Topic: BITMAIN Antminer S3 support and OverClocking thread - page 28. (Read 158144 times)

hello i own 2 antminer s3 b5's is it recommended to download the latest firmware? i read mixed results but no definite answer. ty

maybe try this http://www.rubicite.com/Tutorials/AntMinerS3Tutorial.aspx i found it more helpfull

I can not OC antminer S3. Does anyone just help me with the wrong place? Thanks !!! (S3-B6)
http://imagize[Suspicious link removed]ageshack.us/v2/150x100q90/909/1Jqp8B.jpg

You will then certainly see a drop in the stales and and an increase in returns!

I think both are overkills to a degree. From what I have read, @ 250M overclock, the unit consumes about 402 - 410 WATTS, thus it should be OK to run on a 550 WATT PSU with 80% efficiency. I am running one of my S3's on a second-hand server PSU rated at 550 WATTS that cost GBP 10.00 off fleabay, running off just 2 power connections @ 225M. I had to conjure up the power leads, but that did not cost as much as the PSU, and also had to fire up my solder-gun.
I will try out 250M and 256.25M in the next couple of days to check on stability and if it does run OK at the higher overclock speeds, I'll definitely be looking at getting rid of all the other PSU's and reverting to the server PSU's. I need to lower the cost of this "operation" given the fall in BTC, the anticipated difficulty increase, the proliferation of miners and (at least here in the UK) the anticipated increase in power costs this coming September!

By the way, the server PSU is the Dell PS-2521-1D

EDIT: Actually, @ 410 WATTS, the calculation is 410/0.8 = 512.5 WATTS, so if one can get a PSU rated at such for a better price, then that would be it, but I think I am content thus far with the Dell server PSU which should be rated higher than 80%

Ping to NL1 server is 35ms, UK1 server I can't even ping

I'm using Cooler Master V850 http://www.coolermaster.com/powersupply/v-series-psu/v850/ with all 4 PCI-E connected.
This PSU is possibly overkill so I will probably switch to Super Flower http://www.super-flower.com.tw/products_detail.php?class=2&sn=10&ID=56&lang=en

Yep, makes sense to connect to a local / nearer server to reduce that latency and get less stales, thus you should get a marginally better return.
I am amazed at the solidity of the S3's when overclocked comapred to the S1's.
By the way, what power supply are you using and are you using all four connections?

Thank you pekatete.
Unfortunately I'm clueless regarding batch as I bought it from third party, and impossible to know from the unit itself.

PS. I changed to nl1.ghash.io server, should be much better ping.

omnik, Yes that looks OK.
My only concern is the DiffR number of 37,000+ though for nearly 2 days that may be OK. This could mean your latency to the ghash servers is high as I have come to find out that this number represents the stale shares (at least on ghash), but since ghash only shows the last 24 hrs then you will not be able to compare now.
Otherwise, looks good! What batch was this S3 if you know?

Hi,

Does this all look OK, this is my 1st asic miner?
Latest firmware...



Thanks.

Status update:
I used the latest (3rd) software update and both my miners performed poorly compared to performance with the second release. In addition, one miner that had been very stable, and while it never pulled 504 GHs on250 other than momentarily, did deliver ~480. Now with this release it would consistently drop an ASIC at 250 and 235 as well. My average at any frequency dropped to sub-430. I tried reflashing, and no change in performance. So I decided to go back to release two and now performance is back to being stable at 480-482 GHs. No ASIC are being dropped. That was about 8 hours ago, but I saw ASIC drops within the first few minute and now even the hash averages are consistent with prior performance.

I am hesitant to change back my other miner, since performance is at least consistent with the prior release.

Anyone else experiencing the same?

that's actually the easiest way to do it

since your using 4 connectors, none..

it is more if you only use 2 connectors, then its something to do with the quality of the cables..

with 4 connectors your inside of specs for the cables in regard to pcie specs

in your case, you have to look at your psu.. it might be pushing it to overclock your antminers...

I'm running two batch 5's off of one thermaltake smart 800W, 4 PCI-E connectors.

How significant of a risk do I run by overclocking my batch 5's to 250MHz? Temperature of the S3's are fine (44/43) but I'm not sure how real the risk of burning out the PCI-E cords is.

ok,
I edit and add the missing lines and it works perfectly . 
thank you again.

Ah, it might be this bug:

"There is a problem, when run cgminer as a daemon, it will use 100% cpu. Hope anybody can revise this."

https://github.com/alzhao/openwrt-cgminer-antminer

Does cgminer run at 100% cpu for anyone else?

Just flashed.. good to go so far.
Thanks JakeTri for the info

I like it

I hope they do it soon .. I'm waiting for this to be able to play with code and see if it is possible to set frequency for each board from the miner.

NO ~!!!

So when I put the freq to 250 I get 1 x so I decided to tame it down to 237.5 and its doing fine..
My hash rate on ghash is jumping from 200GH/s to 400GH/s is that normal?