Intel ULP SoCs

tangey said:
The first sentence of my post is:-
Samsungs first tizen phone will run a quad-core Intel Soc.

Which was because the headline of the article I linked to is:-
"Samsung's First Tizen-based Smartphone is Powered by Quad-core Intel Processor"

I did include "if true" in my posting, implying it has to be Moorefield, assuming the article is correct in calling out Intel as the SOC.
Click to expand...
Yeah, sorry, I meant to write "Why would it have to be Intel" rather than Moorefield. You are indeed correct that if it's Intel then it very likely is Moorefield (quad core and 2.3GHz frequency match Moorefield specs).

But the Intel premise is just an hypothesis so the rest of my previous still stands IMHO :)
 
Laurent06 said:
Yeah, sorry, I meant to write "Why would it have to be Intel" rather than Moorefield. You are indeed correct that if it's Intel then it very likely is Moorefield (quad core and 2.3GHz frequency match Moorefield specs).

But the Intel premise is just an hypothesis so the rest of my previous still stands IMHO :)
Click to expand...

Makes no sense. Since when has Tizen the Java-like sandbox like Android ? AFAIK, Tizen (please correct me if I'm wrong) a native ARM implementation/fork/port of Linux with some UI pasted on top.

Though Wikipedia seems to suggest that apps are based on Qt and HTML5 (Javascript)
 
Intel announced the "Core M" line of CPUs for fanless tablets.
Details are scarce, but the chips have a 10W TDP and people are assuming they're based on Broadwell because of having "Core" in its name.

Maybe not an ULP chip, but this definitely belongs in the "Handheld" forum.
 
ToTTenTranz said:
Intel announced the "Core M" line of CPUs for fanless tablets.
Details are scarce, but the chips have a 10W TDP and people are assuming they're based on Broadwell because of having "Core" in its name.

Maybe not an ULP chip, but this definitely belongs in the "Handheld" forum.
Click to expand...

People aren't just assuming. Intel has told everyone it is based on Broadwell. Check their PR/press/news blog thingy here:http://newsroom.intel.com/community...-connected-devices-fuel-next-era-of-computing
Intel said:
The world's first 14nm fanless mobile PC reference design from Intel is a 12.5" screen that is 7.2mm thin with keyboard detached and weighs 670 grams. The innovative design is based on the first of Intel's next-generation 14nm Broadwell processors, purpose-built for 2 in 1s and will be in market later this year. Called the Intel® Core™ M processor, it will deliver the most energy efficient Core processor in the company's history.
Click to expand...
 
Sounds more like a 300MHz frequency but I can't tell for sure with that fillrate.
 
Ailuros said:
Sounds more like a 300MHz frequency but I can't tell for sure with that fillrate.
Click to expand...


That Dell unit is coming up as a dual-core Intel, so should be Merrifield G6400@533Mhz

The following is an Asus bench that has just appeared, which shows as quad-core Intel, so this should be Moorefield, G6430@533Mhz.

http://gfxbench.com/device.jsp?os=Android&api=gl&D=Asus+MeMO+Pad+8+AST21+(K015)&benchmark=gfx30

Theoretically, Moorefield should outperform ipad air by a significant major in graphics tests, given its clock advantage, but these show that thus far it trails badly (still ahead of baytrail graphics mind you).
 
Last edited by a moderator:
The Dell is definitely a Merrifield judging from Alex's comments in the blog and it carries a G6400 with a base frequency of 457MHz and burst clock of 533MHz.

I assume that's also a Merrifield? http://gfxbench.com/device.jsp?benchmark=gfx30&os=Android&api=gl&D=Asus MeMO Pad 8 AST21 (K015)

Either way as Alex says considering for the 6400's the lack of additional FP16 ALUs (and probably also the lack of FB compression) those results are quite normal against the A7/G6430. I guess we haven't seen yet any Moorefield/G6430@533MHz results yet?
 
Ailuros said:
The Dell is definitely a Merrifield judging from Alex's comments in the blog and it carries a G6400 with a base frequency of 457MHz and burst clock of 533MHz.

I assume that's also a Merrifield? http://gfxbench.com/device.jsp?benchmark=gfx30&os=Android&api=gl&D=Asus MeMO Pad 8 AST21 (K015)
Click to expand...
If the glbench CPU info is correct (quad core), then that has to be Moorefield, which is 6430. Clock of 2.33Ghz would also point to Moorefield (Merrifield doesn't go that high)

Either way as Alex says considering for the 6400's the lack of additional FP16 ALUs (and probably also the lack of FB compression) those results are quite normal against the A7/G6430. I guess we haven't seen yet any Moorefield/G6430@533MHz results yet?
Click to expand...

See Above.

That was my question on the blog. I alluded to the above Moorefield G6430 poor relative performance compared to the A7, but his answer was centred around compare G6400 to G6430, and also overall soc design targets.
 
Last edited by a moderator:
tangey said:
If the glbench CPU info is correct (quad core), then that has to be Moorefield, which is 6430. Clock of 2.33Ghz would also point to Moorefield (Merrifield doesn't go that high)
Click to expand...

I stand corrected; then it's most likely a sw or hw implementation issue or a combination of the two. If you notice in T-Rex Moorefield is just a bit ahead of the A7 G6430.

Here I wonder if Apple even has a GPU burst clock like Intel; if not it could be play a minor role also.

See Above.

That was my question on the blog. I alluded to the above Moorefield G6430 poor relative performance compared to the A7, but his answer was centred around compare G6400 to G6430, and also overall soc design targets.
Click to expand...

The A7 G6430 is around 18% ahead in Manhattan then. I just checked and it seems Allwinner A80 devices are using newer drivers (I assume those starting with 29 are newer than 27) then the Intel devices. Apple's devices don't have the same driver numbering scheme but they all start with a 27 too.....:?:
 
Ailuros said:
I stand corrected; then it's most likely a sw or hw implementation issue or a combination of the two. If you notice in T-Rex Moorefield is just a bit ahead of the A7 G6430.

Here I wonder if Apple even has a GPU burst clock like Intel; if not it could be play a minor role also.



The A7 G6430 is around 18% ahead in Manhattan then. I just checked and it seems Allwinner A80 devices are using newer drivers (I assume those starting with 29 are newer than 27) then the Intel devices. Apple's devices don't have the same driver numbering scheme but they all start with a 27 too.....:?:
Click to expand...

The A7 also has a 4MB L4 cache next to the GPU cores which should help it.
As far as I know Moorefield/Merrifield has no equivalent and their theoretical memory bandwidth is otherwise equal.
 
Intel already has a deal with Rockchip. This doesn't make much sense.
 
You must log in or register to reply here.
Back
Top