-
Notifications
You must be signed in to change notification settings - Fork 4
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Documentation update for 0.16 #14
base: master
Are you sure you want to change the base?
Conversation
README.txt
Outdated
@@ -71,8 +69,7 @@ General requirements: | |||
- an OpenCL SDK | |||
|
|||
Please note: the AMD APP SDK has been discontinued. If you still want to use it | |||
to compile mfakto, make sure you have version 2.5 or later. You can download | |||
the SDK here: https://community.amd.com/thread/227948 | |||
to compile mfakto, make sure you have version 2.5 or later. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Please consider removing this whole paragraph. Just because mfakto can be compiled with software discontinued 9 years ago, should an ordinary user care? If this library still needs to be mentioned, I would do it closer to the end of this document, not in the beginning. Otherwise it sounds like mfakto has some obsolete dependency, which makes mfakto somewhat obsolete too.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I agree there is no need to keep this section anymore. It seems out of place now.
@@ -71,8 +69,7 @@ General requirements: | |||
- an OpenCL SDK | |||
|
|||
Please note: the AMD APP SDK has been discontinued. If you still want to use it | |||
to compile mfakto, make sure you have version 2.5 or later. You can download | |||
the SDK here: https://community.amd.com/thread/227948 | |||
to compile mfakto, make sure you have version 2.5 or later. | |||
|
|||
############# | |||
# 1.1 Linux # |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The Linux part needs rewriting, as ROCm is not a requirement. I can compile and run mfakto without ROCm. With VectorSize=1
fixed, mfakto becomes usable on Intel GPUs. I'd rather see potential users try mfakto on whatever GPU they have.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
From what I see, there is an AMD_APP_DIR
variable in the makefile that points to an ROCm installation:
Line 57 in ef462f1
AMD_APP_DIR = /opt/rocm/opencl |
It has been there for as long as I can remember. Did you remove this variable before compiling mfakto?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No, I don't need to change or override anything in the Makefile. I just need to install OpenCL headers from the distro.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It seems AMD_APP_DIR
isn't even needed anymore as I was able to compile mfakto in WSL after removing this variable from the makefile. I'll create a separate PR as this one is mainly for documentation changes.
README.txt
Outdated
interim, resulting in duplicate work and wasted cycles. | ||
partial results as PrimeNet may reassign the exponent to someone else in | ||
the meantime; this can lead to duplicate work and wasted cycles. You can | ||
easily avoid this by setting Stages=0 in the INI file. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't think it's a good recommendation, considering that Stages=0
is not the default. Maybe the user should be asked to wait until all stages complete for the exponent.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@brubsby did recently change the default to Stages=0
for mfaktc: primesearch/mfaktc@5757f54
I don't know if he intends to do the same for mfakto, although I do agree this part is unnecessary either way.
README.txt
Outdated
a program called AutoPrimeNet that does so. See the AutoPrimeNet download | ||
page for instructions: https://download.mersenne.ca/AutoPrimeNet | ||
|
||
Another option is to manually submit the results: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I haven't used AutoPrimeNet yet, but my assumption is that is gets assignments too. Then it should be described earlier.
I would group getting assignments and submitting results for every workflow, like:
AutoPrimeNet:
- how to get assigments
- how to submit results
mersenne.ca
- how to get assigments
- how to submit results
and so on.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sounds like a good idea.
Great to see the readme get some love. Something I noticed recently is the VC++ Redistributable version 10 is no longer supported. (11 and 12 are not either.) Readme.txt line 273ish. Per Microsoft:
The link to the latest supported versions is: https://learn.microsoft.com/en-us/cpp/windows/latest-supported-vc-redist?view=msvc-170 |
@jcoxdco Looks like the links now lead to 404 errors. I'll update the readme to point to the new download page. |
A note from https://www.mersenneforum.org/node/11037/page79#post1062989:
Maybe this could be adjusted in this pull request, too? |
@henning-gerhardt I bumped the version in my latest change. |
results.text
toresults.json.txt