Topic: ANTMINER S3+ Discussion and Support Thread - page 293. (Read 710164 times)

I have the same thoughts.  Since heat rises, perhaps the whole surface area of the controller PCB is getting the brunt of the radiating/rising heat from the heat sinks underneath it.  It's one of the factors that made me decide to orient my S3s vertically (vertical airflow).  This way, the controller is at the bottom (cooler) and vertically adjacent to the heat sinks and not above them.  Pictures here.

both have been running for 13h+ now, first one has 57 HW and second 166.

I have the same thoughts.  With me, it seems they stay between 41C-45C, whether hot (daytime) or cold (nighttime).

Has anyone actually tried cooling these S3's off.  I mean I had them in my garage and it was pretty warm
with 12 42-45) running.  So now I have them all in my basement and it is about 18 celsuis down here and they all still say 38,39,40 degrees on them.  Is it just me or do these fans on these things keep them around 39/40 on purpose?

I'm not concerned but i figured that they would run the fans the same and keep the chips cooler if they could?

Any thoughts on this idea?

Does anyone know if the --no submit stale cgminer flag works as a BMT cgminer param?

Hey "lady", have this Spoon Feed.
Just buy a Corsair RM1000. Enough for all your needs.
http://www.corsair.com/en/rm-series-rm1000-80-plus-gold-certified-power-supply

Re: PSU spam

What is up with all this PSU crap in this thread, its like product placement on crack!!! How hard is it for people to figure out what is the right PSU to use for an S3 ?

Everything still looks good but I'm curious about the HW figure on two miners, one of which on BTC Guild shows 768 dupes and the other 1280. Where as the other two miners have less than an HW of 30. All four though seem to be hashing at normal / advertised speeds. Just received all four Fri the 8th and they've been running as long as seen below.

Ty.

Miner 1 Stale / Dupe / Other : 4608 / 768 / 0

Miner 2 Stale / Dupe / Other : 5888 / 1280 / 0

Miner 3 Stale / Dupe / Other : 5376 / 0 / 0

Miner 4 Stale / Dupe / Other : 3328 / 0 / 0

No, there isn't.

Any/all address block(s) has/have to have a broadcast address. The standard is the last octet ends in .255 so for the class C 192.168.0 the broadcast address is .255, same for 192.168.1, 192.168.2, etc. and 172.16.0, 172.16.1, 172.16.2, etc. You have to have a broadcast address on the network for your device to advertise "here I am" so routers and switches will establish communications with your device and recognize your device's MAC address (which is expressed in HEX and is the physical hardware address of your network interface). If your devices on the network can't advertise their presence on a broadcast address such as .255, then your devices will never be seen on the network and you'll get no communications at all.

For example, address ranges run from 192.168.0.0 through 192.168.0.255, with 192.168.0.255 being the broadcast address, so you have 255 usable addresses on a class C (192.168.0.0 through 192.168.0.254). Same goes for 192.168.1,x, 192.168.2.x, 172.16.0.x, 172.16.1,x, etc. Traditionally the x.y.z.0 address isn't used.

The usual server power supply breakout boards should work best with ants, either the 750W, 800W or 2000W one.

They are also the most cost effective solutions as far as I know, and if a power supply should fail it can be replaced with another 20$ used server supply.

Sidehack has them on sale (including good quality pcie cables, and minersource is reselling them afaik).

Gas or Diesel? Mabey natural gas if u overclock both?(sorry but you know that guy pretending to be a girl will ask)

With the S3's it's a bit of a grab bag regarding what will overclock at what rate and using what power supply and how many pci-e connectors you need.

Example: I have one S3 that overclock's at 500-510 GH/s on a CX500M with 1 pci-e connection and a molex->pci-e adapter plugged into only 2 of the power connections for the S3. It has been running fine like this for almost 2 weeks now without having to do anything but monitor it. It also is connected by wifi without any change to the miner itself. Of course it probably helps the router is in the next room and straight line distance through a sheet rock wall is less than 10 feet. So that tells you for one instance an S3 runs just fine on a CX500M overclocked on wifi without any other changes and is stable for almost 2 weeks now.

In 3 other instances, running on a CX500 each, even connecting all 4 pci-e power connections via the 2 pci-e in the CX500 cable bundle and the molex->pci-e and sata->pci-e adapters, they will not overclock and perform extremely badly, so they're all back at stock clock and running on a single CX500 with the 2 pci-e connections and nothing else. Those 3 are connected to the network by hard wire, not wifi but it's a function of location, not if the wifi works or not.

The bottom line: it is highly dependent on the individual S3 itself if you'll actually be able to over clock or not, no matter what power supply you use. Some over clock without any problem and actually run better without additional power, some need the additional power and probably most simply will not overclock at all. As to the individual power supply to use, it's already been stated numerous times a single CX500 or CX600 (not the modular kind, i.e. CX500M or CX600M) is good PER MINER. Even a 500 watt power supply will give you a bit of room to try the over clocking. I've not seen any of my S3's use more than 390 watts even overclocked.

So if you're asking for advice on what power supply to get, and try to save a bit of money / coin doing it, CX500 or CX600, one per miner, is probably the most cost effective solution I've seen. Plus it has the added advantage that if a power supply goes out, you're down 1 miner, not 2.

Get a CX500 or CX600, one per miner. Stop over thinking the issue.

Is there anything I can do about it?

I grabbed a third pair of these.  this time  batch 6 due to ship the 12th of aug.

looking forward to these.

Its called margin of safety, a pci-e connector is made for 150w at 12 volts, it is really the current that matters, so it is made for 12.5Amps, that amount can be pushed significantly for example the Neptune pushes over 30 Amps, but it is not reliable since there are many cases of burnt cables.
Also not all power supply cables are equal some at 18g, others are 16 g, the 16g you only find on premium power supplies, so most people use the cheaper 18g, if you have a lot of current go over an 18g cable it will get hot.
other factor are the ambient temperature and humidity.
So if you overclock and wish be play it safe use the 4 pci-e connectors, thats why it has 4.
if you are using a cheap power supply you should also use 4.
with 16g cables you can get away and use only 2.

an overclock s3 uses 200watt, which is at 16.7A, which is about 5 Amps above the standard, by using all 4 pci-connectors you are using about 8.3A per connector which is very safe.

you do not have to do what knc did, and it is to use a single pci-e connector and create a fire hazard, in this case an 18g cable burns.

The advantage is safety, you completely eliminate the risk of a fire, this is very important since miners most of the time run unattended.

As more info, I had one S3 hashing board (blade) fail the day I set my S3's up.  I worked with Bitmain via email, and they offered to send me a replacement hashing board back once I sent them the failed one with the attached shipping label.  But, they did not attach the promised shipping label to the email.

Also, when I mentioned I didn't want a delay in hashing and would like to get one sent right away, they offered to let me put down a deposit on a whole S3 to be refunded once they received my old hashing board.  Not optimal, and not what I'd expect after the relationship with several rounds of purchases with them, but since I am definitely sending it, I suggested just putting a deposit on the board instead.

Haven't heard back from them for days now.

So, no shipping label, no response, no nothing.  And losing hash time.  I'm far less happy with them than I have been previously.  Poor communication and commitment to us time-sensitive miners.

I know there's different overclock settings but what's the smallest increment one can increase ? Or is it pretty much go by what BM has in the config files or nothing at all ?

Also, can OC / DC be done from within the webui ?

This oughta do it for 2 S3s:

I have been reading and it goes back and forth.. Oh the cx600 sucks.. It has bad components.. It doesn't have 4 PCI etc etc.. There is no definite answer.. Why is it so hard to ask what PSU or PSUs would be good to run two AntMiner s3 over clocked.

Just another shill employed to generate traffic.