Posts by M_M

1) Message boards : Number crunching : Problems with MBv8_8.22r4008_avx2_intel_x86_64-pc-linux-gnu (Message 2033215)
Posted 20 Feb 2020 by Profile M_M
Post:
Any results yet on 4101? Is it stable and accurate? Is it faster then 3712 avx2?
2) Message boards : Number crunching : Keeping your rig running: UPS (Uninteruptable Power Supplies), Power Conditioners etc. (Message 2029026)
Posted 24 Jan 2020 by Profile M_M
Post:
Can we somehow manually merge it in existing Boinc version (if yes, how?) or we have to wait for new Boinc version?
3) Message boards : Number crunching : Weekend Warriors - aka: not crunching Seti 24/7 (Message 2017327)
Posted 31 Oct 2019 by Profile M_M
Post:
I leave PC crunching mostly during off-peak hours (0am-8am), when electricity is less then a third then a peak price, and one of cheapest in EU.

But if I would crunch 24/7, then not only the electricity cost for remaining day period would triple, but I would in much bigger part slide from "green zone" into "blue zone", when additional "~30% cost penalty" is applied. I tried once, and my monthly bill for that month increased substantially.
4) Message boards : Number crunching : Ryzen and Threadripper (Message 2015934)
Posted 19 Oct 2019 by Profile M_M
Post:

Ran Prime95 Small FFT test on my normal amount of cores and was hitting 83° C. That was the same temp I was getting on the custom loop. Now on my normal configuration as before and the normal loading I am sitting around 85-90° C. on the Noctua air cooler. I'm about 10° C. better on the loop. So not as good as the custom loop when it is working correctly. Which is what I expect to be so.


Good custom loop is for sure better then best air coolers, incl. NH-D14/15. One of the main reasons is much bigger cooling area, but also that heat sink with custom loop is typically outside of the case, where the air temperature is at least 5-10C lower comparing to cases even with excellent ventilation. So for apple2apple comparison of cooling efficiency, open case should be used.

Anyhow, back to your case,
1) Are you sure that vents on NH-D15 are running on their full speed when under full load?
2) Have you try to resit a cooler few times and also use a different amount and spread pattern of some good thermal paste, I use NT-H2. Maybe the contact area between NH-D15 base and CPU is not so good in your particular situation, could be related to surface shapes, orientation, or way how you tighten the cooler screws?

BTW, Also we me, all core SETI load with AVX is pushing few C higher temps then SmallFFT in Prime95 with AVX.

And one more thing I forgot to mention, I replaced my middle fan on NH-D14 to high static pressure one.
5) Message boards : Number crunching : Ryzen and Threadripper (Message 2015875)
Posted 18 Oct 2019 by Profile M_M
Post:
You are getting very good temps. Better than my custom water cooling. You must have very good case ventilation.


Reasonably good case ventilation yes, fans are TY-140, 2pcs for upper exhaust and 1 back exhaust, all nearly silent even under full speed.

But also Noctua NH-D14 is a awesome air cooler, price/performance better then all AIO water cooling system, even matching performance with entry-level custom cooling (at lower price point).
6) Message boards : Number crunching : Ryzen and Threadripper (Message 2015649)
Posted 16 Oct 2019 by Profile M_M
Post:
you have to do some sort of high end water loop. that just for the cpu. with a massive radiator

If you want to run the 3900X at something close to its core clock spec, you need water cooling for a 24/7 compute load. I have mine set at a fixed all-core clock of 4150Mhz which is where the cpu would run at on Auto if it had to run compute for 24/7. But I do it at a lot less voltage than what Auto shoves into the chip.


No need water cooling but need a really decent aftermarket cooler. I am running 3900x on auto (comes to ~4.1GHz on all core load) but with negative voltage offset -0,1V, so effectively ~1.23V CPU voltage, fully stable and full performance at just ~72-73C with Noctua NH-D14.
7) Message boards : Number crunching : Boinc on Ubuntu 18 LTS not detecting "on battery" (Message 2014240)
Posted 5 Oct 2019 by Profile M_M
Post:
Thanks Keith.

I see you already also reported to Github, and you started discussion on Boinc message board.
Please also report in Boinc message board your finding where in code this should be fixed, and we shall keep our fingers crossed that it shall be fixed soon... :)
8) Message boards : Number crunching : Boinc on Ubuntu 18 LTS not detecting "on battery" (Message 2014175)
Posted 4 Oct 2019 by Profile M_M
Post:
seems that upower correctly detect when the power is on-battery, but Boinc still doesn't react...

power on-line (idle):
user@ubuntu:~$ upower -d
Device: /org/freedesktop/UPower/devices/ups_hiddev0
native-path: /sys/devices/pci0000:00/0000:00:01.3/0000:02:00.0/usb1/1-5/1-5:1.0/usbmisc/hiddev0
vendor: American Power Conversion
model: Smart-UPS 1500 FW:UPS 09.2 / ID=18
serial: 3S1442X06625
power supply: yes
updated: петак, 04. октобар 2019. 21:47:45 CEST (22 seconds ago)
has history: yes
has statistics: yes
ups
present: yes
state: fully-charged
warning-level: none
time to empty: 1,5 hours
percentage: 100%
icon-name: 'battery-full-charged-symbolic'

on UPS power (idle):
user@ubuntu:~$ upower -d
Device: /org/freedesktop/UPower/devices/ups_hiddev0
native-path: /sys/devices/pci0000:00/0000:00:01.3/0000:02:00.0/usb1/1-5/1-5:1.0/usbmisc/hiddev0
vendor: American Power Conversion
model: Smart-UPS 1500 FW:UPS 09.2 / ID=18
serial: 3S1442X06625
power supply: yes
updated: петак, 04. октобар 2019. 21:50:45 CEST (12 seconds ago)
has history: yes
has statistics: yes
ups
present: yes
state: discharging
warning-level: discharging
time to empty: 1,5 hours
percentage: 96%
icon-name: 'battery-full-symbolic'
9) Message boards : Number crunching : Boinc on Ubuntu 18 LTS not detecting "on battery" (Message 2014098)
Posted 3 Oct 2019 by Profile M_M
Post:
I have a Ubuntu 18.04.3 LTS, Boinc manager 7.14.2 (x64), I have APC SmartUPS SMT1500i.

Ubuntu correctly detects UPS, apcupsd works ok, reporting status correctly, see below.

When going on battery, it correctly displays battery discharging status etc...

Under Options->Computing Preferances I select "Suspend when computer is on battery", but when goes on battery, Boinc still continue to work. Of course, selected is "Run based on preferances".

Any idea?

P.S. Same works correctly in Windows 10 (dual boot), but not in Ubuntu.

root@ubuntu:~$ apcaccess
APC : 001,028,0732
DATE : 2019-10-03 22:06:38 +0200
HOSTNAME : korisnik-ubuntu
VERSION : 3.14.14 (31 May 2016) debian
UPSNAME : SmartUPS 1500
CABLE : USB Cable
DRIVER : USB UPS Driver
UPSMODE : Stand Alone
STARTTIME: 2019-10-03 21:59:33 +0200
MODEL : Smart-UPS 1500
STATUS : ONLINE
BCHARGE : 90.0 Percent
TIMELEFT : 34.0 Minutes
MBATTCHG : 5 Percent
MINTIMEL : 2 Minutes
MAXTIME : 0 Seconds
ALARMDEL : No alarm
BATTV : 27.3 Volts
NUMXFERS : 1
XONBATT : 2019-10-03 22:01:15 +0200
TONBATT : 0 Seconds
CUMONBATT: 96 Seconds
XOFFBATT : 2019-10-03 22:02:51 +0200
STATFLAG : 0x05000008
MANDATE : 2014-10-20
NOMBATTV : 24.0 Volts
FIRMWARE : UPS 09.2 / ID=18
END APC : 2019-10-03 22:06:41 +0200
10) Message boards : Number crunching : I've almost been talked into purchasing a Threadripper 2950X CPU... Pro's & Con's? (Message 2011499)
Posted 10 Sep 2019 by Profile M_M
Post:
First impression I agree, it doesn't seem to be any faster then regular or even sse4.1.... ~same wu execution times for ~same credits.
11) Message boards : Number crunching : I've almost been talked into purchasing a Threadripper 2950X CPU... Pro's & Con's? (Message 2011338)
Posted 9 Sep 2019 by Profile M_M
Post:
Where to find AVX2 linux app AMS version, i.e. for Ryzen 3000?
12) Message boards : Number crunching : Heat generation only on Seti@home (Message 1999978)
Posted 27 Jun 2019 by Profile M_M
Post:
Lets simplify like this:
- with 10% clock increase, 10% more power is needed
- with 10% more voltage (on same clock), power increase is 20%
- with 10% clock increase and 10% voltage increase (so to get 10% faster CPU), power increase is 30%.

So it doesn't make much sense (except for some benchmarking) to increase a voltage, unless you have a really good cooling system and not care about power consumed.

IMO, reasonable overclock is what you can get on stock voltage while keeping a 100% stability....
13) Message boards : Number crunching : High performance Linux clients at SETI (Message 1992718)
Posted 5 May 2019 by Profile M_M
Post:
The most work we have ever been able to download for our work caches is 100 tasks per cpu + 100 tasks per gpu. So the days of work and additional days of work setting is meaningless on modern fast hosts. Maybe still applicable to phones and single board computers like the Raspberry Pi.


It would make sense that size of work cache is related to RAC... so maybe to keep 100+100 as default (minimum, appropriate for slow hosts) and to be increased to typical planned outage period based on RAC for each host. This way, everyone is happy, resources are optimally used, and also no "unnecessary" additional work is sent to hosts... Probably not so difficult to implement the logic...
14) Message boards : Number crunching : High performance Linux clients at SETI (Message 1992707)
Posted 5 May 2019 by Profile M_M
Post:
In Boinc computing preferences it is set (by default) to store at least 10 days of work, however, it looks that Seti project is ignoring this.
It seems that only several hours of work is downloaded and stored in advance (1 GPU WU is about 1 minute with 1080ti).

Is there any workaround for this, since otherwise every downtime longer then a few hours and machine is becoming idle?
This wasn't so apparent with Windows SoG app, since WU processing time was several times longer...
15) Message boards : Number crunching : Setting up Linux to crunch CUDA90 and above for Windows users (Message 1991669)
Posted 27 Apr 2019 by Profile M_M
Post:
Yesterday I pushed myself to install Ubuntu 18.04 LTS on external usb3 hdd and managed Boinc (TBar All-In-One package) with latest 0.98b1 Cuda10.1 to work with nVidia 418 drivers. :) Total time needed about 2hrs, counting from Ubuntu ISO download to all installed and crunching Seti WUs. It was a nice experience I have to say.... Now I have UEFI dual boot, Win10 and Ubuntu, so also chance to learn Linux/Ubuntu a bit more...

It really wasn't difficult - actually went easier then I initially expected even for last 12-13 years I had no any experience with any Linux distribution, but only Windows - back in mid 2000`s I worked with Sun Solaris 9 for some time at work.

0.98b1 seems like a really big improvement in crunching efficiency/speed on my gtx 1070 - according to WU completion times, I would say at 2.5-3x comparing to parameter optimized SoG on Win10.

I would like to thank Petri33 & TBar on awesome work done on customized app development, and also everyone else on guides/instructions/tips&tricks in this and other Linux threads on how to make it work.
16) Message boards : Number crunching : RTX 2080 (Message 1990914)
Posted 21 Apr 2019 by Profile M_M
Post:
Petri you continue to amaze me with what you have done with the Cuda code optimizations.
zi3v was fast.
v97b2 came with ~40% increase in computing.
Now 33% more on top of that.
Just WOW!


Now seems this 0.98 Cuda10 app is probably 3-4x faster on Linux then SoG on Windows (maybe even more) with same GPUs.
And since it is written in Cuda-C, it should be IMO 95% portable between Linux and Windows. So I not really understand why it is so big deal to make similar app version for Windows also?
Wouldn't this greatly improve overall SETI project crunching power, since most of users use Windows and will continue in foreseeable future.
Wouldn't it be much more sensible to invest time in Windows app then investing time for further speedups of Linux special app?
17) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1952668)
Posted 29 Aug 2018 by Profile M_M
Post:
SETI@home tasks work at 67 C but the PC reboots very often. SETI results complete and validate, but I have aborted a lot of them since I don't like reboots.


If whole PC is rebooting, it might be due to CPU/memory, not a GPU. I know that seti cpu apps, if machine or overclock is unstable cause a reboot.

If GPU is overheating or unstable OC, usually only driver reboots, not a whole PC.
18) Message boards : Number crunching : Setting up Linux to crunch CUDA90 and above for Windows users (Message 1952170)
Posted 26 Aug 2018 by Profile M_M
Post:
boom, truly an easy setup for people who dont know linux.

all they'd have to do is:
-burn the iso to a USB drive making it bootable with rufus.
-boot to the USB drive
-load BOINC and input your information
-configure your compute and whatever other settings (same as windows)
-done.


Yes, this would be really great if it can be done...
19) Message boards : Number crunching : Lunatics optimization for Ryzen, any plans? (Message 1901927)
Posted 19 Nov 2017 by Profile M_M
Post:
As the title say, are there any plans for this?

As I understand, Ryzen is pretty much different architecture then Intel, so it would make sense to get optimized path code for it, especially since there are more and more Ryzen and Threadripper systems out there...
20) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1899269)
Posted 5 Nov 2017 by Profile M_M
Post:
[quote]
The software being run has to actually be able to take advantage of that potential performance.
Which is why for a given Nvidia card the SoG application leaves the older CUDA applications way behind. And for the same card the Linux special application leaves the SoG application way behind.

So it is about how well is application is suited for particular architecture but mostly how well is written to use potential performance - and at the end it seems that currently, in general, nVidia is a bit better in SETI, and AMD is bit better in coin mining?

BTW if Linux special app is so much more efficient, why it isn't ported to Windows app? Is app so much reliant to underling OS, since CPU instruction set is the same and GPU drivers are probably very similar?


Next 20


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.