Apple iPad announced
- Thread starter Dresden
- Start date
I found the CPU benchmarks interesting as well
iPad Float : 3968
iPad Int : 26420
iPhone Float : 2352
iPhone Int : 9751
So the fill rate benchmarks are not the only ones that show higher results than the "overclocked iPhone chip" assertion would imply.
Last edited by a moderator:
your iphone numbers are the wrong way round.
iphone Int=9751
iphone float=2352
processor clock has risen from 600MHz to 1Ghz (x1.667)
ipad float has risen from 2352 to 3968 (x1.687)
ipad int has risen from 9751 to 26420 (x2.71)
Float has risen exactly in line with clock.
int has risen by about twice that rate
Could the int increase have anything to do with the supposedly doubling of the memory bus width from 32 to 64 bits ? Would that also go to some way to explain the increased fillrates ? That is the only other known (suggested) difference between iphone and ipad APs.
source:
http://www.embedded.com/underthehood/224201433
That'd be one heck of an ALU test if it did not operate tightly from cache.
Yeah, but it real life you will end up sampling from larger textures than you would do on the 3gs which will make things even worse ( <2x performance, 5x as many pixels to fill)
Thanks for pointing out my editing error!
Increased bandwidth obviously helps realize the increased arithmetic capabilities of the SoC, but when the integer capabilities increases by a factor of 2.7, we are looking at more than a just the clock differences (and the just over factor of two pixel writing capabilities are a bit surprising as well in that context).
Yes, but those tests are written to evaluate the speed of the GPU internals. One would assume that the screen resolution doesn't have an impact on those, except for the second order effect of slightly reducing memory bandwidth because the display unit has to work harder?
I just find it curious the fill rate has doubled (because of a double width memory bus?) on the iPad but that for a number of texturing tests, the iPhone 3GS is quite a bit faster.
The problem is the iphone allows use of this comprimised data. The only real reason apple doesn't put sd ports or usb ports for expandibilty is because its better for them to out data a device due to storage capacity then by new features. ITs also an excuse to charge you an arm an a leg for the same device. Do you really think going from 16 gigs of flash to 64 gigsof flash warrents a $200 price hike. The only reason apple gets away with it is because theycripple your ability to increase its cpacity.
If the average iphone/ipad user can get torrented content or has the ability to hand off a flash drive / memory card then its not hard for that drive to a.so have the jail broken software or to have software to copy out and strip the drm from the i whatever device.
Once again the only reason there are no sd or usb ports is to force users into an upgrade path and to over charge for small flash upgrades.
And they will still be hacked. Jailbroken iphones are as easy as downloading torrents. NOthing has changed.
No flashy advertising is king. User experience and content is to keep people content on the platform. You get people in by the flash of the gee wiz adverts.
Apple has a good scam going that I'm sure all other comanys would love to get in on. For years they were able to keep people ocming back to the ipods because of storage increases. Then they got to such huge capacity amounts no one needed to upgrade after all how many people need 200 gigs of space on a mp3 player with a crappy screen. Then they were able to start again with storage on thier mini /nano/ shuffle devices. Then they moved to the touch and capcity started out so small again that many had to upgrade as they could to put thier collections back on a single device.
Apple will do the same with the ipad. People might say 16 gigs is alot. But when your consuming music , video and apps with the internet included 16 gigs will go very very fast.
My hope is that a more open platform will actually compete with Apple so that it will drive prices down on both the devices and the content.
Good luck with that. JooJoo and tablets from other startups will excite a few geeks but that's about it.
Even a better known Chinese company like HTC can't offer the package of content and services that Apple can. And for service and support, you have to go with Apple, Dell, HP and MS when it gets in the game.
Yea.I was hoping windows mobile 7 would find a good life on a tablet. Hopefully MS can get a good ecco system set up. I much rather be stuck with MS as my evil over lord than apple. MS has the xbox and I already have alot of content from that. Apple doesn't have a real gaming experiance.
Vsync is always enabled on iPad/iPhone. Even though the devices are not reaching 60 fps (which would be equal to 9.81 Mtri/s) at which point the tests would be entirely vsync limited, vsync already has an effect at lower framerates. Many of the geometry tests are running at either 30 fps (4.9 Mtri/s) or 40 fps (6.54 Mtri/s) on iPad/iPhone.
Larger textures for a higher resolution generally do not make things worse as mipmapping should select the right level of detail for 1:1 pixel to texel mapping.
Don't forget the cycles spent in filling the screen on unified shader GPUs. The triangle tests are drawing 163456 triangles per frame (and even less vertices), 1024x768 is 786432 pixels per frame. Fragment processing actually has a significant impact in these tests.
The "texturing tests" you mention aren't actually texturing tests. They measure geometry throughput with texturing enabled.
True but even assuming 1x1 mapping sampling from a 64x64 texture (say on the 3gs) seems to be faster than sampling from a larger texture.
In any case, I have a somewhat related question ... when rendering to a rtt target on the iPhone , it seems that there is a performance hit related to the size of the rttt target regarldess of the actual texel coverage.
What I do is something like this:
Case 1
1. Render a 64x64 quad to a 64x64 rtt target.
2. Map the resulting texture to a bunch of quads rendered to the framebuffer.
Case 2
1. Render a 64x64 quad to a 256x256 rtt target ( 1x1 mapping – using only a 64x64 subsection of the 256x256 target)
2. Again, map a subsection (64x64) of the 256x256 texture to a bunch of quads rendered to the framebuffer.
The second case is significantly slower … I am wondering if this is related to my framebuffer targeted geometry sampling from a larger texture (256x256 vs 64x64) ?
I think in the future the 3GS and iPod touch 3rd gen 3D performance will be known as an anomaly. There are clear indications that the next gen iPhone/iPod touch will get a higher resolution display. Combined with a (slower) version of the A4 this should normalize the 3D performance across all iPhone OS devices within an generation (e.g. 800MHz A4 with a 960x640 display on the iPhone/iPod touch vs. 1GHz A4 with 1024x768 on the iPad).
Yet its all you can eat on the zune store for as low as $13 a month plus you can keep 10 songs a month forever.
Once xbox live fully intergrates with win mo 7 and zune things will change very quickly
Zune music subscription SOUNDS good (at least to me), but so far it's a flop (at least compared to iTunes and even Amazon). And since Apple allows 3rd party music subscription services (spotify etc.) in the App Store, combined with the new 3rd party background audio ability in iPhoneOS 4, the only real advantage Zune on WP7 will have by year's end (WP7 release), is being preinstalled and maybe a little bit cheaper than some other services.
How laggy is the process of controlling your desktop within 20 meters radius ? Is it like using wireless controller good (like wireless mouse) or is it MMORPG bad (like poor framerate and everything becomes jumpy and hard to work with) ?
Also if you match your desktop resolution to the iPad's, you shouldn't need to scroll around right ?
Do stylus work on iPad screen or do you actually need finger ? I want to be able to sketch out design and stuff so probably be using stylus instead of finger.
I read this from his report
Here he meant switching between apps in iPad right ? I should be able to multitask when connected to my desktop right ? I mean I can run whatever applications my workstation can handle right ?
Basically I shouldn't care about the specs of iPad or it's apps since I am just using it as a wireless touchscreen right ?
Anyone know what other options I should be looking into beside iPad ? I wish Wacom would make wireless Cintiq already but that could be another year or two away.
Should be like wireless controller good.
Yep. I think the iPad is 1024x768, so that should still be a workable display size for most cases.
You'll use the finger by default, but there are special stylusses available (they have been available for the iPhone/iPod Touch for quite a while as well).
Like this one:
http://tenonedesign.com/sketch.php
I thought you might lose pressure sensitivity, but apparently even that is still an option. Though I don't know if that would transfer over remote access, that might be a problem.
Correct.
Similar threads
