15th Annual Folding@Home Holiday Season Race: Mark's Marauders win!

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Endgame124

Senior member
Feb 11, 2008
955
669
136
Yeah, even old parts have shot up due to the lack of availability of the new stuff.
That’s why I was able to sell my 1660 super for what I paid for it.

I’m half tempted to sell my 1080ti as well and live without a video card for a few months
 
  • Like
Reactions: Ken g6

VirtualLarry

No Lifer
Aug 25, 2001
56,325
10,034
126
I'm in, after all, that's why I mine on my cards (excluding PrimeGrid races that can use GPUs) all year, to pay for my upgrade(s) for F@H in Dec.

I have:
(main rig)
Ryzen R5 3600
RX 5700 (reference blower model by XFX)
GTX 1660ti 'Gaming X'

(secondary rig 1)
Ryzen R5 3600
RX 5600XT (refurb)
GTX 1650 D5

(secondary rig 2)
Ryzen R5 1600
RX 5600XT (refurb)
GTX 1650 D5

(temp test rig)
Athlon 3000G @ 3.90Ghz (OCed) DDR4-3200
EVGA GTX 1660ti triple-slot

(mining server)
2x MSI GTX 1660ti 'Ventus OC'
3x Gigabyte GTX 1660ti 'ITX single-fan'

Note that I've previously tested the mining server, and the ENTIRE mining server, due to the slow dual-core Celeron CPU and the use of 1x PCI-E mining risers, basically limits the PPD of all five GTX 1660ti cards, to around the output of the SINGLE GTX 1660ti 'Gaming X' card on my main rig. So the 'mining server' will NOT be participating in the F@H race, sorry. (Plus, I still need some residual income during the month, it makes $3-5 / day.)

The main rig, I have no problem throwing on F@H on both GPUs, although I might continue to mine on the CPU.

The secondary rig 1 and 2, are mostly similar, I could put them on F@H, possibly, but I worry a little bit about the power draw and them shutting down, I've had some minor issues in the past. (I have them "power turned" for ETH, and they take less than stock power.)

The temp test rig, only has a 470 Apevia (Gah, yeah, I know, I know) PSU, with the 8-pin GTX 1660ti, connected to a SATA-to-6-pin, connected to a 6-pin to 8-pin. "Power tuned" to mining ETH, power draw is not too bad, and hasn't even crashed or been flaky at all. Seems OK, in current situation. Unsure what F@H would do to that rig, I would probably at a minimum swap in a better-brand PSU with a factory-spec 8-pin PCI-E cable, before turning that one onto F@H.

Edit: Forgot one rig, I pulled out a "mining shell" with a Raidmax Cobra case, a G4560 Kaby Lake 2C/4T CPU, 16GB DDR4, and a pair of then-brand-new Asus dual RX 5700XT cards. The PSU is a Rosewill, I think possibly Valens, 600W, but it only has 2x 6+2-pin PCI-E on one cable. I've got it jerry-rigged, with a PSU adapter cable from the EPS12V 8-pin CPU cable, which has a pass-through and a brand-off for a 6-pin PCI-E, and I've got another SATA-to-6-pin PCI-E too. (Both RX 5700XT cards take an 8-pin and a 6-pin.)

Was looking, Newegg (Novashop) has an RGB PSU (Raidmax) 735W Bronze, I believe, that has 4x 6+2-pin PCI-E. Was thinking of maybe picking up one of those for that rig, for $59.99 + tax this BF. Currently, if I don't "power tune" those RX 5700XT cards, the rig reboots when I start to mine with it. (Not such a good situation, and not good for F@H on that rig either.)

Would prefer an 80Plus Gold from a major mfg in that rig, maybe I can make enough mining between now and Dec. 1st to order one. ($120?)

Edit: Installed F@H on my "secondary 1" rig, running 12 threads of PrimeGrid on the 3600 CPU, and F@H on both GPUs. PPD was 556K, last time that I checked. I do have my RX 5600XT set to 1350Mhz / 860mV, and probably -20% Power Limit. Kind of disappointing. Then again, Polaris RX 570 was only around 330K, and that was last year, I'm sure that the WUs have gotten a little harder since then.
 
Last edited:

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,540
14,495
136
OK, so I am one of the team captains, and I am supposed to name the team. I will do that now, HOWEVER, once the team is formed, I want suggestions from my team as to alternate names. I will abide by the majority.

The only thing that comes to mind (and I may have suggested it before) is

Marks Marauders
 

VirtualLarry

No Lifer
Aug 25, 2001
56,325
10,034
126
OK, I put in a "Make offer" on a Skylake i5 quad-core "T"-SKU (lower-powered, also slower-clocked, and also generally lower-cost), for the mining rig. If they accept my offer of $60 for it (was asking $80 BIN or MO), then I'll put it into my mining rig, and find out if it's the CPU holding back the F@H performance, or the PCI-E x1 risers. Even if I could get 4x GTX 1660ti cards out of the 5x in there to fold, on a quad-core CPU, that would probably be a win, 4Mil PPD then on that rig. Would be worth it for $60 to unleash that capability.

On my main rig, I fired up F@H tonight, doing a WU or two, and the RX 5700 is around 600K PPD, and the GTX 1660ti Gaming X is doing around 1.1Mil PPD. Together, something like 1.7-1.8Mil PPD. Not bad. If I could get that going for the F@H race, AND get my mining rig up to 4x cards @ 1Mil PPD or so each., that would be really nice. Around or over 6Mil PPD, if I used secondary rig 1 as well. (It's only getting 600K PPD for the RX 5600XT, and around 90K PPD for the GTX 1650 4GB D5 card. Would have thought that the GTX 1650 would do better, being Nvidia, but who knows.)
 
  • Like
Reactions: Assimilator1

StefanR5R

Elite Member
Dec 10, 2016
5,498
7,786
136
I suspect the limitations to address are both, CPU and bus bandwidth. Check the PCIe TX/RX rate on a rig with a 1660ti on a 16 lane or 8 lane attachment and with a large enough CPU while folding. I don't recall whether GPU-z or HWmonitor or Afterburner etc. show the host bus utilization. If they don't,
nvidia-smi dmon -d 10 -s pucvmt
does at least. You may have to prepend the path to nvidia-smi.exe to this command. This will show over time what bandwidth a 1660ti uses. You can change the sampling interval with the -d parameter, in seconds.

If you have figures for an unrestricted 1660ti rig, then of course it still does not tell directly what PPD hit there will be if the host bus supplies less than that. Perhaps nearly proportional if the difference to the unrestricted case isn't too large, and disproportional if the difference gets larger (due to quick return bonus getting impacted more).
 

Assimilator1

Elite Member
Nov 4, 1999
24,120
507
126
On my rig at least I don't see any PCIE bus utilisation listing in HWiNFO, although PCIE link speed is listed.
 

biodoc

Diamond Member
Dec 29, 2005
6,261
2,238
136
It would be great if we had more participants, but here's the teams so far. Let me know if I missed anyone. :)

1606215677287.png
 

StefanR5R

Elite Member
Dec 10, 2016
5,498
7,786
136
  • Like
Reactions: biodoc

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,540
14,495
136
I am actually at 39m average, and Pokey is at 40m average. Put Stefan on my team !
 

biodoc

Diamond Member
Dec 29, 2005
6,261
2,238
136
I am actually at 39m average, and Pokey is at 40m average. Put Stefan on my team !

I still have 6 days to finalize the teams. I'll make adjustments as more racers join or when I get better info on racer PPD.
 

biodoc

Diamond Member
Dec 29, 2005
6,261
2,238
136
I updated the team chart to include actual folding data (PPD) for those racers who are actively folding. For those who are unsure about PPD estimates for their GPUs, this database may be a useful guide. Can anyone confirm the accuracy of the database based on the cards you have?

1606257358760.png
 

Pokey

Platinum Member
Oct 20, 1999
2,766
457
126
@biodoc, I use that database exclusively. It is the most comprehensive I am aware of, and based on a lot of data. But everyone must remember that FAH is really hard to pin down points-wise due to the fact we have new owners so to speak, and a plethora of work units, all with different points,plus all the other variables.
For a while, for whatever reason, the RTX2080ti Rev A out produced the RTX3080. (Things are back in order now.)
 
  • Like
Reactions: biodoc

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,540
14,495
136
@biodoc, I use that database exclusively. It is the most comprehensive I am aware of, and based on a lot of data. But everyone must remember that FAH is really hard to pin down points-wise due to the fact we have new owners so to speak, and a plethora of work units, all with different points,plus all the other variables.
For a while, for whatever reason, the RTX2080ti Rev A out produced the RTX3080. (Things are back in order now.)
what are you getting out of your 3080 ??