inserted a flash drive into a USB slot, then selected to boot off the SSD

I bought an OCZ drive last year when the drive in my XP machine died. Never got it to work reliably. Finally realised that the disk was SATA III, backward compatible with SATA II ... and that my motherboard was old enough to actually be SATA I!

It wasn't a diskpart error either... the MBR simply refused to present itself as bootable unless another drive was plugged in. After contacting OCZ they admitted this was a problem with their firmware but they recently filed for bankruptcy and refused to fix it since we hadn't purchased a large number of drives.

OCZ have now been bought by Toshiba.
9) Message boards : Number crunching : 32 or 64 (Message 1490975)
Posted 31 days ago by Profile ivan
Thank you for the rapid responses. I would doing normal things, no movie or storing large data bases. But, want to crunch Boinc Projects as fast as I can.

As Juan intimates, even 32 GB would be massive overkill. Here's a snapshot of my home PC, running 7xCPU WUs and 4x GPU (AP only) WUs (2 each on two GPUs):
top - 00:05:44 up 12 days, 3:17, 9 users, load average: 10.34, 10.12, 10.15
Tasks: 248 total, 12 running, 236 sleeping, 0 stopped, 0 zombie
%Cpu(s): 0.8 us, 3.5 sy, 92.6 ni, 3.1 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
KiB Mem: 8130840 total, 7299792 used, 831048 free, 312808 buffers
KiB Swap: 5119996 total, 1924 used, 5118072 free, 4697832 cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
21862 ivan 39 19 49748 47m 2040 R 99.8 0.6 167:09.44 astropulse_6.03
28792 ivan 39 19 106m 38m 12 R 99.8 0.5 2:38.56 setiathome_7.01
6079 ivan 39 19 49608 46m 1956 R 99.1 0.6 547:22.29 astropulse_6.01
24667 ivan 39 19 105m 38m 12 R 98.1 0.5 96:58.04 setiathome_7.01
24906 ivan 39 19 106m 39m 12 R 95.2 0.5 91:01.28 setiathome_7.01
22354 ivan 39 19 107m 39m 12 R 91.8 0.5 153:44.96 setiathome_7.01
27017 ivan 39 19 105m 38m 12 R 91.5 0.5 41:30.63 setiathome_7.01
26966 ivan 30 10 28.2g 47m 23m R 47.2 0.6 16:04.58 astropulse_6.07
24380 ivan 30 10 28.3g 90m 64m R 20.0 1.1 11:11.29 astropulse_6.07
25582 ivan 30 10 28.2g 47m 23m R 14.6 0.6 16:59.37 astropulse_6.07
22790 ivan 30 10 28.3g 90m 64m R 8.7 1.1 13:58.59 astropulse_6.07

top says I'm using 7.3 GB, but that includes various caches; the monitoring tool gkrellm says I'm actually only using about 2 GB. You can also discount the virtual memory (VIRT) column for the GPU tasks; there may be that much reserved but it's not actually used or even allocated.
In summary, I'd find it hard to believe that a normal home PC running s@h would ever need more than 16 GB of RAM, and as in my case 8 GB is more than likely to be enough. Any RAM over and above what's needed to keep the tasks in memory is effectively wasted.
10) Message boards : Number crunching : A quick power supply question before I commit (Message 1489516)
Posted 34 days ago by Profile ivan
And also the 850W would be able to handle dual GTX 650 Ti cards, which is the ultimate goal here.

The GTX 650Ti draws a maximum of 110W, add another 10-20W if it's an overclocked model. Even a quality 550W PSU would be capable, 650W overkill.

I've got 2 systems- an E6400 with a GTX 460, and an i7-2600 with 2*GTX 750Tis. With 2 LCD 22" monitors, total power consumption is around 500W. That's for both computers and monitors, in total.

My home machine has a GTX 560 and a GT 640, with an i7-4770K CPU. Its PSU is currently drawing between 250 and 270 W (so maybe 220 W to the system after PSU inefficiency) crunching 7x MB and 4x AP (two per GPU) WUs.
+------------------------------------------------------+
| NVIDIA-SMI 331.38 Driver Version: 331.38 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 560 Off | 0000:01:00.0 N/A | N/A |
|100% 78C N/A N/A / N/A | 359MiB / 1023MiB | N/A Default |
+-------------------------------+----------------------+----------------------+
| 1 GeForce GT 640 Off | 0000:06:00.0 N/A | N/A |
| 69% 66C N/A N/A / N/A | 132MiB / 1023MiB | N/A Default |
+-------------------------------+----------------------+----------------------+
11) Message boards : Number crunching : Panic Mode On (87) Server Problems? (Message 1485085)
Posted 44 days ago by Profile ivan
Yup, something must be terribly wrong with that UL server. It wasn't many days since it died the last time.

Check it out...marvin is back.
Maybe some APs? Maybe I can use my Mac again?
Maybe...

Yeah, noticed that too. I'll go against my AP only principles for some time though, and start crunching MB only, even if AP comes back now. Just to see how low my RAC will drop from the level it is at now.

Tja, but server status page seems to be b0rked, it's over 30 minutes stale (as of a minute ago).
12) Message boards : Number crunching : Panic Mode On (87) Server Problems? (Message 1484549)
Posted 45 days ago by Profile ivan
What about the CPU time? Once you report, the task isn't running anymore? I just wonder what the difference between running time and CPU time is then.

The running time is the time that accumulates on your wall-clock while the job is running; the CPU time is the time that the CPU is actually processing the work-unit. CPU time will almost certainly be smaller than run-time[1] because of overheads that the CPU has to process instead of your job while the wall-clock is still ticking. So if you do a lot of other processing, say word-processing or game-playing, your run-time might be considerably greater than the CPU time. An extreme example can be seen with some GPU processing -- the time spent by the GPU isn't counted against CPU time, so if the GPU is doing most of the work the CPU is either idle or doing other stuff so you can have an elapsed time in minutes or hours (however long the GPU took) but CPU time measured in seconds.

[1] If we ever get into multi-threading or other parallelisation techniques, CPU time can exceed run-time because more than one core runs at once. I had an extreme example of that recently when I ran 180 parallel threads at once on a Xeon Phi -- elapsed running time was seven minutes or so, accumulated CPU time was well over twenty hours!
13) Message boards : Number crunching : Panic Mode On (87) Server Problems? (Message 1483318)
Posted 48 days ago by Profile ivan
Uploads are running again.

...not really...

Sat 01 Mar 2014 17:59:44 GMT | SETI@home | Started upload of 22my13aa.30113.1472.438086664200.12.76_1_0
Sat 01 Mar 2014 18:00:20 GMT | SETI@home | Finished upload of 22my13aa.30113.1472.438086664200.12.76_1_0

Close enough for Government business!
14) Message boards : Number crunching : Panic Mode On (87) Server Problems? (Message 1483315)
Posted 48 days ago by Profile ivan
Uploads are running again.
[Edit]Mustn't have noticed that the thread had overflowed to a new page...][/E]
15) Message boards : Number crunching : An unimagble of number of #x (Message 1483313)
Posted 48 days ago by Profile ivan
Too much beer last night.

It's OK, you can say olut here!
16) Message boards : Number crunching : Panic Mode On (87) Server Problems? (Message 1483309)
Posted 48 days ago by Profile ivan
OK, since AP probably will be unavailable for some time, I did finally bite the bullet, and installed the apps for MB v7, CPU as well as GPU. (manually of course, no installer for me, as long as I still have an intact brain.)

MB v7 will be my backup project, until AP is available again. :-)

I'll probably have to do the same tomorrow, enabling GPU MB work on my home (Linux) machine. petri sent me the mods to do it a few weeks ago but the machine has been unstable, dying irregularly (at least three times while I was sat at the console). Latest change has seen it stable for several days -- when I got the new motherboard and CPU I got 1600 MHz memory to go with it, but the BIOS only wanted to run at 1333 MHz so I manually set it to 1600; I've now let the BIOS have its way but it's still too soon to say "Cured!" as, for example, I haven't had much GPU AP work since. But, with no APs to crunch, the GPUs are idle and my RAC is dropping like a feather in a vacuum...
17) Message boards : Number crunching : Linux vs Windows Performance (Message 1482328)
Posted 50 days ago by Profile ivan
Windows (32 bit) is running SSSE3 MB app. My linux client (64 bit) is only running SSE2 MB app. What I've discovered is that /proc/cpuinfo only has a flag for ssse3 not sse3. My script to detect the correct MB app is looking for "sse3" which does not exist. I did this to match the file names and did not realize the "sse3" mb app can be "used with sse3 or higher" and never noticed ssse3 which appears later in the flag listing.

Anyway, first thing I need to do is rewrite the script to look for "ssse3" at which point it will install the ssse3 app not the sse2 app. Will fix that before investigating the power settings, but thank you for the lead.

SSE3 is sometimes flagged by the CPU as PNI (Prescott New Instructions).
18) Message boards : Number crunching : CLOSED - SETI/BOINC Milestones [ v2.0 ] - XXVI - CLOSED (Message 1481497)
Posted 53 days ago by Profile ivan
Just registered 300 million for BOINC Combined.


Well done, Richard. Sometime Wednesday I should move up another notch on the S@H leader-board, but it's a big stretch to the next victim after that.
19) Message boards : Number crunching : nvidia GT630 rev 2 (Message 1476040)
Posted 66 days ago by Profile ivan
i have also some questions - what is with VRAM on boinc projects and seti - 1 Gb is always more than enought, or better is get 2 Gb version?
on that cards ( 630,640,650) is any gain to run more than one task in parallel? if yes, what is about optimal number?
any other advances?

Well, I'm currently running two astropulse jobs on both my GTX 560 and my GT 640. nvidia-smi sez:
+------------------------------------------------------+
| NVIDIA-SMI 331.38 Driver Version: 331.38 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 560 Off | 0000:01:00.0 N/A | N/A |
| 92% 75C N/A N/A / N/A | 344MiB / 1023MiB | N/A Default |
+-------------------------------+----------------------+----------------------+
| 1 GeForce GT 640 Off | 0000:06:00.0 N/A | N/A |
| 58% 61C N/A N/A / N/A | 132MiB / 1023MiB | N/A Default |
+-------------------------------+----------------------+----------------------+

The 560 is my display card, so it's got some overhead for that, but both are using well under half the available memory. Similarly with my Win7 laptop, currently running 1xAP + 1xMB on a GT 555M and using ~350 MB (its display is probably running on the inbuilt Intel GPU at the moment); IIRC MB workunits take up a bit under 256 MB. I doubt you'll see any advantage running more than two jobs simultaneously, but by all means try it and see.
20) Message boards : Number crunching : Как заставить работать Intel GPU? (Message 1474642)
Posted 68 days ago by Profile ivan
Just wondering, if even with the google translator i can´t understand what you are talking about in Russian. Now imagine if we finaly receive an ET call, how we will do to understand it? Maybe someone will need to create an enigma like project to try to decode the message?

Saw an interesting snippet the other day (The Register?). Google translate uses artificial intelligence rather than language knowledge per se to do its translations. But Google are finding more and more pages on web-sites that use Google Translate to provide multi-lingual offerings. So Google Translate is starting to be poisoned by its own output!


Next 20

Copyright © 2014 University of California