Passion/Dragon VS Moto Droid BENCHMARKS THREAD...What Do YOU Think????

[Android_LIFE]

I have seen alot of benchmark chatter in comparison of the Passion/Dragon to the Moto Droid in different topic threads...

I guess its only appropriate to open up a benchmark discussion...

I know none of us can actually physically compare the two side by side, but
Lets just assume the rumored specs of the Dragon/Passion are real.
or compare the snapdragon processor to whatever. LETS GO!

Looking forward to hearing what everyone has to say...








DISCUSS!!!!!!

 

[Mykpfsu]

As far as pure CPU power the snapdragon should beat all comers with 2100 million instructions per second besting the Iphone 3G s which only does 1200.
Graphics seems to be a toss up. it might depend on which Snapdragon processor is used. But the Snapdragon should be able to push out about 22 million triangles a sec with more powerful versions kicking it up to 80 million. The droid pushes out either 7 million or 14 with the Iphone 3G s pushing 7 million (contrary to popular belief it does not push out 28 million).

 

[wonderbread]

iPhone 3Gs GFX is much faster than that, what is your source for 7mil? Your other numbers are what I've seen before in multiple sources.

 

[Mykpfsu]

iPhone 3Gs GFX is much faster than that, what is your source for 7mil? Your other numbers are what I've seen before in multiple sources.

You can look at the glbenchmarks (though that shouldnt be the end all be all). No benchmark tops 7 mil. People seem to keep thinking the Iphone 3gs has the PowerVX 535 when more likely it has the 520 (tops off at 7 mil). The only proof of the higher chip is a single forum post and a driver file. The driver file though could just be there if the 3G S is a gateway device as the CPU can run faster and records DVD quality video (though the 3G S doesnt)

AnandTech: Understanding the iPhone 3GS

Apple iPhone 3G S - OpenGL ES performance and system information

 

[YankeeDudeL]

As I read every hypothetical theory on the Lock Ness Monster of phones, I can't help but wonder, "What does it matter?" I'm typing this from my Droid, so I guess this is another win for Droid.

 

[vaxick]

It won't be that drastic of a difference from the Droid. The CPU in the Droid can hold its own against the Snapdragon CPU. It isn't going to be a drastic jump in speed over it at all. It'll be faster of course, but too many people are affixed on the CPU numbers instead of the actual architecture.

 

[sooper_droid12]

All you need to know is what Qualcomm said: The Snap/Imageon combination will likely see a 5% improvement in power over similar OMAP3 implementations.

I've had this discussion with you before Myk. If all the phone was doing was crunching integers, then sure, the Snapdragon would be awesome. However, it is essential that the capability of these Smartphones be able to push graphics and that's what the GLBenchmark demonstrates. Should it be the end all be all? No, but it should be heavily considered.

It is THE measure of a phone's ability to render graphics. And last time I checked, that's why these phones have huge screens. The JBenchmark should also be considered (for Java). That being said, the SGX GPU coupled with the OMAP3, while down on the numbers is a very potent combination. This is why this SOC chip is in the best smartphones and mobile GAMING platforms today.

Fact of the matter is, Snap/Imageon was toasted by the OMAP3/SGX combination. You can downplay the GLBench all you want, but facts are facts.

 

[sooper_droid12]

I will give Snap/Imageon this much... 5% improvement in performance (manufacturer statement) with better battery life (likely). Snap is extremely efficient with batteries. It is likely that Acer went with an under-clocked Snap just to extend battery life beyond the ubiquitous 6-7 hour talk-time the benchmark Smartphones hold to today (i.e. iPhone, DROID, etc.)

 

[wonderbread]

New video drivers for snapdragon based HD2 results in 500% 3d performance increase!
Chainfire driver patch improves HTC HD2 OpenGL performance by 500%

 

[sooper_droid12]

New video drivers for snapdragon based HD2 results in 500% 3d performance increase!
Chainfire driver patch improves HTC HD2 OpenGL performance by 500%

I guess it was non-optimized drivers. With the optimized drivers, the Snap/Imageon is equivalent in performance @ 1GHz/600Mhz to the OMAP3/SGX @ 600MHz/440MHz when it comes to the GLBenchmark scores, while still lagging behind in the Texture Sizing and Triangles, by 2-3 fold.

Still not impressive to me, especially since this has to be clocked at 1G's to match.

 

[sooper_droid12]

For all those Passion-owner wannabes, let's hope that HTC delivers with the right drivers because without 'em, it appears certain apps will be slow and stutter, according to driver developer. He tested it on several apps including Opera Mobile and Sense UI.

 

[Mykpfsu]

I guess it was non-optimized drivers. With the optimized drivers, the Snap/Imageon is equivalent in performance @ 1GHz/600Mhz to the OMAP3/SGX @ 600MHz/440MHz when it comes to the GLBenchmark scores, while still lagging behind in the Texture Sizing and Triangles, by 2-3 fold.

Still not impressive to me, especially since this has to be clocked at 1G's to match.

Funny since you keep going on about the Iphone 3G S which is hacked to run the GLbenchmark scores.

iPhone 3GS Smokes Competition in OpenGL Benchmarks - Gadgets by ExtremeTech

"Kishonti encountered some snags when benchmarking the 3GS. The iPhone maxes out at 60 frames per second, which initially capped the 3GS's performance on some benchmarks."...."initially"? I wonder what changed. It certainly couldnt be from overclocking could it?
Not to mention where does it say the HD2 was overclocked?

 

[sooper_droid12]

Maybe a patch or it to run properly on the tests like the recently released HD2 GLBenchmarks that wonderbread posted? After a patch optimized the phone for the tests, HD2 running Snap all of a sudden picked up 100-200% improvement. It doesn't necessarily mean that it was hacked. Even so, even if the processor in the iPhone can be overclocked, I think the upper-limit is 600MHz... 620-ish tops?