Author

Topic: [ANN][BLC] Blakecoin Blake-256 for GPU/FPGA With Merged Mined Pools Stable Net - page 167. (Read 409641 times)

legendary
Activity: 1509
Merit: 1030
Solutions Architect
I have been working on a block chain explorer. It is a fork of Bitcoin Abe. Only minor adaptations were required for Blakecoin.

The source is available here: https://github.com/knorrly/blakecoin-abe

I am running a demo site: http://blakexplorer.dyndns.org:2750/chain/Blakecoin

The demo site will only be temporary and is not able to handle a big workload. Would someone be willing to host it officially? blakecoin.org would be a good candidate Wink .



I will put it up on a server no problem thanks for working on it Smiley
newbie
Activity: 6
Merit: 0
I have been working on a block chain explorer. It is a fork of Bitcoin Abe. Only minor adaptations were required for Blakecoin.

The source is available here: https://github.com/knorrly/blakecoin-abe

I am running a demo site: http://blakexplorer.dyndns.org:2750/chain/Blakecoin

The demo site will only be temporary and is not able to handle a big workload. Would someone be willing to host it officially? blakecoin.org would be a good candidate Wink .

sr. member
Activity: 384
Merit: 250
Well now that I've been mineing for several days. It's settled down to ~350MH/s per core on the CM 1. I' 'm pretty happy with that.:-D

I'll try to slot some CM1's into the build queue so as to get you a slightly faster hash rate. I've been concentrating on the ztex recently as there has been rather more interest (squeaky wheel gets the grease).

To whit: https://www.dropbox.com/s/1ffqdaj1dowkd0j/ztex_ufm1_15y1-v06ad-t6-ucf-150MHz-fmax-157.bit is marginally faster by about 4MHz to 8MHz (not the 10MHz or more suggested by the fmax increase).

Its going to get progressively more difficult to improve on this, unless I can get that third core to route which has been a dismal failure to date, so I think I'm going to take a timeout and do a litecoin build for the ztex as interest is hotting up over at the litecoin forum in the commercial offerings, and I'd like to get my ztex alternative out there (only perhaps 66kHash/s or thereabouts, so not great, though I can't be sure until I've built it).
legendary
Activity: 1470
Merit: 1001
Use Coinbase Account almosanywhere with Shift card
Well now that I've been mineing for several days. It's settled down to ~350MH/s per core on the CM 1. I' 'm pretty happy with that.:-D
sr. member
Activity: 384
Merit: 250
The clock resolution is 4MHz (rounds down) and the valid range is 100MHz to 250MHz.

No! without knowing anything about MHz, Ztex Boards, Mining... that's pure math!  Grin

Yup, you caught me out good and proper. Though it still is the valid range for input; the 250 just gets rounded down to 248MHz when setting the clock speed (notwithstanding a -1 that also gets applied, so the speeds may actually be off by 4MHz across the board, but that's by the by).

Have you tried it out? Do you like it?  Grin
sr. member
Activity: 266
Merit: 250
The clock resolution is 4MHz (rounds down) and the valid range is 100MHz to 250MHz.

No! without knowing anything about MHz, Ztex Boards, Mining... that's pure math!  Grin
sr. member
Activity: 384
Merit: 250
I have updated the blake fpga cgminer https://github.com/kramble/FPGA-Blakecoin-Miner/tree/master/cgminer/cgminer-3.1.1

This removes some of the debug output which has caused such confusion (overflow and checknonce messages) and added a new command line option to set the ZTEX clock speed manually. Its probably best to let it run in automatic (default) mode at first to get a feel for which settings may be required should you decide to use it.

The new windows binary is at https://www.dropbox.com/s/6ydm22w7rclzhj2/cgminer.exe (to upgrade just copy it over the old version in cgminer-3.1.1-blakefpga).

When running on linux you will probably need to run as sudo, and don't forget to update bitstreams/ztex_ufm1_15y1.bit with the blake version from https://www.dropbox.com/s/vk3k5sb64b8641o/ztex_ufm1_15y1-v06ad-2core-ucf-140MHz-fmax-147-fixed.bit or you'll only get a bunch of HW errors instead of shares (yep, I made that mistake too  Embarrassed )

The updated README is as follows ...
Quote
Blakeminer patches for https://github.com/ckolivas/cgminer version 3.1.1

I don't think its appropriate to fork the repository in this case so I will
just supply patches to specific release versions.

Compilation
Open the official repository (link above) in your browser, click the "branches"
dropdown menu in the middle left of the page, select the "tags" tab and scroll down
to the version required (3.1.1) and select it. Download the zip (right side of page)
and unzip the archive.

Copy my patch files (from this github folder) into the official cgminer 3.1.1 folder,
replacing existing files as necessary. Build as normal, see the README or windows-build.txt

Prebuit windows binary at https://www.dropbox.com/s/6ydm22w7rclzhj2/cgminer.exe
Dependancies (DLL) at https://www.dropbox.com/s/xa01f9hhakpsexv/cgminer-3.1.1-blakefpga.zip

To use the windows version, unzip the dependancies then move cgminer.exe into the folder.
Copy the bitstream folder from the official cgminer-3.1.1 distribution.
Replace ztex_ufm1_15y1.bit with the BlakeCoin bitstream.
Edit the RUNBLAKE.BAT script and set the username/password to be the same as the rpcuser
and rpcpassword set in your blakecoin.conf. Change localhost if running blakecoind on a
separate machine. You will need to install the WinUSB driver using zadig which is available
at http://sourceforge.net/projects/libwdi/files/zadig/

For linux users, run as follows (this is the same as RUNBLAKE.BAT)
sudo ./cgminer --disable-gpu --url localhost:8772 --userpass username:password 2>log.txt
You can probably change the UDEV rules to avoid the need for sudo.

Notes:
Should you have problems, redirect the stderr output to log.txt (this is done
automatically in RUNBLAKE.BAT) and examine this for messages.
Do not use this for GPU mining as it will not work.
Do not enable the --debug switch as it crashes, --verbose works OK though.
Cainsmore CM1 will be detected as icarus, you may need the -T switch if it crashes.
Use the -S option eg.  -S \\.\COM20 -S \\.\COM21 -S \\.\COM22 -S \\.\COM23
It will not work for my current lancelot bitstream, use the python miner instead.
Only the ztex 1.15y board is supported. Frequency management is automatic using the
same algorithm as bitcoin. It can be overriden by the --ztex-clock option as follows
--ztex-clock 180:200   sets initial clock of 180MHz, max of 200Mhz
--ztex-clock 204:204   fixed clock speed of 204MHz
--ztex-clock 180:192,184:196,180:204,192:212   set individual fpga device speeds
I don't know if this will work for multiple boards, but its done the same way as
the icarus options so with luck it will be OK.
The clock resolution is 4MHz (rounds down) and the valid range is 100MHz to 250MHz.
If --ztex-clock is not used the default range is 172MHz to 220MHz.
legendary
Activity: 1509
Merit: 1030
Solutions Architect

this is due to not enough cpu for wallet and mining on the same machine and it results in losing connections, this happens more with cgminer than reaper or kramble's python miner

my advice is to just mine on the weak cpu machine and host the wallet on another machine and just mine to it over your local network and blakecoind is better for using less resources than blakecoin-qt so you could just try using blakecoind and see if that get same results if it is not possible to host the wallet on another machine

i do not cpu mine, only qt and cgminer on a fpga is running, the cpu (Celeron dualcore 1.2GHz) is idle at 5-7%

any cgminer e.g fpga or opencl seems to stress the wallet if it is running on same machine

have you tried to use blakecoind and does that get same result?

Sorry i am a n00b by mining without Pool, Who do i setup this?
i have a server there i can install the QT client in a VM maschine.. maybe this works better....  :/
config -> listening on= 0.0.0.0 or 255.255.255.255 ?

EDIT: right now the diff goes over 2K  Tongue

do you have the wallet on more than 1 machine?

if you have a stable wallet on other machine then just edit blakecoin.conf and add something like this: rpcallowip=192.168.0.*  (this allows connections from your lan) replace ip with your lan ip range, you can also just add a single ip like this: rpcallowip=192.168.0.3

then you mining rig does not need to run the wallet  Cheesy

just change the mining command line to point to the ip of the machine with the wallet running e.g cgminer --disable-gpu --url 192.168.0.4:8772 --userpass yourusername:yourpassword

or try using the blakecoind via the command line

here is the list of commands that work with Blakecoin and Bitcoin via the command line
https://en.bitcoin.it/wiki/Original_Bitcoin_client/API_calls_list

on windows in a command window with the directory of the wallet: c:\blakecoin>   I can run: blakecoind getinfo
on linux in a bash window with the directory of the wallet: /media/blakecoin> I can run: sudo ./blakecoind getinfo

both blakecoin-qt and blakecoind use same blakecoin.conf and data but you can only run one at a time

like I said the QT client is slower than blakecoind and all the server nodes just use blakecoind on ubuntu 12.04 x64 and they run for weeks without restart  Cool

edit:
also if you wanted to allow more than 10 connections add this to your blakecoin.conf: maxconnections=50

edit2:
example conf for lan mining

listen=1
maxconnections=50
gen=0
rpcallowip=127.0.0.1
rpcallowip=192.168.0.*
rpcuser=yourusername
rpcpassword=yourpassword
rpcport=8772
server=1
daemon=1
addnode=162.243.133.80
addnode=162.243.14.130
addnode=146.185.135.24
addnode=184.171.247.23
member
Activity: 93
Merit: 10

this is due to not enough cpu for wallet and mining on the same machine and it results in losing connections, this happens more with cgminer than reaper or kramble's python miner

my advice is to just mine on the weak cpu machine and host the wallet on another machine and just mine to it over your local network and blakecoind is better for using less resources than blakecoin-qt so you could just try using blakecoind and see if that get same results if it is not possible to host the wallet on another machine

i do not cpu mine, only qt and cgminer on a fpga is running, the cpu (Celeron dualcore 1.2GHz) is idle at 5-7%

any cgminer e.g fpga or opencl seems to stress the wallet if it is running on same machine

have you tried to use blakecoind and does that get same result?

Sorry i am a n00b by mining without Pool, Who do i setup this?
i have a server there i can install the QT client in a VM maschine.. maybe this works better....  :/
config -> listening on= 0.0.0.0 or 255.255.255.255 ?

EDIT: right now the diff goes over 2K  Tongue
legendary
Activity: 1509
Merit: 1030
Solutions Architect

this is due to not enough cpu for wallet and mining on the same machine and it results in losing connections, this happens more with cgminer than reaper or kramble's python miner

my advice is to just mine on the weak cpu machine and host the wallet on another machine and just mine to it over your local network and blakecoind is better for using less resources than blakecoin-qt so you could just try using blakecoind and see if that get same results if it is not possible to host the wallet on another machine

i do not cpu mine, only qt and cgminer on a fpga is running, the cpu (Celeron dualcore 1.2GHz) is idle at 5-7%

any cgminer e.g fpga or opencl seems to stress the wallet if it is running on same machine

have you tried to use blakecoind and does that get same result?
member
Activity: 93
Merit: 10

this is due to not enough cpu for wallet and mining on the same machine and it results in losing connections, this happens more with cgminer than reaper or kramble's python miner

my advice is to just mine on the weak cpu machine and host the wallet on another machine and just mine to it over your local network and blakecoind is better for using less resources than blakecoin-qt so you could just try using blakecoind and see if that get same results if it is not possible to host the wallet on another machine

i do not cpu mine, only qt and cgminer on a fpga is running, the cpu (Celeron dualcore 1.2GHz) is idle at 5-7%
legendary
Activity: 1509
Merit: 1030
Solutions Architect
count me in  Grin

addnode=94.79.142.190

EDIT:  And again blakecoin-QT fails to update... After restart it works again...

With this Bug i need a restart every Difficulty change!

EDIT2: i have 10 connections... after diff change it goes down to under 4

I think this is due to not enough cpu for wallet and mining on the same machine and it results in losing connections, this happens more with cgminer than reaper or kramble's python miner, it also happened with the cpu miner if you used all the cores on the same machine as the wallet

my advice is to just mine on the weak cpu machine and host the wallet on another machine and just mine to it over your local network and blakecoind is better for using less resources than blakecoin-qt so you could just try using blakecoind and see if that gets same results if it is not possible to host the wallet on another machine
member
Activity: 93
Merit: 10
count me in  Grin

addnode=94.79.142.190

EDIT:  And again blakecoin-QT fails to update... After restart it works again...

With this Bug i need a restart every Difficulty change!

EDIT2: i have 10 connections... after diff change it goes down to under 4
member
Activity: 93
Merit: 10
The "checknone failed" errors are going away then the speed stabilized itself so no update needet Smiley

But all 10 Blocks i found over night are invalid, and after a restart of blakecoin-QT, CGminer says new difficulty... letz see wats happening now.

btw, Stable speed without HW errors now @ 1,54GH  Cool
legendary
Activity: 1470
Merit: 1001
Use Coinbase Account almosanywhere with Shift card
Every block found since 17:00 18NOV13 has been bad.

Well I exited my blakecoin-qt and restarted with a -regen. all bad blocks no longer show in my transactions. We'll see how it goes.


edit

So far so good 3 blocks found and thet all have many confirmations so I think I'm ok now.
sr. member
Activity: 384
Merit: 250
Thanks, i have 1,488 GH speed
Core 1 184 HW: 0
Core 2 196 HW: 196
Core 3 204 HW: 25
Core 4 200 HW: 77

after 1:30 Hours working.

under BTC it goes up to 224 and HW Error <0.7%

Good, its settled down then. I did a fixed 180MHz build (link upthread) that you may want to try out, though you could probably get away with slightly faster (the slow core 1 device may just have been a quirk this time round). BTC is a different algorithm and its a complete coincidence that the blake clock speeds are so similar.

I am working on a faster build, but the xilinx tools are not playing nice (and its takes hours for each attempt). In between these I'm also trying to get a 3 core build to route (it should fit eventually as its only using 77% of the LUT, but its all been negative so far, and the final clock speed will probably be disappointing).

Enjoy
member
Activity: 93
Merit: 10
Thanks, i have 1,488 GH speed
Core 1 184 HW: 0
Core 2 196 HW: 196
Core 3 204 HW: 25
Core 4 200 HW: 77

after 1:30 Hours working.

under BTC it goes up to 224 and HW Error <0.7%
legendary
Activity: 1470
Merit: 1001
Use Coinbase Account almosanywhere with Shift card
Do others get a lot of orphan blocks?
Both on GPU and now on FPGA I seem to generate as many orphans as good blocks

I get hardly any (touch wood). Though I am in the UK so BlueDragon's servers are local.
My config ...
addnode=162.243.133.80
addnode=162.243.14.130
addnode=146.185.135.24


Since I was seeing this on GPU I'm pretty sure it has nothing to do with the FPGA Cheesy

I have 4 nodes listed
addnode=162.243.133.80
addnode=162.243.14.130
addnode=146.185.135.24
addnode=184.171.247.23

first 3 are the same

Would it help at for me to check peers and add nodes
94.79.142.190:8773
122.108.173.195:8773
etc
sr. member
Activity: 384
Merit: 250
now i get many "checknone failed" ^_^'

Its because the automatic frequency management has reached max speed. They should quiet down after a few minutes (the speed will drop slightly). If they don't then I can do a custom fixed frequency build for you. I run mine at around 200MHz but its had a PSU hack done at some point, so is probably running overvolted.

Actually I think I'll just disable the frequency code and put a couple of fixed frequency variants up on dropbox to choose between. Give me an hour or so (I'll update this post with links).

OK give this one a try https://www.dropbox.com/s/m5q5ag80ulpzf9o/cgminer-311-2013-11-19-180MHz.exe

Just rename it cgminer.exe and pop it into cgminer-3.1.1-blakefpga. Its a fixed 180Mhz build, but I can tweak this up or down as needed (it really ought to be a command line parameter, but that's going to be harder to implement).

If you get a bunch of "checknonce failed" when it starts up, just quit and retry it (it happened to me a couple of times in testing, as I said earlier, I'm not a professional programmer and this is experimental code, "it may eat your hamster..." etc).
member
Activity: 93
Merit: 10
now i get many "checknone failed" ^_^'
Jump to: