Topic: My second ZEC + XMR+ ETH thread builds info links thoughts and photos. - page 36. (Read 147923 times)

citronick,

I've tried to have a look on Z170 MBs on Amazon - I could not find any with 4 full size PCI slots

Could you please point one?

I agree. KISS (Keep It Simple Stupid). takes very little down time due to a crashed rig/card to negate any savings from an overly complicated rig or crazy card settings like extreme BIOS flashes etc.

i know many people super tune cards or rigs for a specific algorithm but i prefer a rig that can switch with a simple .bat file and not worry about psu overload or cards dropping out because they are too wound too tight for just one coin.

keep it consecutive, keep it running 24/7

See this link. Pack of 10. Ensure to confirm USB cable length.
https://www.aliexpress.com/item/10pcs-lot-PCI-E-PCI-E-Express-1X-to-16X-Riser-Card-USB-3-0-Extender/1679698470.html?spm=2114.13010608.0.0.Rhx98E

Why build rigs small?

Big PSUs are expensive. More GPUs more power, more heat, more power trips, more unknowns. Risers are finicky too. You need more cooling solutions - buy more fans. So any bigger you need more space.... so rent? lease? Cabling works? So you see monster rigs need alot more additional resources to keep it alive.

Although I still keep my 6 and 7 riggers, its just because I already invested in 1600w EVGA PSUs. Never again - its too expensive and harder to ROI, unless you have ASICs....

Now all my newer PSUs are max 1000w, cheaper 750-850w are ideal for 4xGPUs. Compact Z170 simplifies deployment. BIOSTAR Racing is a cheap and good mobo IMHO. Thanks to Philip who introduced this gem of board. Similar Z170 from ASrock and Asus are also good boards.

You need to save time.... to figure out and getting that one or two 6 x GPU rigs to work will take some much time, frustration and troubleshooting you lose mining time accumulating coins, cashing in if you can get good profit OR keep and wait, to trade & sell high in Poloniex for example.

I guess everyone will go through a unique mining experience, to be profitable spend little time troubleshooting and get mining to work fast. With this 4 slotters and killer miner app like Claymore's ZCASH, XMR and ETH miners... you already have speed. To be ahead of the game, you need accumulate and reach ROI as fast as possible.... any coin after that will be revenue less expenses = pure profit. Mining at any later stage under increased difficulty will get you less profit, longer ROI or even no ROI. So Timing is Everything.

I rather have 10 small rigs that I know will work and stable, VS. 5 x monster rigs that loses a card every now and then because of high power draw, overheated GPUs or intensity too high in Claymore's miners. No problems pushing Claymore to max settings on a 4xGPU rig.

The next step in the horizon is would they be a generic GPU "chip or ASICS" like BTC's A7 or S9, X11 or Scrypt miners. Personally, I dont think it will happen for ASIC-like generic GPUs, unless AMD puts RX480 chips x 100 on a special extension board.It will be good to have 1 big-ass miner that.... something like Panda Miner is trying to do. Then all you need is a couple of these and you are set.... until then I will be happy 4-slot rigs.

These are just my own humble opinions - I am sure many of you out there may have other approaches on how to be profitable in mining. Otherwise, forget mining and go straight to trading in Poloniex - you can also make good money this way without buying a single GPU.

Sorry for long post.

Citronick, could you provide a link to a good Chinese supplier for the USB3 PCI risers ? dealing with the seller "approved by Citronick" will save time to many of us .

Another point I find difficult to understand : how can a 4GPU mining RIG be profitable ?! it's already very hard to ROI a 6GPU H81 rig !
Only 4 GPUs on a more expensive MB like the Biostar should be almost impossible to ROI, no ?

Yeah, most of what I have read concurs that the best method is to use powered USB risers along with the extra molex power to the Mobo in order to distribute the power draw of the GPUs more evenly. My understanding is that using molex to the Mobo without the powered risers has indeed fried a few mobos as apparently it can result in way too much power being drawn directly through the Mobo. I am sure some have made it work with ribbon risers but I will stick solely with powered USB as that's how my first rig is built and it's been running well for 3+ months with no hint of a power distribution issue.

Point of clarification on running the 6xGPU mod... should i go ahead and run it on my rig now... with only the 4 working GPUs hooked up and then try to install the 5th GPU?  Or, should i attempt another install of the 5th GPU FIRST and THEN run the mod?

I admit there is an appeal to go super compact and riser-less.  Depending on how things go with this rig, I may consider something like that if and when I build a 3rd rig.  But, I am not jumping into that unless I start to see some market confidence and price rises in ZEC, ETH and/or XMR.  I just have a hard time getting past the fixed base cost of each new rig... meaning for instance that two 6-GPU rigs would be much cheaper than building three 4-cards rigs. But, at a certain point, if the 6-card rigs encounter too many problems and down-time, that advantage shrinks.

Finally, thanks for the tip on ETH.  I actually had my first rig mining ETH for a couple of months until ZEC came out and i pointed it to that.  I have all 4 of the GPU's on that system modded BIOS using the 1500-strapped method so actually, running that rig on ETH only pulls moderately more power than running ZEC.  Granted, that's only a 4-card rig with only 1000W PSU, but even running full bore ETH i was around 650W so plenty of headroom.  I am highly considering pointing that rig back to ETH shortly.  As for the 2nd rig, I suspect it could easily 5 GPUs on ETH and most likely 6... ASSUMING i can get the GPU strap-modded the same way.  I already tried it on my first GPU and for some reason it didn't work so I had to flash it back to the factory specs.  But, since it seems that that strap-mod benefits ETH mining a lot more than ZEC mining, I haven't prioritized getting back to that yet, at least until I can get my 5th/6th card issue fixed.

Thanks again for the help and please let me know what you recommend on running that 6xGPU mod before or after I try to install the 5th GPU again.

Thanks Citronik for the thoughtful and helpful information... i really appreciate it! I will respond to all your points as some i have done and some yet to do:

1- I do have the additional molex plug powering the mobo. I knew I would need to do this going in so after getting that message I think after installing GPU #2, I plugged that additional molex into the mobo.

2- I am using all powered USB risers at this point, in x1 slots 1, 2, 3 & 5. I got most of them from the same Amazon seller and didn't have any problems with 3 of them on rig #1 and the first 4 on this rig.  However, I do have 1-2 extra ones lying around, along with 3-4 additional USB-3 cables, so I imagine it's a good idea to try all of those before proceeding with further trouble-shooting.

3- My PSU is a Rosewill 1200W Quark Series Platinum.  With 4 GPUs running on Claymore ZEC miner V8, I am hitting total system wattage peaks in the 600-610W range. This includes the CPU/Fans/etc, which use about 60-70 watts, so even adding 2 more 470's would put me at just under 900W max... so, less than 75% of its total capacity.  Based on this, I don't believe it is a total power issue causing my problems (unless I am really missing something).

4- I'm already on W7 and would prefer to avoid W10 if possible, so I am considering that registry hack you mentioned.  I actually read about it over on the ZEC mining forum... at least I assume it is the one you are talking about.  Would you mind checking to see if this is the same one?

https://lbr.id.lv/6xgpu_mod/

Since this involved messing with the registry I don't want to take any chances, but if this is the same version you are talking about and you say it's relatively straight forward to implement, I will give it a shot.

Hoping/assuming this mod will solve the error 43 issue in slot #6, I still have that pesky issue with the lone x16 slot on this board.  Have you ever experienced anything like that?  (not being able to get past post with anything plugged into it via USB riser, including setting PCIe to Gen1 (from auto) in the BIOS).  Note that I tried to use this slot several times before getting my rig up to 4 GPUs, so I don't believe it's the number of GPU limit issue causing this problem.

I also read that some people have had success by turning off the on-board video in the BIOS. Not sure that was meant to fix this exact issue, but maybe it's worth a try?  Not sure exactly how to do that... will probably take me at least an hour to find as does most anything I try to find in the BIOS lol.

Thanks again for the tips... hoping to get this rig up to 5 GPUs soon, ideally in the x16 slot, so I feel comfortable enough that it will take 6 GPUs to buy the 6th.

gabryrox -

I have many H81 BTC mobo with 6xGPU and 7xGPU configs with no problems.

In my case, I got my multi GPU tutorial from BBT - Bits Be Trippin.

You need to ensure that this particular mobo (v1 and V2) has the molex powered directly from PSU.

In addition, all GPUs need to have powered risers.

The errors you are getting, IMHO are power related issues... there just not enough power for 6 GPUs.

Also your PSU need to be large enough to cater for all 6 x GPUs plus you mobo, etc. with 20% to spare (good rule of thumb to ensure safe zone and also longevity of your PSU).

I had a few suppliers from China via AliExpress on the powered risers - ensure your get the version 3 with USB3 type - I have very little issues with those. Avoid the ribbon type risers because the solder works on those are back-alley quality and are a fire-hazard.

Best build for multi GPU is Windows 7 in my opinion, you may need to use the 6xGPU registry mod hack utility to get rid of the Error Code 43, but no rocket science, it's quite straightforward. If you want lazy shortcut and want Windows then just go for Win10 to get all 6xGPUs visible in Device Manager.

For mining (not gaming), 16x PCI is irrelevant, the USB risers are 1x (...4x ??) and enough to carry DATA between OS and the GPU. Unlike high end games, it will utilize the PCI bandwidth on all cylinders to use all the cool features of the GPU. In mining, a big chunk of the GPU's greatness is likely not used because mining need at best the core GPU cycles and the high speed memory. Thats why miners are modding the ROM and taking out the unnecessary stuff -- this ritual is popular in hard core mining.

Afternote:
While I have the multi-GPU config still in my mini-farm, all my new builds are riser-less on based on Biostar Z170 4xPCI slotters. Keeping it simple as possible.

Powered (Molex/PCI-E) ribbon cable 1x-16x risers from Padrino

So, I checked and I actually got both my BTC R2 mobos from NewEgg back in early Nov, not Amazon as I had previously stated.  I also got a notice from someone posting in the ASRock technical support forum that I should open an actual service ticket for these issues, so I just did that  couple hours ago. Fingers crossed but based on things I've seen in a few places, this really seems to be an issue with the R2 version since there are hundreds if not thousands of the original version of this mobo running 6 GPUs.

Out of curiosity, what kind of risers are you using on your rig(s)?  Aside from the x16 slot problem, I am now wondering of trying to hook up the 5th GPU via a ribbon riser instead of a powered USB riser.  I also thought about trying to plug #5 directly into x1 slot #6.  Not sure but I think there might be room since it is at the end of the mobo. not sure where my head was on that lol. obviously cannot plug into mobo since this is an x1 slot lol

I will post more info on these tests after trying them, as well as if i hear anything back from ASRock tech support.

The mobile RX 480 should be more efficient than the normal RX 480, so the price should be high. I think the $2k price is competitive.

I feel the pain as I am in the same boat as you. I have no idea why it isn't working properly. I was F***** piss about it. I bought two used board on amazon, both acting the same way. I think they are both Version 2.

I should get a new replacement this week. (new from newegg)

I will let you know with the new board

yes it should have 8 mobile rx 480 gpus inside it.

this is almost certainly a data center /warehouse miner .

 a guy with power to spare and no room.

At 2000 usd it is fairly well priced

I just need to know software power sound heat stats  to know how good it is.

Hi Philip, looking forward to see your review of the Pandaminer when it reaches you.
Am I correct to assume that this is an aggregator board / box for AMD GPUs up to 8 x GPUs?
 

 A quote From this thread



https://bitcointalksearch.org/topic/pandaminer-b3-pro-new-batch-releases-with-only-800-1750-1668461






The thread is an interesting read.

what's up with this dummy plugs? ... since i started mining in 2013 i didn't have to use one.

the only time i used a dummy dvi plug when i was trying to change the clocks in amd catalyst per card, when i plug the dummy to second card it will become the default display and then i can change clocks in that card (using amd catalyst).....that was waaaay back in the past until i figured out that i can change clocks using mining software and apps like msi afterburner and sapphire trixx..

i'm suspecting radmin (remote access software i'm using)...i don't know...you guys can take a look and test it, this might be the one that saved me from dummy plugs..maybe because it installed "radmin mirror driver" in display adapters in device manager..

this desktop pc i'm using (with fury card as my primary gpu), i tested plugging a rx 480 with usb riser (as a second card) and it hashes with 84 H/s (ZEC) or something, i forgot the exact numbers but it is really low hash....one difference i can point out to my mining rigs is this desktop pc has no "radmin mirror driver" installed because it is a client not a server (it is the pc i am using to access the mining rigs in my LAN)

test it guys and share here if it worked.

Hmmm that's disheartening to hear.The thing I don't get is that there are probably 1000's of these boards running 6 GPUs worldwide and never seen anyone mention these issues on the youtube videos.  I mean, these boards are designed for mining and miners natural expect all 6 PCI slots to work fine, but seems like more and more problems popping up... could it be only with the R2.0 versions?  Anyway, I posted a thread to tech support in the ASRock forum querrying about the same 2 issues so I will pass along any info i get there if it works.  I also have a 2nd one that I havent even cracked the box on yet since I got 2 from Amazon when they first came back in stock about 5 weeks ago. Beginning to wonder if that was a mistake.

 You, sir, are fairly new school.

 *I* am old school - all of my switches are 2xPS2 for KB/Mouse + VGA, or adaptered from older AT KB + serial mouse + VGA switches.

 I'm finding it's usually cheaper to adapter from HDMI to VGA (and is a LOT cheaper to adapter from older cards with DVI-I to VGA) and get the common LOW COST VGA switches, though I might start moving to 2xUSB + VGA switches as MB with 2x PS2 type connectors are getting far to uncommon.


 Most recent rig "rebuild" with the pair of Sapphire RX470 cards I got today was a pain - ended up having to swap some stuff around to free up an extra HDMI -> VGA adapter to use as a "dongle" - one thing I like about all of my NVidia cards is that NONE of them need that stupid stuff, they just WORK no dongles or adapters needed even under WinDoze.


 Down side - they'e backplate cards so they won't work in a 4-card on MB build, though they're fine otherwise.
 They also run somewhat on the warm side.

 Up side - inexpensive and they're pulling 21 MH/s (ETH) or 205ish sol/s (ZEC) out of the box - and they actually work with Claymore v9.1 (they had constant issues with "crashout before connection established" issues with both 9.0 AND 8.0).


 Don't get me started on Wattman - AMD has a VERY VERY LONG WAY to go before they can even get into shouting range of Afterburner, ESPECIALLY on fan control.
 Wattman is TRASH - I've ended up just deleting that "new control panel version" junk out entirely and upgrading to Afterburner 4.3.0

That's pretty much where I am with my 2 H81 Pro BTC, pass the 4 cards, it's nightmare. 4 is sometime fine, but I can't push the card. 3 seems to be fined with a lot of OC.

I run 4 on one and 2 on the other one.

Tested pretty much everything possible.

I'm getting a brand new one this week. I shall see.

Well, it seems I am jinxed in my efforts to get a 6-GPU system up and running lol.  Granted, my first attempt was on an unproven X99 Mobo, so I am willing to take that as a learning experience but now I am having trouble getting past 4 GPUs on a ASRock BTC Pro mobo that I know for a fact there are probably 1,000's out there running 6 GPUs smoothly.

Anyway, here's the latest.  I was running stable with 4 GPUs in PCIe slots 1, 3, 4 & 5.  Although Phil sent me a x16 ribbon riser, I can't fit it in with my current rig config, so I tried what someone else recommended on this thread... changing PCIe to Gen1 in the BIOS.  After about 45 minutes of searching through the BIOS, I finally found this and set to Gen1. Rebooted once, then powered off and plugged in GPU #5 to the x16 slot via USB Powered Riser.  Powered on and once again, system wont even get past post. 

However, since I haven't bought my 6th GPU yet, i decided to give up temporarily on that x16 slot and plug GPU #5 into the 5th x1 slot.  Hooked everything up exactly the way the other 4 GPUs are, booted up and it seemed to be having problems.  It seemed to stutter or restart a couple of times rather than just 1 as with my other GPU installs.  Finally got to windows startup screen and I logged in.  But here's where the problem shows up...

The pop-up about installing the driver for device pops up but instead of saying driver installation complete in a few seconds, it hangs up for a bit then returns a message saying that device driver could not be installed correctly.  I rebooted a few times, including once where I completely unplugged and replugged the GPU, but got the exact same problem.

When I go into Device Manager, this 5th GPU shows up with the dreaded exclamation point in yellow triangle. When I click on properties is says that windows disabled this device due to error code 43.  I googled this error and employed the recommended fix as follows:

- right click on bad GPU
- select properties
- go to driver tab
- uninstall driver

- right click on Display Devices
- choose Search for new hardware
- this finds the 5th GPU
- but, when I attempt to install the driver again it doesn't work. it tries and for a brief moment, the 5th GPU shows up in device manager without the error triangle, but a screen flicker later, it is back to appearing with the error 43 and will of course not even be recognized by Claymore.

So, has anyone experienced this issue with GPU?  Would be very happy to hear some suggestions on what to try to fix it. Thanks!