Skip to content
This repository has been archived by the owner on Apr 24, 2022. It is now read-only.

cu-parallel-hash option is still visible although it has already been removed #2017

Open
pmalek opened this issue Aug 8, 2020 · 2 comments · May be fixed by #2129
Open

cu-parallel-hash option is still visible although it has already been removed #2017

pmalek opened this issue Aug 8, 2020 · 2 comments · May be fixed by #2129

Comments

@pmalek
Copy link

pmalek commented Aug 8, 2020

Describe the bug
cu-parallel-hash is still visible in the help while it has already been removed (in #1808) which causes an error when using it.

./ethminer_0.18.0/bin/ethminer --help-ext cu


ethminer 0.18.0
Build: linux/release/gnu

CUDA Extended Options :

    Use this extended CUDA arguments to fine tune the performance.
    Be advised default values are best generic findings by developers

    ...
    --cu-parallel-hash  UINT {1,2,4,8} Default = 4
                        Set the number of hashes per kernel

The same happens on 0.19.0-alpha.0

To Reproduce
Steps to reproduce the behavior:

  1. Use --cu-parallel-hash option on the command line
  2. Observe Error: The following argument was not expected: --cu-parallel-hash

Expected behavior
No error

Environment:

  • Operating System: Ubuntu 18.04
  • Hardware: GTX 1060
  • Ethminer Version 0.18
jarednielsen added a commit to jarednielsen/ethminer that referenced this issue Jan 19, 2021
@jarednielsen jarednielsen linked a pull request Jan 19, 2021 that will close this issue
@4-FLOSS-Free-Libre-Open-Source-Software

Problem still present. Misleading help

ethminer --help-ext cu

ethminer 0.19.0-17+commit.23d537a0
Build: windows/release/msvc

CUDA Extended Options :

Use this extended CUDA arguments to fine tune the performance.
Be advised default values are best generic findings by developers

--cu-grid-size INT [1 .. 131072] Default = 8192
Set the grid size
--cu-block-size UINT {32,64,128,256} Default = 128
Set the block size
--cu-devices UINT {} Default not set
Space separated list of device indexes to use
eg --cu-devices 0 2 3
If not set all available CUDA devices will be used
--cu-parallel-hash UINT {1,2,4,8} Default = 4
Set the number of hashes per kernel
--cu-streams INT [1 .. 99] Default = 2
Set the number of streams per GPU
--cu-schedule TEXT Default = 'sync'
Set the CUDA scheduler mode. Can be one of
'auto' Uses a heuristic based on the number of active
CUDA contexts in the process (C) and the number
of logical processors in the system (P)
If C > P then 'yield' else 'spin'
'spin' Instructs CUDA to actively spin when waiting
for results from the device
'yield' Instructs CUDA to yield its thread when waiting for
for results from the device
'sync' Instructs CUDA to block the CPU thread on a
synchronize primitive when waiting for results
from the device

ethminer --cu-parallel-hash=8


ethminer 0.19.0-17+commit.23d537a0
Build: windows/release/msvc

Error: The following argument was not expected: --cu-parallel-hash=8


@xuxiaowei13
Copy link

xuxiaowei13 commented Jun 7, 2021

same thing. the help should be updated.
image
and it's said to be able to improve the performance of 1060 in the doc.....
image

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

3 participants