[Guide] Dogie's Comprehensive ASICMiner Blade Setup

[bclcjunkie]

i had a similar issue whenever i switched to "High" mode. At the end it turned out to be insufficient power flow so i got it resolved after changing to 850w Seasonic. Probably more than sufficient for 2 blades but i decided to go for more ampers in case i wanted to get another blade..
Also get a PSU from reputable vendors like Corsair, Seasonic. The first PSU i got was thermaltake 530w and i believe it was of low quality and not pumping enough juice as advertised...

you can confirm power issue with connecting just one blade and hitting "High", if it runs stable without xxx then you may want to upgrade your PSU.

Is there a way to correct it?

Same thing happens with mine.  I find that when I do an update/restart in high clock mode, I get the x on one of the chips.  However, if I switch into low clock mode and then do the update, and then switch to high, all the chips display 0.  Try that?  Otherwise, Friedcat has said this is a voltage issue that is not terribly important if it still hashes at the normal rate and only affects one chip.

[1715249398]

1715249398

[1715249398]

1715249398

[1715249398]

1715249398

[1715249398]

1715249398

[pikeadz]

i had a similar issue whenever i switched to "High" mode. At the end it turned out to be insufficient power flow so i got it resolved after changing to 850w Seasonic. Probably more than sufficient for 2 blades but i decided to go for more ampers in case i wanted to get another blade..
Also get a PSU from reputable vendors like Corsair, Seasonic. The first PSU i got was thermaltake 530w and i believe it was of low quality and not pumping enough juice as advertised...

you can confirm power issue with connecting just one blade and hitting "High", if it runs stable without xxx then you may want to upgrade your PSU.

Is there a way to correct it?

Same thing happens with mine.  I find that when I do an update/restart in high clock mode, I get the x on one of the chips.  However, if I switch into low clock mode and then do the update, and then switch to high, all the chips display 0.  Try that?  Otherwise, Friedcat has said this is a voltage issue that is not terribly important if it still hashes at the normal rate and only affects one chip.

Interesting, hadn't thought of that.  I am powering two blades with a corsair CX750.  They both hash just fine when I do the workaround, but I wonder if they will hash faster if I get a separate PSU for each one.  I am getting a reliable 24,500 to 25,000 Mh/s long term average rate on any pool I use.  The blades themselves in the config screen say they are hashing at 13,050 or so, so I don't know if it's worth it to tinker or just let them run.  Every second counts with these difficulty jumps.

[mitty]

I'm running my blades on a brand new 1000w Seasonic platinum PSU and tested the voltage and it seems to be fine; also I put the blades directly in front of a running air conditioner and the heatsinks are warm, not even hot right now so I don't think that's the issue. Also I feel like if heat was an issue then all four wouldn't fail simultaneously. I'll try a different PSU later today and report back. Thanks everyone.

[WinTame2012]

I'm running my blades on a brand new 1000w Seasonic platinum PSU and tested the voltage and it seems to be fine; also I put the blades directly in front of a running air conditioner and the heatsinks are warm, not even hot right now so I don't think that's the issue. Also I feel like if heat was an issue then all four wouldn't fail simultaneously. I'll try a different PSU later today and report back. Thanks everyone.

I doubt you can read. You've been told above that AC is not a solution. You need good airflow through the heatsink's area and back side where chips are located.

[mitty]

I'm running my blades on a brand new 1000w Seasonic platinum PSU and tested the voltage and it seems to be fine; also I put the blades directly in front of a running air conditioner and the heatsinks are warm, not even hot right now so I don't think that's the issue. Also I feel like if heat was an issue then all four wouldn't fail simultaneously. I'll try a different PSU later today and report back. Thanks everyone.

I doubt you can read. You've been told above that AC is not a solution. You need good airflow through the heatsink's area and back side where chips are located.

I have the blades mounted vertically in a PVC pipe rack such that sufficient amount of airflow is going over the front and back of each board.  Additionally the boards ran fine for about two days straight with the same exact setup.  Actually the ambient air temperature (and hence the air being blown directly onto the blades via box fan) got a few degrees cooler the day of the problem since I shut down some other equipment.  The blades are no longer resetting when connected to the stratum proxy on my other computer, but efficiency remains at ~50-60%.
I'll post a pic as soon as I can but my internet here is currently giving me issues.

I understand this is a user-based help topic on the Internet but there's no need to be rude.

[bclcjunkie]

CX750 is pretty solid and should be more than enough to power 2 blades... your numbers seem to be about what i get on slush too..

Average hashrate in last 10 rounds: 25256 Mhash/s

blades report 13k or so with 99% efficiency, but takes about an hour to reach that level..

when i had this problem i just worked per blade to isolate the issue also i followed dogie's guide to make sure cable thickness was proper... i think your blades are fine, you may want to try swapping cables too to make sure no issue there... my blades draw power from 2 PCI-E ports..

i had a similar issue whenever i switched to "High" mode. At the end it turned out to be insufficient power flow so i got it resolved after changing to 850w Seasonic. Probably more than sufficient for 2 blades but i decided to go for more ampers in case i wanted to get another blade..
Also get a PSU from reputable vendors like Corsair, Seasonic. The first PSU i got was thermaltake 530w and i believe it was of low quality and not pumping enough juice as advertised...

you can confirm power issue with connecting just one blade and hitting "High", if it runs stable without xxx then you may want to upgrade your PSU.

Is there a way to correct it?

Same thing happens with mine.  I find that when I do an update/restart in high clock mode, I get the x on one of the chips.  However, if I switch into low clock mode and then do the update, and then switch to high, all the chips display 0.  Try that?  Otherwise, Friedcat has said this is a voltage issue that is not terribly important if it still hashes at the normal rate and only affects one chip.

Interesting, hadn't thought of that.  I am powering two blades with a corsair CX750.  They both hash just fine when I do the workaround, but I wonder if they will hash faster if I get a separate PSU for each one.  I am getting a reliable 24,500 to 25,000 Mh/s long term average rate on any pool I use.  The blades themselves in the config screen say they are hashing at 13,050 or so, so I don't know if it's worth it to tinker or just let them run.  Every second counts with these difficulty jumps.

[WinTame2012]

I'll post a pic as soon as I can but my internet here is currently giving me issues.

Isn't it connected to the low efficiency of your blades, what do you think?

[mitty]

I'll post a pic as soon as I can but my internet here is currently giving me issues.

Isn't it connected to the low efficiency of your blades, what do you think?

My blades are running in my apartment on a business cable broadband connection, wired via two switches into the router connected to the cable modem.  My work (where I am now) internet connection is sent through a proxy hundreds of miles away and we're forced to use IE8 on top of that, so not surprisingly many websites (including imgur and tinypic) don't work as expected. 

Got the image of my setup uploaded here: http://s3.postimg.org/66bgweyyr/Resized_Image_1370933050829.jpg

[WinTame2012]

I'll post a pic as soon as I can but my internet here is currently giving me issues.

Isn't it connected to the low efficiency of your blades, what do you think?

My blades are running in my apartment on a business cable broadband connection, wired via two switches into the router connected to the cable modem.  My work (where I am now) internet connection is sent through a proxy hundreds of miles away and we're forced to use IE8 on top of that, so not surprisingly many websites (including imgur and tinypic) don't work as expected. 

Got the image of my setup uploaded here: http://s3.postimg.org/66bgweyyr/Resized_Image_1370933050829.jpg

Got it. Just look at your setup and compare it to the one in OP. You definitely need better cooling, even if it's not related to the problems you experience now.

[mitty]

I'll post a pic as soon as I can but my internet here is currently giving me issues.

Isn't it connected to the low efficiency of your blades, what do you think?

My blades are running in my apartment on a business cable broadband connection, wired via two switches into the router connected to the cable modem.  My work (where I am now) internet connection is sent through a proxy hundreds of miles away and we're forced to use IE8 on top of that, so not surprisingly many websites (including imgur and tinypic) don't work as expected. 

Got the image of my setup uploaded here: http://s3.postimg.org/66bgweyyr/Resized_Image_1370933050829.jpg

Got it. Just look at your setup and compare it to the one in OP. You definitely need better cooling, even if it's not related to the problems you experience now.

Yea, I ordered a bunch of 120mm fans only to find out that they were "ultra quiet" and hardly moved any air.  I returned those and ordered new ones but the new ones haven't shown up yet.  I'll try switching a blade to a different power supply and see if that makes a difference.  I just find it really strange that they don't work at all anymore on one stratum server, but work with ~50% efficiency on a different stratum proxy with the exact same settings.  Also the fact that I didn't knowingly make any changes to my setup to cause the issues.

Strangely enough I checked my pool stats page and found my hashrate to be about twice what it was this morning. (this morning the page was reporting ~16 GH/s for 4 non-working blades and about 7.5 GH/s worth of working BFL singles, and about 20 minutes ago it said 30 GH/s total).  However it's now decreasing again.  I'll post anything else I discover...  thanks for your help.

[WinTame2012]

I'll post a pic as soon as I can but my internet here is currently giving me issues.

Isn't it connected to the low efficiency of your blades, what do you think?

My blades are running in my apartment on a business cable broadband connection, wired via two switches into the router connected to the cable modem.  My work (where I am now) internet connection is sent through a proxy hundreds of miles away and we're forced to use IE8 on top of that, so not surprisingly many websites (including imgur and tinypic) don't work as expected. 

Got the image of my setup uploaded here: http://s3.postimg.org/66bgweyyr/Resized_Image_1370933050829.jpg

Got it. Just look at your setup and compare it to the one in OP. You definitely need better cooling, even if it's not related to the problems you experience now.

Yea, I ordered a bunch of 120mm fans only to find out that they were "ultra quiet" and hardly moved any air.  I returned those and ordered new ones but the new ones haven't shown up yet.  I'll try switching a blade to a different power supply and see if that makes a difference.  I just find it really strange that they don't work at all anymore on one stratum server, but work with ~50% efficiency on a different stratum proxy with the exact same settings.  Also the fact that I didn't knowingly make any changes to my setup to cause the issues.

Strangely enough I checked my pool stats page and found my hashrate to be about twice what it was this morning. (this morning the page was reporting ~16 GH/s for 4 non-working blades and about 7.5 GH/s worth of working BFL singles, and about 20 minutes ago it said 30 GH/s total).  However it's now decreasing again.  I'll post anything else I discover...  thanks for your help.

You want use fans with CFM >70 and as high static pressure as you can afford.

Just a final note: check your home network against round-trip latency to the pool(s) you use. Sometimes Internet topology changes and ping from your place goes from 35 ms to 140 ms in a blink of eye.

And make sure you've configured your pool's worker to either use auto-tune for share's difficulty or specify it manually to 16..32 if your pool doesn't support auto-tuning.

Happy mining!

[mitty]

You want use fans with CFM >70 and as high static pressure as you can afford.

Just a final note: check your home network against round-trip latency to the pool(s) you use. Sometimes Internet topology changes and ping from your place goes from 35 ms to 140 ms in a blink of eye.

And make sure you've configured your pool's worker to either use auto-tune for share's difficulty or specify it manually to 16..32 if your pool doesn't support auto-tuning.

Happy mining!

Yea, I didn't bother to pay attention to CFM for my first order of fans but made sure I got 120 CFM fans this time.  I was going to get higher but was afraid the family in the apartment above the basement where my stuff is located would hear/be bothered by the noise, haha.

So I just checked my pool stats again and it looks like everything is running at full speed again.  I have no idea what happened... I'm starting to think it must be some sort of strange network issue since I haven't even been home all day so I'm assuming nothing mining-related has been changed.  

Thanks again for your replies! (and sorry to spam up this topic)

Now I'm just waiting for friedcat to reply back to my email so I can order another. :p

Update: Looks like the problem is back :/
I'm wondering if the ethernet controllers all went bad 

[Cobega]

somebody able to check this here?
https://bitcointalk.org/index.php?topic=262443.new#new

I am not sure if the problem dogie
"Hashrate turns to zero and all "x": Either the power module is broken or the whole blade has some problems. If you bought more than one blades. Try using other functional ones' power module to test if it becomes OK. If so please ask us to send you a new power module. If not, please return it back and we will send you a new one (we cover the shipping fee)."

matches with turning the blade to "clock: high" as I do.
Running my blades in "low" mode chips say: "OOOOOO...." turning them to high makes them all to "XXXXXXXXX..."

Kind regards;

[americandesi]

somebody able to check this here?
https://bitcointalk.org/index.php?topic=262443.new#new

I am not sure if the problem dogie
"Hashrate turns to zero and all "x": Either the power module is broken or the whole blade has some problems. If you bought more than one blades. Try using other functional ones' power module to test if it becomes OK. If so please ask us to send you a new power module. If not, please return it back and we will send you a new one (we cover the shipping fee)."

matches with turning the blade to "clock: high" as I do.
Running my blades in "low" mode chips say: "OOOOOO...." turning them to high makes them all to "XXXXXXXXX..."

Kind regards;

Looks like the chips are not getting enough voltage.
Try increasing the voltage by turning the screws as mentioned in the image below. Please make sure that you are monitoring the voltage using the multimeter.
Keep it not more than 1.2 V

[Cobega]

thanks for the tipp.
I will start doing this tomorrow, but the ethernet controller is definitely broken.
Is it okay, to increase the V of all my 3 blades up to 1.2V per chip?

[americandesi]

thanks for the tipp.
I will start doing this tomorrow, but the ethernet controller is definitely broken.
Is it okay, to increase the V of all my 3 blades up to 1.2V per chip?

Should be. what PSU are you using and are all the 3 blades powered through a single PSU.?

[Cobega]

current config:

BE Quiet 600W Straight Power > one blade + 8 (10) fans -> 10 when the third blade is running
BE Quiet 400W Straight Power > one blade

Turning two blades on one pvo won't work. PVO starts and turns off after 2 seconds.
Any idea how to fix this?-> If the answer is to buy a new PVO which? a 600W or again a 400W?

Kind regards;

[initforthemoney]

I got mine in, and it it got here before some of the parts I'd wanted to use for power got here.  Doh!

So.. I hacked a bit.  Used a PVC pipe rail system, to keep them vertically straight and well separated.

I tried powering 2 off a single "accessory" cable bundle on a PSU, but the wires would get rather warm, uncomfortably so.

However, 1 blade per accessory bundle was no problem.  These, coupled with 2 PCIe (you can hack up a 6 or 8 pin PCIe extender similar to what is shown here) plugs each powering a blade and everything is up and stable - nothing running hot.

Cooling thus far has been a cinch - a box fan keeps them all running stable, and most just warm to the touch.  The two bookending the cluster though do get rather hot, and I have plans to pick up some high-cfm 120mm fans to act as blowers which should work well.

I will be interested to see how the rack they have works cooling wise, it appears to be bottom up cooling which would be difficult to make work inside a cabinet.  If there is an easy way to mount fans on the back though, pretty much everything is better with a few delta 200cfm fans chugging away


Pics

Initial testing - did 2 boards like this, this is the picture of the first successful mining run
https://lh4.googleusercontent.com/-Qfvq6rTkBl4/UZXA_PF2vhI/AAAAAAAAkvk/D-GLp7BJ1jc/s800/IMG_4381.JPG

Good photo of the power boards:
https://lh4.googleusercontent.com/-3CBV3AyUTOo/UZXA56l7KjI/AAAAAAAAkvQ/llqRhOkYSkY/s800/IMG_4377.JPG

Initial try at the temporary rack
https://lh6.googleusercontent.com/--V0N6UaMnF4/UZXBC1HJoQI/AAAAAAAAkv8/a_7QbnqbjRU/s800/IMG_4384.JPG

Completed and blades installed, starting to cable
https://lh5.googleusercontent.com/-6DXJPqfCo-Q/UZXBHY9ytgI/AAAAAAAAkwU/HoH8uv3xfoo/s800/IMG_4387.JPG

Completed and mining away
https://lh6.googleusercontent.com/-30onFChION0/UZXBayGh80I/AAAAAAAAkxo/UeuR89iPwpE/s800/IMG_4397.JPG


phil21@topgun:~$ ./bestatus.pl
10.x.x.40 - Mhash: 13023 efficiency: 099.34% up: 0d,05h,42m,39s
10.x.x.41 - Mhash: 13115 efficiency: 099.99% up: 0d,05h,42m,41s
10.x.x.42 - Mhash: 13058 efficiency: 099.54% up: 0d,05h,42m,41s
10.x.x.43 - Mhash: 13042 efficiency: 099.46% up: 0d,05h,42m,41s
10.x.x.44 - Mhash: 13037 efficiency: 099.41% up: 0d,05h,42m,41s
10.x.x.45 - Mhash: 13138 efficiency: 100.17% up: 0d,05h,42m,25s
10.x.x.46 - Mhash: 13181 efficiency: 100.51% up: 0d,05h,42m,25s
10.x.x.47 - Mhash: 13074 efficiency: 099.75% up: 0d,05h,11m,54s
10.x.x.48 - Mhash: 13079 efficiency: 099.73% up: 0d,05h,11m,49s
10.x.x.49 - Mhash: 13035 efficiency: 099.42% up: 0d,05h,42m,25s
Total: 130782 (13078.2 avg) [Avg Effic: 99.73%]

What inch of piping did you use 1/2 inch? What did you use to cut? Anything you would have done differently?

[zulover]

hi , ive followed your for setting up the blades. we switched in the psu and a green l.e.d. flashed a few times and then went off after that the ethernet cable litup and flashing. next we tried to log in to the config page with the ip address iyou guide but could not reach it kep going to search results, what are doing wrong?

thanks

also recieved mail from the previous owner saying the following ip 192.168.0.111:8000 or 112,113,114 but you have to add :8000 as Port

[mbbc]

I'm currently running two blades over slush's mining proxy. What are your cmd lines? Do you use the -rt (real target) and -nm (no midstate) options?