Nebuchadnezzar
Legend
This is not the first time I've heard this, Samsung is apparently pushing 14nm hard to Apple.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Next-Gen iPhone & iPhone Nano Speculation
- Thread starter Arun
- Start date
anexanhume
Veteran
His word also directly contradicts Ben Bajarin's source who says it's dual core.
No doubt Apple would be back if Samsung's foundry has a clear advantage for a future process generation. In the end, it's business.
The 5C feels like a pretty big miss and the 5S alone doesn't address a lot of what the high-end market is looking for. I wonder how long until Apple releases larger sized iPhones.
I got a chance to get a feel for the Galaxy Mega recently. I figured that might be pushing the maximum size I'd want for a phone, but I was surprised to feel that it still left quite a bit of room in the palm of my hand and in my pocket for still-larger sizes. I'd definitely be interested in Apple exploring 6+ inch screen diagonals.
The 5C feels like a pretty big miss and the 5S alone doesn't address a lot of what the high-end market is looking for. I wonder how long until Apple releases larger sized iPhones.
I got a chance to get a feel for the Galaxy Mega recently. I figured that might be pushing the maximum size I'd want for a phone, but I was surprised to feel that it still left quite a bit of room in the palm of my hand and in my pocket for still-larger sizes. I'd definitely be interested in Apple exploring 6+ inch screen diagonals.
french toast
Veteran
The iphone 5s is the very first apple product I would even consider, the large button no has a good use and justifies the large gap dedicated to it at the bottom of the phone.
I still think the top and bottom bezels sre too large for the screen size...but it at least matches and is better than the 3.5inch screen before it..that was ridiculous.
The gold colour option also strikes a cord with me as it makes the phone look like a expensive peice of jewellery or something.
On a side note I hope A7 is a quad core..that will accelerate apps and games that are coded for it as ios is usually the lead platform for development and android gets ports.
I still think the top and bottom bezels sre too large for the screen size...but it at least matches and is better than the 3.5inch screen before it..that was ridiculous.
The gold colour option also strikes a cord with me as it makes the phone look like a expensive peice of jewellery or something.
On a side note I hope A7 is a quad core..that will accelerate apps and games that are coded for it as ios is usually the lead platform for development and android gets ports.
regarding samsung pushing 14nm @ apple, there was a digitimes article on friday which was based on this exact supposition.
http://www.digitimes.com/news/a20130913PD209.html?chid=2
http://www.digitimes.com/news/a20130913PD209.html?chid=2
anexanhume
Veteran
Anandtech's 5S review is up. Dual core CPU, G6430 GPU: http://www.anandtech.com/show/7335/the-iphone-5s-review
Big memory bandwidth improvement and arithmetic performance. L1 is 64/64, L2 still 1 MB with much less latency. Cores are called "Oscar", not Swift.
Big memory bandwidth improvement and arithmetic performance. L1 is 64/64, L2 still 1 MB with much less latency. Cores are called "Oscar", not Swift.
As mentioned, with the A7's GPU identifying itself only under Apple branding to GfxBench, determining the underlying Rogue configuration will be more challenging. While I had been hoping Apple wouldn't continue optimizing so much for area on their iPhone line of SoCs and therefore use a G64xx, and while a low clocked G6430 seems a tempting configuration to consider, the few performance numbers we have so far are, as of now, leading me to believe it's a G62xx (I'll assume a 623x) at around 340 MHz.
Support for or evidence against that theory should come to light when the pixel fill rate numbers are revealed since the surprising consequence of that GPU configuration would be a lower fill than that of the iPhone 5...
Fill rate almost double the iPhone 5
Nebuchadnezzar
Legend
Amazing IPC, big surprise there.. GPU seems to be as expected.
anexanhume
Veteran
Sounds like they really attacked the memory subsystem for improvements just like Qualcomm did in Krait revisions.
french toast
Veteran
Sounds like they really attacked the memory subsystem for improvements just like Qualcomm did in Krait revisions.
Just wow.
Makes you think whether apple did completely redo the whole core, as wide as cortex a15?
A couple more benchmarks
http://reviews.cnet.com/smartphones/apple-iphone-5s/4505-6452_7-35781409-2.html
Last edited by a moderator:
incredible increase in performance over x3.5 improvement in GL2.7 over the iphone5, in fact it's 20% better than the ipad4 !
Its clear from the review that the overall performance improvement is coming from big increases in shader throughput. Triangle low level results are actually down on the A6, but it looks like the existing triangle performance was overkill. The balance seems much better in Rogue.
Fillrate figures however seem weird to me.
If they are accurate, the phone5s fill rate of 3.3 Gp/s is now 65% more than the ipad4's 2.1Gp/s, even though the ipad4 has x4 the pixels ?? It would imply a substantially higher GPU clock rate than many were expecting.
And to be honest, IMG GPUs were already very generous with fillrate compared to the competition.
I'm wondering is there any possibility that the new GPU/driver is confusing the fillrate benchmark into giving false figures, possibly including some effective fillrate ?
Its clear from the review that the overall performance improvement is coming from big increases in shader throughput. Triangle low level results are actually down on the A6, but it looks like the existing triangle performance was overkill. The balance seems much better in Rogue.
Fillrate figures however seem weird to me.
If they are accurate, the phone5s fill rate of 3.3 Gp/s is now 65% more than the ipad4's 2.1Gp/s, even though the ipad4 has x4 the pixels ?? It would imply a substantially higher GPU clock rate than many were expecting.
And to be honest, IMG GPUs were already very generous with fillrate compared to the competition.
I'm wondering is there any possibility that the new GPU/driver is confusing the fillrate benchmark into giving false figures, possibly including some effective fillrate ?
Quite amazing performance increase just going from 32-bit A32 to 64-bit A64!
I guess the move to 64-bit wasn't as useless as some would have thought around here.
Looking at the review, it seems that 32-bit performance has had major overhaul in itself.
Looking at the review, it seems that 32-bit performance has had major overhaul in itself.
Yeah, still impressed with the dual-core A7 holding up to quad-core solutions, e.g. Qualcomm S800 and Intel Z3000 series.
Although you can see it suffers in multithreaded workloads, which still isn't as important in mobile spaces compared to single-threaded performance.
Sounds like they really attacked the memory subsystem for improvements just like Qualcomm did in Krait revisions.
I thought they slayed that beast last year with Swift? The way they are inceasing IPC year over year is ridiculous, reminds of the increases we saw with GPUs in their heyday. How much of that low hanging fruit is there left to pick I wonder.
Incredible advances in IPC over Swift, given similar lithographic nodes and power draw (and seemingly same clocks)!
Sorcery and witchcraft both, indeed. I would never have thought it possible given the constraints. Where will we be in a year and a half on TSMC 16nmFinFET? And if they so choose, on 300mm2 and 35W power budget?
Remarkable.
Sorcery and witchcraft both, indeed. I would never have thought it possible given the constraints. Where will we be in a year and a half on TSMC 16nmFinFET? And if they so choose, on 300mm2 and 35W power budget?
Remarkable.
Each new performance figure has been like a little plot twist in determining the graphics configuration. In the end, the pixel fill rate along with some expected correlations peg this as something like a 425 MHz or 433 MHz G64xx (I always assume a G643x variant due to Apple's willingness to spend extra die for better throughput).
What's a little hard to reconcile is how MediaTek is scoring so well in their benchmarks with only a G6200. Obviously, some of the earlier assumptions made about the MT8135's clock rate were off by a lot. The Anandtech article didn't provide anything really compelling to claim that the 5s is using a four cluster Rogue over a two cluster core, but its pixel fill would require eight TMUs and not four and therefore G64xx by my understanding.
What's a little hard to reconcile is how MediaTek is scoring so well in their benchmarks with only a G6200. Obviously, some of the earlier assumptions made about the MT8135's clock rate were off by a lot. The Anandtech article didn't provide anything really compelling to claim that the 5s is using a four cluster Rogue over a two cluster core, but its pixel fill would require eight TMUs and not four and therefore G64xx by my understanding.
Incredible advances in IPC over Swift, given similar lithographic nodes and power draw (and seemingly same clocks)!
Sorcery and witchcraft both, indeed. I would never have thought it possible given the constraints. Where will we be in a year and a half on TSMC 16nmFinFET? And if they so choose, on 300mm2 and 35W power budget?
Remarkable.
That's the most interesting question, I think. Apart from x86, if [strike]Oscar[/strike] Cyclone can scale up to 2.0GHz or more, there's not much binding Apple to Intel for laptops.
Edit: thanks for the correction.
Last edited by a moderator:
Similar threads
- Replies
- 367
- Views
- 51K
- Replies
- 871
- Views
- 229K
- Replies
- 2
- Views
- 8K