Pandora Vs. Eeepc 701 Processor Speed? Only For Curiousity

[doublebeta]

Hey Pandorians! (too bad there isn't a Greek city named after Pandora, unless there is?)
I did a quick search, i didn't really find anything, so... I'd like to know if anyone has done a real comparison between processor speed of the EeePC 701 (Intel Celeron M at 600MHz by default) with the Pandora (OMAP3530 or somethin, 500-800MHz, 600 optimal or something?), with both at 600MHz. Like with a program like hardinfo. I know they'll be different, they're a completely different architecture. I just wanna know WHICH is the fastest(I'd assume the Pandora, the OMAP is newer, and it's not x86 ). And it's a fairly fair test since they both run at 600Mhz by default (A lot of people are running the Pandora at 500Mhz right now, i'm confuzed about the default now). I actually own an EeePC, and before I cracked the screen, Xubuntu 9.10 ran incredibly fast on it, with Firefox 3. And since people have said the Pandora is kinda slow with FF3, i was wondering.
And, though I'd assume it bloody could for a modern day processor, even tho the Nintendo DS cant(according to the author of QuakeDS, IIRC???), does the Pandora do hardware floating point math?

Remember, it's only curiosity that drives me to ask this question, I'll still love the Pandora. So no telling me it's crap so you can get a spot closer in the Batch 2 queue

This post has become far longer then it should have, I apologize, and please ask for you to put out your torches, and put down your pitchforks, thanks. Love you too. andora2ut4:

 

[Pleng]

Firefox 3 zips along for me. Mind, I have my CPU pretty much permanently clocked to 750

 

[mali]

I've removed the duplicate thread. This question is nigh on impossible to answer correctly. Apples and oranges, you know
There are some nBench results somewhere including Atom. OMAP3 sucks at floating point, other than that it's comparable to Atom clock for clock.

 

[Asiyura]

OMAP3 sucks at floating point, other than that it's comparable to Atom clock for clock.

I knew I should have kept the external FPU from my 386 lol

 

[mali]

^ We have NEON though

 

[PokeParadox]

They seem similar... but it's hard to say.

 

[Asiyura]

Mali, thanks for clearing things out, found some quick and simple explanations about NEON and VFP here.

Are we already able to use the full potential of those? (like in the kernel)

 

[mali]

It's been a long while since I read something about that matter. The compiler was crap and proper use of NEON instructions had to be done via handcrafted ASM. I'm sure the experts did some hacking in the meantime to utilize it better, but I don't frequent the Devs' Corner much, as I don't understand their magician language anyway

 

[Yod4z]

NEON for AMD is like SEE for Intel.

 

[iampanis]

The Celeron should be MUCH faster. The ARM Cortex-A8 is similar to Atom clock for clock, and I've heard that a Pentium III is roughly twice as fast as that. I'd imagine the Celeron being a fair bit better than the Pentium III. Someone who actually knows what they're talking about should be able to explain better (in order vs. out of order execution, cache sizes, etc.).

 

[Exophase]

The Celeron should be MUCH faster. The ARM Cortex-A8 is similar to Atom clock for clock, and I've heard that a Pentium III is roughly twice as fast as that. I'd imagine the Celeron being a fair bit better than the Pentium III. Someone who actually knows what they're talking about should be able to explain better (in order vs. out of order execution, cache sizes, etc.).

I dunno where you heard that P3 is twice as fast, maybe Core 2 is. The Celeron will still be faster though. Aside from CPU performance the Eee probably has a wider (64-bit vs 32-bit) and lower latency memory bus too (OMAP3's is looking to be pretty slow).

 

[iampanis]

I dunno where you heard that P3 is twice as fast, maybe Core 2 is.

I probably just have poor memory, and/or heard from someone else with poor memory.

EDIT: Wikipedia claims that "The performance of a single core Atom is about half that of a Pentium M of the same clock rate." That's likely what I was thinking of.

 

[Vitel]

The Celeron should be MUCH faster. The ARM Cortex-A8 is similar to Atom clock for clock, and I've heard that a Pentium III is roughly twice as fast as that. I'd imagine the Celeron being a fair bit better than the Pentium III. Someone who actually knows what they're talking about should be able to explain better (in order vs. out of order execution, cache sizes, etc.).

http://talk.maemo.org/showpost.php?p=309782&postcount=6

 

[doublebeta]

0.o 12 replies already??? Thanks all!

Thanks for your replies and thoughts, it's all very interesting. Seems to be a bit of conflicting ideas here, if the Atom is faster than the much older, notably inefficient Celeron (unless i'm confusing here) then Intel isn't trying very hard to make an 'efficent, power saving' processor (...or it's all bull because it's, according to that link, equivalent of the P4, even P3, and the P3 is damn oold man!...) . I got a few replies talking about the Atom, and it was quite interesting to hear that the Atom and Cortex-A8 were clock-for-clock the same speed, except for possibly floating point operations on the OMAP. I'll eat my hat (starting a meme on gp32x ftw) if a P3 is faster than the Pandora (void if Pandora is underclocked/P3 is overclocked)

For the person who was talking about bus width, the Celeron M used in the EeePC 701 was 32bit. And I don't *think* the OMAP is 64bit, in fact I remember reading, vaguely, it being 32bit. if it is 64bit, please let me know.

Really I don't care which is truly faster, because the Pandora doesn't get bloody hot on your lap when you use it and doesn't eat Homer Simpson loads of power. And the Pandora is stil the fastest device that will fit in your pocket that is legal for a civilian to even KNOW about (Attn CIA/FBI/something: i don't know about any devices, i'm taking a guess)

Thanks for removing the dupe mali, hit enter when the box was unfocused :\.

FPU on the unused IO ports anyone? (I'm half-joking here, i'd be crap to fit in the case)

EDIT: Single precision, 32-bit? Not a problem, that should be plenty of resolution for what we're actually gonna use (games), then again, what I did ask was how long it would take to do a mathematical operation, which would likely use double precision floating point...heh. I'm contradicting myself.
EDIT: Good thing I don't actually have a hat (that I can actually be assed to find) because http://www.gp32x.de/board/index.php?/topic/45583-pandora-some-actual-performance-numbers/page__hl__bytemark__st__30 says that the Pentium 3 is just a little faster. But possiblynot clock for clock. they actually look the same. OMAP still kicks the crap outta PIII/Atom/Celeron because like I said, it doesn't draw 100 Amps of current and attempt to emulate a deep fat fryer.

 

[Vitel]

For the person who was talking about bus width, the Celeron M used in the EeePC 701 was 32bit. And I don't *think* the OMAP is 64bit, in fact I remember reading, vaguely, it being 32bit. if it is 64bit, please let me know.

Cortex-A8 specifications are actually very close to Pentium3-EB. They're both 32-bit processors, L1 cache is 32k, L2 cache is 256k, bus clock is 133 (P3) vs 166 (Cortex).
I think it would be interesting to make some performance tests of these CPUs at the same clock.

 

[droneb]

I've removed the duplicate thread. This question is nigh on impossible to answer correctly. Apples and oranges, you know
There are some nBench results somewhere including Atom. OMAP3 sucks at floating point, other than that it's comparable to Atom clock for clock.

WOW, been away some time from the boards,

nice to see that you became a MOD.

well gotta go away again so i forget about pandora until they reach 2k Mark.


BACK ON TOPIC

this question is hard to respond, because you cant compare planes and trains
they are both vehicles, but they suit different uses.

 

[ari64]

I've removed the duplicate thread. This question is nigh on impossible to answer correctly. Apples and oranges, you know
There are some nBench results somewhere including Atom. OMAP3 sucks at floating point, other than that it's comparable to Atom clock for clock.

"OMAP3 sucks at floating point", other than that you forgot:

L1 cache is smaller than Atom.
L2 cache is smaller than Atom.
Memory bus is slower than Atom.

 

[mali]

^ I remember the discussion about Mupen64+'s performance on Atom To defend myself, I said "apples and oranges"

 

[Exophase]

Cortex-A8 specifications are actually very close to Pentium3-EB. They're both 32-bit processors, L1 cache is 32k, L2 cache is 256k, bus clock is 133 (P3) vs 166 (Cortex).

The CPU architecture is completely different. Cortex-A8 is a dual issue in-order design while Pentium 3 is a triple issue out-of-order/speculative execution design. Cortex-A8 also has a deeper pipeline, and as I already mentioned, half the bus width. Of course, you can talk about things the A8 has going for it too - more powerful ISA, arguably slightly stronger SIMD. Really you can't use the metrics you have to say they're even remotely similar; there have been budget/mobile x86 CPUs with a lot of cache and decent buses that have performed like crap.

 

[Vitel]

The CPU architecture is completely different. Cortex-A8 is a dual issue in-order design while Pentium 3 is a triple issue out-of-order/speculative execution design. Cortex-A8 also has a deeper pipeline, and as I already mentioned, half the bus width. Of course, you can talk about things the A8 has going for it too - more powerful ISA, arguably slightly stronger SIMD. Really you can't use the metrics you have to say they're even remotely similar; there have been budget/mobile x86 CPUs with a lot of cache and decent buses that have performed like crap.

That's the point: to compare two different architectures with similar spec parameters.
And I didn't understand your statement re ISA. CISC is more powerful AFAIK.