Cannot allocate big buffer - Not enough graphics memory?

Question

I am using Claymore's Dual Ethereum v9.5

Here is my config file:

setx GPU_FORCE_64BIT_PTR 0
setx GPU_MAX_HEAP_SIZE 100
setx GPU_USE_SYNC_OBJECTS 1
setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_SINGLE_ALLOC_PERCENT 100
EthDcrMiner64.exe -epool eu2.ethermine.org:4444 -ewal 0x9C57aE119dd13EE9Cd444Cf96091f1e84446.rig-epsw x

I keep getting this error: "CUDA error: cannot allocate big buffer for DAG"

From what I've read this happens when graphics memory is too low. I use a Nvidia 750Ti with 2GB.

How can I fix this ?

Answer 1

to form an answer:

@richard's comment in the previous is right, the DAG file's size is the source of your problem, your GPU needs to load it before start mining. Since mid-2016 it is no more possible to mine using a 2GB graphic card while the DAG file has exceeded 2GB. we will reach 3GB in April 2018.

Answer 2

Check you PAGE file size - it should be more than 16000 MBs. Keep minimum 16000 MBs and maximum 20000 MBs. A majority of the miners rely on virtual mem to keep the process quick and it requires a good amount of size allocation.

I have 4x gtx 1060 6GB and was have this error until i changed the page size/virtual memory

Answer 3

When your video cards meet or exceed current DAG requirements, your pagefile is within spec and everything else has been functional... Launch 2 instances of the miner at the same time. This forces a DAG recreation on the second instance. Just kill the miner that is still choking. There will be a few errors thrown about

server: bind failed with error: 10048 (port for remote management is busy, use different -mport value), next attempt in 20sec...

Everything seems to work from there on-ward. I speculate it has something to do with DAG cache and epoch increases, but /shrug

Claymore v9.5 with latest nVidia drivers... I can post logs if needed.

If this helps you Eth: 0x7e5c4Bb4343f0313cfddd56dD4A9c21F43DFDefD

Answer 4

Quite an obvious thing: this error might be also caused by the lack of free space on the hard drive.

What needs checking:

1. Amount of free space on disk
2. Video card memory size
3. Page file memory size

Answer 5

I had this problem and expanding the virtual memory and trying to spam two instances at the same time didn't work.

My fix, which fixed this immediately, was to add "-lidag 0" to the command line. This creates the DAG buffer at the lowest intensity which apparently help with stability.

From the Readme!!!.txt:

-lidag low intensity mode for DAG generation, it can help with OC or weak PSU. Supported values are 0, 1, 2, 3, more value means lower intensity. Example: "-lidag 1". You can also specify values for every card, for example "-lidag 1,0,3". Default value is "0" (no low intensity for DAG generation).

I hope this helps somone as it took me a long time to figure out myself.

If this doesn't work, try -eres 0