Cairnsmore1 - Quad XC6SLX150 Board - page 23.

ebereon

sr. member

Activity: 397

Merit: 500

Quote from: kano on September 08, 2012, 10:03:32 AM

Does that include WU: to handle >1 difficulty shares?

If not, then you need to update it very soon, pools are already starting to have > 1 difficulty shares (EMC) and I know Ozcoin is planning to do it soon.

Cgminer also has WU: which is the number of 1difficulty shares worked on per minute (which is greater than or equal to U:)
So cgminer shows 2 numbers, U: which is the number of share a minute accepted by the pool and WU: which is the number of 1 difficulty shares a minute worked on by the miner

My file is only to display the stats, counting is done by MPBM and I think TheSeven did already counting in the share difficulty. The screenshot I posted is running with EMC diff10 server and it's counting it correctly.

eb

kano

legendary

Activity: 4634

Merit: 1851

Linux since 1997 RedHat 4

Quote from: ebereon on September 08, 2012, 08:49:48 AM

Just a reminder, you can see U/m also with MPBM like you know it from cgminer.

Download the file statsgadget.js from here and put it in the MPBM directory "...\mpbm\modules\theseven\webui\wwwroot\static\statsgadget". You don't need to restart MPBM, just reload the WebUI in your browser.

Like this:
http://content.wuala.com/contents/ebereon/Shared/bitcoin/mpbm_stats.JPG

eb

Does that include WU: to handle >1 difficulty shares?

If not, then you need to update it very soon, pools are already starting to have > 1 difficulty shares (EMC) and I know Ozcoin is planning to do it soon.

Cgminer also has WU: which is the number of 1difficulty shares worked on per minute (which is greater than or equal to U:)
So cgminer shows 2 numbers, U: which is the number of share a minute accepted by the pool and WU: which is the number of 1 difficulty shares a minute worked on by the miner

ebereon

sr. member

Activity: 397

Merit: 500

Just a reminder, you can see U/m also with MPBM like you know it from cgminer.

Download the file statsgadget.js from here and put it in the MPBM directory "...\mpbm\modules\theseven\webui\wwwroot\static\statsgadget". You don't need to restart MPBM, just reload the WebUI in your browser.

Like this:

eb

daemonic

newbie

Activity: 49

Merit: 0

I would just like to again thank everyone for their hard work on everything getting us to this point. After following ebereon's little guide ive tuned my cairnsmore1's to run at a nice level;

This is powered directly from an old ATX psu (200W Max on 12V @ 8A (which should surely be 12V*8A=96W??))

However i have a small issue where I occasionally lose all 4 com ports (Win 7 64bit directly connected to pc) at random intervals. I have to disconnect the usb and power cycle the cairnsmore1's then reconnect the usb to get them back. The idea is to move these cairnsmore1 to my datacentre, to run off of one of my linux machines, so this is not an ideal situation if it continues to happen.

Ive changed the usb cable which seems to have helped somewhat (as you can see 20 hours from the image above), and ive ordered one of those recommended 35 port usb hubs to help also, is there anything else i can do? (I guess worst case i can remove the 5v from the usb cable and hook the atx up to the remote power bar).

hm

member

Activity: 107

Merit: 10

thank you yohan, makomk and ebereon for the info. Apparently, the physics classes I took in school 15 years ago were not too interesting for me.

The four boards I am running seem to work OK .
Concerning the starting problems of board #0017, I didn't test it thoroughly, and it is possible that the board just needs some time, after power-up and start of mining software, to start hashing correctly. The last time, it did'nt hash right away but had many invalids for the first four hours, then the invalids decreased considerably.
There still are periods when board #0017 produces more invalids, but not as much and long as in the first four hours. After a quick search through the logfile I can say that most of the time, it runs OK and doesn't produce more invalids then the 0600+ boards, leading to an average of <5% invalids.

yohan

sr. member

Activity: 462

Merit: 251

Quote from: hm on September 07, 2012, 04:54:48 PM

Good question. I use an old 230W ATX psu which provides 102W on 12V and 115W on 5V and 3.3V combined.
Could the 5V and 3.3V lines be combined to 8.3V and used as a power source to tap into the unused 115W?

You can't combine the 3.3V and 5V as they usually have a common ground/0V. You will simply short a rail and expect burning wires and flames.

Also be careful of power in individual sections and combining those as a capability. Often there is an overall limit as well as section limits and the overall is usually less than individual sections combined.

102W of the 12V is just about enough to run 2 boards but may struggle on the highest power using bitstreams. Probably ok for 190/200 bitstream but probably not 210/220 ones. The Cairnsmore1 has a normal maximum of about 60W but might go to 70W if we had bitstreams that needed and took 12A.

yohan

sr. member

Activity: 462

Merit: 251

Quote from: yxt on September 07, 2012, 02:48:48 PM

so <14V is safe?

How much power is needed for one board?

As long as the supply is actually 14V or under you should be ok. Thew current at 14V will be a little less than at 12V but you will have slightly higher inefficiency in the linear regulator supplying the Controller.

makomk

hero member

Activity: 686

Merit: 564

Quote from: hm on September 07, 2012, 04:54:48 PM

Good question. I use an old 230W ATX psu which provides 102W on 12V and 115W on 5V and 3.3V combined.
Could the 5V and 3.3V lines be combined to 8.3V and used as a power source to tap into the unused 115W?

No, and you'll almost certainly blow up your power supply and/or start a fire if you try.

hm

member

Activity: 107

Merit: 10

Good question. I use an old 230W ATX psu which provides 102W on 12V and 115W on 5V and 3.3V combined.
Could the 5V and 3.3V lines be combined to 8.3V and used as a power source to tap into the unused 115W?

yxt

legendary

Activity: 3528

Merit: 1116

so <14V is safe?

How much power is needed for one board?

yohan

sr. member

Activity: 462

Merit: 251

Quote from: hm on September 06, 2012, 03:28:27 PM

What is the maximum input voltage? Can I use a 19V notebook psu that provides 3.95A?

I think the weakest component in the 12V input is rated 16V so that is the absolute technical maximum. However watch with power supplies for the tolerance range. 5-10% is not unusual so 15V PSU might be 16.5V and over the limit. There are plenty of 12V 5A supplies around for LCD monitors if you want a brick style at low cost. For efficiency an ATX supply is a good choice for even a smallish rig like yours. Even without our PDB to do the turn on they are fairly simple to force on even if you don't have a hosting motherboard already to power and provide the turn on switch.

ebereon

sr. member

Activity: 397

Merit: 500

Quote from: hm on September 06, 2012, 03:28:27 PM

Can I ignore the invalids, them being below 1%?
What is the maximum input voltage? Can I use a 19V notebook psu that provides 3.95A?
Which bitstream is pre-flashed on boards #0600+? Judging by the stats it has between 180 to 190 MH/s per fpga.

You can ignore that, you will have invalids with makomk bitstreams faster then 177Mh/s, as they are "overclocked". I would go for the faster dcmwd4e and test them. As long as you are below 3% invalids you will benefit from it. Also the ambient temperature plays in. The same bitstreams on my boards have <3% invalids when ambient temps are below 30°C and go up to 6% invalids when ambient temperatur is >32°C. My temperature measure is 10cm away from my 50 boards.

The maximum voltage is 12 Volt discribed here -> http://www.enterpoint.co.uk/cairnsmore/cairnsmore1.html. But yohan will tell you.

My boards (>SN#550) were pre-flashed with dcmwd4c bitstreams(180,190,200) and this one still have dcm problems. When you flash the new dcmwd4e bitstreams, you will see faster results and more stable.

eb

hm

member

Activity: 107

Merit: 10

board #0017 did run very good with 190@p0+1 and 180@p2+3:

I turned it off when I installed two newly delivered boards. I haven't turned it on again, I'm one psu short ... When I've found a psu and turn the board on again, I hope I won't have to power it up ten times again as last run.

The new boards #0620 & #0621 are running just fine:

Can I ignore the invalids, them being below 1%?
What is the maximum input voltage? Can I use a 19V notebook psu that provides 3.95A?
Which bitstream is pre-flashed on boards #0600+? Judging by the stats it has between 180 to 190 MH/s per fpga.

LazyOtto

sr. member

Activity: 476

Merit: 250

ty, yohan.

I'll do an initial test to see if the master/slave works. Should be pretty straight-forward to see that.

Expecting the boards today. DHL shows "Out for delivery".

yohan

sr. member

Activity: 462

Merit: 251

Quote from: LazyOtto on September 06, 2012, 09:35:55 AM

But this is still a situation where you only know which version is currently installed by flashing, or attempting to flash, it.

There is no other way? Such as via some query? Or some other response string/artifact?

No direct simple way to tell at the moment but one difference to tell between 1.3 and 1.5 is if master/slave works. It's not in 1.3 but is in 1.5. I think on your boards there is a better chance you have 1.5 but otherwise it is 1.3.

kano

legendary

Activity: 4634

Merit: 1851

Linux since 1997 RedHat 4

Quote from: LazyOtto on September 05, 2012, 08:06:54 AM

It is known that the BFL Single units don't do well with p2pool.

Do the CM1's have any issues with it?

No, the Icarus code (which also runs the CM1) in cgminer handles LP's well.

In one respect it is better that GPU's, since you can abort work at any time you wish, but since it doesn't interrupt as soon as the LP arrives, rather some time in the 0.1s after it, that turns into a (very) small loss compared to GPU mining.

So basically on an LP with Icarus/CM1 you waste 1/2 of 0.1s i.e. 0.05s ... on a 10s LP that's 0.5% ... quite small.

The problem with BFL is that number on p2pool is something like ~25% ... quite large

steamboat

hero member

Activity: 648

Merit: 500

Quote from: LazyOtto on September 06, 2012, 09:35:55 AM

But this is still a situation where you only know which version is currently installed by flashing, or attempting to flash, it.

There is no other way? Such as via some query? Or some other response string/artifact?

according to yohan there isn't a way to tell currently. posted ~august 6 i believe in this thread.

edit: https://bitcointalksearch.org/topic/m.1076651

LazyOtto

sr. member

Activity: 476

Merit: 250

But this is still a situation where you only know which version is currently installed by flashing, or attempting to flash, it.

There is no other way? Such as via some query? Or some other response string/artifact?

ebereon

sr. member

Activity: 397

Merit: 500

Quote from: LazyOtto on September 06, 2012, 09:18:14 AM

Is there a way to determine which version the controller is currently flashed with?

1.3:
Only flashable with SWITCH6 off

1.5:
Only flashable with SWITCH3 off AND SWITCH6 off

Glasswalker:
You should see it on the different LED's

eb

LazyOtto

sr. member

Activity: 476

Merit: 250

Is there a way to determine which version the controller is currently flashed with?

Topic: Cairnsmore1 - Quad XC6SLX150 Board - page 23. (Read 286370 times)