Samsung Galaxy S6 hype thread

[lopri]

Exactly. ARMv8 is a significant reason behind Apple's A7 and A8 performing as well as the Snapdragon 800-series chips of their era (in certain cases, better) despite half the cores and lower clock speeds. Going 64-bit earlier than everyone else just happened to be a nice side benefit.

It sounds like the S810 and Exynos 7 Octa will get a nice boost from ARMv8 and pull ahead of the A8, although that's somewhat expected given that they're coming half a year later.

While I share the thrust of your observation, I take issues with the bolded part: I think we can rise above the misleading hype (largely a creation by Apple and select "journalists" like Anand Lal shimpi who is now an Apple employee). Snapdragons series and A-series SOCs are built from different philosophy, to account for different OS'es needs, and by different financial calculations by different corporations.

I agree that Apple's engineers created a wonderful chip in A7, beating Qualcomm with 64-bit SOC first, and highly refined their architecture in A8 to a degree Qualcomm could not match with theirs. But frankly this recurring theme of "Apple can beat XXXX with fewer cores at lower frequencies" meme needs to die and here is why.

When you compare SOCs, you need to look at the size of the chip, both the single-core performance and multi-core performance, how each architecture better suits target OS, and eventually how large the gap is in perf/watt.

Apple went with highly optimized, fat cores with humongous cache (by mobile standard). Why? I think the No.1 reason is that they believe it to be a better solution for the iOS devices. Whether such an SOC will equally benefit Android is an open question, but I have my doubt. Exhibit No.1: NVIDIA's Denver. It smashes every ARM SOC including A8 in single-core performance. Wonderful, I thought, but why aren't NV following up on it? Instead, they are going with big.LITTLE for their next SOC, Tegra X1. User reviews are mixed on Denver's performance on the Nexus 9. Perhaps the nature of Android (Java VMs, multitask-oriented) makes it better suited for a quad-configuration with medium-sized cores, than an SOC with fat dual-cores?

Also, on balance, S800/S805 and A7/A8 are rather well matched. Similarly sized, similar performance. A7/A8 trump S800/S805 in single-thread performance, but S800/S805 hold slight advantage in multi-core performance and GPU performance. Again, this is not that surprising if you consider what their target products were. And If there is an advantage to A8 in perf/watt, more advanced manufacturing process (20nm v. Snapdragon's 28nm) must have a role there, in addition to Apple engineer's excellent design. I am not trying to diminish what Apple achieved here. Not at all. I just think the misconception spearheaded by an Apple employee (Anand Lal Shimpi) need not be repeated in order to appreciate the excellence of A7/A8 SOCs.

Another point I speculate is not an important one for us - users/enthusiasts/consumers - but there may be financial reasons why it makes more sense for Qualcomm to manufacture quads while Apple can take a risk with duals. Apple make A-series SOCs for themselves, and themselves only. Considering the vertically integrated Apple's business model, and considering iOS products' market position (dominating, that is), Apple can perhaps afford to pay for more expensive, low-yield dies to TSMC. Fat dual-cores face bigger chances of defects assuming no harvesting, and as far as I know there is no die-harvested A8 variant chips. Qualcomm, on the other hand, do not sell consumer products - all they sell are chips. So for them it might make more sense to design smaller cores that are modular, in case of potential low yields and following necessity of die-harvesting.

If you take all these into account, the big picture emerges. Apple created an exceptional chip well suited for their own devices. Qualcomm also make exceptional chips that serve multiple segments of the market. Apple's SOCs may be more sophisticated and a step ahead when it comes to 64-bit adoption, but majority of performance metrics of Apple and Qualcomm's chips are where one might expect them to be within the existing technical boundaries. There is no miracle or a "quantum leap" here.

 

[oobydoobydoo]

While I share the thrust of your observation, I take issues with the bolded part: I think we can rise above the misleading hype (largely a creation by Apple and select "journalists" like Anand Lal shimpi who is now an Apple employee). Snapdragons series and A-series SOCs are built from different philosophy, to account for different OS'es needs, and by different financial calculations by different corporations.

I agree that Apple's engineers created a wonderful chip in A7, beating Qualcomm with 64-bit SOC first, and highly refined their architecture in A8 to a degree Qualcomm could not match with theirs. But frankly this recurring theme of "Apple can beat XXXX with fewer cores at lower frequencies" meme needs to die and here is why.

It's not a meme, it's generally accepted as a fact that Apple's dual core SoCs manage to compete with Qualcomm and Samsung's quad and octo-core SoCs. There are many, many benchmarks attesting to this. Unless, I don't think you are arguing that Apple's SoCs aren't dual core?

When you compare SOCs, you need to look at the size of the chip, both the single-core performance and multi-core performance, how each architecture better suits target OS, and eventually how large the gap is in perf/watt.

Apple went with highly optimized, fat cores with humongous cache (by mobile standard). Why? I think the No.1 reason is that they believe it to be a better solution for the iOS devices. Whether such an SOC will equally benefit Android is an open question, but I have my doubt. Exhibit No.1: NVIDIA's Denver. It smashes every ARM SOC including A8 in single-core performance. Wonderful, I thought, but why aren't NV following up on it? Instead, they are going with big.LITTLE for their next SOC, Tegra X1. User reviews are mixed on Denver's performance on the Nexus 9. Perhaps the nature of Android (Java VMs, multitask-oriented) makes it better suited for a quad-configuration with medium-sized cores, than an SOC with fat dual-cores?

iOS and android are both Unix based operating systems. They have the same memory architecture, use the same CPU archictecture, and use much of the same hardware. A chip that runs one OS will run the other just as well.

Why are you telling us special instrucitons on how to compare SoCs? Everyone here has seen the benchmarks, and almost everyone here has an experience with Apple products. You appear to be attempting set a different goal post for nvidia/Qualcomm than Apple. Performance is performance, and the Nexus 9 was universally panned by critics. It was buggy, and it's processor used virtualization and gimmicks to compete with Apple's A8X. They are completely incomparable, it was even recently announced nvidia has decided to drop denver from future products because it was poor in comparison to their vanilly ARM cores. If anything, Denver shows us how hard it really is to create a dual core SoC that can compete. Nvidia couldn't do it.

"When we compare two SoCs" we should also compare the products that are used in, and I think in that there is really no comparison between A8X and Nvidia Denver. One is in tens of millions of excellent devices, the other is sitting in a few hundred thousand devices that are still on shelves, because nobody bought them, because they are awful devices.

Also, on balance, S800/S805 and A7/A8 are rather well matched. Similarly sized, similar performance. A7/A8 trump S800/S805 in single-thread performance, but S800/S805 hold slight advantage in multi-core performance and GPU performance. Again, this is not that surprising if you consider what their target products were. And If there is an advantage to A8 in perf/watt, more advanced manufacturing process (20nm v. Snapdragon's 28nm) must have a role there, in addition to Apple engineer's excellent design. I am not trying to diminish what Apple achieved here. Not at all. I just think the misconception spearheaded by an Apple employee (Anand Lal Shimpi) need not be repeated in order to appreciate the excellence of A7/A8 SOCs.

Another point I speculate is not an important one for us - users/enthusiasts/consumers - but there may be financial reasons why it makes more sense for Qualcomm to manufacture quads while Apple can take a risk with duals. Apple make A-series SOCs for themselves, and themselves only. Considering the vertically integrated Apple's business model, and considering iOS products' market position (dominating, that is), Apple can perhaps afford to pay for more expensive, low-yield dies to TSMC. Fat dual-cores face bigger chances of defects assuming no harvesting, and as far as I know there is no die-harvested A8 variant chips. Qualcomm, on the other hand, do not sell consumer products - all they sell are chips. So for them it might make more sense to design smaller cores that are modular, in case of potential low yields and following necessity of die-harvesting.

If you take all these into account, the big picture emerges. Apple created an exceptional chip well suited for their own devices. Qualcomm also make exceptional chips that serve multiple segments of the market. Apple's SOCs may be more sophisticated and a step ahead when it comes to 64-bit adoption, but majority of performance metrics of Apple and Qualcomm's chips are where one might expect them to be within the existing technical boundaries. There is no miracle or a "quantum leap" here.

I would agree that S805 and A7 are well matched. S800 and A8 is a different story. Qualcomm is offering good performance, but certianly isn't walking all over Apple like they did in 2012 (when it was A6 vs S600). Apple has chosen to focus on single thread for a good while now, and it seems to have paid off in real world performance terms. Their dual core SoCs punch way above their weight. I don't know how much higher ARM cpus can go above 2.7Ghz but seeing as intel is having trouble going much higher, I doubt Qualcomm will do better. They need to start developing higher IPC cores, now.


64 bit is a gimmick in marketing terms but Apple managed to back it up with real performance increases that people falsely attributed to 64bit. Now, in everyone's minds who has used a 5s or iPhone 6 and who doesn't know tech, the reason it is fast is that it is "64bit". So if Qualcomm comes out with a 32bit device, it just looks bad, whether it should or not. Personally I think Qualcomm should look bad, especially given the showing Samsung is giving with its 64bit octocore SoC. Samsung are already beating Apple so why should Qualcomm get a pass?

 

[lopri]

I apologize for posting a wall of text, an off-topic one at that, which prompted an even larger wall of text. I will be brief, and will not make another off-topic post after this.

@oobydoobydoo: I have a couple of questions for you.

1. Why didn't Apple raise the A7's frequency for the iPad Air, so that iPad Air's performance doesn't suffer? One would have thought Apple could have easily raised the clock frequency from mere 1.4 GHz to something like 1.9 GHz, right?

2. Why did Apple add another core for the iPad Air 2, instead of raising clock frequency of the A8? I would have thought the A8 had some headroom for higher clocks after reading evaluations like yours around the web.

 

[npaladin-2000]

Guys, how did Snapdragon vs. Exynos turn into another Apple vs the World thread?

 

[oobydoobydoo]

Guys, how did Snapdragon vs. Exynos turn into another Apple vs the World thread?

I apologize for taking us on a tangent, however in my opinion Samsung is currently beating Apple's offering with this Exynos 7420. It is pertinent to the discussion because it says a lot about Samsung to be able to accomplish this. They are coming back in a big way, and they are showing that they also have engineering prowess that, at least in CPU design, appears equal to Apple's. It will be very interesting to see

If we assume that Qualcomm was already matching or beating Apple, then the 7420 looks pretty ordinary and the S810 looks insanely terrible. I don't think the 7420 is ordinary at all. It's pretty great, imo.

1. Why didn't Apple raise the A7's frequency for the iPad Air, so that iPad Air's performance doesn't suffer? One would have thought Apple could have easily raised the clock frequency from mere 1.4 GHz to something like 1.9 GHz, right?

Raising the frequency was unecessary, the iPad Air already outperformed every other tablet on the market upon release. Raising the frequency also raises the power draw and consequently also raises the heat that must be disappated. One look at the Nexus 9, which runs at a blistering 2.5Ghz (despite official specs stating 2.3Ghz), to see what happens when you must dissappate a lot of heat and use a lot of power to keep frequency up.

2. Why did Apple add another core for the iPad Air 2, instead of raising clock frequency of the A8? I would have thought the A8 had some headroom for higher clocks after reading evaluations like yours around the web.

This is also for power consumption. Lots of cores running in parallel at low frequency are more power efficient than a few cores running at very high frequency. The third core was added, in my opinion, in anticipation of A8X (or its successor) running a full OS requiring multitasking. A dual core SoC is always a compromise.

 

[dawheat]

Let's not even use the word beat. It looks like the Exynos 7420 will be competitive with the A8 in single threaded performance which is very positive.

http://www.sammobile.com/2015/02/02...tations-for-march-1st-hints-at-curved-design/

So invites are out for March 1 - while it implies a curved variant, I'm pretty confident that the standard S6 will not be curved. It feels too early cost wise for a curved display to be the standard version.

Also the renders seem to be consistent in the S6 being slightly taller than the S5 but thinner and narrower. I'm a little surprised the S6 is going to be taller, especially if it looses IP67 certification as some rumors have said. However, another rumor has been for a touch based fingerprint scanner and not the sliding one currently used. So it could be due to a taller home button.

 

[Commodus]

Let's not even use the word beat. It looks like the Exynos 7420 will be competitive with the A8 in single threaded performance which is very positive.

http://www.sammobile.com/2015/02/02...tations-for-march-1st-hints-at-curved-design/

So invites are out for March 1 - while it implies a curved variant, I'm pretty confident that the standard S6 will not be curved. It feels too early cost wise for a curved display to be the standard version.

Also the renders seem to be consistent in the S6 being slightly taller than the S5 but thinner and narrower. I'm a little surprised the S6 is going to be taller, especially if it looses IP67 certification as some rumors have said. However, another rumor has been for a touch based fingerprint scanner and not the sliding one currently used. So it could be due to a taller home button.

Correction: invitations for March 1st are supposedly out. All the stories I've seen tie back to a single post from a well-known Vietnamese site. It could be legit... or could be wishful thinking.

Me, I'm skeptical of a lot of these rumors. Samsung has a reputation for keeping many details close to the vest, and there's a good chunk of poorly sourced leaks out there. The main thing that seems to be creeping up is that Samsung may have ditched Qualcomm (at least temporarily) for the GS6.

 

[Graze]

iOS and android are both Unix based operating systems. They have the same memory architecture, use the same CPU archictecture, and use much of the same hardware. A chip that runs one OS will run the other just as well.

Wrong!
iOS is based on MacOSX which is Unix.
Android is Linux which is NOT Unix based but Unix liked.

Yeah running the OS kernel fine is one thing but what does that have to do with application performance when the two go about it mostly completely differently(unless of course you write your apps for Android NDK)?

 

[lopri]

@oobydoobydoo: I agree with the answers you provided, which makes me think there is not a large gap between our views. I am only opposed to Apple's misleading marketing and some reviewers who want you to believe that A7/A8 are competitive "despite half the cores and lower clock speeds." (see the original quote in my post above) They are trying to give an impression that somehow A7/A8 are competing with fewer than half the resources against competition. From there some people are led to believe that Apple could flip the switch tomorrow, make a super quad-A7/A8 that clock in 2.5 GHz, and cure cancer. (or so I've heard)

Neither is true as you succinctly explained, and A7/A8 are more than competitive because of, not in spite of, the way they are designed. No one made such a claim about the Nexus 10 in comparison to the Nexus 4 two years ago. (Wow! Dual-A15 beats quad-S4 Pro! Amazing!.. NOT) I just wanted to point out the deceptive nature of Apple's marketing. Nothing more, nothing less.

http://www.sammobile.com/2015/02/02...tations-for-march-1st-hints-at-curved-design/

I hot-linked the image.

It is hard to imagine such a curve on a phone at this time. Anyone more creative than me?

 

[dawheat]

I hot-linked the image.

It is hard to imagine such a curve on a phone at this time. Anyone more creative than me?

http://www.phonearena.com/news/Does...lhouette-of-the-Samsung-Galaxy-S-Edge_id65496

Shows one possibility of the edge.

 

[Commodus]

While I share the thrust of your observation, I take issues with the bolded part: I think we can rise above the misleading hype (largely a creation by Apple and select "journalists" like Anand Lal shimpi who is now an Apple employee). Snapdragons series and A-series SOCs are built from different philosophy, to account for different OS'es needs, and by different financial calculations by different corporations.

I agree that Apple's engineers created a wonderful chip in A7, beating Qualcomm with 64-bit SOC first, and highly refined their architecture in A8 to a degree Qualcomm could not match with theirs. But frankly this recurring theme of "Apple can beat XXXX with fewer cores at lower frequencies" meme needs to die and here is why.

When you compare SOCs, you need to look at the size of the chip, both the single-core performance and multi-core performance, how each architecture better suits target OS, and eventually how large the gap is in perf/watt.

*snipped for the sake of focus*

I suppose I agree with you on a core level -- it's not just about having hardware that's theoretically powerful, but how well it works in tandem with the software. I'm not sure that you could say that Apple is getting "small" yields, though. If you look at iPhone and iPad sales, Apple would have needed 95.9 million processors just for this quarter alone. Most of Qualcomm's Snapdragon customers don't need that many chips in an entire year!

Also, I think you're leaning too heavily on the conspiracy theories around Anand's departure. The guy knows a ton about processor architecture and what you need to make a fast CPU; it's more likely that Apple hired him because of that, not because he was saying what the company wanted people to hear. (Remember, he uncovered things Apple rarely wants to publish about its chips, and was kind to competing processors as well.) A lot of the suspicion about his integrity originates in the diehard Android/Windows fan's "no one can ever say anything good about Apple without being paid off" school of thinking.

 

[lopri]

http://www.phonearena.com/news/Does...lhouette-of-the-Samsung-Galaxy-S-Edge_id65496

Shows one possibility of the edge.

Not sure how I feel about that possibility. I did not particularly care for the Note Edge's design, although I have not handled one in person.

 

[oobydoobydoo]

Wrong!
iOS is based on MacOSX which is Unix.
Android is Linux which is NOT Unix based but Unix liked.

Yeah running the OS kernel fine is one thing but what does that have to do with application performance when the two go about it mostly completely differently(unless of course you write your apps for Android NDK)?

Well, since my point was that they are similar, "Unix-like" works perfectly. Here is an excript from the wikipidea article for "History" on the Linux page. Coincidentally, it starts with the words "The Unix operating system...". Must be a coincidence.

http://en.wikipedia.org/wiki/Linux

The Unix operating system was conceived and implemented in 1969 at AT&T's Bell Laboratories in the United States by Ken Thompson, Dennis Ritchie, Douglas McIlroy, and Joe Ossanna.[22] It was first released in 1971, initially written entirely in assembly language as a common practice at the time. Later, in a key pioneering approach in 1973, Unix was re-written in the programming language C by Dennis Ritchie (with exceptions to the kernel and I/O). The availability of an operating system written in a high-level language allowed easier portability to different computer platforms.
With AT&T being required to license the operating system's source code to anyone who asked (due to an earlier antitrust case forbidding them from entering the computer business),[23] Unix grew quickly and became widely adopted by academic institutions and businesses. In 1984, AT&T divested itself of Bell Labs. Free of the legal obligation requiring free licensing, Bell Labs began selling Unix as a proprietary product.
The GNU Project, started in 1983 by Richard Stallman, had the goal of creating a "complete Unix-compatible software system" composed entirely of free software. Work began in 1984.[24] Later, in 1985, Stallman started the Free Software Foundation and wrote the GNU General Public License (GNU GPL) in 1989. By the early 1990s, many of the programs required in an operating system (such as libraries, compilers, text editors, a Unix shell, and a windowing system) were completed, although low-level elements such as device drivers, daemons, and the kernel were stalled and incomplete.[25]
Linus Torvalds has said that if the GNU kernel had been available at the time (1991), he would not have decided to write his own.[26]
Although not released until 1992 due to legal complications, development of 386BSD, from which NetBSD, OpenBSD and FreeBSD descended, predated that of Linux. Linus Torvalds has said that if 386BSD had been available at the time, he probably would not have created Linux.[27]
MINIX, initially released in 1987, is an inexpensive minimal Unix-like operating system, designed for education in computer science, written by Andrew S. Tanenbaum. Starting with version 3 in 2005, MINIX became free and was redesigned for use in embedded systems.

Why people here must make pedantic arguments in support of points that are clearly wrong is beyond me. More than that, you are rude about it.

Obviously application performance will be dependent on a multitude of things, including perhaps GPU and or dsp performance, so you are correct that there will be differences. My point is that they are fundamentally alike and I can think of no reason why an ARM SoC would run Android well, and iOS poorly or vice versa. Clearly you know the history of Linux and should know this stuff, and personally I think you do and are just being difficult because you dislike the idea of acknowledging Apple's achievement in the creation of A7.



Lopri:

@oobydoobydoo: I agree with the answers you provided, which makes me think there is not a large gap between our views. I am only opposed to Apple's misleading marketing and some reviewers who want you to believe that A7/A8 are competitive "despite half the cores and lower clock speeds." (see the original quote in my post above) They are trying to give an impression that somehow A7/A8 are competing with fewer than half the resources against competition. From there some people are led to believe that Apple could flip the switch tomorrow, make a super quad-A7/A8 that clock in 2.5 GHz, and cure cancer. (or so I've heard)

Neither is true as you succinctly explained, and A7/A8 are more than competitive because of, not in spite of, the way they are designed. No one made such a claim about the Nexus 10 in comparison to the Nexus 4 two years ago. (Wow! Dual-A15 beats quad-S4 Pro! Amazing!.. NOT) I just wanted to point out the deceptive nature of Apple's marketing. Nothing more, nothing less.

I also think we mostly agree. One thing I would like to point out, you said that Apple marketing is pitching the idea that A7 performs better "despite the cores and lower clockspeed", but Apple marketing never even mentions that their processors are dual core or what is even inside them. That was all revealed by Anandtech, and correct me if I'm wrong but I've never seen any official Apple marketing saying anything about cores or clocks or samsung. All of that comes from people's subjective impressions of A7 performance (it's good) and their own somewhat misguided ideas about cores and S800.




You're right that Apple isn't somehow "twice as good" as Qualcomm at chip design because their processors run well at low clocks and with low core numbers. Apple have chosen different compromises in the design of their SoC, and both have their strengths. Using big, but fewer cores is one strategy that is also used by Intel, with a lot of success. Samsung and Qualcomm have chosen an equally interesting and promising route of a higher number of smaller cores. But don't blame Apple marketing, they don't even advertise core counts nor will they mention any other competing designs.

Not sure how I feel about that possibility. I did not particularly care for the Note Edge's design, although I have not handled one in person.

I really hope Samsung isn't going to do something as gimmicky as a curved screen, they have much more important ways they can improve their design. They now have an ostensibly cool-running, fast chip that won't be a power hog so they can make the phone thinner and more ergonomic, maybe even more so than the iPhone. Curving the screen just because they can is a really lame way to "innovate", and I think Samsung is capable of much better work than this.

I would like to see a 4.8" 350+ ppi bezelless screen on a 5mm thin metal device that weighs less than 100 grams. That would be an impressive feat of engineering, and if they could make it look good, then even better. I think it's possible.

 

[CakeMonster]

Curved screen increases the size of the phone itself while not increasing the REAL useful screen size. I hope there is a version without it, with a bezel as thin as the S4 (the S5 disappointed with a larger bezel).

 

[Anubis]

if they make the G6 curved like that its gonna fail harder then the S5
I mean id like us to sell more curved glass but I don't think people are gonna buy it

 

[lopri]

If they are doing a curved glass they should do it right. Half-baked look/function of the Note Edge style curved glass won't cut it.

 

[Skurge]

http://www.gsmarena.com/samsung_galaxy_s6_metal_frame_revealed_in_photos-news-11058.php

 

[Oyeve]

http://www.gsmarena.com/samsung_galaxy_s6_metal_frame_revealed_in_photos-news-11058.php

Horrible. Why in the world would samsung WANT to make a iphone 6 lookalike phone? I hate the ip6 look. I think I will stick with my Note 4 and bypass this S6. I usually get every S since the first one but I am not liking the new look if true. Hell, I didnt like the S5 and sold that sucker long ago.

 

[CakeMonster]

3Gb RAM is a disappointment. I doubt it's 5" as the article says. Makes little sense to go down in size from the S5. I'm still mostly waiting for the camera specs.

On a side note, the color schemes have been ugly as sin since the S5. Not exactly my first priority but I don't get it.

 

[lopri]

http://www.gsmarena.com/samsung_galaxy_s6_metal_frame_revealed_in_photos-news-11058.php

I am liking it. I do not think it looks like iPhone 6 at all. If anything I would say it looks more like the iPhone 5 than the 6 due to the flat-milled sides Sony's Xperia Zs. Well, actually it took the good parts of the iPhone 5 and iPhone 6 designs. Middle portion of the sides is flat which means surer grips than the iPhone 6's round sides, yet the edges are round so that they will not feel as sharp in the hand like the iPhone 5's beveled edges. We don't know what the front and the back are going to be like, but I think the ergonomics is heading to a right direction if this is real. One thing I am sad about is the lack of USB Type C connector.

(In what kind of environment was this pic taken..? Is it what I think it is? D: )

And realistically there aren't going to be drastically different looking phones until new materials develop (e.g. flexible) or our hands evolve to become something else in a short period of time.

I mean, at first I thought the iPhone 6 looks like the S5 from the front and the HTC One from the back.

 

[lopri]

It looks more like my Z Ultra's frame, but with wider middle and curves around it instead of chamfered edges.

 

[poofyhairguy]

http://www.gsmarena.com/samsung_galaxy_s6_metal_frame_revealed_in_photos-news-11058.php

Not bad, but I will miss the MicroSD if it is gone.

The new rumor I heard was a VERY light Touchwiz. Like almost stock Android. That plus Exynos and I burn an upgrade.

 

[dawheat]

Not bad, but I will miss the MicroSD if it is gone.

The new rumor I heard was a VERY light Touchwiz. Like almost stock Android. That plus Exynos and I burn an upgrade.

I find the lack of mSD pretty hard to believe considering the all-metal A3 and A5 have a pop-out mSD slot.

 

[WelshBloke]

I find the lack of mSD pretty hard to believe considering the all-metal A3 and A5 have a pop-out mSD slot.

I really, really hope they keep both the swappable battery and the SDcard. I don't want every handset to be functionally the same.

Oh, and keep the hard home key and other capacitive soft buttons.

Looks wise IDRGAS.

 

[lopri]

I think removable battery is a lost cause at this point. Hopefully Samsung will keep it for the Note series. Then again, knowing Samsung, there will be Galaxy S6 Active/Crown/Prime/Swappa/eBay/etcwtfbbq editions so one of them is bound to have removable battery support.

I think the today's leak may as well be real. The metal side frame looks clearly evolved from that of the Note 4. Erase the paint job from the Note 4's frame and add curves, you get the S6's frame.

Plus, I think that the pictures were taken in such an unseemly place is also a strong circumstantial evidence that they are real. Presumably the person had to take some risk sneaking it out to take those pictures. (I would not further speculate where the person might have hidden the frame in transit to the location)