0

I've set up a new Linux box (Xubuntu based off 18.04) with 64G of RAM and two AMD R9-280 GPUs. I'm trying to run hashcat (latest from hashcat.net) on it but am running into an out of memory error on the GPU. 

hashcat -m 1800 -a 3 shadow
hashcat (v5.1.0) starting...

OpenCL Platform #1: Advanced Micro Devices, Inc.
================================================
* Device #1: Tahiti, 2373/3037 MB allocatable, 14MCU
* Device #2: Tahiti, 2334/2984 MB allocatable, 14MCU
.....
Applicable optimizers:
* Zero-Byte
* Brute-Force
* Uses-64-Bit

Minimum password length supported by kernel: 0
Maximum password length supported by kernel: 256

Watchdog: Temperature abort trigger set to 90c

Initializing device kernels and memory...
Initializing OpenCL runtime for device #1...
CreateCommandQueue(): CL_OUT_OF_HOST_MEMORY

clinfo also gives some strange results:

Global free memory (AMD) <printDeviceInfo:75: get number of CL_DEVICE_GLOBAL_FREE_MEMORY_AMD : error -33>

The test shadow file I am checking is only two lines long so its size is not an issue. I am the only user on the system and nothing else is running during the test. I suspect the issue is related to the amdgpu-pro drivers but I don't have a good starting point. Does anyone have an idea what could cause this problem? Thanks!

jmq
  • 1,559
  • 9
  • 21

1 Answers1

1

I had this problem with the 7970 and the R9-280x on 18.04. I wanted to use these particular cards for their double precision capability.

The included link shows that several environment variables need to be set. This solves my problem on 18.04 with Radeon™ Software for Linux® 19.50.

1). Can add new script in /etc/profile.d/amdgpu.sh

export GPU_FORCE_64BIT_PTR=1
export GPU_USE_SYNC_OBJECTS=1
export GPU_MAX_ALLOC_PERCENT=100
export GPU_SINGLE_ALLOC_PERCENT=100
export GPU_MAX_HEAP_SIZE=100

https://community.amd.com/thread/233040

ginzburg
  • 26
  • 1