[Guide] Dogie's Comprehensive ASICMiner Blade Setup - page 64.

ahitman

sr. member

Activity: 302

Merit: 250

I noticed some weird behaviour today with my blade. I am running stratum proxy on bitparking to feed my blade. Today the pool went down and my gpu miners running cgminer switched to its backup pool, where the blade just stayed on the proxy with 0mhs and resetting every 2 minutes. It does have a backup pool set up and the switch server button makes the blade work fine on the second pool, but this did not happen automatically.

Anyone know how this can be fixed? It seems like whatever algorithm cgminer uses to switch pools is better than what the blade has built in. Is there any way to update the firmware or any of the parameters?

This is what stratum proxy is showing while the blade is trying to get work:

Quote

Unhandled error in Deferred:
Unhandled Error
Traceback (most recent call last):
File "twisted\web\server.pyo", line 156, in process

File "twisted\web\server.pyo", line 191, in render

File "twisted\web\resource.pyo", line 216, in render

File "mining_libs\getwork_listener.pyo", line 163, in render_POST

--- ---
File "twisted\internet\defer.pyo", line 134, in maybeDeferred

File "mining_libs\worker_registry.pyo", line 37, in authorize

File "stratum\socket_transport.pyo", line 93, in rpc

stratum.custom_exceptions.TransportException: Not connected
2013-05-24 20:32:51,382 ERROR proxy # Connection to upstream pool timed out

dogie

legendary

Activity: 1666

Merit: 1185

dogiecoin.com

Quote from: sveetsnelda on May 24, 2013, 03:10:52 PM

Quote from: dogie on May 24, 2013, 11:06:43 AM

This is heat based.

It happens even on low clocks and with a large box fan blowing very cool air at the blades. I don't think it's heat based. In fact, after further testing, they still reboot at the same time even when connected to different proxies/pools.

It's just not likely that two blades with overkill cooling are overheating & locking up every ASIC chip at the exact same time. it's next to impossible, really. Furthermore, if it were heat based, I'm sure i'd get much less than an hour out of them after the reboot.

Feel around the entire board, including the chips on the back.

dogie

legendary

Activity: 1666

Merit: 1185

dogiecoin.com

Quote from: pikeadz on May 24, 2013, 02:03:03 PM

Overall I am very happy with this blade. Today, however, I noticed it restarted itself for no reason. This was after hashing perfectly at 13,000 mh/s for the last 3 days or so. Is that normal? It restarted and is slowly working its way back up to 13,000 mh/s again on Slush's pool. Should I be concerned?

If it lasted 3 days then its neither heat nor power. Its just a bug, hence friedcat offering the new power delivery and network boards for stability.

sveetsnelda

hero member

Activity: 642

Merit: 500

Quote from: dogie on May 24, 2013, 11:06:43 AM

This is heat based.

It happens even on low clocks and with a large box fan blowing very cool air at the blades. I don't think it's heat based. In fact, after further testing, they still reboot at the same time even when connected to different proxies/pools.

It's just not likely that two blades with overkill cooling are overheating & locking up every ASIC chip at the exact same time. it's next to impossible, really. Furthermore, if it were heat based, I'm sure i'd get much less than an hour out of them after the reboot.

pikeadz

full member

Activity: 238

Merit: 100

Overall I am very happy with this blade. Today, however, I noticed it restarted itself for no reason. This was after hashing perfectly at 13,000 mh/s for the last 3 days or so. Is that normal? It restarted and is slowly working its way back up to 13,000 mh/s again on Slush's pool. Should I be concerned?

piit79

sr. member

Activity: 336

Merit: 250

Time for Plan ฿

Quote from: dogie on May 24, 2013, 11:08:58 AM

Be careful, 20awg is smaller than 18

I know. I'm not sure, it might actually be 18 AWG. These are good quality Akasa extensions.

dogie

legendary

Activity: 1666

Merit: 1185

dogiecoin.com

Quote from: piit79 on May 24, 2013, 04:08:40 AM

Just in case anybody's interested in my installation

Two 120mm fans on the heatsink in push/pull config, 140mm fan at the back blowing at the chips. All three fans are powered from one HDD molex connector

I'm using an old-ish ATX PSU with just one PCI-e power connector so I'm powering the other blade from a 4-pin CPU power connector. I purchased 40cm extension cables (male-female) for both (PCI-e power & 4-pin CPU), removed the female connectors on the other end and attached the blade power connectors. This way the blades plug into the high-amp PSU connectors nicely. At 20 AWG 5 amp per cable is not a problem at all provided that there is good contact on the blade connector end.

Be careful, 20awg is smaller than 18

dogie

legendary

Activity: 1666

Merit: 1185

dogiecoin.com

Quote from: mjmvisser on May 23, 2013, 09:11:25 PM

Hi, just chiming in with my experience with power distribution to my BE after seeing all the melted wire posts.

Instead of using a molex splitter, I used one of these:

Cut the 6-pin connector off the end and remove the third (not yellow, not black) wire from the molex plugs. Don't cut it -- just reach in with a pair of tweezers or needle-nose pliers and gently squeeze the pin at the base and you should be able to slide it right out.

You'll now have two molex connectors that you can wire to the power plug: two yellow to the + side, two black to the -. Double-check with a multi-meter before you plug it in -- always double-check! You can get a multi-meter for less than $10.

Use two molex connectors on the same feed from your PSU. I've been running like this for 2 days now at "high" clock and the cables are barely warm. Don't skimp on fans, either. I have 4, two against the heat sink and another two on the other side of the board. I'm going to add two more on the back of the power board, too, it's quite warm.

Absolutely no different to the recommended setup, its still 2 lives and 2 neutrals [actually compared with my 2 lives and 4 neutrals]. The problem with burnt cables is some brands are selling knockoff cabling which isn't sufficient.

Quote from: sveetsnelda on May 23, 2013, 09:24:44 PM

Has anyone had an issue with their blade(s) where they'll mine for a while (minutes or hours) and reboot on a regular basis? Both of mine will run for an hour or so, stop hashing for a couple minutes, and reboot (triggers the failsafe). It's great that they reset automatically, but I'm losing a few minutes of hashing each hour. I'm running them through slush's stratum proxy and they maintain ethernet communication the entire time (the web page stays accessible and they're connected to the proxy).

I tried moving one to another pool (under another instance of the proxy using a different port) and it still does the same thing, although they now reset at different times. It seems like a stratum proxy or pool issue, but I'd expect it to pick back up after a disconnect (if that's what's happening). When the blades go down for a couple minutes, my GPUs keep hashing (although they aren't using the proxy).

I'm going to start digging into it, but I wanted to check to see if someone else ran into the same thing.

This is heat based.

WinTame2012

sr. member

Activity: 315

Merit: 250

Official sponsor of Microsoft Corp.

Quote from: piit79 on May 24, 2013, 10:52:22 AM

Quote from: WinTame2012 on May 24, 2013, 10:46:28 AM

Quote from: slastar on May 24, 2013, 05:20:50 AM

carefully with these adapters, they have copper cables with smaller φ so they heat up more then aluminium molex-sata adapter or molex splitter

Always believed that copper is better conductor than aluminium...

That is true in general. However, it also depends on the cross-section (gauge) of the cable/conductor.

What slastar meant probably was that these adapters have fairly low gauge wires. I also personally think that the molex HDD connector doesn't make a very good contact so the connector will heat up as well.

It's just your assumption. I'v read his post and didn't found he refer AWG. Given equal cross-section copper is much better so I'd write posts to avoid any misleading.

piit79

sr. member

Activity: 336

Merit: 250

Time for Plan ฿

Quote from: WinTame2012 on May 24, 2013, 10:46:28 AM

Quote from: slastar on May 24, 2013, 05:20:50 AM

carefully with these adapters, they have copper cables with smaller φ so they heat up more then aluminium molex-sata adapter or molex splitter

Always believed that copper is better conductor than aluminium...

That is true in general. However, it also depends on the cross-section (gauge) of the cable/conductor.

What slastar meant probably was that these adapters have fairly low gauge wires. I also personally think that the molex HDD connector doesn't make a very good contact so the connector will heat up as well.

WinTame2012

sr. member

Activity: 315

Merit: 250

Official sponsor of Microsoft Corp.

Quote from: slastar on May 24, 2013, 05:20:50 AM

Quote from: mjmvisser on May 23, 2013, 09:11:25 PM

Hi, just chiming in with my experience with power distribution to my BE after seeing all the melted wire posts.

Instead of using a molex splitter, I used one of these:

Cut the 6-pin connector off the end and remove the third (not yellow, not black) wire from the molex plugs. Don't cut it -- just reach in with a pair of tweezers or needle-nose pliers and gently squeeze the pin at the base and you should be able to slide it right out.

You'll now have two molex connectors that you can wire to the power plug: two yellow to the + side, two black to the -. Double-check with a multi-meter before you plug it in -- always double-check! You can get a multi-meter for less than $10.

Use two molex connectors on the same feed from your PSU. I've been running like this for 2 days now at "high" clock and the cables are barely warm. Don't skimp on fans, either. I have 4, two against the heat sink and another two on the other side of the board. I'm going to add two more on the back of the power board, too, it's quite warm.

carefully with these adapters, they have copper cables with smaller φ so they heat up more then aluminium molex-sata adapter or molex splitter

Always believed that copper is better conductor than aluminium...

slastar

full member

Activity: 147

Merit: 104

Quote from: mjmvisser on May 23, 2013, 09:11:25 PM

Hi, just chiming in with my experience with power distribution to my BE after seeing all the melted wire posts.

Instead of using a molex splitter, I used one of these:

Cut the 6-pin connector off the end and remove the third (not yellow, not black) wire from the molex plugs. Don't cut it -- just reach in with a pair of tweezers or needle-nose pliers and gently squeeze the pin at the base and you should be able to slide it right out.

You'll now have two molex connectors that you can wire to the power plug: two yellow to the + side, two black to the -. Double-check with a multi-meter before you plug it in -- always double-check! You can get a multi-meter for less than $10.

Use two molex connectors on the same feed from your PSU. I've been running like this for 2 days now at "high" clock and the cables are barely warm. Don't skimp on fans, either. I have 4, two against the heat sink and another two on the other side of the board. I'm going to add two more on the back of the power board, too, it's quite warm.

carefully with these adapters, they have copper cables with smaller φ so they heat up more then aluminium molex-sata adapter or molex splitter

piit79

sr. member

Activity: 336

Merit: 250

Time for Plan ฿

Just in case anybody's interested in my installation

Two 120mm fans on the heatsink in push/pull config, 140mm fan at the back blowing at the chips. All three fans are powered from one HDD molex connector

I'm using an old-ish ATX PSU with just one PCI-e power connector so I'm powering the other blade from a 4-pin CPU power connector. I purchased 40cm extension cables (male-female) for both (PCI-e power & 4-pin CPU), removed the female connectors on the other end and attached the blade power connectors. This way the blades plug into the high-amp PSU connectors nicely. At 20 AWG 5 amp per cable is not a problem at all provided that there is good contact on the blade connector end.

drlatino999

sr. member

Activity: 335

Merit: 250

Quote from: sveetsnelda on May 23, 2013, 09:24:44 PM

Has anyone had an issue with their blade(s) where they'll mine for a while (minutes or hours) and reboot on a regular basis? Both of mine will run for an hour or so, stop hashing for a couple minutes, and reboot (triggers the failsafe). It's great that they reset automatically, but I'm losing a few minutes of hashing each hour. I'm running them through slush's stratum proxy and they maintain ethernet communication the entire time (the web page stays accessible and they're connected to the proxy).

I tried moving one to another pool (under another instance of the proxy using a different port) and it still does the same thing, although they now reset at different times. It seems like a stratum proxy or pool issue, but I'd expect it to pick back up after a disconnect (if that's what's happening). When the blades go down for a couple minutes, my GPUs keep hashing (although they aren't using the proxy).

I'm going to start digging into it, but I wanted to check to see if someone else ran into the same thing.

I'm having issues with my proxies, every 2 minutes or so they lock up in turn and the blade switches to the next. Rinse and repeat indefinitely.

WinTame2012

sr. member

Activity: 315

Merit: 250

Official sponsor of Microsoft Corp.

Quote from: sveetsnelda on May 23, 2013, 09:24:44 PM

Has anyone had an issue with their blade(s) where they'll mine for a while (minutes or hours) and reboot on a regular basis? Both of mine will run for an hour or so, stop hashing for a couple minutes, and reboot (triggers the failsafe). It's great that they reset automatically, but I'm losing a few minutes of hashing each hour. I'm running them through slush's stratum proxy and they maintain ethernet communication the entire time (the web page stays accessible and they're connected to the proxy).

I tried moving one to another pool (under another instance of the proxy using a different port) and it still does the same thing, although they now reset at different times. It seems like a stratum proxy or pool issue, but I'd expect it to pick back up after a disconnect (if that's what's happening). When the blades go down for a couple minutes, my GPUs keep hashing (although they aren't using the proxy).

I'm going to start digging into it, but I wanted to check to see if someone else ran into the same thing.

I've run into this yesterday when I've tried to mine on old-fashion getwork pool (w/o any proxy). It have reset after several minutes. And I've ceased my counter-science experiments

sveetsnelda

hero member

Activity: 642

Merit: 500

Has anyone had an issue with their blade(s) where they'll mine for a while (minutes or hours) and reboot on a regular basis? Both of mine will run for an hour or so, stop hashing for a couple minutes, and reboot (triggers the failsafe). It's great that they reset automatically, but I'm losing a few minutes of hashing each hour. I'm running them through slush's stratum proxy and they maintain ethernet communication the entire time (the web page stays accessible and they're connected to the proxy).

I tried moving one to another pool (under another instance of the proxy using a different port) and it still does the same thing, although they now reset at different times. It seems like a stratum proxy or pool issue, but I'd expect it to pick back up after a disconnect (if that's what's happening). When the blades go down for a couple minutes, my GPUs keep hashing (although they aren't using the proxy).

I'm going to start digging into it, but I wanted to check to see if someone else ran into the same thing.

mjmvisser

newbie

Activity: 58

Merit: 0

Hi, just chiming in with my experience with power distribution to my BE after seeing all the melted wire posts.

Instead of using a molex splitter, I used one of these:

Cut the 6-pin connector off the end and remove the third (not yellow, not black) wire from the molex plugs. Don't cut it -- just reach in with a pair of tweezers or needle-nose pliers and gently squeeze the pin at the base and you should be able to slide it right out.

You'll now have two molex connectors that you can wire to the power plug: two yellow to the + side, two black to the -. Double-check with a multi-meter before you plug it in -- always double-check! You can get a multi-meter for less than $10.

Use two molex connectors on the same feed from your PSU. I've been running like this for 2 days now at "high" clock and the cables are barely warm. Don't skimp on fans, either. I have 4, two against the heat sink and another two on the other side of the board. I'm going to add two more on the back of the power board, too, it's quite warm.

WinTame2012

sr. member

Activity: 315

Merit: 250

Official sponsor of Microsoft Corp.

Quote from: eastpk on May 23, 2013, 07:34:54 PM

What pools have you guys been using? I've been on slush but was thinking about testing BTC Guild to see if its any better of there.

I'm trying 50btc and Bitparking right now, both via local Stratum-proxy.

dogie

legendary

Activity: 1666

Merit: 1185

dogiecoin.com

Quote from: piit79 on May 23, 2013, 06:53:09 PM

Quote from: bitjoint on May 23, 2013, 02:40:51 PM

Quote from: dogie on May 19, 2013, 03:22:19 AM

24 Hour Earnings 2.656 *faps*

Hey dogie... still getting 2.656 daily? Which pool are you using?

Wow, how did you achieve that, lucky day? I assume that was with 4 blades?

I'm on 50BTC and on a good day I get 1.1 BTC with 2 blades.

It was lucky on BTCGuild. Theoretical is about 2.4 with 4 blades.

eastpk

member

Activity: 120

Merit: 10

WINSTARS - We are changing the face of gambling

What pools have you guys been using? I've been on slush but was thinking about testing BTC Guild to see if its any better of there.

Topic: [Guide] Dogie's Comprehensive ASICMiner Blade Setup - page 64. (Read 580761 times)