by piotr5 » Sat Mar 14, 2015 9:50 pm
what makes you think there is a market for 64-core? the major reason I'd buy 4 parallellas for is to get the 4GB ram, not for the 64 cores! (I do maths with the computer, so data-size is more important to me than increasing speed of the actual calculations -- even though I don't do statistics.) also obviously the kickstarter campaign didn't sell enough 64-core products to unlock the stretch-goal -- if the market would really desire that, those more expensive products would have easily amassed that much. I have no knowledge about adapteva, maybe they didn't do the right announcements in the right areas of the multicore-market, but in my own opinion it is a correct decision to first focus on software-development and keeping alive the 16-core version. without software it really isn't surprising when demand for 64-core is low, there aren't many people who are capable of multi-core programming. what's worse, the market of 64-core computers is extremely volatile in the sense that only few people/companies are interested in obtaining such a product, and if they decide to they'll spend high amounts of money for that -- if a single of these customers would die or go broke, the manufacturer might go broke too. that's a bad buisness plan! a good buisness plan is alike to what video-cards do: now you can buy hundreds of cores, program them through mesa4 and whatever sdk, applications in games are easily found and therefore it's mainly gamers who buy them. gamers are a stable market, and IMHO parallella should slowly move into that direction too. alternatively I can see a future in the communication-market, a smart-phone with lots of cores has higher capabilities in understanding gestures and words of the user -- again an application best covered by any company which has decades of experience with video or audio processing, just like with games. adapteva has no experience in that direction, programmers of that kind is what's needed now...