• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

[H]ardOCP vs TeAm Anandtech Folding@Home race: December 1st - 8th

biodoc

Diamond Member
Dec 29, 2005
5,357
525
126
We've set up a friendly challenge between our friends at [H]ardOCP and the TeAm on Folding@Home. :)

1) There's no need for individuals to sign up. Only team stats will be followed so anyone folding for both teams will be participating whether they know it or not.
2) Starting/baseline team stats will be recorded on December 1, 12:00 UTC from EOC stats, so only points recorded after that time will count.
3) The race end date will be December 8th, 12:00 UTC
 

TennesseeTony

Elite Member
Aug 2, 2003
3,488
2,105
136
www.google.com
Doc I don't see a pic...just a placeholder icon.

NOOOO. I lost a good man sometime during the night! My shiny new R9-3900X and 1080Ti were powered down this morning.

My thought process was as follows:
  • ??? Why did I power that one down? Hit the power button. Nothing. Flipped the power switch on the PSU, then back on. Yeah, it's booting. Made my coffee, turned around and....It's off again. :(
  • Hit the power button and nothing. Great. PSU perhaps. Hmm, it is a 660w unit, that's good. It is a SeaSonic, that is even better. And it is a Platinum unit, that is GREAT. Surely it is not the PSU.
  • I did change the heatsink from stock last night. The stock cooler was buzzing and shaking the case running wide open, so I swapped in a CoolerMaster 212 LED, with twin 120mm fans. The CM fans included have NEVER lasted more than 90 days for me. Perhaps the fan locked up, showed zero rpm, and the motherboard went into protection mode and shut down?
  • I'll have to deal with this later

Now it is later.
  • I removed it from the server rack (kitchen table) and pulled off the suspect CPU fan, leaving the 2nd one installed (different brand), and plugged the remaining fan into the CPU_1 fan header, rather than CPU_AUX/Pump.
  • Tossed in a Windows SSD, installed RYZEN MASTER suite to get CPU temps. Man these things run hot.
  • Downloaded some LHC, CPU 100% load, 4GHz, total draw of the socket was 120w, 83C.
  • I know this cooler can cool a 120w XEON 14 core down to 56C. I seem to recall some discussion on how these Matisse chips measure temp at a different location or some such, so I know the cooler is capable, the built in protections are capable, I just won't look at the temp again. :D
The main point is so far, so good. Aborted a few LHC tasks, fired up Folding, still on windows (this SSD appears to have been previously paired with this motherboard, Windows is activated). It has run for more than 2 hours now, and all tasks will finish in about one hour. Then I'll put it back on Linux and see if the problem comes back. So I'm down about 1M ppd for the day. :/
 

StefanR5R

Platinum Member
Dec 10, 2016
2,843
2,749
106
My X99A with a Xeon in it (and 3 GPUs) switched itself off during the Rosetta sprint. I believe it was because I had switched the BIOS to XMP profile a couple of days earlier, out of curiosity if it does anything at all with the Xeon. The first one or two power-on attempts made it switch off itself immediately again, but then I could enter the BIOS and switch back to auto RAM settings. --- This was my designated appartment heater, so I'm glad it behaves again.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
18,948
6,271
136
Tony, due to these chips thermal density, they run kind of warm. Mine is on a 240mm aio, and its 74f in the house, but the CPU is running 80c. My E5-2683v3 chips did not run as warm. I have however overclocked mine....
I would seriously recommend a 240mm aio long term. And in linux, psensor works great for temps.
 

VirtualLarry

Lifer
Aug 25, 2001
45,700
4,121
126
My main rig, with R5 3600 CPU (mining), RX 5700 (mining), and GTX 1660 ti 'Gaming X' (Folding), seems to be having some issues with the "Folding".

I don't know what's going on. It start off OK, and F@H was reporting 600K PPD for the first few days / WUs. (I had updated all three of my rigs with GTX 1660 ti cards with updated 441.41 drivers for Win10 64-bit DCH edition. Haven't checked the other rigs yet.)

For some reason, at the start of the race, F@H was allowing my RX 5700 to fetch WUs, and they had valid collection server IPs. But they all immediately errored, and then that folding slot showed a red "Error", until a reboot.

I had a problem, where my PC essentially locked up, and then it BSOD'ed with "DPC_WATCHDOG_TIMEOUT" (some kind of serious system hang).

My GTX 1660ti is on a PCI-E 3.0 x4 slot.

Anyways, I had problems AGAIN, and my F@H PPD was down under 10K PPD, and the GTX 1660ti appeared stalled for some reason.

(*I'm not saying it's NOT the card, or my mobo, or CPU, because it was stalling while mining occasionally too. Temps seemed OK.)

So, I rebooted, re-installed 438.xx. (Grr, can't bring up NVidia control panel, says that I don't have a display attached to an NVidia GPU. I thought NV control panel still worked, with an NV GPU as secondary, without a screen connected? Is this new?)

I'm running F@H again, and installed of re-starting the stalled WU, it fetched another one, and it's showing 600K PPD again.

Part of the issue is, the CPU mining uses all 12 cores, for this new algo that debuted a few days ago, and thus far I have been unsuccessful at trimming it's thread usage down to 10 threads, as I had it before with the previous mining algo.

I'm just at a loss to understand, if the CPU usage of the mining was blocking the Folding, then why, when I stopped the miner, didn't the Folding pick up again? It seemed stalled.

Another theory, is that the RX 5700 fetched a WU, worked a little on it, and then the GTX 1660ti took over that WU, which had been "maligned" by the RX 5700, and didn't know what to do with it, and the folding process hung.

Do different Folding devices, take over stalled WUs that had previously been assigned to other devices?
 

motqalden

Junior Member
Jun 13, 2009
16
16
81
I had a system lock up last night due to random windows update, but it was just my MW / WCG rig so no impact on this race.
 

motqalden

Junior Member
Jun 13, 2009
16
16
81
I finally tamed the temps on my 3900x and by "tamed" i mean its not hitting 90c regularly anymore and mostly around 82c . This is in a case with a TON of airflow and an NHD15S I do happen to think my motherboard is feeding it a bit too much voltage even with -.075 offset... My 3950x is running a 240 AIO and it rarely gets over 70c but it just seems to run way cooler since i had the same cooler on the 3900x and it didn't do much better than the Noctua.
 

biodoc

Diamond Member
Dec 29, 2005
5,357
525
126
I'm just at a loss to understand, if the CPU usage of the mining was blocking the Folding, then why, when I stopped the miner, didn't the Folding pick up again? It seemed stalled.
I'm not sure it will help your problem but you can set FAH priority a little higher. In Fahcontrol got to configure and then click on the advanced tab. There is an option to set FAH priority there (default or slightly higher.

I finally tamed the temps on my 3900x and by "tamed" i mean its not hitting 90c regularly anymore and mostly around 82c
@StefanR5R posted (#207) an interesting table from a German forum where they changed bios cTDP and PPT values to set limits on power draw on a 3900X. On my 3900X, ff I set cTDP to 105 watts and PPT to 105 watts (default is 142 watts), then my CPU freq is ~3.8 GHz under full load. Power draw from the wall drops by 50 watts too from bios defaults. After the FAH race, I'll probably boot into windows and see what the impact is on CPU temps. I'm assuming lowering PPT will lower CPU temps.
 

VirtualLarry

Lifer
Aug 25, 2001
45,700
4,121
126
I'm not sure it will help your problem but you can set FAH priority a little higher. In Fahcontrol got to configure and then click on the advanced tab. There is an option to set FAH priority there (default or slightly higher.
But is that "slightly higher" than idle priority or "slightly higher" than normal priority (what the miner runs at)?

Edit: Yeah, 'doc, I can see the image. Looks good so far.

And with only F@H running on the GTX 1660ti (no mining), I've maintained 600K PPD on this PC.
 

[H]Coleslaw

Member
Apr 15, 2014
120
74
101
Not many of the FAH crew left unfortunately. And the few from those days left are still PO'd at Pande Group. So, this is probably all she's gonna get unless someone just crawls out of the woodwork and surprises us.
 

ASK THE COMMUNITY