Pandora Cpu Question


mali said:
I ask myself, how many of you that ask questions about the horsepower of the Pandora and compare it to x86 desktop, have ever used a handheld of a similar category. The day you start to use it, you will change your mind and think differently. It is mindblowing to see a small device like this doing computational tasks without vibrating, heating, noising and gasping for more juice every 2 hours.
You ask yourself, how is that possible, that can't be, where is the cord to the mains?
ARM tries to keep the same power footprint while raising performance, not the other way around like others do.
You won't be able to do some things you do on a desktop, but the things you can do, you do with a grin on your face ;)
Surely those who have seen the iPhone won't be as impressed? After all, the iPhone does a lot of that without "vibrating, heating, noising and gasping for more juice every 2 hours".
 
Last edited by a moderator:
icurafu said:
And the nvidia score was obtained using TnL to repeaqt the same draw triangle. PowerVR could do they same thing if they didn't have an ounce of integrity.
Thanks, I was wondering why it was so high.
 
Last edited by a moderator:
mazza558 said:
Surely those who have seen the iPhone won't be as impressed? After all, the iPhone does a lot of that without "vibrating, heating, noising and gasping for more juice every 2 hours".
If you'd put an appropriate open operating system on the Iphone, better controls and a higher resolution screen, you could come closer at least ;)
 
Last edited by a moderator:
Either way, the same point is proven, because the Iphone uses ARM and not x86. Don't know about the whole power thing impressing the average Iphone user, but do we really want to do that? :D
 
mazza558 said:
mali said:
I ask myself, how many of you that ask questions about the horsepower of the Pandora and compare it to x86 desktop, have ever used a handheld of a similar category. The day you start to use it, you will change your mind and think differently. It is mindblowing to see a small device like this doing computational tasks without vibrating, heating, noising and gasping for more juice every 2 hours.
You ask yourself, how is that possible, that can't be, where is the cord to the mains?
ARM tries to keep the same power footprint while raising performance, not the other way around like others do.
You won't be able to do some things you do on a desktop, but the things you can do, you do with a grin on your face ;)
Surely those who have seen the iPhone won't be as impressed? After all, the iPhone does a lot of that without "vibrating, heating, noising and gasping for more juice every 2 hours".

The iphone is a handheld of a similar category ...
I've been using a zaurus SL-C3100 and I can't wait to get my hands on a pandora to try the apps I'm running on the zaurus (firefox, gnumeric, abiword, portabase, ...). I'll compare those two, not a desktop and the pandora.
Anyway, try working on a spreadsheet on an iphone and then on the pandora. Maybe they are not of a "similar category" after all ...

Chero.
 
Last edited by a moderator:
I've purchased a Nokia N800 off eBay some time ago as a placeholder for the Pandora, so I have a fair estimate what to expect :)
 
es87 said:
The Pandora's ARM CPU is 500-900 MHz. In terms of raw carrots, how fast can you make a stew?
And this daft question (or something like it ;) ) turned into the usual "Is Pandora more like an P3, Xbox or iPhone?", and "Linux is not Windows" - again.

mazza558 said:
Surely those who have seen the iPhone won't be as impressed? After all, the iPhone does a lot of that without "vibrating, heating, noising and gasping for more juice every 2 hours".
Who cares? Different strokes. Room for all.

Benjiro said:
This is just my opinion ... I'm not saying, if you exclude gaming, that most people can't switch, but, it takes so much time researching, finding the new programs, finding out that some have no equivalent, etc... For a IT person, sure. But 95% of the planet will not switch because they don't see a added value to switching.
But 95% of Windows users don't know how to use the programs they have on their pc. ;)
 
Last edited by a moderator:
Acording to one realistic marketing benchmark, the Cortex A8 is faster clock for clock than an Atom. But the IPC may not scale very well to 900mhz. We would need to test it better. http://www.youtube.com/watch?v=cnsaDc6criM

The SIMD unit seems very fast: can decode an 720p H.264 video at 600mhz with NEON (no DSP used). Atom can't do that at 1.6 Ghz (but a C2D certanly could).

The DDR memory can be relatively slow, but the latency probably is very low, as it is together with the processor, in the same package.
 
Last edited by a moderator:
mazza558 said:
Surely those who have seen the iPhone won't be as impressed? After all, the iPhone does a lot of that without "vibrating, heating, noising and gasping for more juice every 2 hours".
No, they won't be as impressed, but they will be impressed all the same once the UI gels. We're talking OMAP2 vs. OMAP3 differences. ;)


whiskthecat said:
Thanks, I was wondering why it was so high.
Nothing like a little benchmarketing, eh?
 
Last edited by a moderator:
Svartalf said:
Nothing like a little benchmarketing, eh?
There are lies, damn lies and benchmarks!

TaG & Svartalf(and probably others) is right. Most folks who ask the "so how powerful is it?" seem to be coming from a desktop environment and because of all the hype are expecting something as powerful as their desktop in the palm of their hand. Folks who want that should order a Pomegranate. However, then comparing benchmarks and OS's and favorite applications isn't necessarily going to enlighten them either. For these folks it's all about the software (nothing wrong with that!)

I doubt folks who are wowed by the iPhone will be at all impressed with the pandora. They'll likely say "hey, that's neat" and back slowly away from the rabid geek who is showing it to them (that would be me). Just like when you show a geek an iPhone he says "hey, that neat" - but show that same geek a pandora and watch out he might steal it!

:)

I don't know about the PSP and DS crowds - do you think they will be wowed by a pandora? Again, I don't think the average handheld gamer will be. When they open up a pandora and see a keyboard and a full linux OS they won't be "wow that's awesome". They'll probably actually be disappointed "Dude, why did you put a keyboard there? Where's the second screen?". But I'm looking forward to showing it off to my friends (the geeky ones) by borrowing my son's DS and doing a size and functionality comparison between the two.

The pandora is a geeks dream. For heavy PDA/mobility users (all three* of them) it's also a dream. For other folks in the consumer mainstream it's not going to impress them.

Especially seeing as it's only a PIII

;-)

* - OK, so there are actually more than 3 people who are heavy users of PDAs - but not many more! I'm talking beyond basic PIM stuff that can be done on most "smart" phones. I think that all these new MID devices will go the way of the Palm Pilot - they'll be hot for a while because they sure are neat - but then they'll go away when folks realize that they don't really use them. The pandora won't suffer that fate because it is aimed at a niche market (which is thankfully us!)

Sorry for the rambling - but I think it fits in with the rest of the postings and replies.
 
Last edited by a moderator:
jdh2550 said:
I don't know about the PSP and DS crowds - do you think they will be wowed by a pandora? Again, I don't think the average handheld gamer will be. When they open up a pandora and see a keyboard and a full linux OS they won't be "wow that's awesome". They'll probably actually be disappointed "Dude, why did you put a keyboard there? Where's the second screen?". But I'm looking forward to showing it off to my friends (the geeky ones) by borrowing my son's DS and doing a size and functionality comparison between the two.



That depends on the titles we end up having on the thing. The staff at the GameStops around the area are dying to see one. A PSP and DS do NOT have 3D hardware accel. The Pandora DOES and is the first there in the handheld space.
 
Last edited by a moderator:
Svartalf said:
The staff at the GameStops around the area are dying to see one.
That's a great idea - I'll make sure to show mine to the local Gamestop crew and get their take on it.
 
Last edited by a moderator:
Manabu said:
The SIMD unit seems very fast: can decode an 720p H.264 video at 600mhz with NEON (no DSP used). Atom can't do that at 1.6 Ghz (but a C2D certanly could).
do you know by any chance how many independent fp units it has and whether they can do an add/multiply in one cycle? the atom i think has 2 units only (like the P4 of old but unlike the core).

if the A8 was similar in performance than the atom, thot would be incredible. atom has 2-4W TDP, the A8 has 0.3W if i am not mistaken.
 
Last edited by a moderator:
hch said:
do you know by any chance how many independent fp units it has and whether they can do an add/multiply in one cycle? the atom i think has 2 units only (like the P4 of old but unlike the core).

if the A8 was similar in performance than the atom, thot would be incredible. atom has 2-4W TDP, the A8 has 0.3W if i am not mistaken.
It has one scalar FPU (compliant with VFP v3), and it isn't pipelined. A subset of scalar FPU operations can be ran on the NEON unit where it is pipelined. Basically, the FPU on the Cortex-A8 is there for compatibility, for instance if you want double precision support. For computationally intensive programming you're limited to what you can do in SIMD - how it stacks up against SSE3 (or whatever Atom has) in terms of performance I don't really know, but it should at least be a lot more flexible. It seems that it can do a 4-way FMAC in 2 cycles, but issues every cycle. That's actually very good performance.

By the way, if you think two of ANY CPU computational unit is a small number for a CPU in a handheld then your perspective isn't very realistic.
 
Last edited by a moderator:
Exophase said:
It has one scalar FPU (compliant with VFP v3), and it isn't pipelined. A subset of scalar FPU operations can be ran on the NEON unit where it is pipelined. Basically, the FPU on the Cortex-A8 is there for compatibility, for instance if you want double precision support. For computationally intensive programming you're limited to what you can do in SIMD - how it stacks up against SSE3 (or whatever Atom has) in terms of performance I don't really know, but it should at least be a lot more flexible. It seems that it can do a 4-way FMAC in 2 cycles, but issues every cycle. That's actually very good performance.

By the way, if you think two of ANY CPU computational unit is a small number for a CPU in a handheld then your perspective isn't very realistic.
thanks for the information. so do i understand correctly, that the neon instructions are exeecuted in a physically different part of the die than the fpu ones? i always thought that the decoding would be different, but the actual computation was done in the same unit.

and an effective 2 FMAC per cycle is fantastic. so the peak perforance of the A8 is actually 3.6 GFLOPS - wow.
 
Last edited by a moderator:
mazza558 said:
Surely those who have seen the iPhone won't be as impressed? After all, the iPhone does a lot of that without "vibrating, heating, noising and gasping for more juice every 2 hours".
they will be impressed at the amount of freedom they have though after using the most restrictive and locked piece of shi**, I mean hardware, out there... The screen actually having a decent resolution will impress too.

Sorry to come across as harsh but I recently ended up with an iPhone (because the phone selection here is extremely limited) and if I hadn't jailbroken my phone it would have gone out the window a long time ago. Also, if you use the 3G/GPS/Wifi/Games/etc. a lot the iPhone WILL gasp for more juice every 2 hours and probably crash and get stuck several times too in the process... ;)
 
Last edited by a moderator:
So Cortex @ 600MHZ can not be compared to a P3 750MHZ. It's slighty slower, but if an app was optimised for ARM say MAME it would be perform like a 750MHZ maybe faster?. Sorry to sound dumb but just trying to get an idea of speed.

Now the Cortex has adjustable L2 Cache or am I smoking something I don't need?

I thought it could go up to 1MB, would more L2 Cache help with intensive CPU emulation like MAME or DOSBOX?
 
What do you mean by that?

Cortex-A8's can be designed with a configurable amount of L2 cache, but the one in the OMAP3530 is stuck with what it has (256KB). It's not external to the CPU, much less the OMAP. That's what makes it decently fast.
 
joshwaan2k said:
So Cortex @ 600MHZ can not be compared to a P3 750MHZ. It's slighty slower, but if an app was optimised for ARM say MAME it would be perform like a 750MHZ maybe faster?. Sorry to sound dumb but just trying to get an idea of speed.

The performance relies on the optimization of the code. The same code may run differently on A8 and P3, because they are different and not directly comparable chips with different strengths and weaknesses. There won't ever be an honest way to directly compare them.
 
Last edited by a moderator:
Back
Top