Hi. I’m new on the forum.
I compiled and run windowsX64 version of your opensource test_cl_zogminer using cmake for generating the Visual Studio 2013 Community project.
I have an AMD 7950 board and this is the output log, consisting of some opencl compiler warning and the output of the kernel.
Listing OpenCL devices.
FORMAT: [deviceID] deviceName
[0] Tahiti
CL_DEVICE_TYPE: GPU
CL_DEVICE_GLOBAL_MEM_SIZE: 3221225472
CL_DEVICE_MAX_MEM_ALLOC_SIZE: 2214174021
CL_DEVICE_MAX_WORK_GROUP_SIZE: 256
Found suitable OpenCL device [Tahiti] with 3221225472 bytes of GPU memoryUsing platform: AMD Accelerated Parallel ProcessingUsing device: Tahiti(OpenCL 1.2 AMD-APP (2004.6))Printing program log"C:\Users\Giovanni\AppData\Local\Temp\OCL2904T1.cl", line 112: warning:
argument of type “const void *” is incompatible with parameter of
type “void *”
memcpy(&w, src, sizeof(w));
^
“C:\Users\Giovanni\AppData\Local\Temp\OCL2904T1.cl”, line 165: warning:
integer conversion resulted in a change of sign
S->f[1] = -1;
^
“C:\Users\Giovanni\AppData\Local\Temp\OCL2904T1.cl”, line 173: warning:
integer conversion resulted in a change of sign
S->f[0] = -1;
^
“C:\Users\Giovanni\AppData\Local\Temp\OCL2904T1.cl”, line 252: warning:
argument of type “const uchar *” is incompatible with parameter of
type "void "
memcpy( S->buf + left, in, fill ); / Fill buffer */
^
“C:\Users\Giovanni\AppData\Local\Temp\OCL2904T1.cl”, line 267: warning:
argument of type “const uchar *” is incompatible with parameter of
type “void *”
memcpy( S->buf + left, in, inlen );
^
Running List Generation…100 Times :)Iteration: 0 Result: 3Iteration: 10 Result: 3Iteration: 20 Result: 3Iteration: 30 Result: 3Iteration: 40 Result: 3Iteration: 50 Result: 3Iteration: 60 Result: 3Iteration: 70 Result: 3Iteration: 80 Result: 3Iteration: 90 Result: 3The thread 0xc38 has exited with code 0 (0x0).
The program ‘[2904] test_cl_zogminer.exe’ has exited with code 0 (0x0).
Cool! Don’t mind the warnings, we are aware of them and are working on them. The build really isn’t doing much right now outside of the list generation as you see. Stay tuned for more progress, should be able to test sorting soon.
Hey thank’s for the kind words and donating if you do.
Unfortunately no. Because zcash does not have a windows build, this means that our solver won’t have a local process to communicate the solutions to.
Maybe you could in theory rpc the solutions to a separate computer over a network connection to bundle with the block information to broadcast to the network but I’m not sure, We don’t have any plans to build something like that.
First of all, thanks for all the donations, you guys rock!
Now to the important stuff, the update:
I am still waiting for my new PSU, but I have been working on a VM. I hope I can get back up and running soon.
This week I have been working on everything related to getting zogminer running against zcashd. I have made nice progress, and now we will be able to test Zogminer against the testnet when the OpenCL solver is complete.
Still needs a lot of testing, but this was a key step to get a full working miner.
Next I will test everything, and try to integrate some GPU programming tools that @Age provided, to make running OpenCL code easier.
I was actually replying to this (and didn’t quote it)
But, although I have never tried this, VrtualBox does has the option Enable 3D Acceleration:
With this feature, if an application inside your virtual machine
uses 3D features through the OpenGL or Direct3D 8/9 programming
interfaces, instead of emulating them in software (which would be slow),
VirtualBox will attempt to use your host's 3D hardware. This works for
all supported host platforms (Windows, Mac, Linux, Solaris), provided
that your host operating system can make use of your accelerated 3D
hardware in the first place.
Really my idea was to run the GPU miner on Windows and the zcash node on a linux VBox on the same machine.
I can try to port the source of the zogminer (only the miner) on Windows.
Yesterday the porting of the test miner from the repository was very simple, it took me about 30 mins. only little adjusting the headers.
Hey S4Lt! Thanks for dropping by. I just finished the first build of the miner on my laptop and I have pushed it’s nvidia memory bandwidth to the max. I pretty much can’t run the solver without crashing so it’s time to try some bigger hardware.
I’m going over the code a bit today to see if I have a memory leak or if it’s just my hardware isn’t powerful enough. Which if anyone is in LA and can lend me a small rig (maybe just a mobo, decent ram and cpu with an RX card) that would help.
I have a couple of Rigs running Ethos right now. Will install Ubuntu on one. Win10 running zcashd in VM on one. To see what is more stable with the miner later. No RX480 cards yet. But I will be getting a RX480 4GB and RX480 8GB (slightly faster memory) next week to compare if it is worth the extra dollars to get the 8GB version or not. I live in Sweden Europe though. Currently I have 24 cards on 4 rigs. 280X and 290 4GB. And 390 8GB.
I can let you know when I have my cards and give you root SSH if you promise not to burn my cards