note: LINK : fatal error LNK1181: cannot open input file 'OpenCL.lib'

[ghost]

Trying to compile in Rust with --features gpu. Compiles and works fine with CPU. Any ideas?

Win 10 Pro 64
Nvidia 1080ti
Driver 397.93

Thank you,

R

[PlasmaPower]

Install the Nvidia CUDA toolkit then restart

[lephleg]

@PlasmaPower I am also facing this:

error: linking with link.exe failed: exit code: 1181
LINK : fatal error LNK1181: cannot open input file 'OpenCL.lib'

OS:
Windows 10 Pro (64bit)
Graphics Chipset:
Radeon (TM) RX 480 Graphics
Radeon Software Edition:
Adrenalin 2019
Radeon Software Version:
19.7.5

[PlasmaPower]

@lephleg Unfortunately I believe there are also other issues using this with AMD cards. See #33

[lephleg]

@PlasmaPower As weird as it may sound, I am also getting this without passing the --features gpu flag now...

[PlasmaPower]

try --no-default-features

[lephleg]

Success at last! Let's see how far an i5 can get us... :)

[MikeRow]

Tested on i7-7700HQ (no GPU flag): 200k keys/second
Tested on GTX1080 (notebook version): 1.8M keys/second

Good job @PlasmaPower

[MikeRow]

@PlasmaPower As weird as it may sound, I am also getting this without passing the --features gpu flag now...

I had the same troubles on windows 10.
Fix:

• Install Visual Studio 2019 (Community version, free)
• Install Nvidia CUDA toolkit
• restart computer

Compile with GPU flag.

[adrianegraphene]

I am stuck on this exact problem too. I can pass --no-default-features and get a successful build, but only my CPU is used, not my GPU....

If I try passing cargo --features gpu, it overrides --no-default-features and I end up with the same error as OP.

[adrianegraphene]

Ok, after downloading this from github, I was able to install with cargo install nano-vanity --features gpu

However, when I try nano-vanity .robocash --gpu-device 0 --threads 0, I get

Estimated attempts needed: 1099511627776
No computation devices specified

If I try nano-vanity .robocash --gpu, the estimates comes up, followed by a delay, then the following.

Estimated attempts needed: 1099511627776

thread 'main' has overflowed its stack

GPU: AMD RX Radeon 5700 XT
LuxMark v3.1 score: 2061

@PlasmaPower Sorry to be a bug, any tips you can give for this? My CPU would take months to generate an 8 character vanity, but my GPU should be much quicker.... I tried running with --gpu as well to be sure I wasn't just calling the wrong GPU index. I tried compiling first with --features gpu, then installing with --features gpu, but that made no difference to me.

[guilhermelawless]

EDIT: Compilation works, but crashes while opening the device.

Instructions for AMD are quite simple. See https://github.com/GPUOpen-LibrariesAndSDKs/OCL-SDK/releases/latest , download the .exe and install. Contains only a few headers and libs, nothing like the Nvidia toolkit.

It should also set the required environment variables so everything works, but just in case, the libraries are then located in C:\Program Files (x86)\OCL_SDK_Light

[boogerlad]

Adding a system environment variable named "LIB" with a value "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2\lib\x64" worked for me