ANTMINER S3+ Discussion and Support Thread

[jelin1984]

how install 4.6 to s3?
any details?
step by step?

[Count_Frackula]

Could you report hash difference (if any) after 24hr?

I sure will (and that is why I said "still early days ...".

Well, it has been a few days. Can you report any increase in hashing on 4.6.0 in comparison with the stock version?
Thanks.

Yep I can. I got the OK, not so good and bad news I am afraid.
1. The OK news
Upgrading to the latest cgminer clearly reduces the load and that definitely slightly improves the hashing. It may not be significant, but it certainly does.
I have not been able to collate any empirical evidence to this effect (see below), but I have no doubt cgminer 4.6 hashes faster than the stock one.
2. The not so good:
a) I have NOT been able to run a full un-interupted 24hr stretch with cgminer 4.6
b) The S3 that I updated cgminer on was overclocked to 262.5 and had been running very well without any significant HW errors.
3. The BAD
a) Having left the unit to hash as usual, I logged into the GUI the next day and was greeted with HW errors counting in the 100,000 ! In the panic, I immediately clicked the system tab and selected reboot. (I forgot to take a shot of the screen).
b) Yesterday, I gave the unit close observation for the few hours I had and the errors kept cropping up over time. This had thown me off until late last night I noticed a pattern.
The unit will hash as usual after a reboot without any issues and single digit HW errors for 2hrs. Between 2 and 3 hrs, there are two scenarios that can develop.
1. HW errors creep up into double digits and past 3 hrs they'll shoot up at will! A refresh would give any additional number from single to three digits!
2. The other scenario is one of the chips is marked X - when this happens, the HW errors remain normal BUT the hashrate decreases.
In both the scenarios above, I noticed the GUI hash-rate reduced to approx the one I had before updating cgminer at 256.25, so early this morning I decided to reduce my speed to 256.25 and so far, running like a champ! I believe by early tomorow I'll have the numbers to show against my earlier tests (barring another HW error panic ....!).

One thing that is still whirling away in my mind is why the same chip always comes trumps with the X? Also this may have something to do with power, as I am running the S3 off a 550 watt server PSU. If the unit is drawing anything over 460 watts, then that might be the reason.

Finally, I noticed that after updating to cgminer 4.6 if I do a power cycle or a reboot via the GUI, the miner status is screwed up. At times it will say its been up for hours and show a puny avg GH/s (say 12 GH/s avg) whereas I have just rebooted! But when I reboot via SSH (putty) this does not happen.

That's unusual.  I've been running 4.6 for ~2 days now with no issues (clocked at 250).  You can see from my hash results that there's been no significant performance difference: https://www.miningrigrentals.com/rigs/491
That small dip on the 8th was me upgrading.

[Count_Frackula]

how install 4.6 to s3?
any details?
step by step?

1. SSH into the S3 and login
2. Issue the following commands sequentially.

Code:

cd /usr/bin
mv cgminer cgminer.bak
wget http://ck.kolivas.org/apps/cgminer/antminer/s3/4.6.0-140908/cgminer
chmod +x cgminer
reboot

[Meech]

Yeah!  Mine are showing shipped as well.  Hopefully this weekend or monday at latest to the U.S.

[pekatete]

That's unusual.  I've been running 4.6 for ~2 days now with no issues (clocked at 250).  You can see from my hash results that there's been no significant performance difference: https://www.miningrigrentals.com/rigs/491
That small dip on the 8th was me upgrading.

It is unusual, 4.6 has been out for justa acouple of days for starters, and I was running ar 262.75???
But I'd say there certainly is an improvement in the hash-rate, if only (as attested above) attributable to a reduction and hash-variance.

[CHAOSiTEC]

Has the CG miner modification increased power draw along with the increased performance, or has it increased hash without any additional power usage?
If the latter then this may mean most may be able to ROI, which would be nice.

Also.....any idea if this will have a similar affect on S3+ (Theoretically speaking ofcourse!) 

ive just looked at my power meter, and it went from 1888 watt to 1902 watt, this is across 3x antminer S1s and 2x antminer S3s

so, its a slight difference, but neglectable i would say since my hashing has also gone up a little on both S3s

[eoakland]

hi all, does anyone know which pcb is chain 2... is it the side that is connected to exhaust or front fan ?  i have a bad chain on my s3 and need to isolate chain 2 for RMA.  also, does anyone know if you hash your s3 with 1 pcb only will it mess it up ?  i disconnected one pcb, and the fans went bonkers...spinning rapidly.  thanks in advance.

eoakland

[bluecityste]

Has the CG miner modification increased power draw along with the increased performance, or has it increased hash without any additional power usage?
If the latter then this may mean most may be able to ROI, which would be nice.

Also.....any idea if this will have a similar affect on S3+ (Theoretically speaking ofcourse!) 

ive just looked at my power meter, and it went from 1888 watt to 1902 watt, this is across 3x antminer S1s and 2x antminer S3s

so, its a slight difference, but neglectable i would say since my hashing has also gone up a little on both S3s

Thanks....seems like they are running slightly more efficiently then, approximately 40 ghs extra and only 14 watt increase so not earth shattering but a nice extra bonus potentially.

[Rabinovitch]

The few reasons I liked Bitmain have quickly deteriorated. I'll be looking for a more reputable company for future purchases!

Looks like there is no such company like Bitmain. pcfli offers 1Th and 1.5 Th miners by LKETC, but shipment cost (to me in Russia) is 400$! Unf***belieaveble! How could Bitmain provide free shipping for S2 (dimensions and weight are the same as Dragon has), HOW?..

[Biodom]

how install 4.6 to s3?
any details?
step by step?

on page 317

[Phosphorous]

how install 4.6 to s3?
any details?
step by step?

on page 317

https://bitcointalk.org/index.php?topic=671189.6340

[Biodom]

Could you report hash difference (if any) after 24hr?

I sure will (and that is why I said "still early days ...".

Well, it has been a few days. Can you report any increase in hashing on 4.6.0 in comparison with the stock version?
Thanks.

One thing that is still whirling away in my mind is why the same chip always comes trumps with the X? Also this may have something to do with power, as I am running the S3 off a 550 watt server PSU. If the unit is drawing anything over 460 watts, then that might be the reason.

I have one batch1 miner at 237.5 with one ASIC that always has the X, hence it is at 464-465Gh instead of 478Gh, but since it is still better than 441 at default, I let it be.

[pekatete]

Could you report hash difference (if any) after 24hr?

I sure will (and that is why I said "still early days ...".

Well, it has been a few days. Can you report any increase in hashing on 4.6.0 in comparison with the stock version?
Thanks.

One thing that is still whirling away in my mind is why the same chip always comes trumps with the X? Also this may have something to do with power, as I am running the S3 off a 550 watt server PSU. If the unit is drawing anything over 460 watts, then that might be the reason.

I have one batch1 miner at 237.5 with one ASIC that always has the X, hence it is at 464-465Gh instead of 478Gh, but since it is still better than 441 at default, I let it be.

I suppose I could let it be, but it seems to be running per my "normal" now I have turned it down a notch and thus far a touch above the hash-rate I got when overclocked at the higher freq (with 4.6 upgrade).

This, however, is the proof of the pudding (to me at least) that 4.6 is worthy of getting, if not just for the security bug-fixes.

Couple of days of running under 4.6.0 now, and very happy. CPU load lower, and less hashrate variance.

BEFORE:


AFTER:


Thank you Con!

And the proverbial chicken and egg situation:
The smoother hash-rate variance is due to stable submissions thus accurate dynamic Diff allocation, or is it the other way round?

[faetos]

I'm trying 4.6 as well on two batch 2s. So far so good - stable, load is lower and web gui is more responsive (due to lower load). Hashrate  is very stable as well

[stan258]

After updating to 4.6 one S3 would not hash.  Maybe I missed something.  After updating, reboot then save and apply freq in advanced settings unit should fire up?

[Count_Frackula]

After updating to 4.6 one S3 would not hash.  Maybe I missed something.  After updating, reboot then save and apply freq in advanced settings unit should fire up?

Is all your base firmware also up to date?

[thehairymob]

Nope still not working. Follow all commands but had to delete cgminer from ant so it would install new version or I got an error. After that the s3 wouldn't mine so had to reflash it. Working again. Lost some more hair trying this, hehehe.

[stan258]

After updating to 4.6 one S3 would not hash.  Maybe I missed something.  After updating, reboot then save and apply freq in advanced settings unit should fire up?

Is all your base firmware also up to date?

S320140811.bin?

[Count_Frackula]

After updating to 4.6 one S3 would not hash.  Maybe I missed something.  After updating, reboot then save and apply freq in advanced settings unit should fire up?

Is all your base firmware also up to date?

S320140811.bin?

So no.
I would update to the most recent version, uncheck the box to keep settings, then apply cgminer using the following:

1. SSH into the S3 and login
2. Issue the following commands sequentially.

Code:

cd /usr/bin
mv cgminer cgminer.bak
wget http://ck.kolivas.org/apps/cgminer/antminer/s3/4.6.0-140908/cgminer
chmod +x cgminer
reboot

[navigator]

My B8 order shipped, says it will be here friday