Galaxy S4 - Watch Thread....

[Grazehell]

I still see custom kernels for the Note 2 with the Exynos 4412, so how does this lack of "documentation" affect anything?

Just like I use to see a lot of ICS roms for my Tegra 2 powered LG G2X, they were complete shit due to not having properly working drivers.

 

[dandragonrage]

Just like I use to see a lot of ICS roms for my Tegra 2 powered LG G2X, they were complete shit due to not having properly working drivers.

I expect Tegra 4 to be best this gen. Apple and Qualcomm are non-starters in the A15 race, and Samsung community support is garbage (after my E4GT, I'm not touching another Exynos CPU, at least not until Samsung proves itself better at community support). Additionally, Nvidia will be the only chip not using crappy Adreno or PowerVR graphics. Both Adreno and PowerVR drivers are absolute junk - Nvidia will easily beat them there. Where are you, TI?! I miss you!

Now hopefully we can get an Android release based on Linux 3.7, 3.8 or 3.9 soon. And don't expect such big changes to work as easily for custom CPUs. Real ARM CPUs will always see the best support - Linux kernel support has recently advanced significantly for ARM CPUs (yes this will help with the custom cores too, but the more custom, the more of the same issues I have with my E4GT - the more that is custom, the fewer software updates you'll get). I try many custom ROMs - it is important to me.

Watch the real A15 chips steamroll Apple and Qualcomm's clones in performance, too. And besides, I'm simply not interested in CPU architectures developed solely because a company didn't want to pay the licensing fee for the REAL next version.

 

[ChronoReverse]

What. The Tegra 2 and Tegra 3 had inferior GPU's to its contemporaries. What makes you think the Tegra 4 will be any different?

 

[dandragonrage]

What. The Tegra 2 and Tegra 3 had inferior GPU's to its contemporaries. What makes you think the Tegra 4 will be any different?

Not sure how the raw GPU performance itself will compare, but it is Geforce, and Nvidia's Geforce drivers in general kill any Adreno/PowerVR crappy driver. Tegra 4 is not the CPU that I want, per se, but it's got real A15 cores and real graphics with what should be significantly better drivers. Hopefully it uses real Geforce drivers. The Adreno/PowerVR drivers are a joke and extremely lacking in features, whereas Nvidia already has relatively good Linux drivers (though they still don't support Wayland, come on Nvidia, implement that buffer sharing API). Hopefully Tegra 4 comes with some real Linux support and not just whatever someone felt like implementing to get Android/iOS working.

Hopefully my next phone will have Firefox OS and perhaps other OSes released for it as well. I'd like to try some of that stuff. Compatibility is important to me.

 

[ChronoReverse]

What on earth are you talking about. Adreno has open sourced drivers. Nvidia has always had only binary blobs.

PowerVR is well understood since it's just a multicore version of the stuff they've been using for a long time now (graphics is nicely parallelizable).


Feature-wise the mobile Geforces have also been behind. They're not even unified shaders!

 

[dandragonrage]

What on earth are you talking about. Adreno has open sourced drivers. Nvidia has always had only binary blobs.

Just because I like Linux doesn't make me a stupid OSS-only Richard Stallman supporter. Regardless of open or closed source, Adreno drivers are junk and Nvidia drivers are decent. I'd like Nouveau or other OSS Nvidia drivers that are great, but I will take decent closed drivers over junk OSS drivers, at least until those Adreno drivers actually get picked up by someone interested in seriously improving them.

PowerVR drivers are extremely problematic, well documented in the case of Intel Atom chips.

Tegra 4 graphics are supposedly much improved and closer to real Geforce graphics than ever. But hey, we'll see. I said I expect it to be the best, not that I know it will be.

 

[ChronoReverse]

No they're not. Tegra 4's Geforce doesn't even support OpenGL 3.0 unlike the Adreno and the PowerVR has sheer compute power behind it.

Nobody in the ARM world whether Apple or Android seems to have trouble with PowerVR so maybe it's just Intel being shitty?

You keep saying that Nvidia has great drivers while Adreno and PowerVR don't but it doesn't add up when they don't give trouble to Android devs.


In any case, we'll find out whenever Tegra 4 is actually released (no indications it'll be soon).

 

[Medion]

Not sure how the raw GPU performance itself will compare, but it is Geforce, and Nvidia's Geforce drivers in general kill any Adreno/PowerVR crappy driver. Tegra 4 is not the CPU that I want, per se, but it's got real A15 cores and real graphics with what should be significantly better drivers. Hopefully it uses real Geforce drivers. The Adreno/PowerVR drivers are a joke and extremely lacking in features, whereas Nvidia already has relatively good Linux drivers (though they still don't support Wayland, come on Nvidia, implement that buffer sharing API). Hopefully Tegra 4 comes with some real Linux support and not just whatever someone felt like implementing to get Android/iOS working.

Hopefully my next phone will have Firefox OS and perhaps other OSes released for it as well. I'd like to try some of that stuff. Compatibility is important to me.

You are still incorrect. The license cost is for the instruction set, not the MPCore. Qualcomm and Apple have paid up on their licenses. They both took different routes on the MPCore. Apple went with Swift, an A15 derivative with tweaks, but performs like an A15. Qualcomm went with an asymetrical design that allowed them to lower costs and battery drain without sacrificing much performance, and allowed them a large head start on the market. Like you, I will take an A15 over Krait. However, you are mistaken on your reasoning.

 

[ChronoReverse]

Incidentally, it's pretty clear that the A6X and Krait will both have less performance per clock compared to A15 but what's not clear is the performance/power.

A Snapdragon 600 at 1.9GHz for instance might not look bad against a 1.6GHz A15.

 

[T4rd]

Just like I use to see a lot of ICS roms for my Tegra 2 powered LG G2X, they were complete shit due to not having properly working drivers.

Well that's certainly not the case with the Note 2. I've used CleanROM and Beanstown's ROMs and they're both awesome. I haven't tried any others because these have all the features I want. So it doesn't seem to be affected by this "documentation" issue.

They often have to pull binary blobs of certain drivers out of official ROMs to get custom ROMs working for Samsung Exynos phones. That's the case with the E4GT I have anyway. Exynos really is bad as far as documentation and code releases go. It's arguably not even legal for these companies to modify the GPL Linux kernel code without releasing the source to it, but that is what most of them do. Samsung GPS stuff especially sucks to work with and is why a lot of custom ROMs for Galaxy phones have awful GPS performance.

TI has the best documentation, but phones don't seem to interest them anymore

Interesting. Well I say again, I haven't had any issues, to include GPS (that I use almost daily) with custom ROMs on my Note 2, but I've only used Beans kernels included with his ROMs and the stock kernel for CleanROM. I actually had a lot of GPS issues though on my previous phone, which was a Galaxy Nexus with a TI OMAP 4460. A lot of times with that phone I had to wipe the phone to a stock ROM, get a GPS lock, then flash the custom ROM in order for GPS to work properly, else it would never lock on to me within like 3000 meters.

 

[ChronoReverse]

Well that's certainly not the case with the Note 2. I've used CleanROM and Beanstown's ROMs and they're both awesome. I haven't tried any others because these have all the features I want. So it doesn't seem to be affected by this "documentation" issue.

It's probably not comparable anyway since Samsung has at least been good about drivers for anything they released an OTA update for.

Likely he was trying ICS ROMs for a phone that didn't have an ICS OTA. It was a huge issue with Tegra 2 since Nvidia dropped the ball with drivers.

 

[dandragonrage]

Well that's certainly not the case with the Note 2. I've used CleanROM and Beanstown's ROMs and they're both awesome. I haven't tried any others because these have all the features I want. So it doesn't seem to be affected by this "documentation" issue.

Gotta find out how the ROMs are being made before you can say that. There are a lot of roms for my SGS2E4GT, but they all sucked at first because of drivers, and like I said before, some of the custom ROMs ended up having to extract certain parts of the binary kernel provided by Samsung to get all features working. Or many of the ROMs just use Samsung's kernel with their own userspace mods. None of that is anywhere near good enough for someone like me who will also be interested in trying Firefox OS, etc. I want a phone with Linux support, not just Android support, and the first step is avoiding as much custom shit as possible, plus avoiding Adreno and PowerVR graphics. And if I can't get non-Android support to try out, I'm still going to go for the best compatibility that I can, and I also disrespect companies that modify Linux without releasing source code, as that is illegal per the GPL (but extremely hard to enforce), so Samsung can take their Exynos chips and shove them up their ....

I did look up more info about the Tegra 4 GPU, it seems it is indeed still a step away from what I was hoping for. Seems worse than I heard it was going to be :/ Though it doesn't seem that lacking in overall performance compared to PowerVR or Adreno. We'll just have to see how it goes. As I've said a few times, my favorite ARM chip producer was TI, and I miss them.

 

[Hotar9978]

If we can take anything from looking at A9 cores then all A15 cores will more or less perform the same CPU wise. The recent Tegra 4 benchmarks just blows the Krait (Both the S4 Pro and the 600) out of water in performance and I'm sure it beats it in power consumption too. The US definitely got the short end of the stick (again).


Also I noticed that there's an extra sensor on the front. Is that for enabling new features such as the pen-less AirView? I would love that on my Note 2 because as things stand right now taking out the pen just to hover over images and such is very counter intuitive. But if that relies on the new sensor then I guess I'm out of luck.

 

[ChronoReverse]

Either way the driver issues have historically been similar regardless of whether it's a Geforce, PowerVR or Adreno. In the end, they all require new drivers from the vendor and none of them have been that great at releasing them on a timely schedule.

The recent Tegra 4 benchmarks just blows the Krait (Both the S4 Pro and the 600) out of water in performance and I'm sure it beats it in power consumption too.

We don't know that yet because it was a tablet Tegra 4 and we haven't seen what the significantly lower clocked A15 in a phone can do. I still suspect it'll be faster but the margin won't be as large I'd also suspect.

 

[Zorachus]

I am a Nexus fan myself. But will admit the new S4 is a sweet phone. I like how its actually smaller than the S3, but with a larger screen, and crazy thin and still packing a large 2600mAh battery.

The S4 design seems like a more mature, and nicer S3, just polished and better looking, but still a Galaxy phone. And at 7.9mm thin, wow

I can't wait to read reviews on battery life, should be really good, with the new efficient hardware. Lets what CM10.1 ROM's do on this phone.

Is tempting for me, but still pretty happy with my Nexus 4 on Straight Talk-AT&T. Will wait for the Nexus-Five before I decide.

 

[Trimlock]

to everyone saying we got shafted (which is opinion right now, since both platforms offer advantages) we don't know which SoC we are getting.

 

[Medion]

No they're not. Tegra 4's Geforce doesn't even support OpenGL 3.0 unlike the Adreno and the PowerVR has sheer compute power behind it.

Nobody in the ARM world whether Apple or Android seems to have trouble with PowerVR so maybe it's just Intel being shitty?

You keep saying that Nvidia has great drivers while Adreno and PowerVR don't but it doesn't add up when they don't give trouble to Android devs.


In any case, we'll find out whenever Tegra 4 is actually released (no indications it'll be soon).

Adreno has been hit or miss, but PowerVR has been the king of mobile GPUs for several years now. His concern seems to be open source drivers...something Nvidia does NOT do, so I'm not sure what his point is. Even the TI OMAP series, his preferred SOC going by an earlier post, was using PowerVR.

All indications are right now that it's going to be PowerVR > Tegra 4 > Adreno, which is a big step up for Nvidia. Their Tegra GPUs have been very underpowered to this point. They're making progress, just not the top of the mountain yet.

Incidentally, it's pretty clear that the A6X and Krait will both have less performance per clock compared to A15 but what's not clear is the performance/power.

A Snapdragon 600 at 1.9GHz for instance might not look bad against a 1.6GHz A15.

A6X has performance per clock similar to A15, actually. At first I thought it was between A9 and A15, but as more information comes out (not to mention benchmarks), it seems to actually BEAT A15 somewhat. Krait is very near A15 in performance per clock, at 3.3 DMIPS/clock versus the 3.5 of Swift and A15. Still, I prefer A15 over Krait as A15 is symmetrical.

 

[MrCrispy]

Nvidia has a consistent track record of under delivering, lies and promising the world. Tegra 2/3 were both disappointing, whereas the S4 and S4 Pro have both been excellent and the reason they are used in pretty much every phone today. I'm not too keen on Tegra 4.

 

[Trimlock]

Nvidia has a consistent track record of under delivering, lies and promising the world. Tegra 2/3 were both disappointing, whereas the S4 and S4 Pro have both been excellent and the reason they are used in pretty much every phone today. I'm not too keen on Tegra 4.

I'd say I won't defend Tegra2/3 at all since it has had areas of let downs but they have filled a segment the S4/pro doesn't fill, which is the super cheap market.

In that regard, the Tegra4 is the first SoC from Nvidia to not have a crippled memory bus and the GPU on it has a pretty extreme core count behind it. Sadly it doesn't use a unified core, which is a pretty big let down with the SoC. If it still fills a super cheap portion of the segment and have the power to back it up, that would be pretty crazy.

 

[Medion]

Nvidia has a consistent track record of under delivering, lies and promising the world. Tegra 2/3 were both disappointing, whereas the S4 and S4 Pro have both been excellent and the reason they are used in pretty much every phone today. I'm not too keen on Tegra 4.

OEMs go for Qualcomm SOCs for the same reason that people shop at Wal-Mart...it has everything cheap. If you get a Qualcomm SOC, you're getting a very good CPU, an average or better GPU, a radio, amplifier (in most of their designs), a digital-analog converter, and many other components. It's a one-stop-shop for the OEM to cheaply and efficiently integrate almost everything needed to run their device.

By comparison, Texas Instruments made awesome media processors with dedicated chips for things that could have just easily been offloaded to the CPU, Nvidia always cheaped out by crippling features, and Samsung had to rely on Qualcomm for radio tech (though that is quickly closing).

Think of it like a custom built PC. Qualcomm's Krait + Adreno versus Samsung's A15 vs. PowerVR is like comparing an AMD APU versus a Core-i5 and a GeForce GTX 660. Which one is better for gaming? The latter. Which one has more raw power? The latter. Which one is more cheaply integrated into the OEM's design? The former. Which one runs Facebook faster? The average user won't ever notice.

I demonstrated this awhile back. If you take a full suite of specific benchmarks (and not Quadrant composite scores and garbage like that), you see that Krait powers past ye olde A9 handily in single-threaded tasks, but Krait + Adreno wasn't leaving A9 + Mali in the dust in GPU-intensive tasks (but did win in the CPU-intensive tasks). Snapdragon is an awesome design because we just don't need/use heavy horsepower on phones like we do on a desktop, and the average user truly won't notice. But take a look at Windows RT tablets, devices meant to be an initial merge between mobile and desktop apps. Look at how many chose Sanpdragon S4, and how many chose older A9 designs. There's a reason they're going for the older A9 designs over Krait. It's not as powerful as people think.

 

[ChronoReverse]

A6X has performance per clock similar to A15, actually. At first I thought it was between A9 and A15, but as more information comes out (not to mention benchmarks), it seems to actually BEAT A15 somewhat. Krait is very near A15 in performance per clock, at 3.3 DMIPS/clock versus the 3.5 of Swift and A15. Still, I prefer A15 over Krait as A15 is symmetrical.

Doesn't the A15 hit 4 DMIPS/clock for some loads? That's why I consider it a tier above Krait/Swift (plus the 600 version of Krait is supposed to be 15% faster which will put it squarely at Swift levels).


As for the Adreno graphics thing you were talking about, you sort of have to differentiate between Adreno 320 and 225. 320 is really fast, 225 isn't. 320 beats out any A9 SoC's GPU short of Apple's. Adreno was anemic before 320 but it's only a few % behind nowadays.

 

[Medion]

Doesn't the A15 hit 4 DMIPS/clock for some loads?

That's been speculative and unproven. I've heard DMIPS numbers between 2.8 and 4.1 on Swift because Apple isn't giving out any info. Nor should they, because ultimately, all that matters is how well the device works. However, benchmarks show that it performs at least on par with where an A15 would be at the same clock speed and similar memory configuration, so 3.5 is a solid estimate, though still unconfirmed.

A15 has been confirmed to be 3.5DMIPS per clock, but certain custom configuration go over 4.0. Perhaps they referred to Swift?

(plus the 600 version of Krait is supposed to be 15% faster which will put it squarely at Swift levels).

It's 15% faster due to a higher clock speed. IPC isn't changed.

s for the Adreno graphics thing you were talking about, you sort of have to differentiate between Adreno 320 and 225.

I already have. The Adreno 320 is just a 225 at a higher clock rate and paired with faster memory. This is the same thing that Samsung did with Mali 400 (SGS2 to SGS3) and PowerVR did with the SGX540 (Nexus S to Galaxy Nexus), only they didn't change the product name.

Also, I posted a thread awhile back detailing VERY specific analysis of Krait and Adreno 320. You participated it in. You did not agree with my assessment, and I respect that. But, GPU SPECIFIC tests showed the Adreno in near lock-step with the Mali-400, and that did not impress me. Tasks that similarly loaded the CPU and GPU put Snadragon ahead of Exynos 4 though. Basically the tests came out like this:

CPU tests - Krait > A9
GPU tests - Adreno 320 = Mali 400
CPU + GPU tests - Snapdragon > Exynos 4

It's not poor form to interpret that data as Snapdragon being superior to Exynos 4 BECAUSE of the CPU and not the GPU. That was my interpretation. However, the majority of those who disagreed with me in that thread (one person in particular) consistently pointed to one benchmark where Snapdragon won (one of the dual-load tests) as evidence of Adreno being superior. The full analysis does not support that.

Qualcomm's strategy has always been to bring their next generation CPU to market before anyone else because it usually isn't as good as the next upcoming competitor. Their new GPU is usually on par with the previous King. They've even admitted this. When the Galaxy S came out in 2010 and the SGX540 absolutely smashed the Adreno 200, Qualcomm stated that their goal was that the upcoming Adreno 220 would be on par with the SGX540 (and the 205 would be a mid-range offering). Unfortunately for them when 220 came out, the SGX540 got overclocked (OMAP 4 chipsets) and would still beat the eventual Adreno 225, coming out a little behind the 320. Qualcomm's new GPU matched the previous King, but could not compete with products shipping alongside it.

But, I'm getting long winded. If you don't agree, I can respect that. All I can do is post my evidence and insight, and let the reader decide what they want to think. My thought process influences my purchase decisions. You should buy what you want.

 

[ChronoReverse]

It's 15% faster due to a higher clock speed. IPC isn't changed.

I'll leave the GPU part alone but I don't think this bit is borne out by benchmarks.

If you look at native CPU benchmarks comparing S4Pro and the 600 both at 1.7GHz (now that both Optimus G Pro and HTC One are available) , you can see that the differences in MIPS and FLOPS are usually in the 10-15% range.

Obviously I'm speculating but it does seem Qualcomm did some core tweaks as well. So a 15% clock speed bump as well as a 15% core improvement.

 

[Medion]

I'll leave the GPU part alone but I don't think this bit is borne out by benchmarks.

If you look at native CPU benchmarks comparing S4Pro and the 600 both at 1.7GHz (now that both Optimus G Pro and HTC One are available) , you can see that the differences in MIPS and FLOPS are usually in the 10-15% range.

Obviously I'm speculating but it does seem Qualcomm did some core tweaks as well. So a 15% clock speed bump as well as a 15% core improvement.

Would you please post a link to those benchmarks so that I can have a look at them? Currently working on another project so don't have time to search.

EDIT: Caught a break and went looking. Anandtech is speculating on a possible performance gain of 15% at he same clock speeds, while adjusting clock speeds upwards of roughly 15%, for a total gain of 30% in performance.

Qualcomm won't give specifics, but states that IPC hasn't changed (it's the same MPcore). The difference is in memory. Krait 200 (S4 Pro) uses LPDDR2, while Krait 300 and Krait 400 (Snapdragon 600 and 800) use LPDDR3-533/800. Those memory gains will give better performance at the same clock speeds. That is what I suspect to be the difference at this point.

If you take a Core i5 with DDR3-1066, and compare the same Core-i5 with DDR3-1600, and the latter came out faster, would you assume that the latter was a faster CPU, or would you attribute it to the other changes in the system? We'll see the same on the GPU side of the equation. While both S4 Pro and 600 both use Adreno 320, the faster memory will boost the GPU performance a bit as well.

 

[ChronoReverse]

Here's the initial benchmarks for the HTC One. The most notable one is CF-Bench since it has native benches for a variety of things (don't use the overall score).

http://rk00064.blogspot.ca/2013/03/htc-one-antutu-quadrant-standard.html

Can't find the 1.7GHz S4Pro at the moment but at least you saw the Anand comment.

That's true, simply alleviating a bandwidth limitation can improve the performance. In any case, the point still stands that the 600 will be getting 3.5 DMIPS/clock. The core is capable of it, perhaps it was just limited by bandwidth.

 

[Medion]

In any case, the point still stands that the 600 will be getting 3.5 DMIPS/clock. The core is capable of it, perhaps it was just limited by bandwidth.

No, it won't. Dhryestone MIPS is a limitation of the MPCore. If IPC was limited by bandwidth before, it is now alleviated. This means that the core was not reaching its theoretical limit of 3.3 (and still might not be). It could have been at 1.8, 2.1, 2.7, etc. But, Krait has a hard limit of 3.3, and that does not change. The core is only capable of 3.3. That is a factual limit. Well, unless you think that Qualcomm is wrong about their own product lineup

 

[ChronoReverse]

Eh. Isn't dhrystone still a synthetic benchmark even though it's so old? Or have they gotten to the point where they just release theoretical DMIPS performance instead?

 

[sethk]

I apologize if this has already been answered, but in the US, do we know which CPU the AT&T version will have, and whether it's generally expected to be easy to root and return to stock Jelly Bean?

 

[ChronoReverse]

The US version will be the Snapdragon 600 version.

Typically, Samsung phones (especially the Snapdragon models) are fairly easy to mod so I wouldn't worry about getting a Cyanogenmod.

 

[Zorachus]

So rumors are the off contract price from U.S. carriers will be $579.

But will wait for reviews first. Really curious on real world battery life with the U.S. versions.

I would guess with the 2600mAh battery life should be really good, but is it a little better than the S3 battery life, or like super better, and near Note 2 battery life ?

 

[onedestinazn2]

I would guess with the 2600mAh battery life should be really good, but is it a little better than the S3 battery life, or like super better, and near Note 2 battery life ?

I would expect it to be a noticeable upgrade considering the S2 -> S3 was only a 250 mAh increase and I noticed a good increase (even with 5" screen) and this would be a 500 mAh from the S3. Purely speculation but if I had to guess I'd say its closer to "super better/Note 2" levels than "a little better".

 

[Zorachus]

I would expect it to be a noticeable upgrade considering the S2 -> S3 was only a 250 mAh increasd I noticed a good increase (even with 5" screen) and this would bbattery mAh from the S3. Purely speculation but if I had to guess I'd say its closer to "super better/Note 2" levels than "a little better".

I'd be happy if the Galaxy S4 is in middle, battery life wise, as my current phone the Nexus 4, which gives me 4-1/2 to 5 hours On Screen time, and overall 13h to 15 h total battery life.

I also had the Note 2 for several weeks, returned due to large size, but great phone, the N2 would give me, typically 7h to 8h On Screen time, and overall 18 to 22h total battery life on workdays.

So I would be looking for the S4 to get like 6h On Screen time, and overall 15h to 17h total battery life. Does that sound doable ?

 

[Ocellaris]

I'd be happy if the Galaxy S4 is in middle, battery life wise, as my current phone the Nexus 4, which gives me 4-1/2 to 5 hours On Screen time, and overall 13h to 15 h total battery life.

I also had the Note 2 for several weeks, returned due to large size, but great phone, the N2 would give me, typically 7h to 8h On Screen time, and overall 18 to 22h total battery life on workdays.

So I would be looking for the S4 to get like 6h On Screen time, and overall 15h to 17h total battery life. Does that sound doable ?

The iPhone 5 can easily do that with a MUCH smaller battery so it should be doable. Otherwise huge /facepalm for Samsung.

 

[sun_bean]

S4 should do better then 6 hrs on screen time.

 

[onedestinazn2]

The iPhone 5 can easily do that with a MUCH smaller battery so it should be doable. Otherwise huge /facepalm for Samsung.

The iPhone 5 and S3 had very close battery life from my usage and general reviews. Yes, the iPhone 5 had a much smaller battery but the screen was also smaller too so I wouldn't "/facepalm" Samsung just on that alone.

 

[Megalith]

I still can't stand the skin. Is it difficult to get stock Android on a Galaxy?

 

[Zorachus]

The Nexus 4 is $349. And rumored off contract price for the Galaxy S4 is $579.

So for costing 65% more, is the S4 a 65% better phone over the N4 ? I would not say so, yes the Galaxy S4 is a better phone than the Nexus 4, no doubt, but spec wise is not that much better, maybe a 1/3rd better phone in my opinion. Based on these noob, simple calculations the S4 should cost like $465

But then we are not factoring in the Nexus-Five, which will be the Nexus equivalent to the S4, the N5 should have equal, or most likely better specs than the S4, and still only cost like $349/$399.

Look at this way, last year in 2012, would you rather pay $649 for the Galaxy S3, or $349 for the Nexus 4 ? I will take the N4 all day long over the S3. Not just due to pricing, but the N4 is just a better phone, not by much, but still little better in my opinion.

So in 2013, would you rather have the S4 @ $579, or the Nexus-Five which most likely won't be over $400.

 

[Adam.C]

I will take the stock android google phone all day long. You are guaranteed to get updates and first at that. That in itself is worth it. Plus stock android just runs better than bloat added on top. IMO . Of course you can root, but it can be a hassle.

 

[brettjrob]

microSD on the N4 and it wouldn't be close, but given the reality of the situation, I'd probably shell out for the S3/S4. Kind of a shame, really.

 

[onedestinazn2]

Some people needed/wanted LTE which wasn't an option on the Nexus 4.

I am actually one of those that enjoy the new Touchwiz. I've loaded stock android all the time and it's just boring other than the fact that you can claim you're a "android purist" imo. I've not notice a performance hit from having the recent touchwiz as oppose to stock. Samsung has been pretty good about updating their phones, though they won't get it before Nexus phones obviously.