I think the entire pallet went for like $4,300 or so which is like $123 per unit and you don't know condition so probably not worth it anyways.

May I ask why only 120v circuits?  If you are deploying new, why didn't you run 240v circuits to the machines for higher density with lower amp draw?  From what I can see, it looks like you had them add a 200a panel but then put in like 30 20a circuits with six of them being tandem (cheater) breakers.  That's a total possible draw of 720a breaker wise but only in a 200a panel?

I've been deploying 20a 240v circuits and I terminate them with 2 15a 120v (each outlet getting a separate hot from the 240v circuit) and one L6-20R 20a 240v receptacles.  I then plug in a 16a 240v PDU that then powers all the big iron at 240v and allows me to have misc 120v items powered through the same circuit but through the 120v receptacles.  

If you are pulling 800w on the servers in yescrypt, at 120v that's getting close to 7a.  Technically that means 2 servers per breaker if you want to stay around 80% of line limit as 3 servers at full tilt should technically just about trip the breaker.  If it was 240, you could technically be powering 5 servers per breaker and be just over 80% load on the 20a circuit.

I'm sure you know all this, just curious as to the rationale behind the decision.  If you ever get a chance to respond to my PM, it would be appreciated!  Obviously you've been busy!

The inability to stomach a loss, aka greed. 

3 phase power?

Were you able to pickup that pallet of z400s?

Video card prices seem to have hit their peak and I'm seeing prices "slowly" slide just a bit, especially on ebay.  I'm hoping this recent bloodbath in prices has scared away a bunch of people so card prices can return to normal and I can fill all these empty PCIe slots...   

 They WERE 700 to 900... They are now close to 1500 and even an 63xx CPU is going for a premium thanks to someone we know.  Heck, they are even being advertised as miners on Ebay now!

<3

What's your opinion on the state of card prices at the moment?  Just curious as to your outlook on them.

I've been doing a ton of electrical wiring myself.  I had to pull 200a service 500' away to the container I'm retrofitting as a server farm.  Fun stuff! :/ lol

When you go all in, you really go all in don't you!  No wonder the prices sky rocketed!

Actually, it's not.  The CPU has no AES...

Senseless, if you were going through ES samples, were you the guy bidding against me on some CPUs last night on Ebay?  Didn't win a single lot...  

Spinx, thanks!  I'm now hashing at 1650H/s on the G7's CPUs and it's got 4 1060s in it doing almost another 1200H/s.  I can't add anymore GPUs to it because my breakout boards haven't come so I can use the extra 1200w supplies to power cards, AND I don't have anymore cards to install into it!    I will however be moving things around to eliminate systems as I can.  I was using the standalone xmrig for CPU only and I was never able to get the thread attribute to configure correctly.  I moved to the integrated stak and ran it with --noNVIDIA to keep my GPUs on zcash and all is well.  Interesting to note, xmr-stak AUTO did the config for me and I didn't have to manually input the one you posted!

Tomorrow I'm picking up a lot of various mostly G7 and a couple G8 servers.  Going to be busy times ahead and I'm going to need GPUs!

My next thought is how can I optimize even more.  My current G7 has 64GB of RAM in it but it's only using 5GB of it with everything running full tilt.  I have all 8 cassettes installed with 2 DIMMs per cassette.  Since none of this work is RAM intensive, I'm thinking of pulling out 4 cassettes and 1 DIMM from the remaining cassettes to see what it does to power consumption.  This would effectively put me at 4GB per CPU, would improve airflow through the case and hopefully drop off 25-50w of power pull.

So last night I got home around 2AM and noticed that the DL580 I had just left wasn't showing up in the zcash pool.  I thought the breaker tripped again because I couldn't reach out and remote control it so I went to sleep because I wasn't going back to fix it.  I get back to the server late this morning and find it still running.  I hadn't checked the tracker for ITNS and it showed the server working all night long.  Somehow this box was still doing work in the EWBF miner but it wasn't registering with flypool even though it was submitting results?  I restart the client and it's now registering as an active worker.  No changes to the config file, just a restart.  Is this common or was this a fluke?

I'm currently getting about 1.25KH/s on the CPU's with CPU usage set to 92.  This gives me 30T which so far has given me the best bang for KH/s w/o impacting the GPU side.  The chips have 32MB L2 and 96MB L3.  I tried setting # of cores to half of L2 but that was only 650H/s or so.

Currently on Server '16.  

This thing draws some juice.  In order to test the load w/ only one killawatt, I had all 4 power plugs plugged into a strip.  Under full load it tripped the built in breaker in the strip.  I've got the server back on 2 120v circuits to run but now I need another killawatt to really know how much power things thing is consuming.

240v upgrades soon.  I already have some 16a PDUs but I'm trying to hold off with them until I start deploying in the container so I don't do things twice.

After much weeping and gnashing of teeth, what should have been a painless procedure that has swallowed the better part of a day to figure out is finally complete.  The DL580 G7 is rocking the E7 8837's.  All in all, it required what should have been easy.  A BIOS update.  Thanks HP.  <waves middle finger>

Spinx, what settings are you using for the miner?  I'm currently seeing about 1K/s w/ the default config from launching it and letting it figure itself out.

The DL580 is back under it's own power and running 4 1060s.  Even with the drive controller disabled and the array physically disconnected, I'm pulling 1085W with all 4 PSU's plugged in.  If I unplug 2 PSUs, it drops to about 1025W w/ the 1060's +175/500.  Stock vcore so power usage only went up 25W or so w/ OC.  That's not exactly efficient.  LOL  Box currently has X7542s so no AES and only like 240H/s for a total of 1280W of power draw with EWBF cuda miner going too.  The penalty draw for the 4 CPU's and 8 memory cartridges has to be supplemented by CPU work and currently, with the CPUs I have, I do no work.  If the 8837's I have that don't boot can do 1600H/s like stated prior, that would definitely help offset the increased energy cost.  I have 4 1060's in a Z97 setup that I'll throw the meter on tonight or tomorrow to see what it's pulling to compare.

My biggest fear is I'll get a bunch of rigs ready to go and then have no cards to fill them.  : |

Currently have 4 1060s in the DL580 G7, 2 in the case and 2 via risers.  If I hadn't shorted out 2 pairs of wires on the SAS power cable, I'd probably have a few more via risers but at this point, I've basically run out of cards other than a 6950 and a 780.  All my new cards are up.  Now to find more cards! lol)

I've tried 2008R2, no go.  I'm currently on Server 2016 using drivers for 2008 and 2012 plus Win 10 64 bit drivers for the Nvidia cards.

Great minds must think alike.  I bought a couple 12V to 5V converters myself.  It's how I'm adapting the power feed that was for the SAS array to drive cards based off the onboard PSUs.  I'm going to place an order from China as I can get the for a fraction of the cost, just have to deal with the wait.  I ordered enough to cover me for a server or two for now.

Wait, I just realized those are lipo converters from the RC world, you fly too?

Cleared NVRAM and I'm back in business.  My ROM date is from 2011.  Would you mind when you get a chance to check what version yours is?  What OS are you running btw?

I had 3 cards running in the box and one being powered by a riser.  For cooling reasons I'm about to fire the box with 2 risers and 2 in the case.  (I'm short 4 pin molex power for the risers.)  I've been doing some heavy adapting and outside of me blowing a dual 12v rail by accident.  I'm getting to be fairly confident about running a bunch of GPUs in the box but of course, only time will tell! 

I just attempted the same.  System still wouldn't boot with the 8837's.  I tried 2 old CPU's with 2 of the new, that didn't work either.  Then I left just 2 of the original CPUs and couldn't get it to boot.  I've put the original CPUs back in and now I get a long and short beep indicating a fatal ROM error: system ROM not properly configured.  Screen never comes up.  Now I'm down 4 workers just because I wanted to mine ITNS on the CPUs...  I'm hoping I can get this guy back up.  I was powering the cards pretty easily off the factory PSUs after figuring a few things out.

Tuesday I go pickup 28 servers.  Next week is going to be super busy...

Even Microcenter is getting in on the gouging game.  1070ti's in stock but I'm not going to pay $939.99 + tax for one! 

I wasn't able to make any headway on my DL580 when it came to getting the better CPUs in but I can say it is currently mining with 3 cards in it.  I have figured out how to adapt some cables and am powering the 3 cards from the onboard power connectors.  I also managed to repurpose the micro sata connector that fed the optical drive to run an SSD.  One of my goals with all the server iron is to remove as much stuff as possible without making the servers get upset.

I've been seriously "ghetto rigging" things since I've been waiting on equipment to show up.  Of course, I spent the better part of yesterday trying to repurpose some older ATX PSUs to get the cards up and running, only to walk out of the location at midnight to find a box containing 2 PSU's that were supposed to be here today.  Sigh, life would have been easier if I had had those! lol

Spinx,

I tried to send you a PM but you don't accept them from noobs like me.    I'm hoping you might be able to give me some advice.  I'm having some problems getting my DL580 to boot w/ the 8837 chips installed.  I fired her up and let her sit for 20 minutes but while on, it never booted.  I also didn't get any error codes on the LCD indicator on the board, just 00.  Did you have to do anything special to get the CPUs to work?  I've read about some incompatibility problems based on board revisions etc.  I've also read something about updating the CPLD.

Any advice you have would be great.

Thanks!

I just picked up a ton of old iron at auction.  Spent more than I wanted to, but I think I just became a mini hosting facility.    I took every single lot they had, now to make a trip to pick it up, hopefully tomorrow or Friday.

I started testing on a different server that is readily available.  I was basically able to strip it of a ton of unnecessary stuff that would alllow me to condense it's size if I removed important boards out, or repurpose the front of the case where the SAS drive housing was to housing the cards themselves.  Unfortunately, the 2 cards that were supposed to show up today from amazon are delayed until next week (hopefully by then!) so no testing today.  SMOS doesn't CPU mine, at least not out of the box so I'm getting a winimage ready.  But USB sticks so slow... lol

I have 9 1060s coming tomorrow so I'll hopefully I'll be able to test the G7 that is supposed to show up later today along with the other server I already have up and going.