Topic: Avalon ASIC users thread - page 147. (Read 438724 times)

I got eloipool working now using his guide. 

The -blocknotify bit was tricky as I happened to be running the daemon on a Windows VM, and eliopool on Linux, but my old friend xinetd helped to bridge the gap.

what's the stratum pool address for ozcoin from the US?  stratum.ozco.in:3333 isn't working for me.

I have my restart timer set at 60 seconds from power off.

When my batch 1 unit gets into that state where all temps and fanspeeds are zero (and indeed the fans are stopped), I have to turn off the power supply switch for 20 seconds or so.  When I power it back on, it's happy again.

Just soft-rebooting the wr703n is not enough.

@elasticband: If you can still flash firmware, maybe it's a good try to flash 0225 firmware, I had been running that firmware for months without restart of the machine, it has minimum overhead and the web interface is very responsive

And you should check the power and data cable on each module, sometimes they get loose

1. About cables - connecting more cables is always better idea to spread the load. As minimum you need three and they are enough
2. About brick - Never do two things at once - a jeneral rule one step forward so you can get back I am very sure that it was loose USb. When avaon is powered up the wr703 boots after 5 secs or so. Menwhile green led is flashing. when 703 finds USB (or cgminer starts) led goes off. So it was usb cable or cgminer not starting or having bad cgminer options left when you downgrade/upgrade and so on
as you see avalon is not there not found

ssh to unit and post dmesg output pls

if you unchecked the "keep settings" box, your device network settings will be back to default. try to connect on 192.168.0.100 and see if it responds

Hi invader, and loshia, thanks for your insight. I just wanted to verify that bolded part above.. Do you mean that I should connect to all 3 PCIe connectors on the power distribution board?

My batch 2 miner came with 850W PSU and was only connected to 2 of the connectors on the PDB. I have since upgraded the PSU and I have an extra PCIe connector that I could connect to that board.

Should I do this? Have you tried it? Does it make any difference either way?

Thanks.

USB!
Cable is unpluged or loose

And connect all cables as advised - mandatory

Better to connect all the cables and use separate for each pci-e connector. This way the load to each one will be less - you're going to draw 600+ W from PSU, do i need to tell you more?

I also replaced PSU to AX860 in my batch 1 unit.

I also unplugged "blue shitty fan" due to not doing it's purpose when big fans (even standart) going to high rpm, and inserted a piece of paper to not let the air go back. Also i used power directly from PSU, as new fans can draw 4A@12v each at full speed, so only PWM and SENSE pins are connected to controller. Im a bit out of luck to have new powerful fans with inverted PWM control, but it's a software issue in general so im planning to edit the sources and build cgminer with additional option for that case for testing. In general, changes would be minimal - edit cgminer.c, add an new option and edit driver-avalon.c functions temp_rise(),temp_drop() avalon_init_task() function by adding something like xor(0xA0,fan) - i'll explain : i noticed that at 0xA0 which is 100% fan is really stops, so it must be xor(0xA0,0xA0) = 0x00 to run inverted PWM fan at 100%. Then, if we want fan at 20% we should now use xor(0xA0,0x20) = 0x80. The final goal is to be able to run inverted PWM fans as long as usual ones, automatic detection based on rpm speed feedback would be also nice, but later. Only problem that i can't build new image myself with modified cgminer permanently as ckolivas put only image without sources, so i must backport and merge his changes with public avalon image sources. But for testing i just put my modified cgminer binary in /tmp and see whatsup.

I also did this, well.. not just moved the one module closest to the PSU, I moved them all to the next module (same as invader)

had a batch one witch has been struggling to get above 66ghs during the day and it goes higher at night..... the two things i had noticed it had a different style PSU, could tell by the back switch and the PSU area got hotter than my others.

Opened it up today to find a 650w bronze PSU..... should run better now with a 860w i hope.


Has anyone thought about moving the module closest to the PSU to the space for the fourth module?

Mine was stable for 2.5 days @350 MHz then cgminer restarted, but also without problems.

I think the original fans @ around 800 rpm makes no sense temps are rising imediatly when running so low.
50% min fanspeed would be fine
or I'lo try to play around with the temp target to get them not so slow

Cgminer restarted after 3.5 days at 325Mhz, quite good stability 

Hi all!
Changed today standart avalon fans to TFC1212DE, and noticed some strange behavior.
To be short, when temperature is going over threshold ( --avalon-temp ) fans is starting spinning less (instead of more) until fully stops! Looks like inverse PWM here. Using latest 20130627 firmware.
Maybe there would be also useful to add an options to cgminer to enable inverse pwm control and fixed target rpm speed.

EDIT:

If i use --avalon-temp 0, then fans stops right after cgminer start, applied a "workaround"
with options --avalon-temp 60 --avalon-cutoff 55, fans keep stable at 3840 rpm now.
Looking now at sources, to make inverse pwm a little changes is needed in temp_rise(), temp_drop()
functions, and a new option like --avalon-inverse-pwm

i bought it from bitmit

What was the transaction number for your payment?

Yes however I rely on them to baseline my bitcoin hardware performance

bitparking.com
actually zero fees on BTC and NMC + DVC + IXC merged mining

very simple interface but it works;)