I compiled firominer on Artix Linux with -DETHASHCUDA=OFF and -DETHASHCPU=OFF. Then, in order to benchmark my GPU, I ran firominer -Z 400000 (as suggested here), which gave me around 680 Kh/s IIRC. On this same page, the RX 480 has been benchmarked at 5.4 Mh/s, so I thought that was weird.
Next, I wiped my Linux install and worked to compile firominer on FreeBSD (had to patch a few things, some features are missing but it works). This is what I still have on my computer right now, here are the logs:
# firominer/build/firominer/firominer -Z 400000
firominer 1.2.4
Build: freebsd/release/gnu
i 14:37:20 Selected pool localhost:0
i 14:37:20 Established connection to localhost:0
i 14:37:20 Spinning up miners...
cl 14:37:20 cl-0 Creating buffers
cl 14:37:20 cl-0 OpenCL OpenCL 1.1 Mesa 20.2.3 not supported, but platform Clover might work nevertheless. USE AT OWN RISK!
cl 14:37:20 cl-0 Using PciId : Radeon RX 580 Series (POLARIS10, DRM 3.35.0, 13.0-RELEASE, LLVM 10.0.1) OpenCL 1.1 Mesa 20.2.3 Memory : 8.00 GB
i 14:37:20 sim Epoch : 307 Difficulty : 4.29 Gh
i 14:37:20 sim Job: 9c10d580… block 400000 localhost:0
i 14:37:22 sim Using block 400000, difficulty 1
cl 14:37:23 cl-0 Pre-compiled period 400000 OpenCL ProgPow kernel
cl 14:37:23 cl-0 Loaded period 400000 progpow kernel
cl 14:37:23 cl-0 Generating DAG + Light : 3.95 GB
cl 14:37:23 cl-0 Creating light cache buffer, size: 54.37 MB
cl 14:37:23 cl-0 Creating DAG buffer, size: 3.90 GB, free: 4.05 GB
cl 14:37:23 cl-0 Loading kernels
cl 14:37:23 cl-0 Writing light cache buffer
cl 14:37:23 cl-0 Pre-compiled period 400001 OpenCL ProgPow kernel
m 14:37:25 0:00 A0 0.00 h - cl0 0.00
m 14:37:30 0:00 A0 0.00 h - cl0 0.00
m 14:37:35 0:00 A0 0.00 h - cl0 0.00
m 14:37:40 0:00 A0 0.00 h - cl0 0.00
m 14:37:45 0:00 A0 0.00 h - cl0 0.00
m 14:37:50 0:00 A0 0.00 h - cl0 0.00
m 14:37:55 0:00 A0 0.00 h - cl0 0.00
m 14:38:00 0:00 A0 0.00 h - cl0 0.00
cl 14:38:02 cl-0 3.90 GB of DAG data generated in 39839 ms.
m 14:38:05 0:00 A0 0.00 h - cl0 0.00
m 14:38:10 0:00 A0 0.00 h - cl0 0.00
m 14:38:16 0:00 A0 848.58 Kh - cl0 848.58
m 14:38:21 0:01 A0 848.58 Kh - cl0 848.58
m 14:38:26 0:01 A0 848.55 Kh - cl0 848.55
m 14:38:31 0:01 A0 848.55 Kh - cl0 848.55
m 14:38:36 0:01 A0 848.03 Kh - cl0 848.03
m 14:38:41 0:01 A0 848.03 Kh - cl0 848.03
m 14:38:46 0:01 A0 848.91 Kh - cl0 848.91
m 14:38:51 0:01 A0 848.91 Kh - cl0 848.91
m 14:38:56 0:01 A0 848.92 Kh - cl0 848.92
m 14:39:01 0:01 A0 848.92 Kh - cl0 848.92
m 14:39:06 0:01 A0 848.43 Kh - cl0 848.43
m 14:39:11 0:01 A0 848.43 Kh - cl0 848.43
m 14:39:16 0:01 A0 847.43 Kh - cl0 847.43
m 14:39:21 0:02 A0 847.43 Kh - cl0 847.43
m 14:39:26 0:02 A0 848.70 Kh - cl0 848.70
m 14:39:31 0:02 A0 848.70 Kh - cl0 848.70
There is a small improvement, but nothing close to what I expected from an RX 580.
I doubt this issue comes from the OS / kernel / libraries, since I get similar results on Linux and FreeBSD, but if you want me to try again on a more conventional distribution, please let me know.
I compiled firominer on Artix Linux with
-DETHASHCUDA=OFFand-DETHASHCPU=OFF. Then, in order to benchmark my GPU, I ranfirominer -Z 400000(as suggested here), which gave me around 680 Kh/s IIRC. On this same page, the RX 480 has been benchmarked at 5.4 Mh/s, so I thought that was weird.Next, I wiped my Linux install and worked to compile firominer on FreeBSD (had to patch a few things, some features are missing but it works). This is what I still have on my computer right now, here are the logs:
There is a small improvement, but nothing close to what I expected from an RX 580.
I doubt this issue comes from the OS / kernel / libraries, since I get similar results on Linux and FreeBSD, but if you want me to try again on a more conventional distribution, please let me know.