Bragging rights, for aspects of the design which may or may not eventually make any sense, has always been the case for products sold to the innocent. I've read that beyond eight cores, memory management becomes so cumbersome that it's hard to get any better performance. Not sure if that's still the case, but it does suggest that migrating to as many as eight makes sense.
What also make sense is to aim for the same amount of processing power in these handhelds as one gets now from PCs. That's when one could conceivably start thinking in terms of docking the handheld to do the real work, and doing away with the separate PC entirely. Looks to like this is where we are heading. The problem is that until now, the battery life of powerful but portable computers has been pretty bad. I wouldn't be surprised in the industry is working hard to fix that problem.
Also, I don't know about 8 cores, but without a doubt 4 cores are a huge improvement over a single or even dual cores, operating at the same clock speed. I've seen that myself, with a single core 2.8 GHz PC vs a qud core 2.8 GHz PC. Like night and day, EVEN THOUGH the applications running on these PCs are not multithreaded.
Qualcomm called eight cores dumb. I find that odd, given how MediaTek is a pretty smart company and the details of the True Octa-Core architecture have yet to be fully revealed. Rushing to judgment before examination of the implementation - where the proof of viability truly lies - is just plain ignorant.
Thx Junko for starting a discussion on the performance of Multi Core processors. This is an extremely technical / sensitive topic. It requires further in - depth reporting. But I wonder if EE Times / EDN would devote the resources needed to bring out the critical details as most of this Octa / Hexa ... hype is being enabled by ARM.
I've found it interesting in dealing with two nearly outdated cell phones (samsung S3 and Motorola Photon Q, both about a year old) the significant lags one experiences with what seems to be "normal use." In the progression of a day, I find that either phone can get pretty laggy. In fairness, we use lots of apps. Most of the phone apps like leaving themselves in memory and learning as much as possible about you. To get an idea, go into the developers options and set "show CPU useage." While this is interesting, then set the background processes to "no background processes" or possibly "at most 2 processes." The phone becomes instantly snappy. However, some google operations won't work. My simple minded take on all of these is that more cores can be helpful with todays phone apps doing such a good job of getting in you business!
One thing to remember is that this is the innovation bed for silicon at the moment. Whether 8 cores is overkill or not, the same innovation achieved from these devices will enable future embedded applications, microservers, and the Internet of Things. Personally, I am more interested in what other applications outside of handsets adopt these solutions and why.
With more mulit-core processing, how much power is compromized? 16 hours battery life for mobile device is also very essential. Also, how does OS and software response to this multi-core beyond quad-core?
@Jim: As I am sure you know it's really the infrastructure world which is pushing the boundries in multicore with basestations, routers and servers packing mnay more than eight cores at the high end, including some many core chips from folks like Tilera and etc.
But the tools to extract parallelism seem to be particular to specific workloads and architectures. Ideas may get shared but it seems loike each sector has to re-implement them.
A Book For All Reasons Bernard Cole1 Comment Robert Oshana's recent book "Software Engineering for Embedded Systems (Newnes/Elsevier)," written and edited with Mark Kraeling, is a 'book for all reasons.' At almost 1,200 pages, it ...