Galaxy S4 benchmarks leak

[blackhand1001]

What happened with this Apple paying them thing? Google yields no results of interest.



I wouldn't count on it. He has said that the Adreno 320 is faster a million times even though it only performs better in a single benchmark, being handily destroyed by the A6 in all the others.

It doesn't get destroyed by the a6. It is 1-4 fps behind in the native resolution test as guess what, it has a higher resolution. In the offscreen test which makes the resolutions equal it beats the a6.

And the margin that the s4 pro beats it in cpu tests (twice as fast in many) is much larger than the gpu tests it loses in. The s4 pro also has a much better radio (integrated lte with simultaneous voice and data support on cdma networks on both 3g/lte. The iphone can't do svdo on either. Its the only phone that can't do it on LTE.)

The s4 pro is faster than the sgx543mp4 which is has one more core than the sgx543mp3 and twice the memory bandwidth. The s4 pro also has a much smaller silicon area especially the adreno 320 portion.

 

[DeathChill]

You picked the EXACT ONLY test where the Adreno 320 does *slightly* better than the A6's GPU. Here's the rest of those charts from the same page (that you of course conveniently ignore):

I also think they're testing a reference board deal where there aren't the thermal constraints that a real shipping device (i.e. the iPad 3 they used in their test) have.

That is a comparison to the iPad 3. We're on the 4 now and the performance has improved even further. I have no idea how posting iPad 3 benchmarks relates to the A6.

The iPad 4 is scoring 40 FPS in Egypt HD and 47 FPS in Egypt HD Offscreen. So, the one test where the Adreno 320 does better doesn't even exist anymore because the iPad 4 scores higher. I don't know what to tell you, but you obviously don't want to listen when I've posted all of the benchmarks multiple times showing that you are wrong.

EDIT: Just in case anyone thinks I'm just showing benchmarks where the A6 looks better, look at this page:

http://www.anandtech.com/show/6472/ipad-4-late-2012-review/4

There's not a single benchmark where the A6 (either iPhone 5 or iPad 4) isn't on top in terms of GPU.

 

[Technarchy]

The beat down

 

[blackhand1001]

You picked the EXACT ONLY test where the Adreno 320 does *slightly* better than the A6's GPU. Here's the rest of those charts from the same page (that you of course conveniently ignore):

Image

Image
Image

I also think they're testing a reference board deal where there aren't the thermal constraints that a real shipping device (i.e. the iPad 3 they used in their test) have.

That is a comparison to the iPad 3. We're on the 4 now and the performance has improved even further. I have no idea how posting iPad 3 benchmarks relates to the A6.

The iPad 4 is scoring 40 FPS in Egypt HD and 47 FPS in Egypt HD Offscreen. So, the one test where the Adreno 320 does better doesn't even exist anymore because the iPad 4 scores higher. I don't know what to tell you, but you obviously don't want to listen when I've posted all of the benchmarks multiple times showing that you are wrong.

EDIT: Just in case anyone thinks I'm just showing benchmarks where the A6 looks better, look at this page:

http://www.anandtech.com/show/6472/ipad-4-late-2012-review/4

There's not a single benchmark where the A6 (either iPhone 5 or iPad 4) isn't on top in terms of GPU.

The ipad 4's gpu is not suitable for a phone which is why its not a valid comparison. The one test that combines all these elements into how fast the gpu actually renders is egypt hd and the adreno 320 beats the a6 and a5x in that test.

Look at the optimus g. The nexus 4 during this test had the initial buggy kernel with throttling issues. The optimus g did not.

 

[DeathChill]

The ipad 4's gpu is not suitable for a phone which i why its not a valid comparison.

You posted a comparison using the iPad 3.

 

[Technarchy]

The ipad 4's gpu is not suitable for a phone which is why its not a valid comparison. The one test that combines all these elements into how fast the gpu actually renders is egypt hd and the adreno 320 beats the a6 in that test.

The iPhone 5 A6 smacks around the S4 with ease. The A6X is even more potent .

 

[matttye]

Dont believe anything till we are closer to release of the GS4. It will be fast, that i am sure of but dont really care what the benchmarks are. I care more about the overall size of the device and if it is smooth and fast and has good battery life.

Doesnt have to be the fastest phone going, just fast enough for me.

The S4 will be the fastest Android device, make no mistake for a while anyway!

 

[nick_elt]

Edit. Deleted.

 

[blackhand1001]

The iPhone 5 A6 smacks around the S4 with ease. The A6X is even more potent .

http://browser.primatelabs.com/geekbench2/1653449

http://browser.primatelabs.com/geekbench2/1652987

Not in the cpu department.

On the tablet front the nexus 10 destroys the ipad 4's cpu even worse.

http://browser.primatelabs.com/geekbench2/1475328

A6 Processor floating point performance 1851
Exynos 5 Processor floating point performance 4057

 

[mattopotamus]

I may be wrong, but I always thought comparing benchmarks of something running IOS and something running android kind of made them irrelevant?

 

[DeathChill]

The ipad 4's gpu is not suitable for a phone which is why its not a valid comparison. The one test that combines all these elements into how fast the gpu actually renders is egypt hd and the adreno 320 beats the a6 and a5x in that test.

Image

Look at the optimus g. The nexus 4 during this test had the initial buggy kernel with throttling issues. The optimus g did not.

You also used the results from a tablet to compare against the iPad 3 so I updated it to include the iPad 4.

They fixed the thermal throttling by putting it in the freezer. That even gives it a slight advantage.

You're using a single benchmark to declare the Adreno 320 the winner when it loses badly in every other benchmark.

The Nexus 4 with the S4 Pro also scores below the iPhone 5 in almost every test (both browser and GPU):

http://www.anandtech.com/show/6440/google-nexus-4-review/3

I may be wrong, but I always thought comparing benchmarks of something running IOS and something running android kind of made them irrelevant?

It is for things like Geekbench but I do believe GPU stuff is comparable.

 

[siiip5]

I may be wrong, but I always thought comparing benchmarks of something running IOS and something running android kind of made them irrelevant?

Ironically, we see the Apple camp using and praising benchmark scores on the iPhone 5 against a phone (S3) that is half a year older. Maybe we will see this same courtesy when the S4 comes out (a half a year newer phone than the iPhone) and the S4 blows the doors off the iPhone? Somehow, I doubt it though. I have a feeling benchmarks will somehow be irrelevant again or claims of unfairness because the S4 is newer and shouldn't be compared to the ip5.

 

[matttye]

Ironically, we see the Apple camp using and praising benchmark scores on the iPhone 5 against a phone (S3) that is half a year older. Maybe we will see this same courtesy when the S4 comes out (a half a year newer phone than the iPhone) and the S4 blows the doors off the iPhone? Somehow, I doubt it though. I have a feeling benchmarks will somehow be irrelevant again or claims of unfairness because the S4 is newer and shouldn't be compared to the ip5.

It couldn't possibly be that people have differing opinions on the subject, nay, we shall discuss this at the next Apple fan meeting.

 

[Technarchy]

Ironically, we see the Apple camp using and praising benchmark scores on the iPhone 5 against a phone (S3) that is half a year older. Maybe we will see this same courtesy when the S4 comes out (a half a year newer phone than the iPhone) and the S4 blows the doors off the iPhone? Somehow, I doubt it though. I have a feeling benchmarks will somehow be irrelevant again or claims of unfairness because the S4 is newer and shouldn't be compared to the ip5.

The s4 pro is in the Nexus 4...

 

[Vegastouch]

The S4 will be the fastest Android device, make no mistake for a while anyway!

And thats fine. If it is faster tha my GS3 which is already fast then great. All this posting of benchmarks to show which is faster for catagories that i dont even know what they are for is ridiculous. What the hec is a fill test and a offscreen test mean and why should i care?

Who cares if an iPhone loads a couple seconds earlier than my GS3 or the GS4. The iP5 still doesnt do as much and has a puny screen.

 

[blackhand1001]

And thats fine. If it is faster tha my GS3 which is already fast then great. All this posting of benchmarks to show which is faster for catagories that i dont even know what they are for is ridiculous. What the hec is a fill test and a offscreen test mean and why should i care?

Who cares if an iPhone loads a couple seconds earlier than my GS3 or the GS4. The iP5 still doesnt do as much and has a puny screen.

The s4 pro is faster where it matters anyway which was the actual egypt hd test which is a more real world example than fill rate tests.

In cpu tests it demolished the a6.

 

[siiip5]

The s4 pro is in the Nexus 4...

Seriously? You believe I am referring to a chip? All I can say is, "wow... Just... Wow."

 

[Vegastouch]

The s4 pro is faster where it matters anyway which was the actual egypt hd test which is a more real world example than fill rate tests.

In cpu tests it demolished the a6.

Well thats my point. You(not you literally) can post all kinds of scores to show something being faster but is a fill test worthy? I still dont know what it is.

Im not sure what a egypt hd is either. I just go by how it works when i use it and dont worry about scores. The one score i know that the GS3 does and GS4 will blow the iP5 away is 4.8, 4.99 to 4.

 

[mattopotamus]

Well thats my point. You(not you literally) can post all kinds of scores to show something being faster but is a fill test worthy? I still dont know what it is.

Im not sure what a egypt hd is either. I just go by how it works when i use it and dont worry about scores. The one score i know that the GS3 does and GS4 will blow the iP5 away is 4.8, 4.99 to 4.

Plus you are talking about phones that are already speced better than their useful life. The s3 will be out dated because of software not hardware limitations.

 

[The Game 161]

According to a PDF file from South Korean financial services company Mirae Asset Group, Samsung’s upcoming top-selling device will have a 5-inch 1920x1080p display, Android 4.2 and 2 GB of RAM. It is also stated that it will not feature that super-charged octa-core Exynos 5410 processor we saw at CES. Instead, the Galaxy S4 should have an Exynos 5440 quad-core Cortex A15 processor.

The processor may not have twice the cores compared to the Samsung Galaxy S3, but it does go up to Cortex A15 from Cortex A9. This should still improve performance significantly. Plus, really… do we need an 8-core processor right now? Don’t forget sometimes more cores doesn’t mean better performance, like we saw with the dual-core Snapdragon S4. This SoC outperformed Tegra 3 processors with half the cores.

We are not sure how accurate this information is, so take it with a grain of salt. This South Korean financial company could potentially have close contact with Samsung, though. It is worth to at least keep the leak in mind as we approach any announcements.

http://phandroid.com/2013/02/16/galaxy-s4-rumors/

So it will be 5 inches and not 4.99?

 

[MRU]

The S4 will be the fastest Android device, make no mistake for a while anyway!

If it matches the performance of the A6 gpu in benchmarks it'll be grand, but it does make me laugh a little when people give out about the iPhone being dual core and 'falling behind' android, when actually android is only catching up with gpu performance.

Now OS - yes!, the iPhone has fallen behind - but general benchmarks and performance overall - not really.

I think there are a lot of people who assume 'screen physical size' and 'more cores' equals = more powerful, especially when the bickering of fanboys kick off.


Either way, I wont be buying an S4, I just don't like Samsung's touchwiz or general design ethos.

I'm looking forward to see what HTC do with the M7 and new Sense 5 tomorrow, but have been burned a few times before with HTC's software rollout policy and customer relationship.

But they do always have lovely screens, and that's the one grey area that I'm doubting in my pre-order for the Xperia Z as more reviews come in.

----------

The processor may not have twice the cores compared to the Samsung Galaxy S3, but it does go up to Cortex A15 from Cortex A9. This should still improve performance significantly. Plus, really… do we need an 8-core processor right now?

But it wasn't really 8 core performance. It's more 4 cores at high speed, and 4 cores at less power and lower clock speed that the phone then dynamically utilises to get best performance / battery life dependent on task at hand...

What the benefit of a physical processor that switches to another over a processor that simply step downs clock speed dependant on task is the real question...

 

[siiip5]

What the benefit of a physical processor that switches to another over a processor that simply step downs clock speed dependant on task is the real question...

Heat?

 

[MRU]

Heat?

But doesn't a cpu that steps down also produce less heat ? Hence the reason your macbook doesn't get warm surfing the web, checking email but does when your rendering or compressing a dvd etc....

So i'm not sure whether there would be that much heat benefits either.

I know there must be some benefits, otherwise what's the point in the technology ?

 

[TheHateMachine]

But doesn't a cpu that steps down also produce less heat ? Hence the reason your macbook doesn't get warm surfing the web, checking email but does when your rendering or compressing a dvd etc....

So i'm not sure whether there would be that much heat benefits either.

I know there must be some benefits, otherwise what's the point in the technology ?

Not sure. Maybe these alternate cores are of a different design to allow them to be far more power efficient than regular cores that just get under clocked and undervolted. Maybe they could be utilized as companion cores. Maybe they just wanted to do it differently. I have a feeling it is probably the first scenario I listed. Seems the most likely. Although I vaguely remember reading something about the companion core thing. Cannot find a source though.

 

[blackhand1001]

Not sure. Maybe these alternate cores are of a different design to allow them to be far more power efficient than regular cores that just get under clocked and undervolted. Maybe they could be utilized as companion cores. Maybe they just wanted to do it differently. I have a feeling it is probably the first scenario I listed. Seems the most likely. Although I vaguely remember reading something about the companion core thing. Cannot find a source though.

That's exactly what they are. Cortex a7 which is actually newer than a9 and a8 designed specifically to be lower power than those two and to be used as companion cores.