Join the MicroCenterOfficial Folding@Home Team: #257944

Five Days to 500M with nearly 5 Hours to spare.
I learned some interesting things, in what I`d call an aggressive sprint mode to 500M as my final & 3rd challenge for July.
 

50!!

Great job everyone keep it going.

Well since I was one of the winners, yeah! Let's do another one! 

I think most people probably like to feel appreciated for their efforts, at least I did. I would like to see more participation in this type of event.  I suspect it is smart business for you too. Speaking for myself, I spent some $$ at my local Micro Center that I probably would not have otherwise to help with my folding quest.  (And enjoyed every bit of it.)  Thanks for all of this to you and the rest of the Micro Center Team.

I certainly appreciate the prize! But now I have a hardware question: the machine I have has an Nvidia GT-730, maximum power draw 38 watts. I'd like to replace it, but the machine also has a PSU rated at only 180 watts, and I don't plan to replace both. I'm thinking Nvidia GT-1030 (GDDR5), maximum power draw 30 watts, should be a drop-in replacement.  Might not quite double the GT-730's performance.

But Nvidia's website suggests a minimum PSU output of 300 watts (vs. 350 for the GT-730). Can I make this replacement and not really worry about it?

Just incase you haven't looked at the FAH site lately you may want to take a look. Some of the new updated info for the projects and in my opnion a great press release that really needs to be wide spread. The Moonshot project from the looks of it will not be short on work units anytime soon and is making great strides. Take a look when you have the time there's some great information there.

JBug55 said:

I recently added a new GPU to the mix (added a GTX-1660 Super to the GTX 1650 I already had) and was pleasantly surprised by the results after just a few days.  I joined the ranks of over 1m ppd and Top 20 Producers.  I also learned that Linux really appears to be more efficient than Windows for this application ( no, I'm neither a Windows nor Linux fan-boy or basher). With a small sample size, I am getting  more ppd by about 15% - 20% with Linux.

Yes the 1660 Supers are nice, my 1660 Super pulls in almost as many points per day as my 1660 TI.  I have to admit I did only buy the TI because it was an open box deal (my daughter loves the LED lights on it to LOL).  What flavor of Linux are you using just out of curiosity?  It has been seen by a number of people that Linux does manage to earn you more points.

ChrisHarshman said:

JBug55 said:

I recently added a new GPU to the mix (added a GTX-1660 Super to the GTX 1650 I already had) and was pleasantly surprised by the results after just a few days.  I joined the ranks of over 1m ppd and Top 20 Producers.  I also learned that Linux really appears to be more efficient than Windows for this application ( no, I'm neither a Windows nor Linux fan-boy or basher). With a small sample size, I am getting  more ppd by about 15% - 20% with Linux.

Yes the 1660 Supers are nice, my 1660 Super pulls in almost as many points per day as my 1660 TI.  I have to admit I did only buy the TI because it was an open box deal (my daughter loves the LED lights on it to LOL).  What flavor of Linux are you using just out of curiosity?  It has been seen by a number of people that Linux does manage to earn you more points.

I'm running Mint (19.2 cinnamon I think).  I chose it as an older version that still supported Python 2. Originally as a "no muss no fuss" entry point, if Linux didn't work out, I wouldn't have a lot of time invested.  It has a familiar look and feel to me as a mostly Windows person, and has  been rock solid so far. So I have no motivation to update to the latest and greatest.  Works so well I put up a second machine without even considering a different Linux. I like standard software across platforms for ease of support, clearly a reflection of my years in corporate IT. Plus, sometimes we "old retired guys" don't want thinking too hard getting in the way of our golf time.

FYI... both my cards were open-box buys as well.  I usually shy away from open boxes, but not at Micro Center. I go back a long way with them, all the way back to the first time I walked into their shop in Westerville, Ohio back in the early 80's.

Here`s one list
https://docs.google.com/spreadsheets/d/1vcVoSVtamcoGj5sFfvKF_XlvuviWWveJIg_iZ8U2bf0/pub?output=html#

ryau said:

This makes me want to set up a linux box now... although I haven't touched linux in maybe 20 years, so I'm sure a lot has changed. It'd be cheaper in the short term to throw another 2070S in my machine, since it performs pretty well already.

I started with Mint, but ended up on win10 so that I could get the monitoring and power/temp capping features of MSI Afterburner, plus the dynamic custom GPU fan profiles it offers.  I seldom Fold on CPUs due to the low efficiencies and severe thermal issues of some current CPUs.   
I`d suggest also that you don`t have identical cards as it`s near impossible to eliminate what I call ToxicWUs & the work server that will have a tendency to latch onto your system.  During my recent 500M sprint I was `forced` to split the system with dual 2060s mid-effort for this reason.  Initially, I felt guilty about rejecting some WU that were choking my systems, so I left the 2x1660, 2x1050Ti, & 750Ti to digest whatever they were feed.  Watching a system go from 1.8M > 3.8M, or a 2080Ti go from 1.7M > 4.2M, well, I soon realized I was also doing the `most good` too with this approach**.   My 2060s+2070s runs at or near the level of the 2080Ti & often higher for those looking to match-up gear.  Me, I`m waiting to see what will arrive in the Fall & resisting the temptation to add more gear.  My $0.02 HTH.
**the FAH researchers might not agree?
Also, it`s your gear, your power, your time, your $$$... participating at a level that fits you, big or small, participating is what matters.
UPDATE: after writing this, the 2080Ti was at 1.5M PPD, it`s now at 4.04M.  I hate to see good gear being poorly utilized.

The nvidia-smi command line tool is a bit clunky to use at times, but it does allow me to tweak things like Afterburner would, including setting power limits on the GPU, so that has helped on those 90+ degree days.

@ryau I had a spate of low yield WU that tookover the dual-2060s, which forced me to change the GPU stack between two systems.  I haven`t seen WU loading issues recently.  In general, things have been `glassy` smooth going, running in the ~11M PPD range.  But you can`t turn your back too long or it`ll be in 8-9M range.   I`m retired so I can devote more time than I should to Folding.  I`ve got some things coming up that will get my attention & I`ll be returning the 2080 I borrowed, so I`ll be nudging Folding to the side at a sustainable level. 

Anyone else having issues getting decent WU's.  I am running 6 RX580's and they are all currently busy doing minutia running at 636 wats total power draw.  Basically I am running at about 20% of what I should be running.  Seems like these WU's should be going to slower CPU devices.  Is is FAH servers causing my problem or is there something I can do to get better utilization of my hardware?

JADO said:

Anyone else having issues getting decent WU's.  I am running 6 RX580's and they are all currently busy doing minutia running at 636 wats total power draw.   Is is FAH servers causing my problem or is there something I can do to get better utilization of my hardware?

Vanilla just ate a huge response.   
Found partial:
See my comments above about Toxic WUs & work servers latching on systems with identical GPUs.  With six (any) matching cards it will be tough to fix as the bad WU just stack up if you remove a slot, which is how I fix it.  Unfortunately, they (FAH) don't actively manage utilization, but a feature request has been issued.  It got percolated up to their GitHub.    
https://github.com/FoldingAtHome/fah-issues/issues/1479
The Covid Moonshot project is chock full of low atom count compounds with low GPU utilization WUs.

Update: the missing post showed up.  Sorry about the double posting.