Vanitygen: Vanity bitcoin address generator/miner [v0.22]

[luv2drnkbr]

I'm worried about running my video card at full for long periods of time because I don't have any kind of cooling other than the chasis fan and the fan in the card itself.  When I run the ocl version, it gives me a timeframe until the probability approaches 100%.

My question is, can I run it for a bit, then quit and run it later, or will it re-find the same old tries?  Do I need to let it run until it finds it, or can I for example close it at 1% and then just look for 1% each day until I find it?

Sorry if this is a stupid question.  I understand what probability means and that it's just brute force guessing, but the fact that it gives me a time frame makes me wonder if it has a certain starting seed, which would mean re-starting it over and over, doing a little each day, would not be effective in finding a solution.

So, do I need to run it all the way through or can I just do a little bit each day?

[bulanula]

I'm worried about running my video card at full for long periods of time because I don't have any kind of cooling other than the chasis fan and the fan in the card itself.  When I run the ocl version, it gives me a timeframe until the probability approaches 100%.

My question is, can I run it for a bit, then quit and run it later, or will it re-find the same old tries?  Do I need to let it run until it finds it, or can I for example close it at 1% and then just look for 1% each day until I find it?

Sorry if this is a stupid question.  I understand what probability means and that it's just brute force guessing, but the fact that it gives me a time frame makes me wonder if it has a certain starting seed, which would mean re-starting it over and over, doing a little each day, would not be effective in finding a solution.

So, do I need to run it all the way through or can I just do a little bit each day?

Interested in this too. Hope we get a comprehensive answer.

[Red Emerald]

It should be fine to start and stop the process. Just like mining, previous hashes have no effect on later hashes. Even if you manually set the seed, then you still don't get the same hashes.  Your calculations for time to solve will be less accurate, but the probability to find the key is the same no matter how many times you have already tried.

EDIT: See the link from deepceleron below

[deepceleron]

I'm worried about running my video card at full for long periods of time because I don't have any kind of cooling other than the chasis fan and the fan in the card itself.  When I run the ocl version, it gives me a timeframe until the probability approaches 100%.

My question is, can I run it for a bit, then quit and run it later, or will it re-find the same old tries?  Do I need to let it run until it finds it, or can I for example close it at 1% and then just look for 1% each day until I find it?

Sorry if this is a stupid question.  I understand what probability means and that it's just brute force guessing, but the fact that it gives me a time frame makes me wonder if it has a certain starting seed, which would mean re-starting it over and over, doing a little each day, would not be effective in finding a solution.

So, do I need to run it all the way through or can I just do a little bit each day?

Interested in this too. Hope we get a comprehensive answer.

There is an answer, just five pages back: https://bitcointalk.org/index.php?topic=25804.msg611858#msg611858

[DeathAndTaxes]

I seem to be having a problem w/ regular expression option.

I want to use a regular expression but only at the beginning of the address.

For example:
I assumed ^1TCa[1-9] would look for addresses which match

1TCa1....
1TCa2....
1TCa3....
etc

however it returns things like:
16Xo9HWAexrC1TCa21T51vhUBCesyZ58KKU

I skimmed through the thread and searched for the word "regular" so if it was already covered I must have missed it.

It looks like in the thread people are searching minus the "1" so can regular expression searches not be limited to the front of address.

The long version of what I am looking for is something like this

1TCaXXXX
where XXXX is a 4 digit number sequence (obviously only 1 to 9 due to base 58 limits).

[BTCurious]

It's about 58 times more likely to find addresses that start with your pattern. You could do a filtering step afterwards…

[DeathAndTaxes]

It's about 58 times more likely to find addresses that start with your pattern. You could do a filtering step afterwards…

Yeah that is an option but that is a lot of filtering.  I was just searching for 1 numeric digit as a test.  Ultimately I would want a 4 digit or 5 digit sequence. 

The 4 digit address space is 58^4 and the valid range is 9^4 so only 1 in 1724 will be valid. 
The 5 digit address space is 58^4 and the valid range is 9^5 so only 1 in 11,115 will be valid. 


I am assuming that based on your answer it isn't possible to use regular expressions and limit the search to start of string.   If that is the case I guess I could modify the code.

[BTCurious]

I meant: Just use vanitygen to generate addresses with your digits. You will get a lot of addresses which start with your pattern, and only some that have your pattern somewhere in between. It doesn't matter how long the digit sequence is. Then afterwards you can filter the resulting list.

I haven't tried regex matching with vanitygen, so I don't know if it can work. What you describes sounds like a bug.

[DeathAndTaxes]

I meant: Just use vanitygen to generate addresses with your digits. You will get a lot of addresses which start with your pattern, and only some that have your pattern somewhere in between. It doesn't matter how long the digit sequence is. Then afterwards you can filter the resulting list.

I haven't tried regex matching with vanitygen, so I don't know if it can work. What you describes sounds like a bug.

Hmm.  I got you.  So if the entire pattern is 8 digits and the address is 34 digits then there is equal chance of it finding the pattern starting @ digit 2, 3 ...... 26.  We only want the patterns which begin @ digit 2 so one in 26 should be good.

I am having a slow morning this morning.  I had to re-read your post twice before I got it.

[BurtW]

After reading more carefully what you really want is:

    \A1TCa[1-9]{4}

I just tried it and it works.

[BTCurious]

I meant: Just use vanitygen to generate addresses with your digits. You will get a lot of addresses which start with your pattern, and only some that have your pattern somewhere in between. It doesn't matter how long the digit sequence is. Then afterwards you can filter the resulting list.

I haven't tried regex matching with vanitygen, so I don't know if it can work. What you describes sounds like a bug.

Hmm.  I got you.  So if the entire pattern is 8 digits and the address is 34 digits then there is equal chance of it finding the pattern starting @ digit 2, 3 ...... 26.  We only want the patterns which begin @ digit 2 so one in 26 should be good.

I am having a slow morning this morning.  I had to re-read your post twice before I got it.

Actually, now I'm confused. It should return addresses starting with "1TCa" much more often than having "1TCa" in the middle, because the first character is always a "1". If it doesn't, then something's weird.

[DeathAndTaxes]

After reading more carefully what you really want is:

    \A1TCa[1-9]{4}

I just tried it and it works.

Thanks for your help.  \A instead of ^ for start of string.  I likely wouldn't have got that.

[DeathAndTaxes]

DOH just realized you can't use regular expression w/ GPU.  Given that regular expressions are next to useless.  A good CPU = 200 KH/s.  A good GPU = 20MH/s. 

[BurtW]

Is this true?  I have never set up for GPU vanity generation but I always just assumed that you could - if you want to - use regular expressions on the GPU version.  Of course all that checking of the results would slow you down but that is the price you would pay.

[BkkCoins]

Ya, I tried regex with GPU too and it just went off to la la land. Doesn't work and has a message about using the CPU for regex.

So I guess you could run it with -k (keep going) and pipe the output thru grep or something and collect matches, or use them to trigger killing the process. Or just collect them in a file and grep it later. I haven't tried this.

[BurtW]

So in D&Ts case match for exactly 1TCa, store them in a file, grep for those that happened to have 4 digits after the 1TCa offline later.  Sounds like a very large file with a few good ones in it

[BkkCoins]

This seems to work ok,

Code:

$vanitygen -k 1T |grep -A1 -P 1T[1-9]{2}.*
WARNING: Built with OpenSSL 0.9.8o 01 Jun 2010
WARNING: Use OpenSSL 1.0.0d+ for best performance
Difficulty: 1353
Address: 1T51oxW5TodSMEHoL8j7r9Ztuv4tVXTuu
Privkey: 5JnQkMpSHPQp34XmDi79nzUNsEYLcudD6a9gVwJ4wWjheQWGcCD
--
Address: 1T95SQJawGuQHVjqXhHdsNn27iEX3rrag
Privkey: 5JANavEjXEp34SEPVPgMM3BhqhgQq49Zgi2ZLNdxboUfKRUudNR
--
Address: 1T77PtZF5wFqDfy17QbLVUk532JWHbouB
Privkey: 5KSAeZTGCmVLXraPsPU3DmwXG4x3FFm3jtGYLHiyMx2eRDSLWar

I made it easier so I could see results quicker but the actual cmd would be,

oclvanitygen -k 1TCa |grep -A1 -P 1TCa[1-9]{4}.*


-A1 for including next line too.(with key).
-P to use Perl regex which I know unlike POSIX regex.

Or to save in a file, and view on screen,

oclvanitygen -k 1TCa |grep --line-buffered -A1 -P 1TCa[1-9]{4}.* | tee -a list.txt

--line-buffered is needed because grep buffers it's output.

Edit: I forgot you would need a -d  arg for oclvanitygen to select device. Also, I tried this on GPU and it said for low difficulty it would be faster on CPU. Seems it is too. But it's faster if you make it more difficult for the GPU, eg.

oclvanitygen -k 1TCa1 |grep -A1 -P 1TCa[1-9]{4}.*

Weird.

[malevolent]

Hello everyone, I have a few (noob) questions

- Anyone know how to use all GPUs for oclvanitygen?
And why if I select GPU #2,for ex.:

oclvanitygen -r -d=2 123

Only my primary GPU (GPU #0) will be used? In my case this is a slower GPU.

- Why with -r only the CPU can be used (vanitygen prints "WARNING: Using CPU pattern matcher")?

- Are there any better/faster address generators out there?

- Why I can only generate addresses starting with "1"? Aren't there anymore addresses left starting with "3"?

- Why am I getting higher speeds without "-r" on CPU?
Without: ~350kkeys/sec
With: ~75kkeys/sec
Isn't it supposed to be the other way around (expression can be placed anywhere in the address instead just at the beginning)? Or am I getting something wrong?

- I have a Q9300 undervolted to 0.975V and underclocked to 2GHz to save power (stock 2.5GHz and sth like 1.1 or 1.2Vcore), before mining this used to be primarily a gaming rig, and I had this CPU running @ 3.5GHz and 1.32Vcore, what speed can I expect if I decide to O/C it back to 3.5GHz? I think it'll increase linearly but not sure and will probably do it unless I find a way to generate addresses with the "-r" option.

- Any way to generate shorter addresses (shortest as possible, 25 base58 chars?)

Thanks.

[BkkCoins]

oclvanitygen -r -d=2 123
Only my primary GPU (GPU #0) will be used? In my case this is a slower GPU.

Take out the equal sign. -d 2

- Why with -r only the CPU can be used (vanitygen prints "WARNING: Using CPU pattern matcher")?

See docs. GPU isn't good at regex expression evaluation.

- Are there any better/faster address generators out there?

I haven't heard of any other vanity gens with GPU support at all.
This one is kind of a miracle it exists at all considering the work involved.

- Why I can only generate addresses starting with "1"? Aren't there anymore addresses left starting with "3"?

Because ALL bitcoin addresses always start with 1. No such thing as a 3 address.

- Why am I getting higher speeds without "-r" on CPU?
Without: ~350kkeys/sec
With: ~75kkeys/sec

I guess the regex eval process slows it down a lot.

- I have a Q9300 undervolted to 0.975V and underclocked to 2GHz to save power (stock 2.5GHz and sth like 1.1 or 1.2Vcore), before mining this used to be primarily a gaming rig, and I had this CPU running @ 3.5GHz and 1.32Vcore, what speed can I expect if I decide to O/C it back to 3.5GHz? I think it'll increase linearly but not sure and will probably do it unless I find a way to generate addresses with the "-r" option.

See post above, explains how to do regex without -r

[malevolent]

Because ALL bitcoin addresses always start with 1. No such thing as a 3 address.

https://en.bitcoin.it/wiki/Address :

A Bitcoin address, or simply address, is an identifier of approximately 34 alphanumeric characters, beginning with the number
1 or 3
An example of a Bitcoin address is 37muSN5ZrukVTvyVh3mT5Zc5ew9L9CBare

So there's no way to make use of all GPUs simulatenously?
Or maybe I could be running 3 instances of oclvanitygen each for each GPU to increase the probability of finding an address and it's priv key? Would that be the same?