Ars quotes CNET UK saying iPad 2 performance remains mostly unchanged

[sigamy]

Read the article and the comments. Lots of holes in the reporting here and it seems the Cnet UK article has been pulled.

http://arstechnica.com/apple/news/2...costs-remain-largely-unchanged-for-ipad-2.ars

 

[HXGuy]

The article says ...at least as far as Web-browsing performance is concerned.. And this has already been discussed...the iPad 2 seems to be about 1.6x as fast in Javascript tests as the iPad 1 running the same iOS version.

To me, 1.6x is a pretty good increase and I wouldn't consider it "mostly unchanged". If a webpage takes 10 seconds to load on iPad 1, it will now take 6 seconds. If it takes 20 seconds, now it will take around 12 seconds. Pretty big difference, no?

And if you compare it to iPad 1 on 4.2.1 (which is what everyone was running, so it's relative to the end users experience), the iPad 2 is 4x as fast. So that same page that took 10 seconds now takes 2.5 seconds or the 20 second page, 4 seconds now.

 

[No1up]

The article says ...at least as far as Web-browsing performance is concerned.. And this has already been discussed...the iPad 2 seems to be about 1.6x as fast in Javascript tests as the iPad 1 running the same iOS version.

To me, 1.6x is a pretty good increase and I wouldn't consider it "mostly unchanged". If a webpage takes 10 seconds to load on iPad 1, it will now take 6 seconds. If it takes 20 seconds, now it will take around 12 seconds. Pretty big difference, no?

And if you compare it to iPad 1 on 4.2.1 (which is what everyone was running, so it's relative to the end users experience), the iPad 2 is 4x as fast. So that same page that took 10 seconds now takes 2.5 seconds or the 20 second page, 4 seconds now.

Ok ok, wait. Explain this again? I'll try and keep up ---> chemistry student lol

 

[Coukos34]

The article says ...at least as far as Web-browsing performance is concerned.. And this has already been discussed...the iPad 2 seems to be about 1.6x as fast in Javascript tests as the iPad 1 running the same iOS version.

To me, 1.6x is a pretty good increase and I wouldn't consider it "mostly unchanged". If a webpage takes 10 seconds to load on iPad 1, it will now take 6 seconds. If it takes 20 seconds, now it will take around 12 seconds. Pretty big difference, no?

And if you compare it to iPad 1 on 4.2.1 (which is what everyone was running, so it's relative to the end users experience), the iPad 2 is 4x as fast. So that same page that took 10 seconds now takes 2.5 seconds or the 20 second page, 4 seconds now.

They are measuring javascript performance only there. Let's take it easy on equating that to overall web rendering performance. I will be disappointed to find out it is a dual core A8 though. That's kinda weak and hopefully not true

 

[HXGuy]

I will be disappointed to find out it is a dual core A8 though. That's kinda weak and hopefully not true

Again, does that really matter? It's just specs talk again. If the iPad 2 performs well, and a fair amount better than the iPad 1, does it really matter what chip is inside? I don't even know what the difference between the A8 and A9 are, other than I assume the A9 is newer due to the higher number associated with it.

I think these specs are really a mental block for people. If it said "A9" on the chip but it was really an A8 chip, would people feel better about it? The point is how well does the end product work.

 

[Coukos34]

Again, does that really matter? It's just specs talk again. If the iPad 2 performs well, and a fair amount better than the iPad 1, does it really matter what chip is inside? I don't even know what the difference between the A8 and A9 are, other than I assume the A9 is newer due to the higher number associated with it.

I think these specs are really a mental block for people. If it said "A9" on the chip but it was really an A8 chip, would people feel better about it? The point is how well does the end product work.

Yes actually it does. The difference between it being an A8 or A9 will directly effect performance. If you don't know what they are, then it's kind of hard to have an opinion really. Unfortunately the more I think about it as well, it makes sense for it to be a dual core A8. If it was a dual core A9, Steve would have certainly said it was more than 2 times as powerful. It also makes sense given the performance figures comparing javascript performance to the ipad 1. This is kind of sad news actually. I know that all that really matters is the actual performance of the iPad itself (real-world), but that will hinder it's ability to hang in the future. I guess it doesn't really matter that much when the iPad 3 will prob be the one worth waiting for. Just gonna have to sell the iPad 2 next February

 

[admanimal]

I think these specs are really a mental block for people. If it said "A9" on the chip but it was really an A8 chip, would people feel better about it? The point is how well does the end product work.

This is very true. Steve said the iPad 2 is 2x as fast as the iPad 1. Given that he also told us that the iPad 2 is dual core, the most logical conclusion is that each iPad 2 core is equal in speed to the iPad 1 CPU, and any single threaded processing task should take roughly the same amount of time. That is what we have known since Wednesday. The fact that it might be an A8 instead of an A9 is essentially irrelevant once we know the performance characteristics.

 

[Coukos34]

The fact that it might be an A8 instead of an A9 is essentially irrelevant once we know the performance characteristics.

This is true to a degree. Truth is though, if it is a dual core A8, it means it can be up to two times the performance, and real world will be more like 1.5, which correlates to what we have seen so far with comparing JavaScript performance benchmarks. I was hoping for real world performance to actually be double. That is the difference I'm referring to

 

[Pressure]

They are measuring javascript performance only there. Let's take it easy on equating that to overall web rendering performance. I will be disappointed to find out it is a dual core A8 though. That's kinda weak and hopefully not true

This is true to a degree. Truth is though, if it is a dual core A8, it means it can be up to two times the performance, and real world will be more like 1.5, which correlates to what we have seen so far with comparing JavaScript performance benchmarks. I was hoping for real world performance to actually be double. That is the difference I'm referring to

The ARM Cortex-A8 is not available in a dual-core configuration, so the logical explanation would be that it is a Cortex-A9 MPCore (dual-core).

Have in mind that the SunSpider JavaScript test is most likely not multithreaded, which would explain the "low" performance increase.

 

[HXGuy]

This is true to a degree. Truth is though, if it is a dual core A8, it means it can be up to two times the performance, and real world will be more like 1.5, which correlates to what we have seen so far with comparing JavaScript performance benchmarks. I was hoping for real world performance to actually be double. That is the difference I'm referring to

Steve did specifically say "up to 2x" for the CPU speed and "up to 9x" for the video.

 

[sigamy]

The ARM Cortex-A8 is not available in a dual-core configuration, so the logical explanation would be that it is a Cortex-A9 MPCore (dual-core).

Have in mind that the SunSpider JavaScript test is most likely not multithreaded, which would explain the "low" performance increase.

+1. Can't see Apple spending the time and money to create a dual-core A8 variant. But I guess it is possible.

Also agree on the SunSpider point. Seems like a case where benchmarks can be "just a number" and not really meaningful in real world use cases.

I'm looking forward to some good old basic Pepsi Challenge type tests where we put iPad 1 and iPad 2 (and Xoom) side by side and load a number of sites.

 

[HXGuy]

Now I wish I would have ran some baseline tests with my iPad 1 before I shipping it out so I could run the same tests on the iPad 2 once I get it.

 

[Coukos34]

The ARM Cortex-A8 is not available in a dual-core configuration, so the logical explanation would be that it is a Cortex-A9 MPCore (dual-core).

Have in mind that the SunSpider JavaScript test is most likely not multithreaded, which would explain the "low" performance increase.

Hmm didn't know it wasn't available in dual core. That's encouraging. Also your point about JavaScript performance not being multithreaded is also a good point as well. If this were true though, wouldn't Steve be able to say more than "up to two times the performance"? A dual core A9 should be more capable than that, especially in steve's terms (9 times the graphic performance as an example). Couldn't apple have made a dual core configuration? They modified the original A8 to create the Apple A4

 

[admanimal]

From Twitter:

They are Cortex A9s ~1ghz. Apple's ARM license is insufficient to hack an A8 to be multicore or make their own core.

@jrk is a PhD student at MIT who also does work for AMD and Intel.

 

[blackNBUK]

This is true to a degree. Truth is though, if it is a dual core A8, it means it can be up to two times the performance, and real world will be more like 1.5, which correlates to what we have seen so far with comparing JavaScript performance benchmarks. I was hoping for real world performance to actually be double. That is the difference I'm referring to

Sunspider is a single threaded test so I would expect a dual A8 to show little to no improvement over a single A8. A 1.5x improvement seems likely to me to be the result of using the improved Cortex A9 cores.

Expecting a real-world doubling of performance is pretty unrealistic, there simply isn't enough code that really takes advantage of multiple cores. Even on my MacBook, it's really only audio and video encoding that will pin both cores win open.

 

[Coukos34]

Sunspider is a single threaded test so I would expect a dual A8 to show little to no improvement over a single A8. A 1.5x improvement seems likely to me to be the result of using the improved Cortex A9 cores.

Expecting a real-world doubling of performance is pretty unrealistic, there simply isn't enough code that really takes advantage of multiple cores. Even on my MacBook, it's really only audio and video encoding that will pin both cores win open.

I understand that and wasn't saying I expect double the performance necessarily. I was more or less saying that you would think a dual core A9 would potentially (in the ultimate situation) have more than double the performance potential and thus would expect Steve to spin it that way. I know that dual cores does not equal double performance, but with his claim of 9x the graphics performance, I figured the CPU potential would be dramatized in the same manner.
Anyway guys, thanks for the info on the A9. I'm hoping it's true

 

[Branskins]

Those Ars commenters don't seem too happy with that article! Were they a little to quick to use CNet as a source?