Posts by MarkJ

21) Message boards : Number crunching : Ryzen 3000 head count? (Message 2004052)
Posted 24 Jul 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
I’ve got a Ryzen 3600 that replaced a Ryzen 1700. Three more on order.
22) Message boards : Number crunching : Ryzen and Threadripper (Message 2002992)
Posted 17 Jul 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
I assume you switched the XMP profile to 3200 and its still running at 2400?

XMP profile #1 is shown as 3200MHz CL16, profile #2 as 3000MHz CL15. The memory is HyperX Predator 3200Mhz CL16. I couldn't get it to select either XMP profile. I could set the memory speed to pretty much any value it had on the drop-down list but it goes back to 2400MHz when its rebooted. I've raised a bug with ASUS. While its not as fast as it could go at least its working.

Talking about out of stock, I ordered my first Ryzen 3600 on 7/7 almost as soon as they put them online. The next day they were out of stock. I went to order another 3 machines two days ago and they've been back-ordered. When I spoke to the sales rep they said they have 100 cpu's coming in at the end of the month. Motherboards are also out of stock.
23) Message boards : Number crunching : Ryzen and Threadripper (Message 2002978)
Posted 17 Jul 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
Have anyone reported some struggles with 3700X (Zen 2 series overall)
My 3700x started to reboot sporadic and only got worse and worse.
It started with a reboot in Ubuntu, tried to get it going, got it to work but then it was the same in Windows too, sporadic reboot

Sorry to hear of your troubles. My 1700’s sometimes just lock up and I have four of them, never have found out why.

So far my 3600 has been behaving well. The only issue I have had is with the mobo treating my memory as DDR4-2400 when it’s actually 3200. I’m blaming ASUS for that one and expect they’ll iron out issues with the bios in due course. Mother board is a X570-P running it on optimised defaults apart from fans which are set to turbo.
24) Message boards : Number crunching : Ryzen and Threadripper (Message 2002376)
Posted 13 Jul 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
Got my 3600 assembled and running. Unfortunately it picked up a pile of Milkyway (GPU) work which its working through now. Once that is done I will get it doing Seti work. Its this one

Motherboard is an ASUS Prime X570-P and CPU is a Ryzen 3600 (without the X).

It doesn't recognise my HyperX Predator DDR4-3200 memory correctly, thinks its DDR4-2400. I can manually set the speed to 3200 but it doesn't seem to allow me to select the XMP profile for it. As soon as I reboot its back to 2400MHz. According to the BIOS SPD info the memory has two XMP profiles, one at 3200MHz cl16 and the second at 3000MHz cl15. I downloaded the latest BIOS from ASUS site (0804) and updated it. but that made no difference. Have logged a bug report with them.
25) Message boards : Number crunching : [Guide] Debian 9 Stretch Linux with nVidia 418.x drivers and CUDA 10.1 (Message 2002226)
Posted 12 Jul 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
If you’re doing headless crunchers you can also use BOINCtasks to add projects and control settings. It’s a lot easier to see what they’re up to for day to day operations.
26) Message boards : Number crunching : Ryzen and Threadripper (Message 2002083)
Posted 11 Jul 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
I’ve got myself an ASUS X570-P with Ryzen 5 3600 to build as a GPU cruncher. It’s destined to replace an X370-Pro/Ryzen 1700. The GPU (GTX1660Ti) will be swapped into the new build.
27) Message boards : Number crunching : "Best" Am4 motherboards for Boinc/Seti crunching? (Message 2002080)
Posted 11 Jul 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
ASUS have an X570 workstation motherboard but it’s only got 3 PCIe x16 slots (which run at x8/x8/x8) but hey they’re PCIe4 so twice as fast as the previous generation.

https://www.asus.com/au/Motherboards/Pro-WS-X570-ACE/
28) Message boards : Number crunching : [Guide] Debian 9 Stretch Linux with nVidia 418.x drivers and CUDA 10.1 (Message 2002078)
Posted 11 Jul 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
I install the nvidia-kernel-dkms package rather than nvidia-driver:
apt install -t stretch-backports nvidia-kernel-dkms nvidia-opencl-icd-y

Also to save time I have the account* files on a USB key and copy them into the BOINC data directory /var/lib/boinc-client that way you don’t have to tell it to attach to the projects.
29) Message boards : Number crunching : Don't know where it should go? Stick it here! (Message 1997917)
Posted 12 Jun 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
I am looking for information about the use of a proxy with BOINC. I looked in the Wiki and got some,
but I noticed several errors, and the page has a reference to a Google group that maintains the page, but
that lands me in a page that is in Vietnamese. Yes, I live in Vietnam, but I don't read/write/speak it.

My question is: who to contact more directly, or at least an English webpage, to report errors in the Wiki?

Any help would be much appreciated.

I'm not the one to contact about the wiki. They moved the original mailing lists to google groups so that might be why you end up at a google page, and it probably thinks because you're in Vietnam that you can read the language, Maybe if you drop me a PM with the bits you think are wrong and I can post them to the list on your behalf.

I run Squid with BOINC and I don't have any issues with it. It works well with Einstein and their locality scheduler and its good at caching OS updates under Debian. Most of the other stuff just flows through it but there is no caching benefit as the files are usually unique.
30) Message boards : Number crunching : High performance Linux clients at SETI (Message 1997532)
Posted 9 Jun 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
After 2 weeks of running Mint my 2500K has made it to 80th spot by RAC (it'll hit a 70K RAC today) and a 4th day with over 100K worth of work returned while I've made it to 102nd by total RAC.

You know the next thing you'll have to do is replace those 1060's with something faster. I replaced 2 of mine with GTX 1660 Ti's and they seem to really be faster.
31) Message boards : Number crunching : Problems with Boinc Tasks and linux (Message 1997531)
Posted 9 Jun 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
There was no "remote_host.cfg" file and I had to create one using commands which I assume I did correctly as it appeared OK.

Totally out of ideas now!!

1. Create a remote_hosts.cfg to tell BOINC which machines are allowed to connect to it The BOINC event log will show you at startup if it found it and what machines you specified.

2. Create a gui_rpc_auth.cfg with some value in it - Put this value into the password column in BT. BOINC will mention if you don't have one in the event log as part of the startup messages.

3. Make sure the BOINC machine isn't blocking port 31416. A ping from the BT machine should prove if its blocked (ping will use port 80 by default). If it seems blocked try the following command on the BOINC machine "sudo iptables -L" (without the quotes and yes thats an uppercase L) and see what it lists. The computers tab in BT will tell you if it thinks its a password issue under the status column.

4. The manager needs to be given the password (the one in gui_rpc_auth). When I add a desktop launcher I have to add "--password xxx" to the end of the command. After that it will be able to connect to the core client. I am not using the all-in-one version so your mileage might vary.
32) Message boards : Number crunching : Ryzen and Threadripper (Message 1997161)
Posted 7 Jun 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
Back to the X570 boards, apparently the chipset won't work with 1st gen Ryzens. It only supports 2nd gen and 3rd gen. Not that I could think of any reason why someone would buy a new motherboard and stick a 1st gen CPU on it.
33) Message boards : Number crunching : User achievements thread......... (Message 1996630)
Posted 3 Jun 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
Passed 100k RAC for Seti. I haven't been able to run machines 24/7 for a long time due to the weather. The GTX 1660 Ti's make all the difference.
34) Message boards : Number crunching : has anyone retired old hardware for seti or bonic (Message 1995934)
Posted 30 May 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
I regularly replace my kit. The last lot I sold were eight 6th gen i7’s which were replaced with six 8th gen ones. That upgrade gave me an increase in core count while reducing the physical number of machines and wattage due to the new ones having more cores.

The next lot on the chopping block are 1st gen Ryzen’s in favour of 3rd gen ones (when they become available in July).
35) Message boards : Number crunching : GTX 1660 thread (Message 1994863)
Posted 23 May 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
I couldn’t get the EVGA cards I was after, nobody seems to be selling that model in Australia. I ended up getting a couple of ASUS dual slot cards. I then went through some pain with the drivers in Debian. After they fixed that I have them running in a couple of machines, generally overnight. Wattage is a little higher than the GTX1060’s they replaced. Task run times have dropped from 160 sec on the 1060 to 102 sec on the GTX 1660 Ti.

Not many projects support Turing based cards yet. Still waiting on Asteroids and GPUgrid to update their apps. Haven’t tried Einstein but their GPU apps are OpenCL so probably will work.
36) Message boards : Number crunching : Question about the special app for linux (Message 1992956)
Posted 7 May 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
Going off topic here. I recently upgraded one machine from a GTX 1060 to a 1660Ti. I took the opportunity to also upgrade from the CUDA 80 to 101 while I was at it. Below is the output from one of each. Should I be worried the the CUDA 10.1 has decided to use -pfp 1 on the GTX1660Ti while the GTX1060 decided to use -pfp 9? They both have autotune in the command line.

GTX 1660 Ti
unroll limits: min = 1, max = 256. Using unroll autotune.
setiathome_CUDA: Found 1 CUDA device(s):
Device 1: GeForce GTX 1660 Ti, 5914 MiB, regsPerBlock 65536
computeCap 7.5, multiProcs 24
pciBusID = 9, pciSlotID = 0
In cudaAcc_initializeDevice(): Boinc passed DevPref 1
setiathome_CUDA: CUDA Device 1 specified, checking...
Device 1: GeForce GTX 1660 Ti is okay
SETI@home using CUDA accelerated device GeForce GTX 1660 Ti
Unroll autotune 1. Overriding Pulse find periods per launch. Parameter -pfp set to 1

setiathome v8 enhanced x41p_V0.98b1, Cuda 10.1 special

GTX 1060
unroll limits: min = 1, max = 256. Using unroll autotune.
setiathome_CUDA: Found 1 CUDA device(s):
Device 1: GeForce GTX 1060 3GB, 3019 MiB, regsPerBlock 65536
computeCap 6.1, multiProcs 9
pciBusID = 9, pciSlotID = 0
In cudaAcc_initializeDevice(): Boinc passed DevPref 1
setiathome_CUDA: CUDA Device 1 specified, checking...
Device 1: GeForce GTX 1060 3GB is okay
SETI@home using CUDA accelerated device GeForce GTX 1060 3GB
Unroll autotune 9. Overriding Pulse find periods per launch. Parameter -pfp set to 9

setiathome v8 enhanced x41p_zi3v, Cuda 8.00 special
CUDA 8.0 Special version by petri33.
37) Message boards : Number crunching : Raspberry pi 3B issues (Message 1992953)
Posted 7 May 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
I have a big heatsink and fan on the pi with a quality 5V6A psu connecting directly to the gpio pins. The extra current is for future upgrade to add a battery backup and this will then be able to charge and run at the same time..

Normally you would power the Pi via the micro USB port. That gives the added protection of a poly fuse. Powering it via the GPIO pins is not recommended and 6 amps is way more than it would like. The recommended power supplies are only 2.5A and that is to allow for USB devices, they used to use 2A power supplies.

I have a couple of Pi's using a PiDrive 24/7 without any issues, so not sure if your SSD is part of the problem or not. I boot off the SD with them and then the rest runs off the PiDrive. That is the root partition is on the PiDrive and the SD card is only read when it boots up - no writes.

My crunching Pis use SD cards. The latest Raspbian had some SD card fix in it, so that might also effect your reliability. All are powered via the micro USB port (most are using USB chargers), but the two with PiDrives have the official 2.5A power supply.

I've suggest taking it back to a standard SD card setup using the micro USB for power and see if that runs reliably. You can then introduce the SSD into the equation and see if that effects it before going back to the GPIO power setup.
38) Message boards : Number crunching : Opinions requested from home Linux users (Message 1989998)
Posted 14 Apr 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
Can we nail whether this is happening with the distro repository installations, please? That's my concern in this thread, and a question I can't answer for myself. I have read about libcurl4 uninstalling libcurl3, about libcurl4 and libcurl3 (separate installs) not being able to run together on the same system, and about the existence of libcurl34 for (or is that 43?) which combines both.

But have the package managers cracked all that?

The reports I have seen only seemed to be Ubuntu users.. I haven’t had this issue using the packaged BOINC under Debian.

The current release of Debian (called Stretch) has libcurl3. The next release (called Buster) has libcurl4. They don’t have a libcurl34 package.
39) Message boards : Number crunching : Raspberry Pi & Other SBC Computers Discussion Thread :) (Message 1989949)
Posted 13 Apr 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
Looks like Newegg doesn't have the very Noctua fan kit you show on your blog. I did find it on Amazon though.

I got mine on eBay. The seller turned out to be Noctua direct. The local PC and electronics shops didn’t have 5 volt fans at all.

Since I have my Pis in a stack build,

In a stack arrangement the guys usually put larger (like 80mm) fans on the GPIO side of the stack and blow across. Sometimes they have one fan on top of the other which typically will be good for a stack of 5 Pis. The fans are still 5 volts though.

I don’t know if you’ve seen this video from our team founder on Einstein where he builds a PI cluster using a stack arrangement:
Raspberry Pi 3 Super Computer Cluster part 1
Raspberry Pi 3 Super Computer Cluster part 2
40) Message boards : Number crunching : Opinions requested from home Linux users (Message 1989784)
Posted 12 Apr 2019 by MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Post:
I have a fleet of 10 machines all running Debian and the repo version of BOINC (currently 7.10.2 from Stretch backports). I also have about 16 Raspberry Pis. They run a customised version of Debian Stretch known as Raspbian. The Pi’s run headless. All machines are dedicated crunchers.

I have used the manager to shut down the running client a few times in the past but I could live without it. I usually shut the service down using the service command if I have a need. Most command and control stuff I do via BOINCtasks on a Win7 laptop that is my daily driver.


Previous 20 · 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.