Swedish ASIC miner company kncminer.com

[xstr8guy]

It didnt take Elenelen long to ramp right up to 500mhz with simple vrm heatsinks. Imagine what we could do if that cooling tower, thermal blanket, and that stupid bar were all gone, and replaced with an H2O cooler. the vrms, and any heatsink you put on them would then have excellent exposure to airflow.

That's all fine and dandy. But it's a complete waste of time and money this late in the game. Any tiny bit of performance you might gain is eaten up by miner downtime and the cost of additional parts.

If your Neptunes are running... just let them run.

[1713874237]

1713874237

[1713874237]

1713874237

[1713874237]

1713874237

[1713874237]

1713874237

[samsonn25]

Actually it currently takes 10 days with 3.3 TH/s to mine a bitcoin.

Can you imagine what its going to be like for the August/Sept SP30 customers ? Thats a few more diff changes down the road still ...

pretty much screwed like all the rest of the preorders to date...only one that worked great was avalon preorder...everyone thinking they will get that kind of success

I agree, Avalon first batch has been the only mining gear that has paid off way more than it cost. From here on in your lucky to break even at best. Unless someone comes up with a way to produce 1ph for around $10,000.

Not true... Many of us mined around 50 coin per Jupiter... just so you know....
with all the problems I had with Jupiters, it was worth every red cent imho.

... and Yifu Guo was no angel. There were a lucky few who got Avalons early, but make no mistake,
Avalon is just as guilty(or more) when it comes to orders which were delivered late.

http://www.coindesk.com/avalon-asic-customer-bitcoin-refunds-delivery-delays/

https://bitcointalk.org/index.php?topic=294234.0   Avalon chips refund not received - Lawsuit action proposal

http://www.reddit.com/r/Bitcoin/comments/1jwraw/1400000000_in_avalon_chips_missing_and_you/  "Held in customs"  yeah, right.

Whatever gave you the idea Avalon was so "Golden"?  and after a good shafting from Avalon, I bought from BFL.
Jupiter made up for all that, and more.

...but I agree.... Neptune, and every other miner screwed from here on out.... sad

 

There was alot of evidence of Avalon secretly mining weeks/months before delivery.

[opentoe]

I can see why jelin was so frustrated with the AX1200i. It won't even power one Neptune module plus controller... the Over Current Protection (OCP) kicks in and the PSU shuts down.

As annoying as this is, it is a safety feature as it would normally be unusual to draw that much current through a single PCIe cable. So i blame KnC for not putting two PCIe jacks on each module, rather than Corsair for their smart PSU. Bitmain put two PCIe jacks on each S3 blade, but it turns out only one is needed. That's a forward thinking design (in case the BM1382 28nm chip needed the power) - KnC take note.

Corsair link software lets you increase the OCP limit which fixes the problem... BUT the link cable is designed to plug into a motherboard USB pin header, not into a USB port (which makes sense as the PSU is designed to be inside a PC). Damn, I've only got a laptop.

I've got 2-to-1 PCIe cables ordered from Cablez, hopefully they will spread the Amps between two ports on the PSU and get around the OCP.

I'm also trying to source an EVGA 1300W superNOVA G2, but no luck yet finding one in New Zealand.

So my Neptune has been delivered, visually looks ok, but I can't power it on yet.

Months ago I kept saying when everyone starts to get their Neptunes they will have trouble powering them up and keeping them cool. This was the first line on my list that stopped my from pre-ordering one. Yea, I was on a "bitcoin high" and "wanted it first" like everyone else, but you have to put some brains into the picture to really figure it all out. The bottom line and my point here is by the time everyone buys all these power supplies and cooling products you will never get even the chance to break even. Just slap on an extra $500+ for power supplies and then if you had any electrical wiring down in your house that's not cheap either. People need to stop complaining about KNC and start using their brains before dumping out $10k on a pre-order. Trust me, I know the feeling. You see a brand new product and want to be one of the first to order it. Your blood pressure increases, you make typing mistakes on the keyboard because you want to move faster, you move some of that bitcoin around and BAM, make that purchase. Then right after the purchase, go to the forum and post a thread that you just ordered blah blah blah blah blah and can't wait to get it. Start posting what order# you got and the torture starts.

[opentoe]

It didnt take Elenelen long to ramp right up to 500mhz with simple vrm heatsinks. Imagine what we could do if that cooling tower, thermal blanket, and that stupid bar were all gone, and replaced with an H2O cooler. the vrms, and any heatsink you put on them would then have excellent exposure to airflow.

If someone can mount an Corsair H80 on the chip and mount a push/pull fan configuration right on the side of the box you can literally bring the size of that box down to a small shoe box size. I've been using that Corsair H80 on my overclocked Intel 3930K processor for a while now and works great. It is a completely sealed system and keeps the processor cool at all times. If there is a will, there would be a way. The ASIC chips are larger in size, so making it work would take a good DIY'er.

[kingscrown]

soon i will review their cloud mining service on my blog

[faetos]

People need to stop complaining about KNC and start using their brains before dumping out $10k on a pre-order. Trust me, I know the feeling. You see a brand new product and want to be one of the first to order it. Your blood pressure increases, you make typing mistakes on the keyboard because you want to move faster, you move some of that bitcoin around and BAM, make that purchase. Then right after the purchase, go to the forum and post a thread that you just ordered blah blah blah blah blah and can't wait to get it. Start posting what order# you got and the torture starts.

So true. Been there, done that, stopped doing it. If the Math doesn't work out then don't do it.

[Phoenix1969]

It didnt take Elenelen long to ramp right up to 500mhz with simple vrm heatsinks. Imagine what we could do if that cooling tower, thermal blanket, and that stupid bar were all gone, and replaced with an H2O cooler. the vrms, and any heatsink you put on them would then have excellent exposure to airflow.

If someone can mount an Corsair H80 on the chip and mount a push/pull fan configuration right on the side of the box you can literally bring the size of that box down to a small shoe box size. I've been using that Corsair H80 on my overclocked Intel 3930K processor for a while now and works great. It is a completely sealed system and keeps the processor cool at all times. If there is a will, there would be a way. The ASIC chips are larger in size, so making it work would take a good DIY'er.

https://bitcointalk.org/index.php?topic=616090.0
https://bitcointalk.org/index.php?topic=313978.msg5536997#msg5536997
https://i.imgur.com/vf9eekv.jpg
https://i.imgur.com/GXvNz3g.jpg
https://i.imgur.com/oZW5Bfs.jpg
https://i.imgur.com/uhhZlLx.jpg
https://i.imgur.com/IJzV8fA.jpg
https://i.imgur.com/xoHsaiJ.jpg
https://i.imgur.com/tQSoEJI.jpg
  being the actual chip package is the same size, I'd imagine so....
there were others... can't seem to find
* The Jupiters came just as "maxed out".  A true "hardware junkie" can get pretty far. If you increase your hashrate by say 20%,
for say... arguably 500 dollars(less than one twentieth of the neppy price)... you just increased your ROI by a substantial amount.
"Trance2" got 1000mhz, 758 Gh/s @ 50C out of October jupiters with watercooling and a bin file alteration. If you got that from Neppie....
in comparison, most unaltered Oct Jupiters get about 550-560Gh/s...
He increased his ROI by arouund 50% with about 500 bucks.
Sound better now?

Get crackin' Padrino....      

I made a small bin file for those not sure how to apply oc setting via cgminer.sh.

The bin file is loaded via the upgrade tab in miner administration just like you would upload a firmware. On reboot the modification will be gone again and bin file will have to be uploaded again in order to apply oc setting.

The bin file contains a folder with two files:

runme.sh

Code:

#!/bin/sh
cp cgminer.sh /etc/init.d/cgminer.sh
echo "900MHz OC setting copied! Please go back to Miner Settings page and Restart Miner <b>without rebooting</b> by <a href=\"/miner_setting.html\">clicking here</a>."

cgminer.sh

Code:

#!/bin/sh

PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin

use_bfgminer=
if [ -f /config/miner.conf ]; then
	. /config/miner.conf
fi
if [ "$use_bfgminer" = true ] ; then
	DAEMON=/usr/bin/bfgminer
	NAME=bfgminer
	DESC="BFGMiner daemon"
	EXTRA_OPT="-S knc:auto"
else
	DAEMON=/usr/bin/cgminer
	NAME=cgminer
	DESC="Cgminer daemon"
	EXTRA_OPT=
fi

set -e

test -x "$DAEMON" || exit 0

do_start() {
	# Stop SPI poller
	spi_ena=0
	i2cset -y 2 0x71 2 $spi_ena

	good_ports=""
	bad_ports=""

	# CLear faults in megadlynx's
	for b in 3 4 5 6 7 8 ; do
		for d in 0 1 2 3 4 5 6 7 ; do
			i2cset -y $b 0x1$d 3 >/dev/null 2>&1 || true
		done
	done

	for p in 0 1 2 3 4 5 ; do
		i2cset -y 2 0x71 1 $((p+1))
		good_flag=0
		ar="$(spi-test -s 50000 -OHC -D /dev/spidev1.0 0x80,3,0,0,0,0,0,0 | tail -c 13)"
                if [ "x$ar" = "x00 30 A0 01" ] ; then
			good_flag=1
		fi
		ar="$(spi-test -s 50000 -OHC -D /dev/spidev1.0 0x80,2,0,0,0,0,0,0 | tail -c 13)"
                if [ "x$ar" = "x00 30 A0 01" ] ; then
			good_flag=1
		fi
		ar="$(spi-test -s 50000 -OHC -D /dev/spidev1.0 0x80,1,0,0,0,0,0,0 | tail -c 13)"
                if [ "x$ar" = "x00 30 A0 01" ] ; then
			good_flag=1
		fi
		ar="$(spi-test -s 50000 -OHC -D /dev/spidev1.0 0x80,0,0,0,0,0,0,0 | tail -c 13)"
                if [ "x$ar" = "x00 30 A0 01" ] ; then
			good_flag=1
		fi

		if [ "$good_flag" = "1" ] ; then
			good_ports=$good_ports" $p"
		else
			bad_ports=$bad_ports" $p"
		fi
	done

	if [ -n "$good_ports" ] ; then
		for p in $good_ports ; do

			# Re-enable PLL START MOD FW 1.01
			i2cset -y 2 0x71 1 $((p+1))
			for c in 0 1 2 3 ; do
				cmd=$(printf "0x84,0x%02X,0,0" $c)
				spi-test -s 50000 -OHC -D /dev/spidev1.0 $cmd >/dev/null
				# cmd=$(printf "0x86,0x%02X,0x02,0x01" $c)
				# cmd=$(printf "0x86,0x%02X,0x02,0x11" $c)
				# cmd=$(printf "0x86,0x%02X,0x02,0x21" $c)
				cmd=$(printf "0x86,0x%02X,0x02,0x31" $c)
				# cmd=$(printf "0x86,0x%02X,0x02,0x41" $c)
				# cmd=$(printf "0x86,0x%02X,0x02,0x51" $c)
				spi-test -s 50000 -OHC -D /dev/spidev1.0 $cmd >/dev/null
				cmd=$(printf "0x85,0x%02X,0,0" $c)
				spi-test -s 50000 -OHC -D /dev/spidev1.0 $cmd >/dev/null
			done
			# END MOD FW 1.01

			# re-enable all cores
			i=0
			while [[ $i -lt 192 ]] ; do
				i2cset -y 2 0x2$p $i 1
				i=$((i+1))
			done
			spi_ena=$(( spi_ena | (1 << $p) ))
		done
	fi
	if [ -n "$bad_ports" ] ; then
		for p in $bad_ports ; do

			# Disable PLL START MOD FW 1.01
			i2cset -y 2 0x71 1 $((p+1))
			for c in 0 1 2 3 ; do
				cmd=$(printf "0x84,0x%02X,0,0" $c)
				spi-test -s 50000 -OHC -D /dev/spidev1.0 $cmd >/dev/null
			done
			# END MOD FW 1.01

			# disable all cores
			i=0
			while [[ $i -lt 192 ]] ; do
				i2cset -y 2 0x2$p $i 0
				i=$((i+1))
			done
			spi_ena=$(( spi_ena & ~(1 << $p) ))
		done
	fi

	# Disable direct SPI
	i2cset -y 2 0x71 1 0

	# Enable SPI poller
	i2cset -y 2 0x71 2 $spi_ena

	start-stop-daemon -b -S -x screen -- -S cgminer -t cgminer -m -d "$DAEMON" --api-listen -c /config/cgminer.conf $EXTRA_OPT
}

do_stop() {
	killall -9 bfgminer cgminer 2>/dev/null || true
}
case "$1" in
  start)
        echo -n "Starting $DESC: "
	do_start
        echo "$NAME."
        ;;
  stop)
        echo -n "Stopping $DESC: "
	do_stop
        echo "$NAME."
        ;;
  restart|force-reload)
        echo -n "Restarting $DESC: "
        do_stop
        do_start
        echo "$NAME."
        ;;
  *)
        N=/etc/init.d/$NAME
        echo "Usage: $N {start|stop|restart|force-reload}" >&2
        exit 1
        ;;
esac

exit 0

runme.sh will copy a modified version of cgminer.sh set for 900MHz to /etc/init.d/ overwriting the original cgminer.sh

After uploading the bin file, you need to go to the mining tab in administration and pick Restart Miner in order to apply setting.

Content of bin file can be examined and modified using e.g. 7zip

Other oc settings are listed in the modified cgminer.sh like:

Code:

# cmd=$(printf "0x86,0x%02X,0x02,0x01" $c)
# cmd=$(printf "0x86,0x%02X,0x02,0x11" $c)
# cmd=$(printf "0x86,0x%02X,0x02,0x21" $c)
 cmd=$(printf "0x86,0x%02X,0x02,0x31" $c)
# cmd=$(printf "0x86,0x%02X,0x02,0x41" $c)
# cmd=$(printf "0x86,0x%02X,0x02,0x51" $c)

# are disabled settings

Active setting is 231 which corresponds to 900MHz:

Code:

cmd=$(printf "0x86,0x%02X,0x02,0x31" $c)

Lowest disabled setting is 201 which corresponds to 825MHz:

Code:

# cmd=$(printf "0x86,0x%02X,0x02,0x01" $c)

Highest disabled setting is 251 which corresponds to 950MHz:

Code:

# cmd=$(printf "0x86,0x%02X,0x02,0x51" $c)

If you wish to go beyond these ranges, then look up the list of settings posted in various threads here.

Applying any other setting than the active 231 setting require that you modify the cgminer.sh in the bin file.

I take no responsibility for any damage due to oc if such should occur. The 231 setting is quite safe if you have proper cooling, but you will probably need to adjust die voltages to match the higher asic frequency.

Any changes made to voltages in Advanced Settings, will reset frequency to stock default setting and you will have to pick Restart Miner in the Mining tab in order to reapply oc settings.

Download available here:

https://www.dropbox.com/s/8w4dacsp8r1u9eu/1.01.900MHz.bin

Thanks to all those who shared their findings in this great thread.

no actually if you got two advanced settings you can go past 1000mhz i am currently @ 940ghs with 1050 power supply.

hehe, 950 gh/s on a jupiter.

Following old tricks...   same thing could be done with Neppie    

[xhomerx10]

It didnt take Elenelen long to ramp right up to 500mhz with simple vrm heatsinks. Imagine what we could do if that cooling tower, thermal blanket, and that stupid bar were all gone, and replaced with an H2O cooler. the vrms, and any heatsink you put on them would then have excellent exposure to airflow.

If someone can mount an Corsair H80 on the chip and mount a push/pull fan configuration right on the side of the box you can literally bring the size of that box down to a small shoe box size. I've been using that Corsair H80 on my overclocked Intel 3930K processor for a while now and works great. It is a completely sealed system and keeps the processor cool at all times. If there is a will, there would be a way. The ASIC chips are larger in size, so making it work would take a good DIY'er.

 Do you know how much heat the H80 can remove?  I think those ASICs are about three times the power draw of your 3930k... it probably wont be enough.

[xhomerx10]

There was actually "Talk" early on about a dc/dc "Team" who was supposedly working on a custom solution if I remember correctly.... evidently that never transpired.

 Nobody volunteered a Neptune box as a guinea pig and the crowd funding never got off the ground.
I might play around with one of my 4 VRM Saturn modules and see if I can upgrade the VRMs on it.  The problem is, they are GE Critical Power modules so any results wont be conclusive for Neptune owners.

[jelin1984]

thanks for donation btc for my english!!!!

btc adress is

1ALEoQMkQ1Pd5TedE2uc876kHF4KDC5haD

 

[Phoenix1969]

It didnt take Elenelen long to ramp right up to 500mhz with simple vrm heatsinks. Imagine what we could do if that cooling tower, thermal blanket, and that stupid bar were all gone, and replaced with an H2O cooler. the vrms, and any heatsink you put on them would then have excellent exposure to airflow.

If someone can mount an Corsair H80 on the chip and mount a push/pull fan configuration right on the side of the box you can literally bring the size of that box down to a small shoe box size. I've been using that Corsair H80 on my overclocked Intel 3930K processor for a while now and works great. It is a completely sealed system and keeps the processor cool at all times. If there is a will, there would be a way. The ASIC chips are larger in size, so making it work would take a good DIY'er.

 Do you know how much heat the H80 can remove?  I think those ASICs are about three times the power draw of your 3930k... it probably wont be enough.

That depends on the speed of your fan, and ambient temps; but it can dissipate up to about 2500 watts according to a Corsair support specialist, but he couldn't give me solid figures although I think that's a lofty estimate we don't need nearly that.
theres the H-100 as well, but were only talking about a single asic here... about 150-200 watts max per asic (300 watts total draw for the entire cube according to KNC) found here: https://www.kncminer.com/news/news-96 The asic alone may be half that maximum(an estimation)
It can be done with H60's.     How many watts of you think that cooling tower is capable of? It's already pushing it's spec limit in warmer climates.
Surely even an H60 will be far superior to that bulky archaic arctic tower heatsink.

[Easy2Mine]

What is the price of a H60?

[Phoenix1969]

What is the price of a H60?

http://www.amazon.com/Corsair-Series-Performance-Liquid-Cooler/dp/B00A0HZMGA/ref=sr_1_1?ie=UTF8&qid=1404411121&sr=8-1&keywords=h60
so yeah, 5 of those H60's run about 290 dollars shipped.    Gotta be "cost effective" there.

[xhomerx10]

It didnt take Elenelen long to ramp right up to 500mhz with simple vrm heatsinks. Imagine what we could do if that cooling tower, thermal blanket, and that stupid bar were all gone, and replaced with an H2O cooler. the vrms, and any heatsink you put on them would then have excellent exposure to airflow.

If someone can mount an Corsair H80 on the chip and mount a push/pull fan configuration right on the side of the box you can literally bring the size of that box down to a small shoe box size. I've been using that Corsair H80 on my overclocked Intel 3930K processor for a while now and works great. It is a completely sealed system and keeps the processor cool at all times. If there is a will, there would be a way. The ASIC chips are larger in size, so making it work would take a good DIY'er.

 Do you know how much heat the H80 can remove?  I think those ASICs are about three times the power draw of your 3930k... it probably wont be enough.

depends on the speed of your fan, and ambient temps; but it can dissapate up to about 2500 watts according to a Corsair support specialist, but he couldn't give me solid figures
theres the H-100 as well, but were only talking about a single asic here... about 400 watts max per asic(300 according to KNC)
it can be done with H60's.     How many watts of you think that cooling tower is capable of? It's already pushed it's spec limit.
Surely even an H60 will be superior to that bulky archaic arctic tower heatsink.

 Whoa! 2500 watts would suffice

[Phoenix1969]

It didnt take Elenelen long to ramp right up to 500mhz with simple vrm heatsinks. Imagine what we could do if that cooling tower, thermal blanket, and that stupid bar were all gone, and replaced with an H2O cooler. the vrms, and any heatsink you put on them would then have excellent exposure to airflow.

If someone can mount an Corsair H80 on the chip and mount a push/pull fan configuration right on the side of the box you can literally bring the size of that box down to a small shoe box size. I've been using that Corsair H80 on my overclocked Intel 3930K processor for a while now and works great. It is a completely sealed system and keeps the processor cool at all times. If there is a will, there would be a way. The ASIC chips are larger in size, so making it work would take a good DIY'er.

 Do you know how much heat the H80 can remove?  I think those ASICs are about three times the power draw of your 3930k... it probably wont be enough.

That depends on the speed of your fan, and ambient temps; but it can dissipate up to about 2500 watts according to a Corsair support specialist, but he couldn't give me solid figures although I think that's a lofty estimate we don't need nearly that.
theres the H-100 as well, but were only talking about a single asic here... about 150-200 watts max per asic (300 watts total draw for the entire cube according to KNC) found here: https://www.kncminer.com/news/news-96 The asic alone may be half that maximum(an estimation)
It can be done with H60's.     How many watts of you think that cooling tower is capable of? It's already pushing it's spec limit in warmer climates.
Surely even an H60 will be far superior to that bulky archaic arctic tower heatsink.

 Whoa! 2500 watts would suffice

even if he had said 2 to 500   its better imho

[tolip_wen]

Whilst on the subject of waterblocks.

My water loop is running
(not installed yet)
No leaks!
Might burp it once more but then it's time to take the plunge.

My goal is <15 minute pit stop per cube.


YMMV


Pics of 4kW single rail, water cooled nep to follow in a few hours if all goes well.

[Phoenix1969]

Whilst on the subject of waterblocks.

My water loop is running
(not installed yet)
No leaks!
Might burp it once more but then it's time to take the plunge.

My goal is <15 minute pit stop per cube.


YMMV


Pics of 4kW single rail, water cooled nep to follow in a few hours if all goes well.

You are the man! Can't wait to see what you're up to hehe
any firmware tweaks you recommend are always welcome too!

[padrino]

Good to see people getting into the Neptunes just like the Jupiter..

Although I think watercooling is great the way the Neptune is put together and the heat one must dissipate means an i30 with a decent fan provides plenty of heat dissipation.. The VRM and temps on the bottom of the board are two of the most concerning and really the place one needs to focus to keep things under control.

I have built over 30 1Th/s 4 board Jupiters so far (6 boards all run 1.5Th/s) and raising the boards a little and swapping out the fan for a fairly low pressure high CFM fan (keeps the noise down with a low pressure fan) is enough to keep air moving quick enough to keep them where they need to be..

On the Neptune side I haven't put my thermal camera on the boards yet since I don't have one but based on the layout it's a similar problem..

The biggest challenge I've seen is the VRM shutdown, with a hardcoded 48A shutdown I don't know how to push past that but I have replaced VRMs on the Jupiters so I might do it on the Neptune and see what happens, if I get one at some point..

[Phoenix1969]

Good to see people getting into the Neptunes just like the Jupiter..

Although I think watercooling is great the way the Neptune is put together and the heat one must dissipate means an i30 with a decent fan provides plenty of heat dissipation.. The VRM and temps on the bottom of the board are two of the most concerning and really the place one needs to focus to keep things under control.

I have built over 30 1Th/s 4 board Jupiters so far (6 boards all run 1.5Th/s) and raising the boards a little and swapping out the fan for a fairly low pressure high CFM fan (keeps the noise down with a low pressure fan) is enough to keep air moving quick enough to keep them where they need to be..

On the Neptune side I haven't put my thermal camera on the boards yet since I don't have one but based on the layout it's a similar problem..

The biggest challenge I've seen is the VRM shutdown, with a hardcoded 48A shutdown I don't know how to push past that but I have replaced VRMs on the Jupiters so I might do it on the Neptune and see what happens, if I get one at some point..

Yes!
The idea was if we get the big ass cooling tower, bar, and blanket out of the way, we may be able to cool those VRM's and the entire board a bit better.... what ya think?

[padrino]

Yes!
The idea was if we get the big ass cooling tower, bar, and blanket out of the way, we may be able to cool those VRM's a bit better.... what ya think?

I agree.. Since they sheered the aluminum the blankets are going to vary based on the unit, some will work much better than others.. I would like to try replacing the current setup with the same type of bracket on the old Neptunes, possibly some small VRM heatsinks (with 48A the heat dissipation is pretty insane) and a higher CFM fan.. I suspect that will be the easiest option that provides plenty of heat dissipation..

I'm most interested in swapping out the VRMs for the 50A version, I suspect they can really be pushed if that jump is taken... While on the Nov we can drive the VRMs much higher they still don't hit the 48A hard shutdown, in the Neptune the stock config is within a few percent of 48A..