Topic: Faster and Autonomous Large Bitcoin Collider fork - page 4. (Read 6778 times)

I have it running on several machines now.  Pulled the latest bitcoin address with balance file and am looking for anything that has 10btc or more.  Pulling in 142k addresses does knock about 10% off my hashing speed, but it's still much faster than LBC was on a P5000.  It actually uses all of the video card instead of being CPU bound.  If it could use a bloom filter like LBC does, it would be even faster and look for more.

Thanks for doing this.

Thanks Holy_Ship, managed to contact him 

In verbose mode it always complains on CPU, even when working OK and finding everything.
I didn't ask him what this complain means.

If something is not working it is possible that he will fix it for free if you report a bug on github. Because he was doing this work partially for portfolio 

I got it working on Ubuntu.  I was running LBC at the same time, and they were fighting for memory.  Still no go on the Windows version, but it's popping a CPU check error.  Might be because I'm using an integrated card for testing.  Some time this week I'll throw a couple ATI cards in the box and see if those will work through windows.

Thanks for the help and thanks for setting this up!

The command is OK.
Are you really talking about NDV fork of oclvanitygen ?
Do addresses.txt file contain 160 bounty addresses? Case of chars matters.

There is a verbose switch. Post the log.

I found him on upwork.

https://www.upwork.com/mobile/freelancers/~01cb02eecfe0e0725e

Yes, we are going to make a farm.

What is his mail, do you have it, can you PM?

Fired up my Ubuntu and dowloaded the code.  Running an Nvidia card in that box.  Compiled it and ran the same command:

oclvanitygen -D 0:0 -f addresses.txt -o found.txt -k lastkey.txt 3000

Still doesn't find anything. 

What's the command you ran to have those return positive?

I haven't tried intel gpu, also I didn't even try windows.
The command looks fine, although -D 0:0 is not necessary.

On nvidia card I've tested first 30 and 52,53,54,55 puzzles - works like a charm.

You can request updates from NDV directly, if you wish. He promised 30-50% boost after rewriting ECC code for around $300-400.
You can make the change public or personal.

If you are going to use big farm it makes sense. For 1-2 cards (like I have) it's just easier to buy more hardware.

Do you still need funding to improve the speed?

oclvanitygen -D 0:0 -f addresses.txt -o found.txt -k lastkey.txt 3000
Start with 3000
Searching for 160 addresses
81920/131072
[1.70 Mkey/s][total 44564480]

Removed the leading zeros.  Capitalized the numbers. 
Still doesn't find anything.
Can you share a command line that returns a find?

I guess these zillions of leading zeros transform your HEX to some crapy WIF.
Also HEX should be uppercase, at least in Linux.

try on cpu -D 1:1

To reiterate, using
should reveal the addresses for the first 14 puzzle transactions within seconds.  It doesn't.

It's an integrated video card.  It won't be the final machine I'll be using...just testing.

@holy_ship do you have an opinion?

why should you -v replace with -D 0:0 or -D 1:0 it seems like you're running on a CPU card I do not know if it works fine.

Still doesn't find it.  This mean anything to you?

oclvanitygen -v -f addresses.txt -o found.txt -k lastkey.txt 000000000000000000000000000000000000000000000000000006bd3b27c800
Start with 000000000000000000000000000000000000000000000000000006bd3b27c800
Searching for 160 addresses
Device: Intel(R) HD Graphics 4600
Vendor: Intel(R) Corporation (8086)
Driver: 20.19.15.4531
Profile: FULL_PROFILE
Version: OpenCL 1.2
Max compute units: 20
Max workgroup size: 512
Global memory: 1708759450
Max allocation: 427189862
OpenCL compiler flags: -DDEEP_PREPROC_UNROLL -DLOCAL_MEM_SIZE=65520
Loading kernel binary a278d6da1e690bf2f16f62f7f94db316.oclbin
allocsize=427189862, memsize=1708759450
Grid size: 1280x2048
Modular inverse: 10240 threads, 256 ops each
Bitmap size (bytes): 427189862

Start a thread81920/131072

Copy 1280 sequential points to the device
Copy the row stride array to the device
Move the row increments forward
Copy 1280 sequential points to the device
[1.52 Mkey/s][total 2621440]
Move the row increments forward
Copy 1280 sequential points to the device
Wait for the GPU to complete its job
Found -4593 candidadtes, check on CPU
CPU check failed
[1.62 Mkey/s][total 5242880]
Move the row increments forward
Copy 1280 sequential points to the device
Wait for the GPU to complete its job
Found -4593 candidadtes, check on CPU
CPU check failed
[1.65 Mkey/s][total 7864320]
Move the row increments forward
Copy 1280 sequential points to the device
Wait for the GPU to complete its job
Found -4593 candidadtes, check on CPU
CPU check failed
[1.66 Mkey/s][total 10485760]
Move the row increments forward
Copy 1280 sequential points to the device
Wait for the GPU to complete its job
Found -4592 candidadtes, check on CPU
CPU check failed
[1.68 Mkey/s][total 13107200]
save the priv. key to a file
Copy 1280 sequential points to the device
Copy the row stride array to the device
Move the row increments forward
Copy 1280 sequential points to the device
Wait for the GPU to complete its job
Found -4597 candidadtes, check on CPU
CPU check failed

he goes the other way from bigger to smaller. take that 000000000000000000000000000000000000000000000000000006bd3b27c800.

to start multiple cards you can create cmd file bablabla.cmd

so that the window turns off after start /min can add.

Hello,
I have it up and running, but want to test that it is working correctly.

I run

oclvanitygen -f addresses.txt -o found.txt -k lastkey.txt 000000000000000000000000000000000000000000000000000006bd3b27c500

and would expect it to find the private key to 1PiFuqGpG8yGM5v6rNHWS3TjsG6awgEGA1 almost immediately since the private key is 6bd3b27c591.

It doesn't.  It blows past that private key without logging anything.

It is likely I'm doing something wrong, but I can't see the trees from the forest right now.  Can you advise?

Thanks!