Sounds interesting, but, btw, are you aware that sha256 is not an encryption function? It's a hashing function. Encryption is reversible, hash is not.
|
|
|
It is happening to me right now, user 41599, workers: *_0, *_2, *_tst3.
Update: I restarted everything after 25 minutes. Despite what I told you in PM, actually the malfunctioning workers are able to pull jobs trough already opened connections, but they are unable to open new connections to submit job results, those failed attempts to open new connections are visible with netstat as CLOSE_WAIT. If I restart the application, everything seems to get solved. I would guess there is some issue with routers/firewalls, maybe.
|
|
|
Well, gl.. LM4F120H5QR is very buggy, just read the errata, it has some retarded memory alignment issues.. Almost to trow it away first day when I tried to compile something which was not their example.. PS: In case you are looking for alternatives to ARM, there is PIC32, a cheap MCU based on MIPS. Has some bugs too, not as much as TI lm4f120, and runs at 80Mhz. PSS: I was looking over microchip product page, and they have this new board based on new low cost family (I think running at 40Mhz): http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1406&dDocName=en555947Board will be available in November and will cost $109.
|
|
|
It is possible that this platform will be standard design for BTC hardware wallets in the future. Better, at least, take in account the possibility to add security and plan some reliability for the platform, like batter battery backup, option to backup one wallet to a new one, option to add 3G modem, etc. Look at most used software wallet.
|
|
|
That MCU family does not seem designed for secure applications. There are probably 100 ways to read it, despite CRP/ERM/xRR (whatever) level. You should look for a smartcard if you want some protection.
I on the other hand think that sitck and slush made excellent choice. I only have an issue with the original claim that it is impossible to extract the keys from the device after a theft. Just reduce the claim somewhat and you'll be more than fine. The main problem with using "real smartcard" device lies in the unvieldy development process for the "real crypto stuff". At least use a smart card to store secure key while the device is powered off, not the MCU flash. When device is powered on, require some password (add device unique key) and read the secure part from smartcard (or some other secure device).
|
|
|
Slush, would you kindly ask Mr. stick for additional information to substantiate the above claim?
http://www.nxp.com/documents/application_note/AN10968.pdfChapter 3 (page 4) describes security level of the chip we currently want to use. Do you know about some cheap and quick solution how to skip this protection and read the seed from the device? It is probably possible to read memory with high level laboratory equipment, but purpose of seed protection is that attacker need some time to read memory, so original owner can reload the seed to another device and send his coins out of compromised seed. That MCU family does not seem designed for secure applications. There are probably 100 ways to read it, despite CRP/ERM/xRR (whatever) level. You should look for a smartcard if you want some protection.
|
|
|
Ok, its a defensive mechanism. However, even if the worker was flagged, or whatever, its shares were accepted, not rejected, so I had very little clue about whats happening.
PS: No, I'm not using stratum, I think vardiff is bugged, however, I cant tell you exactly how. I need to add some extra logging module and perform more logging. I think the vardiff algs rounds down somehow, especially when the miner client has itself a variable speed (e.g. FPGA with variable clock). Again, right now it is just some feeling.
|
|
|
Man, I think you have a very nasty bug in workers management. If I use de server (176.9.42.247), and I have one worker working (my client connected to it and mining), and I change its name, then hell breaks loose. The old worker will still be available and new worker name wont be usable anymore (anything I send trough it isnt recorded), even if I change it later, create a new worker with this new name after I freed the name. More, any worker which was somehow in contact with this scheme wont be usable anymore. Check user 41599, worker with extension _h3m5.
|
|
|
I had lot of issues with mine, the filesystem kept getting corrupted when I overloaded USB ports. After some investigation I found that USB hub design its crap, you need to bridge polyfuses and add a big, low ESR, electrolytic capacitor, in order to avoid power spikes caused by USB devices. Like this: http://theiopage.blogspot.nl/Also try to add a small ceramic cap (100nF) on SD card power supply pins.
|
|
|
He asked for die size, the silicon inside, not for package size.
|
|
|
Guys, your are definitely crazy, and I do love seeing people which actually read the Book..
|
|
|
As usually I'm in, and guess what, I bring rum for the rabbit.
|
|
|
Barrel connectors are pain in the arse, especially socket ones. You should wait Tom to give you exact specs for what is he using before buying the power converter.
|
|
|
zomg, I also demand pictures with cablepair naked!!!!!!1111
|
|
|
OpenCL is purely graphics.
OpenCL on Altera FPGAs is in beta.
|
|
|
Package size. Do I get half a bounty?
Just out of curiosity, suppose we get our boards and you guys upgrade your chips a year later. Would it be possible to just do chip swaps? Pop off the old one and insert the new, smaller, faster one? If this were intended, the chips would be socketed. Yeah, and with a nice slab of copper in the middle of the socket I'm still trying to imagine how to cool the bottom side of these chips. If there are no traces directly under the ASIC chip, I'd like to cut a square hole under each chip and get at least some copper down there... The hole size would be limited by pressure on the ASIC being applied from the HSF on the top side. The bottom heatsink could simply be a small square about the height of the board's depth above a larger square that adheres flush with board, and with some fins beneath. It'd be like a 2 level pyramid with fins on bottom. This is just my best guess for how I'd modify things on the board if I could magically change the existing design. Copper island and vias can transfer as much as 30% of heat from the chip on the other side of the PCB. Just add heatsinks on the other side.
|
|
|
Interesant. Deci daca am inteles bine rigurile raman la Inaba si le accesezi de aici?
Pai asta face si cu FPGA-urile vandute de BFL, ca sa nu le mai trimita inainte si inapoi.
|
|
|
EMC e a lui Inaba/BFL_Josh. Am stat mult pe pool-ul ala, dar am renuntat pentru ca devenise instabil. Mai mult, pierdeam din hashing (si aveam deja vreo 30GHs) cand intrau useri cu mai mult de 100 GHs.
|
|
|
daca, chiar o sa iasa..
|
|
|
|