I think it more points to them adding the 2 extra GPU cores by increasing the die size, and still being on the same process technology hence no increase in any clocks.
Next-Gen iPhone & iPhone Nano Speculation
- Thread starter Arun
- Start date
I think it more points to them adding the 2 extra GPU cores by increasing the die size, and still being on the same process technology hence no increase in any clocks.
ltcommander.data
Regular
Well with iPhone 3GS and iPhone 4S, I think Apple's shown they can make sales with or without superlative names.
Even if Apple didn't put effort in the CPU, I'm surprised that Apple didn't increase the memory bandwidth to benefit the GPU.
french toast
Veteran
So its still @ 6.4gb/s?..thought they would hook up some lpddr3 to that badass
D
Deleted member 13524
Guest
ipad 3 is getting ~16000 frames in Egypt, versus 10200 in ipad 2 and 7750 in Transformer Prime (Tegra 3).
<60% performance bump for twice the 5XT "cores". Is it getting strangled on low bandwidth? Lower GPU clocks? CPU bottleneck? Adding 5XT cores doesn't increase the performance as much as one would think?
<60% performance bump for twice the 5XT "cores". Is it getting strangled on low bandwidth? Lower GPU clocks? CPU bottleneck? Adding 5XT cores doesn't increase the performance as much as one would think?
Let's face it, 99.9% of people who will buy the iPad3 won't have a clue what CPU/GPU/clock speeds etc etc it has. In fact most of them won't know what CPUs/GPUs/clock speeds are!
As long as it looks good, works well and performs well, that's all that matters.
Apple's advantage is that they have created the modern Smartphone/Tablet market and, as such, they have a huge amount of the mindshare. Android devices are still playing catch-up to Apple devices (in the tablet space especially) and the competing manufacturers have to play the specs game to try and win the sales. For most consumers an iPad is the default tablet - heck, my 8 year old niece is getting an iPad2 from her school shortly (well, leasing one) and they use them and the iPod touch in class. The parents are willing to spend the money on these devices for the kids, but would the situation be the same if, instead of iPads, the school was trying to get the kids to buy/lease Android tablets? I'm guessing that a lot more of the parents would complain about the expenditure in this case.
As long as it looks good, works well and performs well, that's all that matters.
Apple's advantage is that they have created the modern Smartphone/Tablet market and, as such, they have a huge amount of the mindshare. Android devices are still playing catch-up to Apple devices (in the tablet space especially) and the competing manufacturers have to play the specs game to try and win the sales. For most consumers an iPad is the default tablet - heck, my 8 year old niece is getting an iPad2 from her school shortly (well, leasing one) and they use them and the iPod touch in class. The parents are willing to spend the money on these devices for the kids, but would the situation be the same if, instead of iPads, the school was trying to get the kids to buy/lease Android tablets? I'm guessing that a lot more of the parents would complain about the expenditure in this case.
D
Deleted member 13524
Guest
So this is now a topic about replaying apple's marketing mumbo jumbo?
And here I thought this was a topic for the 0.01% of the people who are interested in discussing the technical specs...
Silly me.
A quick look at the lower level results shows that virtually all of them have approximately doubled, including the raw fill rate up from 1GPixels to 2Gpixels, signalling that the clock is unchanged and the x2 cores is giving virtually x2 raw peformance (IMG has indicated >95% linearity).
So it looks to me like its something in the design of the Soc other than the raw GPU performance that's causing the less than x2 increase, possibly bandwidth ?
Also is there ANY CPU element in these tests at all ?
He has a valid point. The Android have no other strategy left than touting specs to differentiate from iPad.
Google talked up Gingerbread and then ICS but the hardware vendors have to show why their brand is better than the other brand running the same software. So some will have USB ports, others will tout quad-cores or whatever -- and even then, the user experience isn't any smoother.
They're not going to market to the .01% because they're not touting specs for specs sake, more the promise of those specs. yielding a better experience.
Apple can fight the specs. game as well as any other company in this space, since they got the supply-chain advantage. But rather than waiting a few months to produce a better SOC, they decided to ship now because of economic reasons.
Google talked up Gingerbread and then ICS but the hardware vendors have to show why their brand is better than the other brand running the same software. So some will have USB ports, others will tout quad-cores or whatever -- and even then, the user experience isn't any smoother.
They're not going to market to the .01% because they're not touting specs for specs sake, more the promise of those specs. yielding a better experience.
Apple can fight the specs. game as well as any other company in this space, since they got the supply-chain advantage. But rather than waiting a few months to produce a better SOC, they decided to ship now because of economic reasons.
french toast
Veteran
Which kind of backs up what i was originally calling for..dual channel lpddr3 @ 12.8gb/s..maybe we will soon be needing dedicated graphics memory?? ala PS VITA...
It almost seems a bit absurd that they would hamper the performance like that after spending the die area?
D
Deleted member 13524
Guest
Maybe the scenario changes for real world results?
Seems like the most probable factor to me too..
AFAIK, nope.
GLBenchmark 1.1 had two CPU tests, but no one has ever tested it with an ipad 3 or an iphone 4s.
ltcommander.data
Regular
Is LPDDR3 even ratified? Moving to LPDDR2-1066 would have been more reasonable.
french toast
Veteran
Yes it is, Tegra 3 supports it, but hasn't shipped with it yet, lpddr3@1500mhz effective.
I think there is the Transformer prime infinity or something like that will be the first to use it..single channel mind.
On another note, benchmarks are up for the one x and looks like the Tegra 3 gpu has been downclocked for power consumption...suddenly s4 is looking alot more interesting for a smartphone
http://www.glbenchmark.com/compare.jsp
Agreed..im very surprised Apple's stayed with LPDDR2-800 and didnt go for at least LPDDR2-1066. In fact i was expecting LPDDR3 but maybe its too early for that. The next iphone will probably be on LPDDR3 for power reasons.
Yes it is, Tegra 3 supports it, but hasn't shipped with it yet, lpddr3@1500mhz effective.
I think there is the Transformer prime infinity or something like that will be the first to use it..single channel mind.
On another note, benchmarks are up for the one x and looks like the Tegra 3 gpu has been downclocked for power consumption...suddenly s4 is looking alot more interesting for a smartphone
http://www.glbenchmark.com/compare.jsp
AFAIK, Tegra 3 does not support LPDDR3, it supports DDR3L which is low power DDR3 and not a variant of LPDDR.
ltcommander.data
Regular
That's DDR3L. Low-voltage desktop/server memory. LPDDRx is specifically designed for mobile embedded applications. For example, LPDDR2 operates at 1.2V while DDR3L operates at 1.35V. LPDDR2 also includes other power savings features. Apple may need more bandwidth but they aren't likely to sacrifice battery life to do it.Yes it is, Tegra 3 supports it, but hasn't shipped with it yet, lpddr3@1500mhz effective.
I think there is the Transformer prime infinity or something like that will be the first to use it..single channel mind.
On another note, benchmarks are up for the one x and looks like the Tegra 3 gpu has been downclocked for power consumption...suddenly s4 is looking alot more interesting for a smartphone
http://www.glbenchmark.com/compare.jsp
EDIT: Looks like Erinyes beat me to it.
The proportion of useful returns on performance compared to the power that's consumed from the CPU starts to nosedive when the clock is pushed too high, so Apple is just trying to stay in the sweet spot as far as frequency goes, whether A5X ended up on 45nm or something better.
Sony must've also profiled that performance:consumption relationship because they chose some conservative CPU clocks for the Vita SoC, too. Of course, in both Apple's and Sony's case, they have the benefit of not having to sell their SoCs to the open market and depend on marketing MHz.
Sony must've also profiled that performance:consumption relationship because they chose some conservative CPU clocks for the Vita SoC, too. Of course, in both Apple's and Sony's case, they have the benefit of not having to sell their SoCs to the open market and depend on marketing MHz.
No need to get your knickers in a twist. I was merely making a comment which seems pretty valid to me.
The A5X is unexciting, as expected. The iPad3 (sorry, just iPad) has the groovy new high-res display and LTE, but little else of interest to techie sorts.
Believe me, I am the 0.1% (you were an order of magnitude out!
) who is interested in the hardware and the reasoning behind the design decisions. Why else would I be posting in a thread about Apple devices when I've never owned any of their hardware or used any of their software, but for Quicktime? (NeXTSTEP doesn't count) The unchanged CPU clocks and likely unchanged process node seems to point to the fact that 6-9 months ago Apple wasn't sure if Samsung was going to have 32nm ready, so they went ahead with another 45nm design just to make sure that there wouldn't be any delay from making CPUs. Plus, with the extra LEDs needed for the Retina display and the "need" for LTE, they just threw in a larger battery to make the battery life problem go away.
The last time there was a process node transition in 2010 to 45nm, Apple could have released the iPad at any time when they had enough SoCs ready for production, but they didn't have such a luxury with the new iPad.
We all know if they had waited 1-2 months, they could have gotten a good supply of 32nm chips from Samsung, bumped the CPU clocks to 1.2-1.5Ghz, used faster memory, gotten access to the lower power MDM9615 LTE baseband, and not had to make the iPad any thicker. But, what can you do, the rumors from the "delay" and lack of an "iPhone 5" last year were so insane that they likely do not want repeat that experience again, even if it didn't cost them one penny in the end.
The last time there was a process node transition in 2010 to 45nm, Apple could have released the iPad at any time when they had enough SoCs ready for production, but they didn't have such a luxury with the new iPad.
We all know if they had waited 1-2 months, they could have gotten a good supply of 32nm chips from Samsung, bumped the CPU clocks to 1.2-1.5Ghz, used faster memory, gotten access to the lower power MDM9615 LTE baseband, and not had to make the iPad any thicker. But, what can you do, the rumors from the "delay" and lack of an "iPhone 5" last year were so insane that they likely do not want repeat that experience again, even if it didn't cost them one penny in the end.
What I do hope is that the greatly increased battery size is something which will stay on for future devices, even if a lower-power LTE chipset is used. An extra 10% weight isn't a big problem, IMHO, compared to the benefits of increased battery life. The question is, will the hardware vendors agree or will they prefer to reduce battery size to cut their costs once again?
More importantly, I'd like to see phones with larger batteries. As the Droid Razr Maxx (already discussed elsewhere in this forum) shows, the weight and bulk of the phone doesn't need to change a great deal to greatly improve battery life. To tell the truth, I was surprised that none of the phones announced at MWC had batteries of a similar size. One or two have a battery around 2000mAh in size, but none are close to the Razr Maxx with its 3300mAh.
It will be interesting to see what the Galaxy S III brings to the table.
More importantly, I'd like to see phones with larger batteries. As the Droid Razr Maxx (already discussed elsewhere in this forum) shows, the weight and bulk of the phone doesn't need to change a great deal to greatly improve battery life. To tell the truth, I was surprised that none of the phones announced at MWC had batteries of a similar size. One or two have a battery around 2000mAh in size, but none are close to the Razr Maxx with its 3300mAh.
It will be interesting to see what the Galaxy S III brings to the table.
french toast
Veteran
It seems i have got a bit mixed up on the memory nomenclature 
it is DDR3L and not lpddr3...yes so your correct the best they could have done is lpddr2 1066 what would the bandwidth calculation be for that set up? an example of the math would be much appreciated
it is DDR3L and not lpddr3...yes so your correct the best they could have done is lpddr2 1066 what would the bandwidth calculation be for that set up? an example of the math would be much appreciated