Hi,

the luck I had with the one system of mine (happy mining PXCs in the moment), is the bad luck of the other:

I try to get neo-gpuminer to work on an Ubuntu12.04-32bit system containing a HD7850. I have a similar 64bit system containing two HD7970 which works fine.

On the 32bit system, scrypt-cgminer is working, so gpu is detected:

[2014-10-23 22:37:36] Started cgminer 3.2.1
[2014-10-23 22:37:44] Probing for an alive pool
[2014-10-23 22:37:46] Pool 0 difficulty changed to 32.000000
[2014-10-23 22:37:47] Network diff set to 44M
[2014-10-23 22:37:51] Accepted 5c6f3272 Diff 43/32 GPU 0
[2014-10-23 22:37:58] Accepted 0814b14c Diff 106/32 GPU 0
[2014-10-23 22:37:58] Accepted 106365f8 Diff 38/32 GPU 0
[2014-10-23 22:38:11] Accepted 22625720 Diff 100/32 GPU 0
[2014-10-23 22:38:18] Accepted abd35f00 Diff 38/32 GPU 0
[2014-10-23 22:38:22] Accepted f72139cb Diff 60/32 GPU 0
(5s):317.6K (avg):291.1Kh/s | A:6 R:0 HW:0 U:10.0/m WU:253.2/m

though fglrxinfo reports:

Error: unable to open display (null)

If I start neo-gpuminer I get the following error message:

Error -61: clCreateBuffer (padbuffer8)

which can be found in ocl.c at line 971:

/* This buffer is weird and might work to some degree even if
 * the create buffer call has apparently failed, so check if we
 * get anything back before we call it a failure. */
clState->padbuffer8 = NULL;
clState->padbuffer8 = clCreateBuffer(clState->context, CL_MEM_READ_WRITE, bufsize, NULL, &status);
if (status != CL_SUCCESS && !clState->padbuffer8) {
        applog(LOG_ERR, "Error %d: clCreateBuffer (padbuffer8)", status);
        return NULL;
}

Anyone any idea what I am doing wrongly?