Topic: [ANN][BURST] Burst | Efficient HDD Mining | New 1.2.3 Fork block 92000 - page 489. (Read 2170648 times)

Not as yet.

In regards to a minimum program size understand that things like "standard I/O" don't apply to AT so basically you would need to look at something closer to cross compiling I think (just like people creating programs for CPUs that don't have an OS - and note that at this stage there is not even a "heap").

Over time the plan is to extend the AT API to include more and more functionality to make it easier for standard systems such as GCC to work with it.

That would be awesome.  Is anyone working  on that?  I was looking at the feasibility of an LLVM backend today, but decided it was probably too awkward because it's claimed here that the binary it produces for a minimal program is 10KB, and the current maximum code size for a burst AT (if I'm reading correctly from MAX_MACHINE_CODE_PAGES in AT_Constants.java) is 2.5K.  I am pretty clueless about compiler technology, though, would love to be wrong.  Does the machine code size depend on the backend you use?

Stay away. Goldman Sachs is involved so that means co-opted by Banksters.

http://www.youtube.com/watch?v=q5FDvzj8YX4

True - but they are also trying to convince everyone to use their own browser (something we are not planning to do).

Also it should be noted that if we get GCC to do AT then you'll have every fancy tool you could ever want at your disposal.

The competition sets a very high bar.

tnx mate

wallet.burstcoin.io

that will do the trick

Glad to know - I will make a note to update all the test cases (as many of the op codes are going to have the same issues). Unfortunately when the test cases were first created the precise syntax had not been worked out (am happy to see that burstdev has been very accurate to what had been intended). This was mostly an issue of the "list" output from the original C++ prototype and the Java version of that (rather than an issue with the specification itself).

The use of @ indicates a variable to be "assigned" whereas the use of $ indicates a variable to be "read" (of course @ could mean read and write but $ should never indicate write and normally the first variable will be the one to be assigned if there are two).

Creating a simple and consistent syntax for a "virtual CPU" is actually not so easy (as I discovered).

I will say that you are in good hands with the Burst developer as he has even corrected myself over inconsistencies with AT.

If you are keen to participate more with AT then feel free to make more direct contact with those involved to become "part of the team" (there is no intention to make ATs "too hard" but we are at the very start of creating tools to make that easier so I hope people can be tolerant while we work on making it more accessible).

Apart from using a "testnet" you could use the C++ prototype, however, it can only emulate API functions with appropriately provided test data (and doesn't implement the A and B 256 bit registers at this stage).

I agree that it would make things easier to have a better "emulator" for testing (but am unfortunately rather too busy at the moment to put a lot more effort into the C++ prototype).

Perhaps this is something that Burst could consider creating (having an Assembler tool is a great start as writing pure machine code is very time consuming).

Is there an easy way to run AT code independently of the blockchain?

Thanks, that worked and produced matching output.

callmejack thank you for your advice but i cannot find the url now...

found it must be here i think http://www.shabal.com/?p=213#more-213

Is this ok ?

THE REASON IS YOU NEED INTERNET. OFFLINE=NO_INTERNET=NO_P2PCONNECTION=NO_SYNC

any working web wallet?
my offline wallet not syncing

replace the shabal.s file with the one for the arm platform.
the url for the shabal.s file is mentioned in the source.
worked for me fine.

Hello All.

Does anyone have non java miner for ARM cpu? i tried to compile dcct's c miner for Linux but no success...

Thanx!

Try my little tool (AutoIt script): https://yadi.sk/d/jlbEbj73djZRK



How it works:
1) Open Blago-miner`s "stat-log.csv"
2) Read best deadlines from every line
3) Compare it with blockchain data (Burst daemon API - "http://localhost:8125/burst?requestType=getBlock&height="&$blockHeight)
4) Write and execute "results.txt"

Now there`s two types of behavior:
1) write only if your best deadline is less then blockchain data
2) write all (append blockchain data to every line for further analysis)

Sample log and source included.

P.S. you can close working script by right-clicking tray icon.

Although I haven't played with that assembler tool myself perhaps try this instead:


(I've just noticed that the test case listings are different to when I use the current C++ prototype)

Oh, it's the way I'm invoking it.  You have to compile the file first, or it does something inscrutable with stdout.  Still not sure it's behaving correctly, though.