Samsung Galaxy S series rumours.....

[Laurent06]

I just got my Octa S4, so if there's any questions, for them up.

Having individual scores for Geekbench2 would be interesting (hatter got a global score of 3533).

I'm also wondering about battery life, but I'm afraid that'd be harder for you to test. So just some feeling compared to other phones you have used would be enough

BTW is that GT-I9500 or SHV-E300S?

 

[Nebuchadnezzar]

Having individual scores for Geekbench2 would be interesting (hatter got a global score of 3533).

I'm also wondering about battery life, but I'm afraid that'd be harder for you to test. So just some feeling compared to other phones you have used would be enough

BTW is that GT-I9500 or SHV-E300S?

http://browser.primatelabs.com/geekbench2/1913380

9500.

 

[Ailuros]

I just got my Octa S4, so if there's any questions, for them up.

Not directly related, but after that newsblurb I decided to press the "postpone" buttom

http://www.xbitlabs.com/news/mobile..._of_Galaxy_S4_Flagship_Smartphone_Report.html

 

[Laurent06]

Thanks.

Here is a comparison against Nexus 10: http://browser.primatelabs.com/geekbench2/compare/1913380/1875306

Points to note:
- memory performance (aka Android memory routines) is significantly worse on octacore
- stream is better on octacore
- single thread score is often better on octacore despiste a slightly lower frequency
- the A15 cores are different: the octacore uses r2p3 while nexus 10 is using r0p4 (I find the minor revision numbers odd and wonder if Geekbench reports them correctly).

 

[french toast]

Thanks.

Here is a comparison against Nexus 10: http://browser.primatelabs.com/geekbench2/compare/1913380/1875306

Points to note:
- memory performance (aka Android memory routines) is significantly worse on octacore
- stream is better on octacore
- single thread score is often better on octacore despiste a slightly lower frequency
- the A15 cores are different: the octacore uses r2p3 while nexus 10 is using r0p4 (I find the minor revision numbers odd and wonder if Geekbench reports them correctly).

By memory performance do you mean ram or cache?
There was a rumour flowing around that samsung dropped the lpddr3 speed down a notch...maybe nubu could confirm whether this is just chinese whispers?

 

[Nebuchadnezzar]

Memory runs at 800Mhz DDR.

 

[Laurent06]

By memory performance do you mean ram or cache?

That is RAM.

 

[french toast]

Thanks guys..nubu is that on par with s600 version do you know?

 

[Nebuchadnezzar]

Thanks guys..nubu is that on par with s600 version do you know?

Are you dyslexic?

As per source code the S600 runs at max 633MHz I think but I can't confirm that.

 

[french toast]

Are you dyslexic?

As per source code the S600 runs at max 633MHz I think but I can't confirm that.

Are you for real?? What do you mean by that?

 

[french toast]

Apparently samsung has some new 2gb sticks of Lpddr3.

According to this article they will run at 2133 mbit/s as apposed to I think 1600 at the moment.

The article states it will be 20% more power efficient due to 20nm process..I assume this is over lpddr2 on an older process.

They will be in devices this year according to the article, which speculates will make it into the galaxy note 3 @ 4gb.

I have my doubts.
http://www.phonesreview.co.uk/2013/04/30/galaxy-note-3-or-s5-release-with-4gb-of-ram-possible/

Edit: rumours of note 3 sporting a 5.99 inch screen, 1080p, exynos octa processor with eagle cores clocked @ 2.0 ghz and kingfisher cores @ 1.7ghz.

The screen seems credible, although could well as be guesswork as everyone expects it to he around this size, processor wise im not sure, can A7s clock that high? If they could what would be the benefit?

 

[Nebuchadnezzar]

After screwing around and figuring out what the hell Samsung did here: the phone is running in cluster migration and not core migration.

This is terrible for battery life. I don't know what the hell the reason for it is. I'm currently trying to get the proper IKS driver to run.

 

[Laurent06]

After screwing around and figuring out what the hell Samsung did here: the phone is running in cluster migration and not core migration.

This is terrible for battery life. I don't know what the hell the reason for it is. I'm currently trying to get the proper IKS driver to run.

Hmm doesn't cluster migration include the possibility to power down some of the cores on the active cluster? If so I'd expect that model to be better on battery life at the expense of latency. I probably misunderstood what you wrote

 

[Nebuchadnezzar]

Hmm doesn't cluster migration include the possibility to power down some of the cores on the active cluster? If so I'd expect that model to be better on battery life at the expense of latency. I probably misunderstood what you wrote

It means that at no moment does an A7 run concurrently with an A15, either all 4 A7's run or all 4 A15's run. They might still get powered down due to CPUIdle, but otherwise, no.

 

[wco81]

Are you guys tools developers?

The switching of cores should be transparent to app. developers, right?

 

[Laurent06]

It means that at no moment does an A7 run concurrently with an A15, either all 4 A7's run or all 4 A15's run. They might still get powered down due to CPUIdle, but otherwise, no.

Is it really better to have an A7 up with its L2 cache for small tasks than an extra A15 core that uses an L2 cache already turned on?

 

[Nebuchadnezzar]

Is it really better to have an A7 up with its L2 cache for small tasks than an extra A15 core that uses an L2 cache already turned on?

That depends totally on the granularity of the power gating on the A15 cores. Is it better to have them shoot on during bursts on the lower load cores at high leakage, or to simply have the A7 cores do the jobs without the power "pressure"?

 

[Laurent06]

That depends totally on the granularity of the power gating on the A15 cores. Is it better to have them shoot on during bursts on the lower load cores at high leakage, or to simply have the A7 cores do the jobs without the power "pressure"?

Yes it depends, and that's exactly why I questioned your initial comment about cluster migeation being terrible for battery life I'd like to see real life comparisons between that various use models of multi cluster.

 

[french toast]

Finally I have managed to nail down a comparison video between the two galaxy s4 versions..i9500 & i9505.
https://www.youtube.com/watch?v=Wt5im3WAZYc&feature=youtube_gdata_player

Basically it seems that the octa version is marginally better on batterylife and beats the snapdragon on most benchmarks except graphics.

Although it appears that sgx 544mp4 is able to pull out better frame rates in some current games, as indicated by gl benchmark 1080p egypt...adreno 320 seems to be faster on benchmarks that simulate heavily shader based games that may come out in the future.

660ms on sunspider is very fast..still side to side comparisons between the two in the video show no discernible difference in performace, whether that be app launch or web browser...only the flight game showed a lag on the snapdragon.

More puzzling was the lag shown in the video on both devices going through the touchwhizz UI, I flagged this up some time ago and was shot down because of pre release hardware, here it is clear to see in shipping products.

As it is clearly nothing to do with hardware performance this must be some kind of software bloat on the touchwhizz, any more bells and whistles from samsung and there might be a danger of touchwhizz turning into the laggy bloated mess htc UI carried untill the one x.

Edit: Here is the article URL.
http://pocketnow.com/2013/04/30/galaxy-s-4-octa-vs-galaxy-s-4-quad-video

 

[Turbotab]

Finally I have managed to nail down a comparison video between the two galaxy s4 versions..i9500 & i9505.
https://www.youtube.com/watch?v=Wt5im3WAZYc&feature=youtube_gdata_player

Basically it seems that the octa version is marginally better on batterylife and beats the snapdragon on most benchmarks except graphics.

Although it appears that sgx 544mp4 is able to pull out better frame rates in some current games, as indicated by gl benchmark 1080p egypt...adreno 320 seems to be faster on benchmarks that simulate heavily shader based games that may come out in the future.

660ms on sunspider is very fast..still side to side comparisons between the two in the video show no discernible difference in performace, whether that be app launch or web browser...only the flight game showed a lag on the snapdragon.

More puzzling was the lag shown in the video on both devices going through the touchwhizz UI, I flagged this up some time ago and was shot down because of pre release hardware, here it is clear to see in shipping products.

As it is clearly nothing to do with hardware performance this must be some kind of software bloat on the touchwhizz, any more bells and whistles from samsung and there might be a danger of touchwhizz turning into the laggy bloated mess htc UI carried untill the one x.

Edit: Here is the article URL.
http://pocketnow.com/2013/04/30/galaxy-s-4-octa-vs-galaxy-s-4-quad-video

As I don't currently own a TouchWiz device, I forget can you alter the animation speed in the settings panel a la Vanilla Android? The home button lag, looks like an exaggerated animation speed.