Samsung Exynos 5250 - production starting in Q2 2012

[lanek]

I hardly ever pay attention to Antutu scores since its more than hard to define what they actually stand for, but that link above doesn't point at any Exynos SoCs. What am I missing?

It point to the result of the next Samsung Galaxy Note III, who will use the new Exynos 5.. so it become easy to compare with the actual score of the Galaxy S4 "exynos version". and so get an idea of the performance difference with the new exynos ..
Ofc if this benchmark result are real. ( and i dont think Antutu is really credible, but thats another question ).

 

[french toast]

Im pretty sure nebu mentioned this already, anyway here is some info (confirmation? ) of exynos 5420 supporting full big_LITTLE MP mode.

Thats the the full heterogeneous version. .it can use any number of the 8 cores at anytime to complete tasks, as opposed to the current agricultural cluster migration set up which lets the os use only 4 at any time, all on or all off.
http://www.cnx-software.com/2013/07...log+(CNXSoft+-+Embedded+Software+Development)
(Check comments at bottom)
The exynos 5410 not only used the inferior cluster migration setup, but the CCI 400 coherent bus was broken, meaning everytime the scheduler switched clusters, the cache had to be flushed, this affected both perfomance and batterylife.
This bug has apparently also been fixed.
http://www.anandtech.com/show/7164/samsung-exynos-5-octa-5420-switches-back-to-arm-gpu

I have also read that samsung has made other subtle improvements to the circuitry that enable faster switching between cores, maybe they were referring to the cci bus bug fix?.

I suppose we need google to enable big_LITTLE MP mode in OS before we see the full benefit of ARMs vision.

 

[Ailuros]

Not that I'd support any SoC like the 5410 not functioning as it should, but on the other hand what would we need 8 simultaneous threads for in the SFF mobile market for the time being? (honest question)

 

[french toast]

Not that I'd support any SoC like the 5410 not functioning as it should, but on the other hand what would we need 8 simultaneous threads for in the SFF mobile market for the time being? (honest question)

Totally agree...I said this before 4 main threads is ample, perhaps overkill for android devices if single core performance is up to scratch.

I think the strengh of it is not using all threads at once..although possible of course, but to combine just the right combinations of big and little cores for what ever job, whether that be just 2 A15s and three A7s for loading up heavy web pages and A7s in the background handling processes and keeping everything smooth and snappy.

Of course you like many others on here already know the pros and cons of big_LITTLE MP.
8 cores does seem a little too much in the way of resources to ever be totally usrfull in a smartphone, I remember saying this exact point some time back on here and got canned for it.

 

[Deleted member 13524]

Not that I'd support any SoC like the 5410 not functioning as it should, but on the other hand what would we need 8 simultaneous threads for in the SFF mobile market for the time being? (honest question)

For some spectacular Antutu scores.

 

[newbieforever]

http://www.engadget.com/2013/07/23/samsung-exynos-5-octa-5420-hands-on/


Cant see the glbenchmark score though.

 

[newbieforever]

edit
Looks like 16fps onscreen(2560x1600),nexus 10 scored 8.5fps

 

[Ailuros]

edit
Looks like 16fps onscreen(2560x1600),nexus 10 scored 8.5fps

I saved the image, pulled it through sharpening filters, zoomed in but can't make anything out of it....anyway first scores should always be taken with a grain of salt. T604 have gained quite a bit in the meantime since its initial appearance in the GLB database.

 

[Ailuros]

http://gfxbench.com/device.jsp?benchmark=gfx27&D=Samsung+SHV-E330S+Galaxy+S4

Galaxy S4 (s800)...a smartphone with no heatsink, no fan? blah....

 

[Wishmaster]

http://gfxbench.com/device.jsp?benchmark=gfx27&D=Samsung+SHV-E330S+Galaxy+S4

Galaxy S4 (s800)...a smartphone with no heatsink, no fan? blah....

Those are the scores I expected from 330
50% higher score than fastest 320 in the database
If qualcomms keeps the pace even mobile kepler could be not enough to save tegra

 

[french toast]

Those are the scores I expected from 330
50% higher score than fastest 320 in the database
If qualcomms keeps the pace even mobile kepler could be not enough to save tegra

This.

This adreno 330 is some seriously good bit of kit, if adreno 420 is another major uarch revision...as its nomenclature amd leaked api suggest, then keplar may not be that revolutionary in a smartphone after all.

Still cant wait to see exynos 5420 have a off screen 1080p stab at gfx 2.7 benchmark with decent drivers.

Edit..Now android 4.3 is on the table and fully supporting halti (open gl es 3.0) I wonder when kishonti will release the relevant benchmark for this new api?

 

[kalelovil]

There is a new Samsung-built Nexus 10 successor to be released soon.
http://arstechnica.com/gadgets/2013...xus-10-successor-is-coming-in-the-near-future

Although there are no internal details yet, it would be a logical place for the Exynos 5420 to debut.

 

[Exophase]

Kind of surprise Google isn't giving another company the Nexus 10 design win since they tend to shuffle those, although I guess with so many products now there's a hard limit to how much they can do that..

 

[Alexko]

Kind of surprise Google isn't giving another company the Nexus 10 design win since they tend to shuffle those, although I guess with so many products now there's a hard limit to how much they can do that..

At some point they might just do it in-house, since they've acquired Motorola.

Maybe tablets too.

 

[newbieforever]

Mp6 for note 3 and mp8 for nexus 10 successor maybe

 

[newbieforever]

http://www.nextpowerup.com/news/213...ynos-processors-rival-qualcomm-and-apple.html
Interesting

 

[newbieforever]

Sammobile link here http://www.sammobile.com/2013/07/29...or-to-have-customized-arm-based-architecture/

Maybe we will see this with exynos 6

 

[Deleted member 13524]

Anand took the hint from Nebuchadnezzar and published an article about the selective GPU clocks:

http://www.anandtech.com/show/7187/looking-at-cpugpu-benchmark-optimizations-galaxy-s-4

It seems he also found similar behavior for the CPU cores, as well as the strings referring to the benchmarks in the drivers.

Auch.

 

[ltcommander.data]

Anand took the hint from Nebuchadnezzar and published an article about the selective GPU clocks:

http://www.anandtech.com/show/7187/looking-at-cpugpu-benchmark-optimizations-galaxy-s-4

It seems he also found similar behavior for the CPU cores, as well as the strings referring to the benchmarks in the drivers.

Auch.

Would have been nice to get some battery life measures to see if it makes a difference there. Especially if the A15 cluster is constantly enabled when these benchmarks are running.

 

[french toast]

Anand took the hint from Nebuchadnezzar and published an article about the selective GPU clocks:

http://www.anandtech.com/show/7187/looking-at-cpugpu-benchmark-optimizations-galaxy-s-4

It seems he also found similar behavior for the CPU cores, as well as the strings referring to the benchmarks in the drivers.

Auch.

Yea it sucks, interesting they do the same for snapdragon on the cpu side as well...I wonder whether the adreno clocks are also affected in this way, also a comparison between other manufacturers such as htc, sony, asus etc would be nice, if not unlikely

Yea it was nice to see anand reading these pages...nebu picked that out first, he also picked out the cci 400 fault and cluster migration only mode with exynos 5410, as well as exynos 5420 coming next and enabling full MP/GDC mode.

Maybe anand is among us...