Why is hardware multithreading not more common in embedded systems? was put on hold for being "opinion-based". Is there any way that the question can be refined to be answerable (by SE standards)?

(I can understand why the question would generate opinion-based answers. Since I was not expecting results from a 100-person-year research effort [which might be a low requirement for an "actual answer" determining the economic and other factors leading to the current state], perhaps I should have recognized that this would be an "opinion-based" question [though if the results of a substantial research project could not be summarized well, the question would be "too broad"]. However, I am curious about the seeming lack of market penetration for multithreading relative to multicore. [Strangely, I was more concerned, after seeing the first close vote, that the question might have been off-topic. As an outsider to embedded system development, I could not guess at how obvious the answer would be.])

Would reducing the breadth from "embedded systems minus networking" help sufficiently? E.g., if only microcontrollers were considered. (Limiting the question to microcontrollers would make the question less interesting to me, but I would still learn from any answers. However, even that question would still seem to be opinion-based.)

Asking about the success specifically of MIPS MT-ASE (i.e., in what types of systems is MIPS MT-ASE seeing use and at what volumes) would seem to be less opinion-based, but answers to that question would only hint at why even that specific ISA has limited uptake for its multithreaded implementations. However, it would at least provide some data on the reception of multithreading in embedded systems.

(Unless someone could get a mole into ARM, Ltd., I doubt a question about why ARM has not defined a multithreading architectural extension could be anything other than opinion-based. Likewise for those controlling development for various 8- and 16-bit processors.)

(I am grateful that it did not attract down votes [even getting two up votes--more than 3% of my reputation!--though based on comments it might not have been sufficiently clear, I think it "shows research effort" :-)].)

Perhaps the question cannot be fixed in any meaningful way and should just be allowed to pass into a permanently closed state.

| |
  • 1
    \$\begingroup\$ Typically when there is that much text in a question, it is because the asker is using the site as a soapbox for some out-of-mainstream ideas. I am not saying this is what you did, but to be honest I did not bother to read your question in all its detail because I did not want to wade through the wall of text. \$\endgroup\$ – The Photon Oct 22 '13 at 18:45

I think this is a interesting discussion, but this site is not well suited to discussions. Also, you did a lot of pontificating in the "question", which always makes writing a answer more difficult. I frequently pass up answering such questions because you have to deal with the OP's assumptions and statements before you even get to the question part.

I think it was correctly closed as being too opinion based. I'm not sure what you seem to be trying to ask/discuss can be modified to be a good fit here without losing your intent altogether.

| |
  • 1
    \$\begingroup\$ Yeah, I thought that might be the case. As to the pontificating, I was trying to show "research effort" and encourage an answer that goes beyond the obvious things that I think I sort of understand (or explicitly address why that thinking is wrong). I have a habit of providing too much context in my questions (trying to defend against down votes, I think), especially when I have a half-(assed)answer (posting such as even a c.wiki answer would be questionable and would discourage other answers if it is really half or more of an answer). Thanks for the feedback! \$\endgroup\$ – Paul A. Clayton Oct 18 '13 at 23:22

Generally, I think questions of the form, "Why does product XYZ not exist on the market?" are a bad fit for the site.

This question can only have a few different answers:

  • The product does exist: see vendor ABC's website. (Then you should have asked first, "Does this product exist?", or even better, "How can I solve this problem?" With a note that you tried looking for a product to do XYZ but didn't find it.

  • What you're asking for isn't technologically feasible at this time. But then the answer would have have complete knowledge of what is and isn't possible, so we're only likely to be able to answer this way in the most extreme cases.

  • No manufacturer has thought that making this product would be profitable. This makes the question a marketing question and not an engineering question. It also requires us to speculate about the thoughts in the heads of every marketing department in the world.

    In particular, if the point of the question is that the product should exist in the market, then there are two problems: First, questions here should not be used to make a point; second, the asker should not be asking about the product here, they should be applying for a patent and/or polishing up a 30 second salespitch for the idea and spending their days going up and down elevators in buildings that VCs often visit.

My point is, no matter which of these answers applys to the "why doesn't XYZ exist?" question, either the question could have been asked in a better way, or the answer isn't going to be available from the other users here.

In your case, I don't know if any of this applies specifically, but you do have other problems:

  1. Question is too long. An answer touching on all your points would be too long for the site. I'd mark this under the "too broad" category for closing the question.

  2. The topic of the question is marketing rather than engineering.

  3. Answering the question requires speculation rather than facts. (I think this is what people were getting at when they checked the "opinion-based" box in the close dialog).

| |
  • \$\begingroup\$ +1 That was helpful. However, I think the question length (point 1) came almost entirely from including speculations toward an answer rather than expressing the breadth of the question. "embedded - networking" is probably "too broad". On 2, price and market segmentation are marketing but cost and demand are more engineering. The question is somewhat on the border, but I was interested in the engineering aspects. On 3, I am too ignorant of the field to know whether the answer is obvious, "good speculative" (my guess), or "bad speculative" (the community's judgment). \$\endgroup\$ – Paul A. Clayton Oct 23 '13 at 20:40
  • \$\begingroup\$ Chatty: Incidentally, one place where multithreading might be more attractive is in FPGA soft cores since the replicated content would mostly be in relatively dense/low-cost block RAMs (compared to replicating whole cores). \$\endgroup\$ – Paul A. Clayton Oct 23 '13 at 20:47
  • \$\begingroup\$ @PaulA.Clayton, rather than speculate about an answer in the text of the question, it's allowed and accepted to submit an answer to your own question. Of course an answer that's only speculation is not likely to be well-received. There isn't really any good way, except in the comments, to post speculations about an answer. I suspect this is by design. \$\endgroup\$ – The Photon Oct 23 '13 at 20:53
  • \$\begingroup\$ Not only are highly speculative answers poorly received, but providing a partial answer serves relatively little purpose (if c. wiki, others would have to sacrifice potential rep. by extending it and a CW answer [if halfway decent] probably discourages good/better answers; if self-answer, the answer is even more likely to be downvoted [with rep. loss]). The one answer given basically confirmed two of the speculative points in the question; the incentive to post such would presumably have been reduced if another answer already included the information. \$\endgroup\$ – Paul A. Clayton Oct 23 '13 at 21:16
  • \$\begingroup\$ As a data point (anecdote), I posted a CW answer on SpaceEx.SE after what I felt was an incomplete answer was posted. It was (surprisingly) upvoted the same as the other answer and accepted, but it was not improved by anyone with knowledge (nor its content refined and included in the earlier answer) and so the question has two inadequate answers. \$\endgroup\$ – Paul A. Clayton Oct 23 '13 at 21:32
  • \$\begingroup\$ It's true that CW answers don't typically attract a lot of "help". But in ee.se, it's not particularly uncommon for one answer to give just a bit of additional information to an existing answer, and it doesn't seem to be particularly poorly thought of to do that. If somebody comes to a page and there's a combination of two answers that tells them what they need to know (instead of one complete answer), I don't think that's a huge problem. \$\endgroup\$ – The Photon Oct 23 '13 at 21:59
  • \$\begingroup\$ Yes, not a huge problem, but ideally the highest scored answer would be the accepted answer and would include all the good content posted in answers to the question--well, ideally such would be a perfect answer from an omniscient perspective :-). The SE system imperfectly manipulates human motivations, but humans are just so complex. Anyway, thank you for the insights. \$\endgroup\$ – Paul A. Clayton Oct 24 '13 at 2:44

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .