OMAP3 roughly equivalent in x86?
 

I know that they are differences between the x86 architecture (CISC) and the ARM Cortex A8 (RISC) inside the OMAP that the next tablet will use.

However, now that there is a lot of talk about ARM netbooks and stuff like that, I was wondering what would be the rough equivalence between both platforms. For example, for generic office apps (let's say, calculations in a big spreadsheet), OMAP 3 is equivalent to an Atom processor? A Core Solo?

Any idea/info about that would be welcome.

Re: OMAP3 roughly equivalent in x86?
 

I have found no MIPS or FLOPS information regarding the OMAP. If you can find it, you can do some comparison to a x86 processor.

Re: OMAP3 roughly equivalent in x86?
 

I couldn't find any info either, that is why I am asking for some help :(

I found some saying that the Atom is roughly equivalent to the original Pentium 4, and then another saying that OMAP 3 is similar to the Atom, but those were comments in blog posts, so the info doesn't serve me well.

Re: OMAP3 roughly equivalent in x86?
 

It's RISC vs CISC, so MIPS would translate to Meaningless Indication of Processor Speed real quick :D FWIW they do about 2 DMIPS/MHz and FLOPS is not applicable as our OMAP3-s are FPU-less,

Re: OMAP3 roughly equivalent in x86?
 

Lol, sounds like Dr. Seuss to me. ;)

I think this is a valid question. Someone load ubuntu onto a beagleboard and various x86 chips, and start benchmarking. Please.


See, I asked nicely.

Re: OMAP3 roughly equivalent in x86?
 

The Cortex A8 performance is about equal to Atom and Pentium 3 at the same clock speed except floating point, which is much less powerful on Cortex. NEON optimizations might help a bit, though. Memory throughput is very high on OMAP3 especially compared to OMAP2.

There are Bytemark results on the web, that can be compared. For OMAP3-CortexA8 search the Open Pandora board(gp32x.com).

Re: OMAP3 roughly equivalent in x86?
 

Here are the benchmarks for the OMAP1 and 2, plus some other mobile processors: http://wiki.hbmobile.org/index.php?t...sor_Benchmarks

Here are the benchmarks for the Pandora (OMAP3): http://www.gp32x.com/board/index.php...temark__st__30

In fact read that whole last thread as some tweaks are made to try to improved the Pandora results.

I don't have nbench comparison results for a PC (but anyone can just compile that and try of course) but do have some dhrystone and whetstone results comparison if that's of interest: http://people.bath.ac.uk/enpsgp/benchmarks/

Re: OMAP3 roughly equivalent in x86?
 

lol. Honestly, all I care is Omap3 >>>>>>>>> Omap2, which means that the next tablet will be a butt kicking wonder. :D

Re: OMAP3 roughly equivalent in x86?
 

This is not true. There is an FPU on the OMAP3. It is optimized for single precision vectorized floating point operations.

Re: OMAP3 roughly equivalent in x86?
 

You mean NEON?
There's still vfp, is it? At least I remember the fremantle gcc generating code for vfp by default.