Next-Gen iPhone & iPhone Nano Speculation

[juicytuna]

... You have a message to deliver, you want it clear and without noise, hence why thes specification delivered are well choose. Nothing wrong there.

My contention is that the criteria for choosing those specs includes having to be at least as good as the competition. It's completely logical and no, there is nothing wrong with it.

I stand corrected, it looks like they did mention dual-core afterall. Laughing at the "in volume" part, back then all "in volume" tablets were Apple!

But they couldn't very well make that claim again with quad-core now.

That one's easy. "World's first 64-bit Quad Core ARM CPU". Tell me how that's not a marketers dream.

 

[juicytuna]

delete

 

[anexanhume]

That one's easy. "World's first 64-bit Quad Core ARM CPU". Tell me how that's not a marketers dream.

They'd have to qualify with "shipping". AMD announced 8 and 16 core versions in June. We've known they were doing 64 bit ARM chips since late last year.

 

[juicytuna]

Considering they are first to market I think shipping is implicit in that claim. But whatever, you get my point.

 

[ltcommander.data]

Dual core was one of their big marketing points with A5, especially because the concept covered both the GPU and CPU for that updated SoC.

Still, I wouldn't use the logic of whether they promoted dual core or not last time as any strong evidence for quad core either way this time.

I think GfxBench scores should turn out decent, though not dominating. Two times the graphics performance was likely in relation to FLOPs, and Rogue seems to deliver more real performance per FLOP than SGX. All together, it'll be a high performing SoC for graphics.

At that time RIM was making a big fuss about how amateur hour is over and that the PlayBook will be dual core compared to the iPad. Maybe Apple wanted to rub it in that despite the RIM's bluster Apple was going to ship a dual core tablet before them.

Although Apple mentioned the A5 was dual core, they made no mention of the move to Cortex A9. The improved single threaded performance was probably more noticeable than the move to dual core and was applicable to existing applications. With Cortex A9 the theoretical CPU performance improvement was also greater than the 2x figure they used. So Apple both exaggerates or underestimates specs when it suits them.

It would be good if benchmarks present both 32-bit and 64-bit results to see how much benefit 64-bit actually provides.

I wonder if the M7 is actually a separate chip? The picture they showed didn't have any package markings as the A7 had so it could just be a cartoon to represent the concept of self-contained functionality, but it's still physically part of the A7. However, if they did move the sensor controllers to a dedicated chip there would be more transistors available for new functionality in the A7 than just the doubled transistor count that Schiller mentioned.

 

[anexanhume]

Brian Klug noticed someone uploaded a graphics bench from 5S.

https://twitter.com/nerdtalker/status/377847764300099586
http://gfxbench.com/device.jsp?benchmark=gfx27&D=Apple+iPhone+5S&testgroup=overall

GFXBench 2.5 Egypt HD C24Z16 - Offscreen (1080p) : 56 FPS
GFXBench 2.5 Egypt HD C24Z16 - Onscreen : 53 FPS

Compared to iPhone 5:
GFXBench 2.5 Egypt HD C24Z16 - Offscreen (1080p) : 29.8 FPS
GFXBench 2.5 Egypt HD C24Z16 - Onscreen : 41.1 FPS

He mentions VSYNC for lower onscreen result. There's your 2X claim. 1 FPS less than Adreno 330 in Snapdragon 800.

 

[juicytuna]

I think not mentioning A9 is a branding thing. Cortex A9 is an ARM trademark. No mention of PowerVR either, nor the Qualcomm radios or the Omniviosion/Sony cameras. They like to keep the branding to purely Apple.


That's a great score for that test. Competitive with Snapdragon 800 based phones.

 

[Deleted member 13524]

Brian Klug noticed someone uploaded a graphics bench from 5S.

https://twitter.com/nerdtalker/status/377847764300099586
http://gfxbench.com/device.jsp?benchmark=gfx27&D=Apple+iPhone+5S&testgroup=overall

GFXBench 2.5 Egypt HD C24Z16 - Offscreen (1080p) : 56 FPS
GFXBench 2.5 Egypt HD C24Z16 - Onscreen : 53 FPS

Compared to iPhone 5:
GFXBench 2.5 Egypt HD C24Z16 - Offscreen (1080p) : 29.8 FPS
GFXBench 2.5 Egypt HD C24Z16 - Onscreen : 41.1 FPS

He mentions VSYNC for lower onscreen result. I didn't know iPhone refresh rate was 53 Hz? Anyway, there's your 2X claim. 1 FPS less than Adreno 330 in Snapdragon 800.

No mention of PowerVR or IMG in the driver section. Is this normal?
BTW, OpenGL version is still 2.0.

 

[anexanhume]

No mention of PowerVR or IMG in the driver section. Is this normal?
BTW, OpenGL version is still 2.0.

https://twitter.com/jomaKern/status/377852030528000000

.@nerdtalker Esp as iP5 struggled on 2.7. Definitely PowerVR based, as the GLconfig mentions PVRTC, & the driver numbering is Powervr esque

 

[french toast]

Brian Klug noticed someone uploaded a graphics bench from 5S.

https://twitter.com/nerdtalker/status/377847764300099586
http://gfxbench.com/device.jsp?benchmark=gfx27&D=Apple+iPhone+5S&testgroup=overall

GFXBench 2.5 Egypt HD C24Z16 - Offscreen (1080p) : 56 FPS
GFXBench 2.5 Egypt HD C24Z16 - Onscreen : 53 FPS

Compared to iPhone 5:
GFXBench 2.5 Egypt HD C24Z16 - Offscreen (1080p) : 29.8 FPS
GFXBench 2.5 Egypt HD C24Z16 - Onscreen : 41.1 FPS

He mentions VSYNC for lower onscreen result. There's your 2X claim. 1 FPS less than Adreno 330 in Snapdragon 800.

That depends on the s800bin/implementation. ..galaxy note 3 has been benched at 68fps on gfx 2.5 off screen.

 

[anexanhume]

That depends on the s800bin/implementation. ..galaxy note 3 has been benched at 68fps on gfx 2.5 off screen.

Yeah, different manufacturers and form factors have different thermal profiles too. Still, same ballpark. Imagine when A7X is >2X that, though

 

[tangey]

Of course the really telling one, gl2.7 isn't there. Grrrrrrrr
SGX has been badly outgunned on shader performance. Really keen to see what rogue can do in comparison.

And now it appears we won't get which rogue core it is from the drivers

Double grrrrrrrrr

 

[wco81]

So there seems to be little doubt that A7 has Rogue cores?

 

[ltcommander.data]

http://www.fastcompany.com/3017153/...built-from-scratch-to-show-off-your-iphone-5s

Well here's some fodder for the ever popular "how close is mobile to console gaming" debates:

“Our characters have four times the resolution of what you’d see on current console games,” he said.

...

That in turn has allowed the team to further close the gap between what can be done on a current-generation console, such as an Xbox 360 or PlayStation 3, and what can be done on an Apple device. "We're rendering a full depth of field blur and bloom pass, a color adjust pass, a vignette pass, and a distortion pass--and then anti-aliasing the whole thing while maintaining a blazing frame rate," said Donald Mustard. "On a device that fits in your pocket."

Of course resolution is kind of ambiguous since they could just be referring to screen resolution which on retina iPad's would be a lot higher than consoles without looking better.

They say they've had 40 million downloads of IB and IB2 for $60 million in revenue. I wonder if that compares favourably to F2P games to see whether there will continue to be a focus on premium mobile games to really push these faster GPUs.

 

[Brilliantdeve]

So there seems to be little doubt that A7 has Rogue cores?

I Tried to find any info in the internet seams No one knows yet

is Apple shamed of what they did use with their smartphone or what ?!

 

[tangey]

So there seems to be little doubt that A7 has Rogue cores?

The driver shows it can do PVRTC. That compression is proprietary to IMG
Apple have said its gles3.0 compliant. the only IMG cores that are 3.0 compliant are Rogue ones.

Interestingly, the drivers don't support PVRTC2

 

[Turbotab]

So there seems to be little doubt that A7 has Rogue cores?

As I mentioned in one of my tweets, posted above, it is definitely PowerVR based, because of the mention of PVRTC texture compression. It can't be using a high clocked 543 MP3, as per the GS4, as there is too large a performance gap between that and the 5S GLB 2.5 score.

In theory it could be using a high clocked 543 MP4 or even 554 MP4, but that would be crazy in terms of die area, and would still get its arse kicked in the more complex GLB 2.7 T-Rex tests.

A G6200 is the sensible choice, and it would allow Apple to simply use a G6400 to double the GPU perf of the next iPad, compared to its predecessor.

 

[tangey]

As I mentioned in one of my tweets, posted above, it is definitely PowerVR based, because of the mention of PVRTC texture compression. It can't be using a high clocked 543 MP3, as per the GS4, as there is too large a performance gap between that and the 5S GLB 2.5 score.

In theory it could be using a high clocked 543 MP4 or even 554 MP4, but that would be crazy in terms of die area, and would still get its arse kicked in the more complex GLB 2.7 T-Rex tests.

It can't be SGX, apple specifically cited GLes3.0 compliance in their presentation.

A G6200 is the sensible choice, and it would allow Apple to simply use a G6400 to double the GPU perf of the next iPad, compared to its predecessor.

Could be, or I wonder are they targeting the one soc for iPad as well, and hence have an G6400 running low clock here, with uprated clocks for the GPU and CPU on the iPad.

 

[ltcommander.data]

The driver shows it can do PVRTC. That compression is proprietary to IMG
Apple have said its gles3.0 compliant. the only IMG cores that are 3.0 compliant are Rogue ones.

Interestingly, the drivers don't support PVRTC2

Isn't it possible that Apple has a broader license to PVRTC? They introduced their own proprietary OES extension modifying PVRTC to add sRGB support (GL_APPLE_pvrtc_sRGB) in the early iOS 7 betas. They have since broadened it to GL_EXT_pvrtc_sRGB in the final iOS 7 betas presumably so that IMG can use it in their own drivers.

Maybe it's too early for a ground up custom Apple GPU with licensed PVRTC for backwards compatibility, but it could be an Apple customized Rogue like how Sony customized the SGX543MP4 for the Vita.

 

[tangey]

Isn't it possible that Apple has a broader license to PVRTC?

It's totally possible, I just don't think that is the likely scenario here.

They introduced their own proprietary OES extension modifying PVRTC to add sRGB support (GL_APPLE_pvrtc_sRGB) in the early iOS 7 betas. They have since broadened it to GL_EXT_pvrtc_sRGB in the final iOS 7 betas presumably so that IMG can use it in their own drivers.

That is also possible, but isn't it more likely the other way round. Why would apple give IMG the rights to use an apple designed extension in generic drivers ?More like IMG did it initially for apple, and then made it generic ?

Maybe it's too early for a ground up custom Apple GPU with licensed PVRTC for backwards compatibility, but it could be an Apple customized Rogue like how Sony customized the SGX543MP4 for the Vita.

I haven't seen anything to suggest who did the customisation ?, but maybe it is generally known it was Sony ?

BTW, I assume just about every embodiment of an IMG IP is customised one way or another by IMG to fit the particular needs of the licencee, wasn't the TI version of 544 called RC1 or some such thing ?

We know apple said its opengles3.0. We know it supports PVRTC (as it has to, to be compatible with all the preexisting apps) . The most simple answer is that it is Rogue. There are other possibilities, but they are more contrived IMO.