Hey guys. About 2 hours ago now I started getting this error:
Allocating/mapping single buffer failed with: clCreateBuffer(-61). GPU can't allocate the DAG in a single chunk. Bailing.
clEnqueueWriteBuffer(-38)
I am mining on an r9 270 2GB using
cpp-ethereum v1.2.2 for Windows on Windows 7 64-bit.
I did some digging and apparently it has something to do with the DAG being one single chunk that is too large for the GPUs memory. I also have an r9 290 with 4GB memory in my system and mining still works fine on that.
When i type "ethminer --list-devices" I get the following output:
Listing OpenCL devices.
FORMAT: [deviceID] deviceName
[0] Hawaii
CL_DEVICE_TYPE: GPU
CL_DEVICE_GLOBAL_MEM_SIZE: 4294967296
CL_DEVICE_MAX_MEM_ALLOC_SIZE: 3019087173
CL_DEVICE_MAX_WORK_GROUP_SIZE: 256
[1] Pitcairn
CL_DEVICE_TYPE: GPU
CL_DEVICE_GLOBAL_MEM_SIZE: 2147483648
CL_DEVICE_MAX_MEM_ALLOC_SIZE: 1408867653
CL_DEVICE_MAX_WORK_GROUP_SIZE: 256
I can see that the DAG file is now 1417674752 bytes so it would seem the DAG has just grown beyond the allocated memory. I have tried deleting and building a new DAG, uninstalling my drivers with display driver uninstaller and re-installing the crimson 15.12 drivers. Also I have tried setting all the following evironment variables, and rebooting, and setting them again:
GPU_FORCE_64BIT_PTR 0
GPU_MAX_HEAP_SIZE 100
GPU_USE_SYNC_OBJECTS 1
GPU_MAX_ALLOC_PERCENT 100
GPU_SINGLE_ALLOC_PERCENT 100
as was suggested in
this thread on github but it had no effect.
I suppose I can try the 16.x drivers (in fact I will do that next) but I used those before and they dropped my hash rate by 25 or 30%.
Is anyone else having this issue on 2GB AMD cards all of a sudden. Any foolproof way of allocating more mem?
Any help would be appreciated. Thanks.
