Topic: Block Erupter Prisma (>=1.4 T/device, 0.75-0.78 W/G, <1 BTC/T, October Shipping) - page 23. (Read 108032 times)

Well done ckolivas, for another time.

Hah, all I did was a code review and merge. I've never seen the hardware in the flesh or virtually.

Another case of beer if you can get them working with p2pool ck.......

Many thanks! Small donation sent have a case of beer on me 

Very Cool! 

Will use your stuff from GitHub for the CubieTruck Build.

In an unexpected but most welcome move, the driver for these was submitted to cgminer. I've reviewed their driver and suggested some changes which they've already resubmitted so the driver is now part of mainline cgminer and will be in the next release due out today.

anyone....im getting bored over in the speculation thread

Ha Yeah, well when people use that phrase they often imply the same outcome as with BFL. I'm sure FC will pull through. Hopefully gen 4 comes soon, as well.

 Nope.  Last time there was a minor delay they gave partial refunds.  They still have exactly two weeks^tm.
I'm just afraid to use that phrase without the tm since it might be owned by BFL and I'm sure they'll be looking for alternate revenue streams now.  Don't misunderstand my humour; I love AM.

Well, at least ASICMiner delievers their gear. Has ASICMiner ever not delivered their gear on time, or a delay of a day or two tops?

Don't worry -


 They'll be shipping by two weeks^tm time

anyone recived tracking number
starting to worry here !

Said yes before, but there turns out many "job not found" errors which causes hash rate loss after long time test.
Now I realize that nicehash has its own modification with stratum protocol, it seems either we patch official cgminer or we fork from nicehash's own branch and add driver for PRISMA.

When using a RPi, I'd recommend to keep a fully configured second RPi ready to go so in case the first one dies it is a simple plugging exercise to get back mining quickly. (Just make a full backup of the SD card once everything is set up and runs as it should).

If you can't spare that extra 35$, at the very least keep a second SD card ready.

Thanks CG,  I'll probably grab an RPI from you and pick up a BBB online for future projects.  The BBB has a seperate Ethernet control chip, for anyone interested--I just finished up researching them.

I don't think an RPi controller is out of the question. It may have limitations, but AM rates the RPi as being able to handle up to 4 Prismas. If you are looking to run more than 4, I might suggest a beaglebone black or 2 RPis.

Thanks for the explanation, @sdjernes. I had no idea about that USB<>Ethernet problem on the RPI and did not think there could be such an issue with the RPI B+.

And yes, so if the AM ethernet controller and a RPI based controller solution are out of question, which low-cost & low-power SoC device capable of running Minera and one or more chains of Prismas would be appropriate?

whainting for updates i-am really interested in prisma!

Does anyone know of a SoC device that is designed in a way that avoids this issue?  A Hummingbird, or BBB?

Cheers.

It is very feesible to put a whole chain on one USB controller adapter to use with a computer (aka RPi). The problem is the RPi has it's Ethernet on a USB port also as part of the internal hub. So what happens is interrupt problems because you are trying to keep your data stream constant on both Ethernet and to the USB connected miners.


Basically it comes down to how well the USB circuits on the board are designed and what else is using it. If you were on a desktop computer you would want to make sure that you put the miners on a separate USB controller port from a hard drive. On the RPi you don't have this option. There is one USB controller port on the processor. This is then used to drive hub chip. Off that hub chip hangs your 2/4 external ports and the RPi's Ethernet controller.

Now if you could re-write the driver for CGminer and build the correct cabling you could possible get more units connected to the UARTs. 


I have mentioned some of the technical reasons above but their is one other reason not to do this.


Single Point of Failure


You don't want to have a single $35 device shutdown your operation. If the miner software crashes in your single RPi setup your entire 11.2Ths is offline until you notice it and then are able to fix it. Splitting that load over several devices allows you to keep most units mining while the one Controller is delt with.