Timeline for Maintenance cost of SIMD programming code base
Current License: CC BY-SA 4.0
7 events
when toggle format | what | by | license | comment | |
---|---|---|---|---|---|
Nov 23, 2022 at 9:11 | history | edited | Glorfindel | CC BY-SA 4.0 |
2 broken links fixed, cf. https://meta.stackoverflow.com/a/406565/4751173
|
Dec 17, 2014 at 13:38 | comment | added | Den | @DocBrown maybe, but his actual question is formulated to be more generic: "...industry consensus regarding the value of clean and simple code for SIMD code...". I doubt there is any (official) consensus at all, but I submit that higher-level languages can reduce the difference between the "usual" and SIMD code, just like C++ let's you forget about assembly, thus reducing the maintenance costs. | |
Dec 17, 2014 at 11:05 | comment | added | Doc Brown | To my understanding, the OP is asking if there exist solutions with widespread commercial usage. Though I appreciate your hint (maybe I can use the lib for a project here), from what I see RyuJIT is far from beeing a "widely accepted industry standard". | |
Dec 17, 2014 at 9:13 | comment | added | Den | @gnat I have actually read that whole paragraph, not just the top-level bullet points, and the poster is not mentioning any general-purpose SIMD libraries, just the computer vision and image processing ones. Not to mention that analysis of higher-level languages application is completely missing, despite no C++ tag and no C++-specificity reflected in the question title. This leads me to believe that while my question will not be considered primary it is likely to add value, making people aware of other options. | |
Dec 16, 2014 at 16:10 | comment | added | gnat | per my reading, the option to use external libraries has been already investigated and addressed by asker: "Libraries with widespread commercial usage do not seem to be heavily SIMD-enabled..." | |
Dec 16, 2014 at 15:46 | history | edited | Den | CC BY-SA 3.0 |
edited body
|
Dec 16, 2014 at 15:38 | history | answered | Den | CC BY-SA 3.0 |