[GUIDE] GridSeed GC3355 5 Chip Setup/power/windows/linux/rpi by UnicornHasher - page 19.

CartmanSPC

legendary

Activity: 1270

Merit: 1000

Quote from: CartmanSPC on April 01, 2014, 06:52:19 PM

Ok, not sure where to post this so for lack of a better place I will put it here for discussion:

KNC is now specifying a MINIMUM of 250 Mh/s for 10K or $40 per 1 Mh/s! Compare that price to performance cost to the Gridseed 5 chip miners (360 Kh/s) and they should be selling for $14.40 each Shocked

Supposedly KNC is shipping Q2/Q3 2014 but by that time they may increase the hashrate yet again (originally specified at 100 Mh/s about two weeks ago?).
If KNC wasn't a reputable company I would call BS.

I find myself hard pressed to order more Gridseeds at $165/piece Huh

Just had a thought...wonder if KNC announced their hash rates to slow down Gridseed sales? Would be interesting to know if it had an affect on the sale of Gridseeds and GPUs. I know it stopped me from buying more.

CartmanSPC

legendary

Activity: 1270

Merit: 1000

Quote from: pallas on April 02, 2014, 07:44:52 AM

Quote from: BotwinBG on April 02, 2014, 07:26:45 AM

Quote from: emoomjean on April 01, 2014, 06:02:36 PM

Quote from: BotwinBG on April 01, 2014, 03:13:03 PM

Quote from: emoomjean on April 01, 2014, 02:01:15 PM

Quote from: BotwinBG on April 01, 2014, 11:59:38 AM

uhm....can someone explain this -

http://imgur.com/GHd4vYv ?

Where was that originally posted?

here

Would you mind posting an updated screencap with a longer run time and details on your config? That's a remarkable hashrate for only 744mhz...

-EMoomjean

There you go , http://imgur.com/eRItNxl .

So how did you achieve that?

And are you using

https://github.com/dtbartle/cgminer-gc3355/commits/master

or

https://github.com/girnyau/cgminer-gc3355/commits/master

?
(BTW, anybody knows which one of the two cgminer repos is better?)

I wouldn't say one is better than the other.

girnyau forked from dtbatle to add some additional freq and a per miner freq option (which is nice).

Since then dtbartle has taken out the hard coded freq and added stepping in increments of 25 and other stuff. Would be nice if he changed it to increments of 5.

Took a quick look at the code but it didn't look obvious to me how to do that.

Code:

		// Support frequency increments of 25.
		pll_f = info->freq / GRIDSEED_F_IN - 1;
		pll_f = MAX(0, MIN(127, pll_f));

If I could figure out stepping in increments of 5 I would fork dtbartle combining the per miner option of girnyau.

jamieb81

sr. member

Activity: 308

Merit: 250

Quote from: suchmoon on April 03, 2014, 12:56:29 PM

Quote from: edonkey on April 03, 2014, 09:06:30 AM

Quote from: suchmoon on April 02, 2014, 10:17:12 PM

Most likely six of your devices disconnected and reconnected, cgminer correctly picked them up but also kept the old zombies. It's harmless, just the hashrate calculation is skewed. As for why this happened - maybe the hub is having issues. If you have a spare hub you can do some testing otherwise don't worry about it.

Thanks for getting back to me. That's a good explanation. I don't have a spare hub at the moment, unfortunately.

I guess what I can do for now is look at the log for disconnect/reconnect events. That would at least confirm what's happening.

Too bad that cgminer and/or Scripta doesn't do a better job of reporting these issues. But as long as it's not trying to give work to zombies, I agree that's it's harmless.

Note that one of my GS units never shows the red LED on. Only the green LED is on. Searching indicates that that means this unit is not receiving 12 volt power.

If this one unit is really not receiving 12 volt power, then it must be powered by the USB bus. The USB hub is a powered Orico P10 U2. But under the hood these Orico's are actually have two hubs with different power characteristics. Perhaps the additional power draw of the bad GS unit is crapping out the sub-hub in the Orico.

I can probably experiment with moving the bad unit's USB cable to the other internal hub in the hopes that it will have less of an impact. Of course the real solution is to replace the unit. I'm working on getting an RMA.

I don't think it can hash at all without 12V. Is it even submitting shares? You should probably just disconnect it until you get a replacement.

It does not work without 12V why would we need PSU's if it would ^^

suchmoon

legendary

Activity: 3654

Merit: 8909

https://bpip.org

Quote from: edonkey on April 03, 2014, 09:06:30 AM

Quote from: suchmoon on April 02, 2014, 10:17:12 PM

Most likely six of your devices disconnected and reconnected, cgminer correctly picked them up but also kept the old zombies. It's harmless, just the hashrate calculation is skewed. As for why this happened - maybe the hub is having issues. If you have a spare hub you can do some testing otherwise don't worry about it.

Thanks for getting back to me. That's a good explanation. I don't have a spare hub at the moment, unfortunately.

I guess what I can do for now is look at the log for disconnect/reconnect events. That would at least confirm what's happening.

Too bad that cgminer and/or Scripta doesn't do a better job of reporting these issues. But as long as it's not trying to give work to zombies, I agree that's it's harmless.

Note that one of my GS units never shows the red LED on. Only the green LED is on. Searching indicates that that means this unit is not receiving 12 volt power.

If this one unit is really not receiving 12 volt power, then it must be powered by the USB bus. The USB hub is a powered Orico P10 U2. But under the hood these Orico's are actually have two hubs with different power characteristics. Perhaps the additional power draw of the bad GS unit is crapping out the sub-hub in the Orico.

I can probably experiment with moving the bad unit's USB cable to the other internal hub in the hopes that it will have less of an impact. Of course the real solution is to replace the unit. I'm working on getting an RMA.

I don't think it can hash at all without 12V. Is it even submitting shares? You should probably just disconnect it until you get a replacement.

suchmoon

legendary

Activity: 3654

Merit: 8909

https://bpip.org

Quote from: dyland on April 03, 2014, 11:05:46 AM

somehow i was given UK plugs with my LA3M order. Huh

I'm in Canada (NA standard plug)--anyone know what my options are for the plugs?

You can use something like this:

http://www.amazon.com/gp/product/B0080SE6H2

Or DIY as jamieb81 suggested.

jamieb81

sr. member

Activity: 308

Merit: 250

Quote from: dyland on April 03, 2014, 11:05:46 AM

somehow i was given UK plugs with my LA3M order. Huh

I'm in Canada (NA standard plug)--anyone know what my options are for the plugs?

cut it and make your own plug on it?

dyland

full member

Activity: 238

Merit: 100

We must become the pitiless censors of ourselves.

somehow i was given UK plugs with my LA3M order. Huh

I'm in Canada (NA standard plug)--anyone know what my options are for the plugs?

edonkey

legendary

Activity: 1150

Merit: 1004

Quote from: suchmoon on April 02, 2014, 10:17:12 PM

Most likely six of your devices disconnected and reconnected, cgminer correctly picked them up but also kept the old zombies. It's harmless, just the hashrate calculation is skewed. As for why this happened - maybe the hub is having issues. If you have a spare hub you can do some testing otherwise don't worry about it.

Thanks for getting back to me. That's a good explanation. I don't have a spare hub at the moment, unfortunately.

I guess what I can do for now is look at the log for disconnect/reconnect events. That would at least confirm what's happening.

Too bad that cgminer and/or Scripta doesn't do a better job of reporting these issues. But as long as it's not trying to give work to zombies, I agree that's it's harmless.

Note that one of my GS units never shows the red LED on. Only the green LED is on. Searching indicates that that means this unit is not receiving 12 volt power.

If this one unit is really not receiving 12 volt power, then it must be powered by the USB bus. The USB hub is a powered Orico P10 U2. But under the hood these Orico's are actually have two hubs with different power characteristics. Perhaps the additional power draw of the bad GS unit is crapping out the sub-hub in the Orico.

I can probably experiment with moving the bad unit's USB cable to the other internal hub in the hopes that it will have less of an impact. Of course the real solution is to replace the unit. I'm working on getting an RMA.

suchmoon

legendary

Activity: 3654

Merit: 8909

https://bpip.org

Quote from: edonkey on April 02, 2014, 06:25:11 PM

I'm on a Raspberry Pi running 10 GS units with the dtbartle version of cgminer that I built from sources. It's been running for about a week.

Today I noticed that Scripta was reporting that I have 16 devices when I only have 10. Further, it's telling me that I have hash power that is impossible.

When I go to the command line and ask cgminer to list the number of units, it comes back with the correct number of 10. However, when I run a little python script I wrote to get status from the cgminer API, it comes back and tells me that I have 16 devices.

I know that I can just bounce cgminer and it should get back to normal. But it would be nice if it would just work.

Again, this is the first time in a week that I've seen this. Although since I've been setting up and tweaking this rig, I don't think I've had cgminer running for more than a couple of days at a stretch. Currently it's been running for about 32 hours.

Any ideas, other than the obvious work around of restarting cgminer?

Most likely six of your devices disconnected and reconnected, cgminer correctly picked them up but also kept the old zombies. It's harmless, just the hashrate calculation is skewed. As for why this happened - maybe the hub is having issues. If you have a spare hub you can do some testing otherwise don't worry about it.

edonkey

legendary

Activity: 1150

Merit: 1004

I'm on a Raspberry Pi running 10 GS units with the dtbartle version of cgminer that I built from sources. It's been running for about a week.

Today I noticed that Scripta was reporting that I have 16 devices when I only have 10. Further, it's telling me that I have hash power that is impossible.

When I go to the command line and ask cgminer to list the number of units, it comes back with the correct number of 10. However, when I run a little python script I wrote to get status from the cgminer API, it comes back and tells me that I have 16 devices.

I know that I can just bounce cgminer and it should get back to normal. But it would be nice if it would just work.

Again, this is the first time in a week that I've seen this. Although since I've been setting up and tweaking this rig, I don't think I've had cgminer running for more than a couple of days at a stretch. Currently it's been running for about 32 hours.

Any ideas, other than the obvious work around of restarting cgminer?

sandor111

hero member

Activity: 616

Merit: 500

The new LightningAsic controller firmware is released, please see: https://bitcointalksearch.org/topic/lightningasic-gridseed-controller-firmware-v3-554542

suchmoon

legendary

Activity: 3654

Merit: 8909

https://bpip.org

Quote from: dyland on April 02, 2014, 01:22:54 PM

Quote from: BotwinBG on April 02, 2014, 01:14:49 PM

Quote from: suchmoon on April 02, 2014, 01:08:18 PM

Quote from: BotwinBG on April 02, 2014, 12:54:50 PM

return GRIDSEED_HASH_SPEED * (double)elapsed_ms * (double)(info->freq * info->chips);

Cheater

hey, it's not like I advertised a new way of speeding up those units

set the chips variable to above 5 by mistake, was a bit puzzled by the results till I started digging.

Hold, on, so what does this mean exactly? That we can get 500+kHash on 744kHz?

No, it's a bug in how cgminer displays hashrate. It seems that the actual hashrate is 5/8ths of that (~300 KH/s)

dyland

full member

Activity: 238

Merit: 100

We must become the pitiless censors of ourselves.

Quote from: BotwinBG on April 02, 2014, 01:14:49 PM

Quote from: suchmoon on April 02, 2014, 01:08:18 PM

Quote from: BotwinBG on April 02, 2014, 12:54:50 PM

return GRIDSEED_HASH_SPEED * (double)elapsed_ms * (double)(info->freq * info->chips);

Cheater

hey, it's not like I advertised a new way of speeding up those units

set the chips variable to above 5 by mistake, was a bit puzzled by the results till I started digging.

Hold, on, so what does this mean exactly? That we can get 500+kHash on 744kHz?

BotwinBG

sr. member

Activity: 588

Merit: 250

Quote from: suchmoon on April 02, 2014, 01:08:18 PM

Quote from: BotwinBG on April 02, 2014, 12:54:50 PM

return GRIDSEED_HASH_SPEED * (double)elapsed_ms * (double)(info->freq * info->chips);

Cheater

hey, it's not like I advertised a new way of speeding up those units

set the chips variable to above 5 by mistake, was a bit puzzled by the results till I started digging.

suchmoon

legendary

Activity: 3654

Merit: 8909

https://bpip.org

Quote from: BotwinBG on April 02, 2014, 12:54:50 PM

return GRIDSEED_HASH_SPEED * (double)elapsed_ms * (double)(info->freq * info->chips);

Cheater

suchmoon

legendary

Activity: 3654

Merit: 8909

https://bpip.org

Quote from: BotwinBG on April 02, 2014, 07:26:45 AM

Quote from: emoomjean on April 01, 2014, 06:02:36 PM

Quote from: BotwinBG on April 01, 2014, 03:13:03 PM

Quote from: emoomjean on April 01, 2014, 02:01:15 PM

Quote from: BotwinBG on April 01, 2014, 11:59:38 AM

uhm....can someone explain this -

http://imgur.com/GHd4vYv ?

Where was that originally posted?

here

Would you mind posting an updated screencap with a longer run time and details on your config? That's a remarkable hashrate for only 744mhz...

-EMoomjean

There you go , http://imgur.com/eRItNxl .

What does the pool show? If my math is correct, based on your accepted shares you are getting between 250-300 KH/s. And one of your cupcakes seems to be dead or maybe just resting. Don't trust cgminer hashrates.

BotwinBG

sr. member

Activity: 588

Merit: 250

Quote from: pallas on April 02, 2014, 07:44:52 AM

Quote from: BotwinBG on April 02, 2014, 07:26:45 AM

Quote from: emoomjean on April 01, 2014, 06:02:36 PM

Quote from: BotwinBG on April 01, 2014, 03:13:03 PM

Quote from: emoomjean on April 01, 2014, 02:01:15 PM

Quote from: BotwinBG on April 01, 2014, 11:59:38 AM

uhm....can someone explain this -

http://imgur.com/GHd4vYv ?

Where was that originally posted?

here

Would you mind posting an updated screencap with a longer run time and details on your config? That's a remarkable hashrate for only 744mhz...

-EMoomjean

There you go , http://imgur.com/eRItNxl .

So how did you achieve that?

And are you using

https://github.com/dtbartle/cgminer-gc3355/commits/master

or

https://github.com/girnyau/cgminer-gc3355/commits/master

?
(BTW, anybody knows which one of the two cgminer repos is better?)

Seems to be a calculation/display error. Try setting chips variable to 8 , like --gridseed-options=freq=744,chips=8,per_chip_stats=1 for instance instead of 5. 8 seems to be hardcoded max for some reason and speed is calculated based on that.

Don't quite speak C , but it looks like in https://github.com/girnyau/cgminer-gc3355/blob/master/driver-gridseed.c
info->chips = MAX(0, MIN(8, info->chips));
and
return GRIDSEED_HASH_SPEED * (double)elapsed_ms * (double)(info->freq * info->chips);
are responsible for that.
Number 8 is present in the gridseed version of cgminer as well. No idea why, I found it by mistake the other day when playing with chips variable.

emoomjean

newbie

Activity: 39

Merit: 0

Quote from: BotwinBG on April 02, 2014, 07:26:45 AM

Quote from: emoomjean on April 01, 2014, 06:02:36 PM

Quote from: BotwinBG on April 01, 2014, 03:13:03 PM

Quote from: emoomjean on April 01, 2014, 02:01:15 PM

Quote from: BotwinBG on April 01, 2014, 11:59:38 AM

uhm....can someone explain this -

http://imgur.com/GHd4vYv ?

Where was that originally posted?

here

Would you mind posting an updated screencap with a longer run time and details on your config? That's a remarkable hashrate for only 744mhz...

-EMoomjean

There you go , http://imgur.com/eRItNxl .

Very cool! I see you're running this in Linux, can you provide any details regarding your config or mods that you've made to the miners? I'm specifically interested in anything out of the ordinary you have done to achieve those results... I'm sure we could probably come up with a community reward for you! I know i'd chip in .5 btc if you could increase my hashrates across the board that significantly!

-Eric

WTParker2

member

Activity: 61

Merit: 10

Texas Proud

How do you get a share of a Titan?

dyland

full member

Activity: 238

Merit: 100

We must become the pitiless censors of ourselves.

Quote from: CartmanSPC on April 01, 2014, 06:52:19 PM

Ok, not sure where to post this so for lack of a better place I will put it here for discussion:

KNC is now specifying a MINIMUM of 250 Mh/s for 10K or $40 per 1 Mh/s! Compare that price to performance cost to the Gridseed 5 chip miners (360 Kh/s) and they should be selling for $14.40 each Shocked

Supposedly KNC is shipping Q2/Q3 2014 but by that time they may increase the hashrate yet again (originally specified at 100 Mh/s about two weeks ago?).
If KNC wasn't a reputable company I would call BS.

I find myself hard pressed to order more Gridseeds at $165/piece Huh

I have a feeling KNC is holding onto something big with respect to the Titan; I expect before they are ready to ship they will drop another doubling of hashrate on us. I am conservative with an August delivery date that being said.

Difficulty rise was a big worry for me, but I believe that the alts will stay alive and a profit-switching pool sort of negates that worry for me.

Currently own 1/10th of a Titan with free hosting full disclosure. If I can get enough BTC from my mining and BTC goes up as well over the next couple months, I will probably order a complete Titan for myself.

Topic: [GUIDE] GridSeed GC3355 5 Chip Setup/power/windows/linux/rpi by UnicornHasher - page 19. (Read 365630 times)