Motorola Atrix 4g Vs. Samaung Galaxy S 2 3D Games

[ohad.e89]

Hey,

I need help, with choosing my next android phone.
and i don't know which phone is the BEST for 3D games, the atrix with the Tegra 2 (Im Nvidia Fan!!!) OR the Galaxy S 2 with Mali mp400 unit?

Please help me decide , im want to buy the phone next week!!!


thanks everyone!!!!!!!!!!!!

 

[Medion]

It hasn't been factually proven yet which is more powerful, though there are people in both camps. They are definitely best of breed and you can't go wrong with either. However, the CPUs backing them up are a different matter.

The Galaxy uses a 1.2ghz dual-core Cortex A9 CPU, 45nm, w/NEON instruction set. Tegra 2 uses a 1.0ghz dual-core Cortex A9 CPU, 42nm, and no NEON instruction set. What does this mean? Slight battery life advantage to the Tegra 2, and slight performance advantage to the Exynos. However, Tegra 2 supposedly takes the duties that would normally be handled by NEON and tries to offload them to the GPU. Apparently this isn't working as well as planned, because Tegra 3 will use the NEON instruction set.

 

[ohad.e89]

thank you very much for the quick answer!!!

so do you think that maybe its way better for me to wait until the tegra 3, will be out?

 

[Medion]

thank you very much for the quick answer!!!

so do you think that maybe its way better for me to wait until the tegra 3, will be out?

Absolutely not. There will ALWAYS be better technology on the horizon. Find a phone that you love, get it, and keep it as long as you want. Remember, these things are like PCs. You don't HAVE to get a new one just because a better one came out. You get a new one because your old one doesn't keep up with your demands anymore.

Android gaming hasn't gotten to the point yet where high-end hardware is needed. My HTC Incredible using ye olde Snapdragon (CPU is awesome, GPU not so much) plays anything that I desire to play, though some higher end games do push it. However, I can easily see this phone outlasting my contract.

So, if you see a phone you want that's out now or will be out shortly, get it. Don't wait for something better, or you'll be continuously waiting. From what I've heard about the developing successor to Exynos, the same issue you have with Tegra 2 and Exynos today will repeat next year with Tegra 3 and "Exynos 2."

 

[Evil.Bonsai]

Ahem!

 

[Shocky]

Mali 400 MP outperforms Geforce ULP and has higher clocked A9 CPU with NEON enabled so it's faster all round. as for if you should wait or not, I can't answer that, I wish I had a crystal ball and could give you all the answers but I can't.

The Galaxy uses a 1.2ghz dual-core Cortex A8 CPU.

It's Cortex A9, the original Galaxy S used the A8.

 

[Medion]

Mali 400 MP outperforms Geforce ULP and has higher clocked A9 CPU with NEON enabled so it's faster all round. as for if you should wait or not, I can't answer that, I wish I had a crystal ball and could give you all the answers but I can't.




It's Cortex A9, the original Galaxy S used the A8.

Thank you, huge typo on my part. BOTH use Cortex A9. I'll edit my post accordingly.

 

[Guamguy]

To the OP,

There is no question that the SGS2 has the more powerful GPU, but there is another reason for choosing the SGS2.

The 480x800 resolution used by the SGS2 is more compatible with games than the qHD resolution of the Atrix. There are some games that will not work on qHD --- they won't list themselves in the Android Market in the first place using a filter.

One has to consider that some games are Tegra2 optimized. That gives a feather for the Atrix. But I would pick the Optimus 2X aka G2X over the Atrix given that both has the same Tegra2, but the Optimus uses the more standard 480x800 screen resolution and is more compatible with more games.

 

[Medion]

There is no question that the SGS2 has the more powerful GPU, but there is another reason for choosing the SGS2.

Do you have anything to back this up? I'm not picking on you, it's just something I'd like to personally see for myself. I haven't seen anything conclusive on it yet.

 

[Roze]

I like my Atrix but if I had to pick the two, I would go with the SGS2.
Other things to consider. It comes with Gingerbread and 1020P recording/viewing out of the box with an unlockable bootloader These three are promised updates with the Atrix...supposedly this or next month (for AT&T). So out of the box, the SGS2 has has a bit more than the Atrix.

Camera on the Atrix is mediocre at best. From my experience, the software for it is not that great. It doesn't compare to my pictures taken from my old Nexus One.

For Navida gaming, you need to be rooted, but there's an app that opens up the GPU so you can play any games on it. (sorry I forgot what it's called) but it seems that the SGS2 that's coming to North America will have the Navida CPU chip.

 

[Shocky]

Do you have anything to back this up? I'm not picking on you, it's just something I'd like to personally see for myself. I haven't seen anything conclusive on it yet.

You mean like GLBenchmark 2.03, Electopia and Nenamark 2 where the Galaxy S II completely dominates Tegra 2?

I'm sure Tegra 2 has some advantages, every GPU is different but in terms of raw performance Mali 400 MP has quite a big advantage.

Also, Tegra 2 oddly enough does not support MSAA, Mali 400 MP can do 4xMSAA with almost no performance hit and 16xMSAA.

 

[Medion]

You mean like GLBenchmark 2.03, Electopia and Nenamark 2 where the Galaxy S II completely dominates Tegra 2?

I'm sure Tegra 2 has some advantages, every GPU is different but in terms of raw performance Mali 400 MP has quite a big advantage.

Also, Tegra 2 oddly enough does not support MSAA, Mali 400 MP can do 4xMSAA with almost no performance hit and 16xMSAA.

You're telling me, not showing me. Can you please link me to any published results? I'm not trying to be a smart-ass, I actually want to see the comparison and learn for myself.

I hadn't seen any mobile GPU where you can turn on anti-aliasing, so if the Mali 400 allows you to do that, then that's quite an achievement. ARM recently went on record stating that a future version aimed for 2013 (possibly late 2012) would support GPGPU computing via OpenCL. Now that is HUGE.

But yea, if you can link me any published comparisons, I'd like to see them. I'm always curious about the testing methods because I do thin kind of thing myself. FYI, the T-Mobile G1 beat the Nexus One in Nenamark. I don't want to ruin it for you, so I'll see if you can figure out why this occurs.

 

[Shocky]

You're telling me, not showing me. Can you please link me to any published results? I'm not trying to be a smart-ass, I actually want to see the comparison and learn for myself.

I hadn't seen any mobile GPU where you can turn on anti-aliasing, so if the Mali 400 allows you to do that, then that's quite an achievement. ARM recently went on record stating that a future version aimed for 2013 (possibly late 2012) would support GPGPU computing via OpenCL. Now that is HUGE.

But yea, if you can link me any published comparisons, I'd like to see them. I'm always curious about the testing methods because I do thin kind of thing myself.

The results for Nenamark 2 and GLBenchmark 2.03 are available on there websites.

GLBenchmark: mobile OpenGL ES benchmarks
Nena Innovation AB - NenaMark results

Electopia results
 
- OMAP34xx (Motorola Droid, SGX530 @ 110MHz): 5fps WVGA
- OMAP3630 (my Defy, SGX530 @ 200MHz): 10fps WVGA
- Hummingbird (Nexus S, SGX540 @ 200MHz): 17fps WVGA
- Hummingbird with Rys' special driver (Nexus S, SGX540 @ 200MHz): 19fps WVGA
- Exynos 4210 (Galaxy S II, Mali400 MP4 @ ??): [B]37[/B]fps WVGA
- MSM7230 (T-Mobile G2, Adreno 205 @ ??): 17fps WVGA
- Tegra 2 (Optimus 2X, Geforce ULV 2 @??): 16fps WVGA
- MSM7227-1 "Turbo" (Galaxy Ace, Adreno 200 @ ??): 10fps 480*320

FYI, the T-Mobile G1 beat the Nexus One in Nenamark. I don't want to ruin it for you, so I'll see if you can figure out why this occurs

Nenamark 1? resolution maybe? Don't know, can't even find any results for the G1.

 

[Medion]

There's something wrong with those Nenamark results. Tegra 2 should not be losing to the Adreno 205 by any amount (nor should there be such a huge discrepancy between two similar Tegra 2-based devices on the expanded list). But then again, Nenamark, as I stated, has always been a poor comparison.

Still, where there's smoke, there's fire. If the Mali 400 is continuously beating the Tegra 2 in benchmarks, there has to be a reason for it. But it's clear to me there's an issue with benchmarking the Tegra 2, because the results are all over the place.

 

[Shocky]

There's something wrong with those Nenamark results. Tegra 2 should not be losing to the Adreno 205 by any amount (nor should there be such a huge discrepancy between two similar Tegra 2-based devices on the expanded list). But then again, Nenamark, as I stated, has always been a poor comparison.

Still, where there's smoke, there's fire. If the Mali 400 is continuously beating the Tegra 2 in benchmarks, there has to be a reason for it. But it's clear to me there's an issue with benchmarking the Tegra 2, because the results are all over the place.

There will always be a margin of error and people also running custom/rooted phones which will effect the results, but Atrix/Droid X2 for example score allot less because there running a higher resolution than the 2X.

I don't see why Adreno 205 should automatically be scoring lower, it's a capable GPU, both Tegra 2, Adreno 205 and SGX540 have simalar performance.

 

[Medion]

There will always be a margin of error and people also running custom/rooted phones which will effect the results, but Atrix/Droid X2 for example score allot less because there running a higher resolution than the 2X.

It still shouldn't be THAT much of a difference. For one, the difference in resolution should only affect fill-rate. The difference between a 960x540 screen (518,400 pixels) and an 800x480 screen (384,000 pixels) is 35% using the 800x480 screen as a base. Yet, the benchmarks results show a larger variance than this. Also, fill-rate usually isn't the issue with these mobile GPUs (outside of the Adreno 200).

I don't see why Adreno 205 should automatically be scoring lower, it's a capable GPU, both Tegra 2, Adreno 205 and SGX540 have simalar performance.

Not quite. The Adreno 205 is 70% more powerful than the Adreno 200. The SGX 535 is on par with the Adreno 205, but the SGX 540 kills it (Qualcomm even stated that their goal with the Adreno 220 was to catch up to the SGX 540). The Tegra 2 is in another league when compared to the SGX 540.

So, it's Tegra 2 > SGX 540 > Adreno 205 = SGX 535 > SGX 530 (200mhz) > Adreno 200 > SGX 530 (100mhz)

However, you're going off of poor benchmarks that don't properly show this. Nenamark doesn't account for this. I will, however, give GL Bench a try as it should deal specifically with OpenGL performance. Nenamark seems to deal more with fill-rate and driver optimization.

 

[Shocky]

It still shouldn't be THAT much of a difference. For one, the difference in resolution should only affect fill-rate. The difference between a 960x540 screen (518,400 pixels) and an 800x480 screen (384,000 pixels) is 35% using the 800x480 screen as a base. Yet, the benchmarks results show a larger variance than this. Also, fill-rate usually isn't the issue with these mobile GPUs (outside of the Adreno 200).

That's on the assumption the performance will scale accordingly, it doesn’t always work that way.

The Tegra 2 is in another league when compared to the SGX 540.

So why does Optimus 3D with higher clocked SGX540 score higher in benchmarks than Tegra 2 then? The numbers don't add up for that to be true.

You can believe whatever you want, but the numbers are not backing up that claim.

For comparision the Galaxy S II actually scores 43 fps in this benchmark.

 

[Medion]

That's on the assumption the performance will scale accordingly, it doesn’t always work that way.

Fill-rate does (assuming all other things are equal). In this benchmark, it looks like the only thing changed was resolution. Yes, fill-rate does scale linearly, so that alone shows how far off these benchmarks are.

So why does Optimus 3D with higher clocked SGX540 score higher in benchmarks than Tegra 2 then? The numbers don't add up for that to be true.

Because these numbers are wrong. I did a massive benchmark comparison with members of this website back in January/February, and the end conclusion was the Android benchmark apps were too unreliable to use as a basis for fact. Results were horrible, such as stock clocked Droid 1 murdering the Droid 2 and Droid X (Quadrant, Nenamark, and Linpack!), the T-Mobile G1 edging out the Nexus One (Nenamark), among some other oddities. The ONLY chipsets that displayed "regular" results were the Qualcomm chipsets in Linpack and Quadrant Advanced, but Nenamark still botched these results.

You can believe whatever you want, but the numbers are not backing up that claim.

And you're choosing to believe numbers that are known to be wrong. This is why I refuse to draw conclusively that the Mali 400 greatly oupaces the Tegra 2. There have been no conclusive benchmarks to show the difference. Nenamark? That bench can't get any numbers straight.

 

[Shocky]

I still havent seen any proof to say they are wrong, seems your just choosing to believe what you want to believe and to hell with the benchmark result.

 

[Medion]

I still havent seen any proof to say they are wrong, seems your just choosing to believe what you want to believe and to hell with the benchmark result.

I'll be honest with you, I can't fault you for that statement. Without producing my benchmark results, I can't "prove" you wrong. With that said, your GL Benchmark post wasn't showing up for me until after my last post, so I wanted to comment on that.

The benchmark is more uniform and seemingly accurate than Nenamark. However, it also reinforces some of my points, such as,

-You should see a linear difference in benchmarks where fill-rate is a factor, and this occurs here with Tegra 2 (roughly 35% difference as there is with fill-rate)
-The pecking order I listed in regards to chipsets was correct (your own graphics shows this). There are exceptions, but they are easily explained. The lone SGX 540, the overclocked one on top, will always outperform lower-clocked versions. However, I'm curious why you have it listed at the bottom that the GS2 scores 43fps in that benchmark...when the chart has it at 16fps? Seems you're doing exactly what you accused me of?

 

[Shocky]

The pecking order I listed in regards to chipsets was correct (your own graphics shows this).

But all things are not equal as some are using the single core A8 or qualcomm equivalent and some the dual core A9 and graphics performance does scale with CPU performance, just increasing my Galaxy S to 1.2GHz would increase the framerate in benchmarks to match Tegra 2.

However, I'm curious why you have it listed at the bottom that the GS2 scores 43fps in that benchmark...when the chart has it at 16fps? Seems you're doing exactly what you accused me of?

That benchmark was done on a pre-release verison, that was not the final version and is not representative of what the Galaxy II actually scores. Anandtech need to retest, but that probably won't happen until it's released in the US.

Since I have one I know exactly what it scores, you can check the results on GLBenchmark website if you don't beleive me.

 

[JonQ]

There's something wrong with those Nenamark results. Tegra 2 should not be losing to the Adreno 205 by any amount (nor should there be such a huge discrepancy between two similar Tegra 2-based devices on the expanded list). But then again, Nenamark, as I stated, has always been a poor comparison.

Still, where there's smoke, there's fire. If the Mali 400 is continuously beating the Tegra 2 in benchmarks, there has to be a reason for it. But it's clear to me there's an issue with benchmarking the Tegra 2, because the results are all over the place.

ive seen a recent benchmark test with galaxy 2 vs, droid x2.
the specs really make a difference for instance the droid x has a qHD screen (960x540) with 512 mb of ram & a 1 ghz dual cpu (A9) & runs on froyo 2.2...
while the galaxy 2 has a super amole plus screen (800x480) with 1 gig of ram & a 1.2 ghz dual cpu (A9) & runs on gingerbread 2.3.3. now consider these scores & the specs.
on quadrant the galaxy s2 scored 3400, while the droid x2 scored 2800.
on untutu the galaxy s2 scored 2300 on the gpu test, while the droid x2 scored 2480.
on linpack the galaxy s2 scored 60 mflops, while the droid x2 scored 48 mflops.
now consider the fact that the galaxy with all the power & low resolution screen, 1.2 ghz dual core & gingerbread 2.3.3 only only scored a bit higher in 2 tests except in the untutu gpu test, so basically a 1ghz dual core with a qHD screen & only 512 mb's of ram gave the galaxy s2 some tuff competition. now if with those low specs if the droid x tegra chipset were to be overclocked 200 mgz more to equal the clock speed of the exynos 1.2 ghz cpu i would bet that the droid x2 would beat it in most of the tests. but theyre basically so close you wont be able to tell the difference & also your qHD droid x2 has to push more pixels (35%) more! & with 512 mb of ram is really incredible. so i bet that the photon for sprint with 1 gb of ram & with the gingerbread 2.3.3 would give the sgs2 a more equal score or beat it, but in reality i think its stronger for the fact that its keeping up with sgs2 while processing more pixels. i believe that nvidia has either updated there drivers or did something because the latest phone to carry the tegra soc's are able to keep up with the sgs2's pretty well.

 

[JonQ]

That's on the assumption the performance will scale accordingly, it doesn

 

[Mad_Sunday]

Hey,

I need help, with choosing my next android phone.
and i don't know which phone is the BEST for 3D games, the atrix with the Tegra 2 (Im Nvidia Fan!!!) OR the Galaxy S 2 with Mali mp400 unit?

Please help me decide , im want to buy the phone next week!!!


thanks everyone!!!!!!!!!!!!

These phones are not 3D phones, they can't play 3D games, only pseudo 3D

Only choices that can play real 3D games at the moment are HTC evo 3D or the LG Optimus 3D

Pete

 

[Roze]

These phones are not 3D phones, they can't play 3D games, only pseudo 3D

Only choices that can play real 3D games at the moment are HTC evo 3D or the LG Optimus 3D

Pete

I think that's what the OP meant.