Are there long-term miners on server PSUs around here ? How long has your server PSUs been lasting ? Average time before going dead ? Got early deffective ones ?
Due to their almost non-existent warranty time (30 days), I am skeptical to choose them over ATX PSUs which have 10 years warranty as I fear I might need to buy numerous ones over the span of few rigs lifetime cycles, then I am not so sure about actual savings, all things considered (reselling of ATX replacement units of never models in the long run)
They should last longer than consumer grade PSU's. These are designed to run 24x7 for years on end in servers after all. well, after query, server PSUs' (refurbished!) resellers are eyeballing a longevity of 1.5 years - 2 years, not more (with some rare cases less than 1 year or above 2 years), past any warranty time in all cases therefore, considering : - twice units of server PSUs favoribly run closer to 50% load rather than 90% (ATX) - zero resale value of server PSUs (dead units) against a resale value of 30%-40% of ATX psus, eventually from a newer model of a replacement during the 10 yr warranty time around year 4-6 - despite a 3% efficiency boost in favor of server PSUs underloaded ...the attractiveness of server PSUs does not look evident actually in my case (considering electricity costs, shipping costs, etc), ATX psus would appear more cost effective from year 1 to year 10, no matter how I manipulate the figures (side story, you don't see a giant cloud mining company like Genesis running their GPU rigs from server PSUs but from ATX. Was it studied ?) Well, actually if you check some of the newer Genesis Mining DC tour videos you will notice them using HP server PSU's to power their rigs. |
|
|
|
Are there long-term miners on server PSUs around here ? How long has your server PSUs been lasting ? Average time before going dead ? Got early deffective ones ?
Due to their almost non-existent warranty time (30 days), I am skeptical to choose them over ATX PSUs which have 10 years warranty as I fear I might need to buy numerous ones over the span of few rigs lifetime cycles, then I am not so sure about actual savings, all things considered (reselling of ATX replacement units of never models in the long run)
They should last longer than consumer grade PSU's. These are designed to run 24x7 for years on end in servers after all. |
|
|
|
I would also set the power limit for the GPU to 80W and remove the globalcore value (the NVIDIA cards will simply then run the core clocks to fit within the power budget).
Interessting. I'll check it. It's the same for the mem clock? You want to set the memory clocks specifically when mining ETH as that is the major determining factor in generating the hash rate. |
|
|
|
How do I calculate the globalmem for NVIDIA cards? In Windows I simply used +750 in Afterburner. After that I switched to nvOC and there I doubled the numbers to +1500 because of Linux.
Now I am testing ethOS and it looks like this is again a different system.
For now I use globalcore 1300 and globalmem 4700 on my GTX 1060 but it will be nice if I can calculate the right amount to get a stable rig.
Personally I set GPU clocks on a per rig basis. Those values appear about right for a 1060 however. I would also set the power limit for the GPU to 80W and remove the globalcore value (the NVIDIA cards will simply then run the core clocks to fit within the power budget). |
|
|
|
Lowering overclock settings seems to be helping   No surprises there  |
|
|
|
guys, trezor vs ledger nano s, which one do you choose?
Ledger FTW. |
|
|
|
Hello. Greetings. I have been mining with 6xP106 on Ethos. OC at +750 mem and power at 80 watts. Have not filled the CC yet. The cards whenever they crash the rigs shows in the panel 'possible pool connection problem, rig is not hashing' with a Red Triangle mark. (You will know the sign if you are familiar with ethos). http://imgur.com/a/6W8c3 The miner shows 0 Hash rate during the error. Then I have to reboot the rigs and they work properly afterwards. The pools, failover and wallet are obviously correct. It has to be noted that it does not happens to all P106 rigs together, if pool connection was the problem it should happen to all rigs. This I thought was happening due to ethminer but then I shifted to Claymore and its happening with it too. the rigs were very stable with Windows though and no such error ever came. No Clue why only on Ethos. Any help or input from the community would be appreciated Thanks Try lowering your memory clocks and see whether that improves stability. |
|
|
|
What is the command "ethos-overclock"?
And don't you need to run "putconf" between the echo and the minestart to force it to use the new conf? or is that done automatically when you run minestart?
Personally, I run these commands in a bash script rather than all at once. That also allows me to add "pauses" like "sleep 5" to pause the script for 5 seconds between commands.
(man I wish ethos documentation went into more detail.)
"ethos-overclock" is the command used to apply any core, memory and power settings that you have specified within your configuration. EthOS runs this automatically upon boot, however if you modify your config after boot and wish to have those updated settings applied this is the command that you need to run. WRT "putconf", this simply fetches any remote configuration file from the URL that you have defined within /home/ethos/remote.conf to /home/ethos/local.conf - you don't need to run "putconf" if you have simply modified local.conf, or replaced it by copying another file to its location. As for running the commands all at once, running them with ampersands between the commands from cron is fine, the commands after the ampersands will only run if the command proceeding it succeeded, so there is no problem there. |
|
|
|
How many are you looking for?
I dont' need right now, im only looking for it, but i think 13-19, only know where to buy, ty. I can supply and ship those to Europe. Send me a PM if you're interested. |
|
|
|
|
How many are you looking for?
|
|
|
|
Hi,
Just few questions...
Q1: It’s possible to switch automatically between two conf? My Need: after 22:00 and until 06:00 I have 20% discount on electricity. I’ll use this time frame to power-up my rig. from globalpowertune 80 to globalpowertune 90.
Q2 : How to update my Nvidia drivers? It’s automatically updated with each new ethos release or should I do something manual?
Thank you in advance for any help
For my first question, I'll try to use Crontab to switch between two remote conf file. I'll post my result here if somebody is interested... That should work, just execute something like this from crontab within a script: minestop && <copy new config file to /home/ethos/local.conf && sudo ethos-overclock && minestart |
|
|
|
How does this function work:
"-Added ability to set voltage for RX series GPUs."
I mean is this better than adding offset to vbios? Did someone checked if power consumption from the wall is lower/higher/same comparing to offset?
They have code to control the voltage regulator modules (there are several types and manufacturers) of the GPU's directly so they can set the voltage to the desired levels. |
|
|
|
Does anyone know where you can buy these cards. From what I understand they are currently only selling these card to mining companies or in bulk. I am looking for a bulk order and was wondering if there was any place to buy these cards?
Thanks, RigbyDigby
I buy them direct from a manufacturer in HK, if anyone wants some of these please let me know how many you're after. Pricing is $300 USD per P106-100 6G GPU. |
|
|
|
With this setup ethOS is working fine.
Still not sure vlt parameter is working for RX580 GPU series.
try with this vlt 49709f 1 1 1 1 1 if its work you can try 0.9 0.9 The voltage needs to be set in millivolts. So, to set 0.9 volts you would specify 900 instead. vlt 49709f 900 900 900 900 900 Most RX 470/480/570/580's should be able to run at 850 mv when mining ETH with the core clock set to 1150 mhz, hence set the config like this: cor 49709f 1150 1150 1150 1150
vlt 49709f 850 850 850 850 850 |
|
|
|
|
Received a few minutes ago via email:
"Hello one and all.
First I want to thank everyone for the positivity we've been experiencing both via online channels as well as at the open house events! I've also received a TON of really supportive energy from the community and I can't thank you enough... it really keeps us all going when there's huge amounts of hard work to do.
Unfortunately, I have to announce a delay in construction which will block us from issuing Batch 3 tokens on time. We've been working tirelessly to bring pods out of the ground, but several conditions related to the construction and planning process have introduced additional delays that are now clearly keeping us from turning up as many pods as planned for Sept and October. The process of building the pods and deploying equipment is going as expected. The real time-impact has to do with permitting, planning and inspections. Our GigaPods were virtually the first purpose-built blockchain processing facilities these people had ever seen, and it has caused small delays at every turn.
The good news is that we are through most of that. We have educated the stakeholders in two counties about our GigaPod construction processes and how the finished product looks and works. I expect these types of delays to become less and less now that it has become familiar to them.
Our construction crews and electricians are getting pod construction down to a science. We will continue to run day and night crews to bring up as much power capacity as possible, as soon as possible.
On the topic of hardware arriving in September: there's significantly less equipment arriving in September than the amount of power we are building, so it becomes a little easier to see through to getting it all running. I expect to have much of the September hardware running soon after it comes in. Because the construction process is not a perfect science, its hard to know exactly when backed up gear will be put into service. To mitigate the situation, GW is seeking any and all alternative solutions to get gear running as soon as possible so that it won't just sit on the warehouse shelves. As a miner myself, I know just how important that is to our customers.
I'm sure that questions and concerns will still linger, so I am planning to make myself personally available on chat channels in order to answer questions whether to the group or to me personally. I do appreciate your continued patience and support. I want you to know that the whole GW team is directly focused on getting tokens issued and miners running as fast as humanly possible.
It’s always hard to maintain communications and transparency during the chaotic startup phase, but we are committed to improving on this front. We are improving our support to clients slower than we and everyone wanted but are progressing on that. Within the next 3 to 4 business days I will begin releasing weekly updates to help everyone track progress along the new timelines. I aim to improve confidence in Giga Watt and significantly reduce concerns you may have going forward.
Best Regards
Dave Carlson, CEO
Giga Watt, Inc."
|
|
|
|
They generally go up lineraly, with teh 1070 and 1060 being slightly more efficient than the higher cards. But really it's all about sticker price and longevity. So with a 1060 you might get 25 hash, 1070 50 hash, 1080 75 hash, Ti 100. Although it's not perfect and again the cheaper ones are generally more efficient, but you need more space and auxiliary hardware
The more expensive cards will hold their value better
Cheaper cards will pay themselves back faster unless you get a deal.
Really it's all about finding cheap cards more than anything
Mining ETH you'll max out around 25 MH for a 1060 6GB / P106-100, 32 for a 1070 and 38 for a 1080 Ti pushing them hard. ETH is never gonna be profitable for nvidia cards So no one cares Well, I can tell you thats bullshit as I'm profiting on mine right now. |
|
|
|
They generally go up lineraly, with teh 1070 and 1060 being slightly more efficient than the higher cards. But really it's all about sticker price and longevity. So with a 1060 you might get 25 hash, 1070 50 hash, 1080 75 hash, Ti 100. Although it's not perfect and again the cheaper ones are generally more efficient, but you need more space and auxiliary hardware
The more expensive cards will hold their value better
Cheaper cards will pay themselves back faster unless you get a deal.
Really it's all about finding cheap cards more than anything
Mining ETH you'll max out around 25 MH for a 1060 6GB / P106-100, 32 for a 1070 and 38 for a 1080 Ti pushing them hard. |
|
|
|
|
1060 6GB or P106-100 6GB GPU's for price/performance/power efficiency.
|
|
|
|
Yeah, I would say that amount of power capacity is a little overkill.  Well...you can technically fit 10 x 1080 ti Founder Edition cards in a rack mount case, and they have been show to pull sustained power consumption in the 3.0-3.2kW range at 100% PL, and even peak to nearly 5.3kw Patrick over at STH has been playing around with some ridiculously cool Supermicro GPU servers https://www.servethehome.com/category/server-parts/server-systems/  I have one of those, good machines. |
|
|
|
1) Read Phil Ma's blog (in this forum) for tips and heads up on component bargains. Among other things, he finds amazing deals throughout the commercial and ebay worlds. He is also someone that discusses money management. You cannot read too much.
What is the URL? |
|
|
|
|
Show Posts
