-
Notifications
You must be signed in to change notification settings - Fork 5k
Intrinsics are not documented #42374
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
Comments
I couldn't figure out the best area label to add to this issue. If you have write-permissions please help me learn by adding exactly one area label. |
My personal recommendation is https://www.felixcloutier.com/x86/index.html - it has all the instructions which you can search by name using Ctrl+F. I'm personally worried about the amount of efforts this would require (there's thousands of intrinsics out there), and how much value it would hold (compared to, say, linking either the Intel intrinsics explorer page or the felixcloutier page). It would be a tremendous amount of effort and it has to be kept up-to-date in case something changes w.r.t the manual which isn't trivial. |
Tagging subscribers to this area: @tannergooding, @jeffhandley |
CC. @terrajobst, has there been any update on being able to reuse the documentation from the architecture manuals? |
@Gnbrkm41 |
@MajorScruffy, the issue isn't in writing some tool to port the documentation, it is in getting permission to reuse it. The existing documentation are largely |
@tannergooding
Writing the documentation manually is clearly overkill, but perhaps links to Intel's documentation for every method would be a good start and I doubt Intel would complain about that. |
Is it fair to assume that most folks likely to use the intrinsics will be advanced enough to be comfortable referring to the hardware manuals? ie., the only info they need is the mapping to the underlying instructions? |
The mapping to the underlying instruction and corresponding native intrinsic is what we have today. Linking back to something like:
Might provide some additional benefit, but that might also require sign-off, stable URLs, or other considerations. CC @terrajobst who should be able to give an update. |
@danmosemsft I agree.
This is already here (xml-doc). And for usage workflow it's quite common to have a look at intrinsics guide. So copy & paste from the xml-doc to the search field is acceptable. |
Perhaps we can add hyperlinks to the relevant page of the Intel's/AMD's intrinsic guide on each functions? Also, one suggestion - perhaps we can provide an easier way to map C-style intrinsic functions (e.g. |
Due to lack of recent activity, this issue has been marked as a candidate for backlog cleanup. It will be closed if no further activity occurs within 14 more days. Any new comment (by anyone, not necessarily the author) will undo this process. This process is part of our issue cleanup automation. |
This issue will now be closed since it had been marked |
As someone who has never worked with intrinsics before, I wanted to have a look at the documentation to see what APIs are available, and what can actually be achieved with them. To my dismay, the documentation for all methods was basically just the declaration of the native function.
Let's take Avx2.Add(Vector256, Vector256) as an example. The documentation is just:
That doesn't tell me anything about what the method does. So if I want to see what it does, I have to copy "_mm256_add_epi64" and paste it on Intel's Intrinsics Guide.
As you can imagine, doing this for every method gets tiresome pretty fast. It would be much more helpful if the .NET documentation was the same as Intel's documentation.
The text was updated successfully, but these errors were encountered: