Topic: Vanitygen: Vanity bitcoin address generator/miner [v0.22] - page 40. (Read 1153678 times)

Hi can you assist with my issue on the program not allowing me to use any of the gpu's while i get that error i posted?

thanks!

What a pity you didn't go back and read the thread.  Does it really matter in the larger picture if you loose a Meg or two of hashes with two or more cards?

Sooo, was the conclusion that no one understand how to fix it,  still?

Hi,

I'm having issues with vanitygen using my GPU's

I'm on WIn 10 Enterprise,

I have 3 rigs i've tried.

Rig1: 5x RX480

Rig2: 5x R9 290X

Rig 3: 6X R9 280X

Not sure if Win 10 is the issue, but I have gotten errors on the RX480 rig, which I have heard simply doesn't play well.

But the other 2 rigs, I can see my gpus, and when I select them in a bat file similar to this:


C:\Users\R2\Desktop\vanitygen-0.22-win>oclvanitygen.exe -D 0:0 1buckrogers
Difficulty: 10054102514374869638
Compiling kernel, can take minutes...failure.
clBuildProgram: CL_BUILD_PROGRAM_FAILURE
Build log:
C:\Users\R2\AppData\Local\Temp\\OCL3888T3.cl:173:19: error: variable in constant address space must be initialized
__constant bignum bn_zero;
                  ^
C:\Users\R2\AppData\Local\Temp\\OCL3888T3.cl:173:19: error: constant address space qualified variables are required to be initialized
2 errors generated.

error: Clang front-end compilation failed!
Frontend phase failed compilation.
Error: Compiling CL to IR
Device: Hawaii
Vendor: Advanced Micro Devices, Inc. (1002)
Driver: 2117.13 (VM)
Profile: FULL_PROFILE
Version: OpenCL 1.2 AMD-APP (2117.13)
Max compute units: 44
Max workgroup size: 256
Global memory: -1073741824
Max allocation: -1073741824
Could not open device '0:0', ignoring
Available OpenCL platforms:
0: [Advanced Micro Devices, Inc.] AMD Accelerated Parallel Processing
  0: [Advanced Micro Devices, Inc.] Hawaii
  1: [Advanced Micro Devices, Inc.] Hawaii
  2: [Advanced Micro Devices, Inc.] Hawaii
  3: [Advanced Micro Devices, Inc.] Hawaii
  4: [Advanced Micro Devices, Inc.] Hawaii
  5: [GenuineIntel] Intel(R) Pentium(R) CPU G3220 @ 3.00GHz

C:\Users\R2\Desktop\vanitygen-0.22-win>


Not sure why it is returning the error on the gpu, but as a side note, if I just run the:

C:\Users\R2\Desktop\vanitygen-0.22-win>vanitygen64.exe 1buckrogers
Difficulty: 10054102514374869638
[548.54 Kkey/s][total 1977344][Prob 0.0%][50% in 402856.1y]

The cpu portion will run, but obviously I would want to use the GPU's, and have many different rigs to choose from, all are running Win 10 enterprise though.

If anyone has some suggestions, please ...i'm all ears.

thanks!

I make mention of this effect some dozen pages back...

This thread has reached a natural conclusion and should be locked so that a more relevant one can be started by somebody else (and no, not me...)

It "Seems" like the oclvanitygen does not increase rate like we are use to mining with GPU cards.  If you get 10Mk/sec with one card, if you double, you get less than 2x10; triple the exact same card you get <3x10.  and the more cards you add, it seems to gain less of the next card you add.

To prove this, in your 4x 280x system, try this:

oclvanitygen -D 1:0 -D 1:1 -D 1:2 -D 1:3 -k 1Bitcointalk
Then Record
Speed All4 =

Then each card

oclvanitygen -D 1:0 -k 1Bitcointalk
Speed D0 =

oclvanitygen -D 1:1 -k 1Bitcointalk
Speed D1 =

oclvanitygen -D 1:2 -k 1Bitcointalk
Speed D2 =

oclvanitygen -D 1:3 -k 1Bitcointalk
Speed D3 =

I bet your speeds
D0+D1+D2+D3 >>>  Speed All4

does anyone know if its possible to search for multiple prefixes when using multiple ECDSA keypairs?

Or is it only possible to generate one of these at a time? When ive searched for my own addresses previously i can obviously seach multiple prefixes but im not sure the method to try and acheive the same reults here.

You should be getting >80Mkey/s with that setup I would think...

Try running a different instance for each GPU and see if you can determine the problem.  One GPU could be locking up, you could accidentally be using a CPU, etc...  

Another thing if you haven't generated a vanity on that machine before, is to generate a short one and make sure you are getting the public/private key combo as some driver combinations don't return valid keys (if anyone has that issue, search this thread for lifeboat).

How is the performance if you only use the primary?

Was using exactly what I Pasted. I am running Win 7 x64. Maybe the 280X are just slow at finding keys?

 Sorry!  I missed that part.

 
 Are you running in safe mode with the -S switch?  That slows me down to 12Mkeys/s

Do you use any switches?

I tried using oclvanitygen.exe -D 1:0 -D 1:1 -D 1:2 -D 1:3 1BLazed

 Sounds a little slow.  I would think you should see almost double that.
My single 7970 runs @ 25 Mkeys/s and the 280X is essentially the same thing.


 Just tested it to make sure

Does 46 Mkeys/s sound right for 4 x 280X using oclvanitygen?

the oclvanitygen code would have to be modified to have suffix search.

Does anyone know how to do that?

This gives the available commands:
./vanitygen -h

You can't.

Does anyone know how to search for Suffixes instead of Prefixes, using OCLvanitygen?

(I know we can use vanitygen -r Bitcoin$ )

Can I have a list of Vanitygen commands? After I made post #2126, I started my Vanitygen again. With normal commands. It is now at 5.9 percent.
The -i command means that it will run a case insensitive search. So if yo u want it to generate a 1JTK wallet with the "-i" command, it might then generate 1JtK instead. It basically just speeds up the generation time, but you're giving up case sensitivity.

If you spend several hours getting oclvanitygen working with AMD Radeon 7xxx on 64-bit Linux - Ubuntu 12.04 and Ubuntu 14.04. In the end only the following is needed:
Ubuntu drivers
vanitygen patch
1 userspace library file from AMD APP SDK v2. I created a minimal instruction text file. Maybe this will save someone and some hours of searching the internet.
What do you should to do next is add a note to remove *.oclbin, since ppl with problems already have a compiled OpenCL binary and it wont be regenerated unless the file is missing.

There are lots of distributed designs out there already, some open source but for CUDA, would need to convert CUDA language to OpenCL language. This would make it easier for those with large farms to utilize an address generator without it being a PITA to setup.Test CPU operation. This command line will generate a Bitcoin addresses beginning with 1ABCD in around a minute or less:
>vanitygen 1ABCD
Test GPU operation. This command line will generate a Bitcoin addresses beginning with 1ABCDE in around a minute, using the first OpenCL device in your system:
>oclvanitygen -d 0 1ABCDE