possible soc updates.


3GB RAM works fine with 3-way interleaving found on some Xeon systems.  See also HP Z800 workstations.  There is precedence for Intel doing that.
That's true but this isn't a XEON and it won't have socketed memory or a strange bus width - more than likely it'll use PoP memory which won't come in exotic sizes.
Ahem:

http://www.engadget.com/2013/07/24/samsung-starts-making-3gb-low-power-memory-for-smartphones/

3GB of low power memory on a single chip - perfect for a Pandora successor - and no 3-way interleaving required.
 
So - educate me...  would that 3GB chip work on anything other than a Samsung SoC - or is the point moot since Samsung declined?
 
You can buy most TI SoCs (including DM3730) from many distributors who have them in stock, although they may only have hundreds or even tens instead of thousands. Same with Freescale.
Yep, the ARM-based Freescale i.MX 6 series SoCs are popping up in various projects recently. Doesn't look totally off as an option for a Pandora successor.
 
Last edited by a moderator:
Has there been a cosmic shift of some kind? Reading an article about Clovertrail+ it states that 'ARM based processors are going to have to find significant improvements in energy efficiency in order to compete'.

I know there are lies, damn lies & Intel benchmarks but based on the difficulty of sourcing current gen ARM Socs for P2 is it time to consider going over to the dark side (x86)?

Additional article here.
 
Last edited by a moderator:
I guess the best "tactic" here is to just wait till devices with Clovertrail appear on the market, everything else will just end up in people getting at each other from entrenched positions.
 
Clovertrail is already in products on the market. The SOCs to watch from Intel will be the Merrifield and Baytrail SOCs .
 
^ long time since I read up on them, but from memory they performed pretty decently. It's hard to say how well as it was clovertrail that was embroiled in the whole dodgy Antutu benchmark fiasco. Ask Exophase about it, he'll be happy to go into further details :p , from memory it was him that brought the Clovertrail Antutu scores into question.
 
Recent PowerVR GPUs support OpenGL and are used both in the x86 CloverTrail SoCs and in several ARM SoCs (e.g. Exynos 5410, OMAP5). But I would avoid PowerVR in any future project, because there are no FOSS drivers and their proprietary drivers tend to suck.
 
FOSS drivers from day 1 would be damn excellent.. but we would be aiming at older tech? Time to juggle.
 
I am ok with older hardware for FOSS drivers, if they implement full open gl (best version 3 or 4).
 
Has there been a cosmic shift of some kind? Reading an article about Clovertrail+ it states that 'ARM based processors are going to have to find significant improvements in energy efficiency in order to compete'.

I know there are lies, damn lies & Intel benchmarks but based on the difficulty of sourcing current gen ARM Socs for P2 is it time to consider going over to the dark side (x86)?

Additional article here.
2GHz is the turbo frequency on that chip meaning if it was operating at 2GHz it was literally doing a benchmark sprint.  The normal frequency on that chip is less then half that at 933MHz.

The claims proved to be wrong in review, and those articles have already been brought up.  Intel and Microsoft pull these stunts semiregularly.
 
Bullshit doesn't even begin to describe the recent CloverTrail+ "study" ABI Research did (I'm sure with no funding by Intel, oh no) with the real details hidden behind a paywall. It's using AnTuTu which is both a terrible benchmark AND was broken by Intel's compiler. Throw it in the trash.

CT+ is good and it's competitive but it's not nearly like this. Silvermont may be a different story.
 
Last edited by a moderator:
Back
Top