I of course understand shares can vary in weight. However this is why I posted asking for others to investigate this, mostly because it's quite late I'm tired and other opinions wouldn't hurt.


Yes you can: https://www.nicehash.com/?p=faq#faqs10
I know it is not properly static but more or a preferred diff, however it is usually working nice to keep a worker diff around the specified limit.

How would you then explain the stales? and as said, we used static diffs and not vardiff.

Please verify this for yourself before making any accusations. A few of us have been wondering if the dev fee charged by Claymore is really 2.5% on his Zec miner releases.
 
We were able to isolate the fees charged by Optiminer, and did indeed verify that the fee was 15% as announced. We used the same method to isolate the fees charged by Claymore. For reference, we used version 4.0 and 5.0 to confirm this behavior. What we found is very surprising.
 
During (and after) the 90 seconds where it mines for the developer we do notice serious inconsistencies:
 
 - We counted 111 submitted shares in 90 seconds per hour (2.5%) that the claymore mines for himself.
 - We counted 15 submitted shares in the immediate 90 seconds after the fee period.

Here is a brief extract of logfiles for the two periods of 90 seconds we referred to. See for yourself:
http://pastebin.com/0EFM8Bat
 
Using various methods, we redirected the "dev fee" workers to Nicehash. We set up static difficulty on the workers and noticed that during the 90 seconds the amount of stale shares increased.
 
What can we conclude from this? User shares outside of the dev fee period could be used inside the dev fee period. This means that claymore's fee could be as high as 5% (or more).
Because a new block can appear at any time during the "accumulation" period, this would explain why we've been seeing stale shares. They are being gathered from the previous block.
 
We invite other users to verify this behavior and confirm this theory.

Edit; confirmed wrong. Nicehash ports on Zec aren't working as we assumed for preferred diffs, the miner does mine its dev fee at a lower difficulty.

I did, just looking for a legit reason not to release it which someone else provided (the fact Claymore has a hard time packing his binary to avoid it from being reversed).

I do confirm seeing noticeable gains (about 5Sol/s per card on certain models such as rx 480 4gb)

Good job! Sticking with silentarmy's miner definitely.  Hopefully further improvements can be made to gain even further hashrate!

Huh.. He did release 1.1 in linux.

Usually when you have a solid source (and he does, and he already produced windows and linux binaries on previous releases) it is easy to compile for all platforms from the same source code. Are you sure you know what you are talking about?

Mining on windows at large scale sucks, there is no other words for it.

I still can't understand why he won't release the linux version.

He's basically turning heads to the biggest miners out there, the large scale farmers.

Serious operations with lots of rigs don't run Windows. He could bank so much with his "dev fee" by providing binaries for linux.

Greetings everyone.

Here is a couple of scripts I made back then to add the ability to nagios infrastructures to monitor antminer devices. It's been tested on a large deployment of S7's without any issues. Basically any device that has similar outputs to cgminer-api running on port 4080 will be compatible.

Those scripts are perfdata enabled so everyone out there who generates graphs from their nagios instances (either through custom graphite environment or pnp4nagios) will be able to generate graphs for the statistics colleted by these scripts.

In order to generate multiple graphs I had made all checks as separate scripts. Also take note these scripts could be easily modified to monitor other types of cgminer compatible devices.

The scripts can check the following attributes:
- Check amount of blades detected
- Check amount of chips detected
- Check fan speeds
- Check hashrate (using the average hashrate, the hash5s (the "live" one) did fluctuate a lot on S7's)
- Check hardware errrors
- Check share rates
- Check temperature

Requirements: Those scripts rely on the use of nc (aka netcat) to poll the api port.

You can view the source and acquire these scripts on our repo:
https://bitbucket.org/hashrate/check_antminer

Enjoy!

Last time I looked S7's went for 450 USD per unit then a bit before being taken off from their website.
However you seem right, you can allegedly still buy large batches of S7's (500 to 1000+ units) from them at a lower price than ours.

Unfortunately still a bit low.

Each machine earns roughly around 91.75$ usd monthly at this time. When buying from Antminer the ROI is more like 5 to 6 months while at 325$ per unit it's of about 3.25 months (without counting power costs obviously).

The market price can't be 200-250$ish, that would be a ROI of 2 roughly months. It'd be worth more to just keep mining with them rather than selling them out for us at that price. Also, prices on ebay are in the 400$ range in most cases.

Nothing really planned yet, however if you want multiple just make an offer and we can negociate.

Sale cancelled. Offers we have received were too low.

Considering our low power rates we can still run them for a while while making decent profit. We'll instead ship those to our next facility located in an area with even lower power rates; they will likely be making profit for a lot longer.

Part of the dashboard ajax doesnt auto update it seems, like the current block. You may look below at the Found blocks table as it does auto update properly

Refresh, it's on block 140 as we speak.

Compiled and stratums started now!

Join us www.suchpool.pw/chemx

We're ready for launch!

You may create your accounts and workers here: www.SuchPool.pw/chemx

Just waiting on source to be released by the devs!

I can vouch. Dealing with him in real life.

I got some for you. I will need to take a day to take good pictures of our stuff, for now here is what I have for the ETH setup. It's located in our smaller farm.
We are currently running 20x 280x and 20x 380, looking to buy 20 more to load both racks fully and use the kits we have bought.

The racks were at temporary location on these pictures, there is no more S7's pushing heat into the GPU's

This is while I was still setting them up (was missing some PCI-e splitters and some risers and was also missing the motherboard trays):



The trays:



A 20-gpu rack of R9 Nitro 380's:



Everything is neat and tidy:

Hi everyone,

Just to add upon secup's pictures, I have a few more of that same farm given that we work together on this venture.

The upcoming pictures are still of Farm #2. We'll sort some pictures of farm #1 and post them soon.

The build was started a little bit over 3 weeks ago and we managed to complete it in order to host all the S7's we have received so far. Of course, a lot more is to come, but for the time being here are some pictures:

First of all, here is a picture of a full Row of S7's along with their transformer:



You can see here we still have a lot of switches for future racks. Our farm network is entirely redundant and built upon Cisco certified products for a robust setup.



Electrical distribution for one row (half a pod):



Here's another row of S7's:



Some overview of two pods. Lot more of those will be built in 2016!



Next pods are in progress, we keep receiving electrical distribution components.



And finally here are some custom sensors I built for the farm so we can monitor temperature, humidity and pressure at all times:



Feel free to share us your comments and thoughts.