SoC: Back and forth!

[Askarus]

Don't let's judge the Snapdragon before we don't have any details.

Being skeptical and discussing it is the right way to go.

Anyways we may not forge that Snapdragon has big.LITTLE with 4 cores up to 2.5 GHz each.

There are advantages above the OMAP and there are disadvantages.

The Snapdragon 805 is a great processor and worth to think of.

The decision should be rational by making a priority list and checking which processor is better in each point.

Also we may not forget that we don't have the go from Qualcom yet.

We were at about the same situation for the Tegra 4, then Nvidia told us that they only allow to use the Tegra if a certificated company does the board design.

We don't know what condition Qualcom has.

 

[Linux-SWAT]

They say you and ED have to cut your hair !

 

[Askarus]

OMAP wins.

 

[Zarkon]

Would either of them (omap/snap) have a shot at decent saturn/yabause performance, or is that still a fair way off?

 

[Linux-SWAT]

Should be decent, although the main problem here is the quality of the emulator itself..

 

[Hồng Thất Công]

Is an exclusive Saturn emulator for the Pandora/Pyra in the works???

 

[Linux-SWAT]

Maybe ptitSeb can take a day or two to write one ?

 

[Askarus]

Maybe Snapdragon 805 could even target Gamecube or PS2 emulation.

But don't know which one is better for emulation.

OMAP or Snapdragon.

 

[Hồng Thất Công]

Maybe Snapdragon 805 could even target Gamecube or PS2 emulation.

But don't know which one is better for emulation.

OMAP or Snapdragon.

Drooling...

 

[fusion_power]

I'd vote for whatever Notaz wants.

Makes sense. ^^I would like to see a direct competition between OMAP and Snapdragon, not sure if Pyra prototypes can be built with each processor to compare. And one big question at the end would also be which one is cheaper? Maybe the starting competition could lead to company side price adjustments if they see they have competition. Pyra gets already popular so let's see how it developes.

 

[Hồng Thất Công]

I'm not sure if ED can show us what PPSSPP and Reicast are like on the Pyra Protoype he used at FOSDEM.  If it's smooth like butter, I would say just stick to OMAP5 for PYRA

 

[Askarus]

I'm not sure if ED can show us what PPSSPP and Reicast are like on the Pyra Protoype he used at FOSDEM.  If it's smooth like butter, I would say just stick to OMAP5 for PYRA

Uses GPU.

No Hardware acceleration yet.

Maybe later.

But I can say with 90% certainty that they will be smooth like butter.

Didn't you see the video with 2 PSX emulators double resolution, Battle for wesnoth, KDE 4, Gimp and Firefox running smooth without any acceleration.

Don't think Snapdragon will help here much.

 

[Hồng Thất Công]

I'm not sure if ED can show us what PPSSPP and Reicast are like on the Pyra Protoype he used at FOSDEM.  If it's smooth like butter, I would say just stick to OMAP5 for PYRA

Uses GPU.

No Hardware acceleration yet.

Maybe later.

But I can say with 90% certainty that they will be smooth like butter.

Didn't you see the video with 2 PSX emulators double resolution, Battle for wesnoth, KDE 4, Gimp and Firefox running smooth without any acceleration.

Don't think Snapdragon will help here much.

Yeah saw the video, but still want to see OMAP 5 slices through PPSSPP and Reicast like butter though

 

[Askarus]

You know it's pretty hard to concentrate at University exams when knowing that we might have a chance to get the Snapdragon.

I need to know what Qualcom says.

Now!

 

[Linux-SWAT]

They say "Cut now !".

 

[hns]

Some first (very rough) benchmark using some self-compiled floating point code (n times doing a 2D coordinate transform):

9.5s    armhf OMAP3 @ 1 GHz (GTA04)

4.5s    i386 Celeron @ 900 MHz (eeePC 701)

1.2s    armhf OMAP5 @ ? (Evaluation Board)

0.6s    i386 Core2Duo @ 2.8 GHz (some desktop)

So the OMAP5 has 8 times FPU performance as the OMAP3 (using exactly the same binary code, which means no better compiler optimization). And 0.5 of a fast Core2Duo.

There are of course better (standardized) benchmarks that also separately test integer and float performance.

If someone has a recommendation for a test (ideally available in Debian Wheezy) I can run it on all machines.

But running GPU tests is very different.

 

[TrashyMG]

^ WoW interesting stats..

Well I'm really looking forward to the OMAP5 if chosen.. and has nothing to do with the money I invested in the OMAP5 EVM dev board.. nothing..

 

[ssokolow]

Not getting my hopes up.

nice, will be cool if the snapdragon makes it in, but will not be a letdown for me if it doesnt.


What about the screen size? will it be widescreen like the pandora, or will it be something like 16:10?

That should really be ask somewhere else, but I think 16:10 would be neat.

While it's not fixed (note the star), the current probably is the same 1920x1080 part used in the Nexus 5.

 

[Askarus]

Some first (very rough) benchmark using some self-compiled floating point code (n times doing a 2D coordinate transform):

9.5s    armhf OMAP3 @ 1 GHz (GTA04)

4.5s    i386 Celeron @ 900 MHz (eeePC 701)

1.2s    armhf OMAP5 @ ? (Evaluation Board)

0.6s    i386 Core2Duo @ 2.8 GHz (some desktop)

So the OMAP5 has 8 times FPU performance as the OMAP3 (using exactly the same binary code, which means no better compiler optimization). And 0.5 of a fast Core2Duo.

There are of course better (standardized) benchmarks that also separately test integer and float performance.

If someone has a recommendation for a test (ideally available in Debian Wheezy) I can run it on all machines.

But running GPU tests is very different.

Luckily the Intel core 2 Duo is still faster.

I spent 250€ for my 3 Giga core.

Do I really want a handheld to beat my Laptop?

YES!

Let's see what's the true Power of OMAP.

Now it would be interesting to do the same with Snapdragon 805.

 

[rohezal]

So the OMAP5 has 8 times FPU performance as the OMAP3 (using exactly the same binary code, which means no better compiler optimization). And 0.5 of a fast Core2Duo.

This is without neon which can do 2 floating point operations at a time, right?