WCG THOR Challenge 2020

Sounds like a cool event. Count me in. In addition to my normal cores, I scraped together an additional 5 cores from my ESXi hosts that probably wouldn't get much use otherwise. Its not much but but I hoping to build a new six or eight core rig before the end of the year so maybe that will be ready for the Christmas Challenge!
 
Only 6 more cores? Are you going to yard sales and swap meets?...haha j/k Glad to see you around. It has been a while.
 
Last status update from Friday on the Threadripper.


kylepolen2017.png
09/11/20 01:45 pm by Kyle

Location: Installation Status: Testing
The Operating System has been installed, updated and configured as requested. Your order is now ready to begin our benchmark suite in order to ensure safe hardware temperatures, system stability and expected performance.


Possibly could ship sometime next week. 🤞
 
Final Week 1 (first round) standing. The team moves to the second round (aka Week 2 challenge) which starts on Sep 21st to Sep 27th which is also based on the total run time.

The top 50 teams from Week 1 will now compete in Week 2 and the top 25 teams will then move to the Week 3 challenge.

1600044865375.png
 
Nice job guys. Use this time to tweak and make necessary adjustments. This is a great challenge to get things fine tuned and ready for the WCG birthday challenge in November. Find the settings that work well and practice bunkering if that is something you have interests in.
 
It's getting close 👍

09/14/20 01:58 pm by Puget Systems
Location: Quality Control Status: In process
As part of our advanced testing, we use thermal imaging. This process gives us crucial information about your computer's temperatures and helps us to improve the reliability of your computer by optimizing how heat is dissipated. The image on the left shows your computer at idle temperatures. The image on the right shows your computer at full load temperatures.
IdleLoad


Will see how it does once it's running Open Pandemics on all cores/threads. Hopefully will have it in hand before the start of week 2 on the 21st.
 
If you really want to see how it holds up in between weeks, run some PrimeGrid work units through it. They will stress it harder than anything WCG can throw at it...lol.
 
Prime 95 was always an old standby CPU/RAM load tester for stability (besides looking for prime numbers). So I'm guessing PrimeGrid would do the same. 👍 Since this company's primary customer base is workstations for content creation, engineering, and scientific companies their range of benchmarks they test with looks to be oriented to that base, mostly content creation. So some old D.C. guy buying a threadripper for D.C. is a little out of their norm. However it means they're going to get input from our side of things as to how their systems stand up under this kind of load, and if they can't we'll then see how receptive they are to the feedback. But good or bad I'm going to be passing it along to the company's founder and president (Jon Bach @jonbach) who I've already had some conversation with when I was asking why I should let them build me a system vs just putting it together myself.One thing Kyle showed us over the years was test the shit out of it and report those findings good and bad, just be fair.

Looking at the list of their benchmarks they ran on it it shows Nbody CUDA, Unigine Superposition 1.0, PugetBench for Photoshop, NeatBench, Lame 3.100, CineBench, V-Ray Next Benchmark, and CrystalDiskMark. Didn't really care much for any GPU results because I got the "cheapest" card I could knowing an Ampere was what would be going in it once they can be purchased. I was specifically interested in the disk tests because I went with a faster SSD due to the system being PCIe 4.0 and was curious if that would show, plus of course Ampere is able to take advantage of it too. Looks like the SSD is doing good with it.

Sequential Read (Q= 32 T= 1)C: (Seagate FireCuda 520 SSD ZP1000GM300...5011.618 MB/s
Sequential Write (Q= 32 T= 1)C: (Seagate FireCuda 520 SSD ZP1000GM300...4294.348 MB/s
Random Read 4KiB (Q= 8 T= 8)C: (Seagate FireCuda 520 SSD ZP1000GM300...1782.624 MB/s
Random Read 4KiB (Q= 8 T= 8)C: (Seagate FireCuda 520 SSD ZP1000GM300...435210.9 IOPS

In any case should mean a few more cores working for the team once it gets here and online.
 
Some do, some don't. I have WCG going all the time at least to a certain degree.
 
I had wanted to run for the first week but temps hadn't dropped yet so I held off. For the next week we're looking at most days of 80F or under so I'm firing my main system back up.
 
Trouble with getting the new system. It's possible it may not happen. I had taken the current system earlier today and set it to not download new units so it could finish out what it had. But due to this new development I've had it return to downloading and processing WU as normal for this next week's contest.
 
Trouble with getting the new system. It's possible it may not happen. I had taken the current system earlier today and set it to not download new units so it could finish out what it had. But due to this new development I've had it return to downloading and processing WU as normal for this next week's contest.
Hopefully not lost in shipping
 
Hopefully not lost in shipping
Haha, in a manner of speaking. It appears to be "lost" at the local Fedex facility sitting on the delivery truck since 3.30am yesterday morning. Supposed to be delivered yesterday, didn't happen. Then they change the status to "delayed due to weather" till Tuesday because it started a light rain in the area last night. This morning it's back to today. So I'm guessing it will be here next week sometime as the possibility for rain is out there.
 
UPS has been delaying a lot of packages at the destination facility a day or two because of capacity issues. That is probably all it is with FedEx too, unless you have actually been having bad weather recently.
 
UPS has been delaying a lot of packages at the destination facility a day or two because of capacity issues. That is probably all it is with FedEx too, unless you have actually been having bad weather recently.

It was clear most of the day yesterday, until the early evening when it started a very light rain, barely getting the ground wet. This was their "weather delay"

System has been delivered and it was raining harder today.


1600715966710.png
 
Week 2, Day 1 Result. We are still in top ten.

1600742687952.png
 
  • Like
Reactions: EXT64
like this
you’ll have to let us know, on a work per watt basis, how the pis hold up against the thread ripper
To achieve comparable PPD/watt as the pis, it is likely that one would need to undervolt and underclock the beast. Won't get that crunching power efficiency at default/stock settings but the beast can return the results much faster at stock ;)
 
My 3950x running 32 threads on OP draws less than 130W from the wall ;)

For the first 4 Thor events, only total run time counts. Might as well dial down the power and keep all cores running, lol.
 
You might be the only person I know who has both a threadripper and multiple raspberry pis running distributed computing.

you’ll have to let us know, on a work per watt basis, how the pis hold up against the thread ripper

I'll compare it with Pi that's been running O.P. the longest and started back in June sometime. Once the TR reaches its point/time value will compare how long it took 350watts to reach the 5 watter.
 
Doesn't look like this is going to take very long. T.R. should tie or just pass that Pi by this reporting time tomorrow.

1600822790454.png



EDIT - Looks this morning and the T.R. is already pretty close. So by this evening's update it should be passed the Pi by noticeable numbers.


1600868931288.png


EDIT 2 - yeah the T.R. blew by the Pi in less than two days.

1600910173207.png
 
Last edited:
It's no Threadripper by any means but just for the hell of it and for the runtime part of the competition I decided to install BOINC on my old Samsung Galaxy S6 I don't use anymore. Hopefully it doesn't catch anything on fire. The phone definitely gets warm when under heavy load.
 
It's no Threadripper by any means but just for the hell of it and for the runtime part of the competition I decided to install BOINC on my old Samsung Galaxy S6 I don't use anymore. Hopefully it doesn't catch anything on fire. The phone definitely gets warm when under heavy load.
Don't forget to test the ultra power saving mode. I do not recall if that stopped BOINC entirely or just slowed it down. Worth a try to reduce heat and still possibly get runtime.

I have an S6 that I replaced the battery in but ruined the screen to where I cannot see it anymore. So, it just crunches until it dies entirely now. I unfortunately did not have the remote software installed and configured on that one....
 
Back
Top