[Roman2025]

So, there are a great number of 9.7" 1024x768 IPS Android devices being released... all at varying price points but I think we can generally say between $220 - $320 safely . The two chips that are being used predominately in China Tabs of this size/price point are the Rockchip RK2918 and the Allwinner A10.

This article does a good job comparing these two chips as well as the AmLogic Single Core A9 that is being used a fair bit as well.

http://blog.slate4su...-allwinner-a10/

I don't agree with the article's final rankings (slight difference of opinion... for the most part I do agree). I will say, it convinced me not to bother with the AmLogic A9 chip. IMHO AmLogic really shot themselves in the foot with that chipset by neutering its L2 cache and building it on a "hot running, power eating" 65 nm process. In chipset design, you want more L2 cache (think of it as being really really really fast ram for chipset level operations... someone feel free to correct me on this if I am wrong... so the more of it you have, the more stuff gets done quickly) and you smaller "process"? Basically, how "big" are the circuits on the chip. The smaller they are, the faster the chip can run while consuming less power and not getting as hot. So, for those who are, imho, crazy enough to actually want to overclock a mobile chipset, you want smaller process size. So, AmLogic going A9 and using a Mali-400 GPU are the two saving graces, but they don't do enough. AmLogic did too much damage in the way they designed the chip to begin with and as a result produced a rather mediocre product as far as China Tab chips go.

This little write-up isn't going to deal with the AmLogic chipset though. I am going to focus on the RK2918 and the Allwinner A10.

I recently Acquired a Cube U9GT2 and an Onda Vi40 Elite.

The Cube uses the RK2918 processor.

The Onda Vi40 Elite.

This is utterly unscientific at this point as I have not run any benchmarks between the two, that might come later. First, I will talk about specs on both of these SoC's and then I will get into which is the better buy and why based on my user experience and how I see the current market. So without further ado, in this corner...

RockChip RK2918 (Cube U9GT2 for Reference)
Cortex A8
1.0 Ghz (most common and probably most stable speed)
GC800 GPU

and in the other corner...

Allwinner/Boxchip/SoChip A10 (They are all the same company I think? Onda Vi40 Elite for reference)
Cortex A8
1.0 Ghz
Mali 400 GPU

----------------

Those are very similar sounding chipsets aside from the GPU... And, as far as performance goes, at least in my experience, the same thing can be said. They are very similar except for Graphics performance. I think the A10 chipset also is better at power management because it has dedicated co-processors specifically for power management. Furthermore, some variants of the RK2918 are higher voltage and some are lower voltage. Usually the higher voltage chips are in bigger tablets (like the 9.7" models) and the Cube U9GT2 is an example of this. Higher voltage generally means higher power draw. As far as I am aware though, all A10 chips are low voltage.

Both chips were designed with Video Playback in mind and Rockchip in particular has a lot of experience in building chipsets for media playback in general. That being said, I think the RK2918 might edge out the A10 a bit in compatability. You will see it decode more file types in hardware than perhaps the A10. As I said though, this is utterly unscientific. This has been the "feel" that I get and my experience thus far with the two processors. Now, the A10 is much much more powerful for video processing. They advertised the Novo 7 Advanced (the first tablet that I am aware of to feature the A10) as being able to hardware decode 2048p video and output that through HDMI. And lo' and behold I think it actually WASN'T just marketing hype because several people tested 1080p 3D video playback and the chipset could handle it. (1080 in 3D is two separate video streams playing at the same time so it would effectively be 2048p.) So, while RK2918 might have an edge in compatability, the A10 still wins for brute force.

UI performance - The two chips are pretty much tied I think. However, in my experience with several different tablet models featuring both chipsets... the A10 seems to be more consistent as far as providing smooth UI performance. RK2918 parts however tend to fluctuate. Most of the newer tabs I have seen with the RK2918 though tend to have very smooth performance.

Web Browsing Speed - So here was the test, I have both my Cube U9GT2 and my Onda Vi40 Elite in front of me. I typed in Ebay, Amazon, Slatedroid, and CG talk and tested load times simultaneously on each for each website (I pressed the enter key on the soft keyboard at the same time). The RK2918 was a tiny tiny smidge faster on several site, and the A10 was a smidge faster on a smaller number of site. This one was a tie.

Web Browsing - Pinch-to-Zoom - A10 won this one, it isn't perfect, but it is smoother than the RK2918. It isn't a huge difference but it is noticeable when they are side by side.

Stability - Rockchip wins this competition. Specifically referencing the Onda Vi40 vs the Cube U9GT2 - There is NO difference. They both seem pretty much rock solid, no freezing, no crashing. HOWEVER, in the greater android Tablet-Sphere (I just made that word up) I have had much better luck with RK2918 devices as far as quality control and stability goes at the chipset level. Let me reiterate, we are talking about chipsets and I am talking about firmware/chipset level stuff irregardless of the rest of the tablet. The Wopad I7 had some of the worst build quality of any tablet I had sold, but the software was rock-solid. The Momo 9C and Novo 7 Advanced were/are both very well designed and well built units. They are also both performance/gaming monsters. However, as a vendor, I have had more returns on those units than any of my other models. Most often for problems like... random frequent crashing that wasn't fixed by firmware re-flashing & Wifi instability. I think, these are both a testament to Allwinner quality control and/or Allwinner firmware bugginess. With that being said, all of the new Ainol tablets (the Elf, the Aurora) and my more recent batches of Ployer Momo 9's have been just fine. But, as it stands, I give the stability crown to Rockchip.


GPU Performance - A10, A10, and yes A10... The Vivante GC800 GPU CANNOT KEEP UP WITH THE MALI-400. This is where I differ with the review in the link I referenced above. They mention that the Vivante GC800/RK2918 combo does better for gaming than the Mali-400 because the GC800 has a much higher clock-speed than the Mali.... This just isn't the case at all though. The Vivante GC800 is decent, but when you are dealing with a 1024x768 or for that matter even a 1024x600 display, it cannot keep up with the Mali-400. Why? Piss-poor 2D rendering is one reason... so games that use large textures (all that pretty stuff) really will start to choke. That means really advanced 2D titles will/could possibly lag although I haven't done much testing in this area. (I think this is in part why even the Android UI on high resolution devices using the RK2918 can sometimes stutter a bit, this might just be poor firmware though too...)

What games have I looked at? Two titles in particular... Shadow Gun & Dead Space... Both of which are extremely graphically advanced titles for Android. I wondered if the Mali-400 would be able to push them at 1024x768. It does, and it does so very very smoothly. The Vivante GC800 will also play them, but they are "chunky" with a low frame-rate. Playable? yes... enjoyable... well, not so much. Also, for Shadow Gun, there is a major graphical anomaly on devices using GC800 GPUs that basically causes the textures on the characters to look all warped and twisted. If you watch heavy gunner being played on the Cube U9GT2, you will see what I am talking about (notice the player character's arm).You will also see that it is very very chunky and the game stutters quite a bit.

Lets talk a bit more about the Mali-400 and why it is so cool and superior for gaming... First, I only recently found this out, but it was actually developed BY ARM. The same folks that develop the reference designs for mobile chipsets that most every else buys rights for and bases their chipsets on. So you have probably heard it said, XYZ Chip is a "Cortex-A8" or the Tegra II is a Dual-Core Cortex-A9.... "Cortex-A8/9" (or for older chipsets, "ARM 11") has to do with what ARM Reference design the chip was built around. So, ARM recently also designed the Mali-400 which also adheres closely to a "mobile standard" that, in the coming months, will make the chipset one of the most compatible for games. The Mali-400 is ALSO used by chip/tablet making giant Samsung in their latest flagship Exynos II SoC. So, even better, is that this will help strong arm PIA (Pain in the Arse) developers like Gameloft to actually develop according to a standard rather than continuing to do their own crap. If you ever wonder why their games only work on a narrow margin of Android devices it is because their devs do things to make their games work better on specific and more proprietary GPU's, specifically the PowerVR chipsets. Why do they do this? Because they are primarily an Apple game developer and Apple chipsets use the PowerVR GPU. So, with Samsung transitioning towards a standard by using the Mali-400, developers like Gameloft will start developing according to that standard if they want to sell games that work on the most popular devices. So... the reason for that rabbit trail... The A10 Chipset is going to be more compatible with more High-End games over the next year because it is similar to the Exynos II in that it uses a Mali-400 GPU. The Exynos II is still a much much faster (and more expensive) chipset, so I am certainly not equating the two. What I am saying is that the Mali-400 GPU is an absolute beast of a processor and it is also going to be a standard for devs who want their games to work on the best devices which means folks that are using an A10 processor, will probably have better luck with game compatibility in the long run.

CONCLUSION:

The RK2918 and the A10 are somewhat evenly matched. But I think, and my experience definitely holds true for this, that the A10 has a significantly more powerful GPU and it affects the entire system and will continue to do so as the GPU is relied upon more heavily as hardware acceleration is becoming the standard for everything. If you already bought an RK2918 powered device, don't feel bad about your purchase. It is a solid platform, reliable, and does provide decent performance.

However, if you are currently "looking"... My 2-cents is don't bother with the RK2918. The A10 is either matched or better than this chipset and I think it is actually the cheaper of the two however, probably not enough to influence the final price a device manufacturer sticks on their tablet. The prices are comparable though and there really is no reason to buy an RK2918 powered device at this point as their are a plethora of options using the A10.

I look forward to everyone's thoughts

Kind Regards,

Roman

http://www.topnotchtablets.com

This post has been edited by Roman2025: 24 March 2012 - 01:13 PM

[Miss Dragon]

Ok, my first thought, you did a great job for this long whiting.

[Tablet_Collector]

Thanks very much for the detailed and well written comparison.

It is a question I have wondered the answer to for some time. I am looking at replacing my zt280 with a 9.7 inch ips tablet and it really seems to be between the A10/mali and rk2918/gc800 cpu/gpu.

Recently I have also been looking at the smartq ten t15 which incorporates a dual core ti omap 4430 cpu and powervr gpu. I would be interested to see how this compares to the A10/rk2918, although probably an unfair comparison

Jez

Sent from my ZT ICS using Tapatalk

[Nickos-V]

Roman, great job with your write up. I agree with the conclusion. AllWinner A10 is the better chip and people should be buying tablets with that. Buyers should pass on RK2918 tablets.

A10 benefits over RK2918 (both have same CPU performance; ie, Linpack scores)
-somewhat more responsive because of multiple co-processors (minor +)
-GPU can handle more 3D graphic intense games better (major +)
-power consumption somewhat better over 3.7v RK2918 devices (minor +)

I will point out that:
GC800 (RK2918) - 4 texture units, clocked around 575MHz, single core GPU
Mali 400MP (A10) - 8 texture units, clocked either 300 (or 250?) MHz, dual core GPU; ie, 2 X 300MHz GPUs.

What happens is, for simple games (less or simpler textures), the RK2918 will handle them slightly faster & smoother because graphics going to 1 GPU core. For A10, graphics go to 2 GPU cores, which slows things down a little but for complex textures, having 2 GPUs with an extra 4 texture units better handles the graphics.

Note: Having less texture units on the GC800 is what negatively impacts its performance.
If you go into Quadrant bench you can check GPU info. ie, Wopad i7: max texture units:4, max texture size:8192, max lights:8

http://reviews.imp3....artq/t15/43.jpg

From above link of GLBenchmark comparison. Notice how:
#1. 8726-M3 & A10 have very close GPU performance? Difference is probably because of clock speed of the GPUs. 8726 uses 250MHz Mali400s & A10 uses 300MHz Mali400s though I have not been able to confirm this info.
#2. GC800 just barely beats out Mali400 (A10) in the standard test (ie, less textures). ie, scores are very close. 19.1 vs 18.8 FPS
#3. Mali400 hits GC800 hard in high quality (PRO) test. (ie, lots more textures to deal with). ie, 19.6 vs 10.5 FPS.
#4. SGX540 does somewhat bad in the S5PC110 (A8) but fairly good in OMAP4 (dual A9s).

So, in terms of better gaming performance on Chinese brand tablets. I would say choose A10 over RK2918.

This post has been edited by Nickos-V: 24 March 2012 - 03:59 PM

[Roman2025]

Nickos-V, on 24 March 2012 - 03:37 PM, said:

Roman, great job with your write up. I agree with the conclusion. AllWinner A10 is the better chip and people should be buying tablets with that. Buyers should pass on RK2918 tablets.

A10 benefits over RK2918 (both have same CPU performance; ie, Linpack scores)
-somewhat more responsive because of multiple co-processors (minor +)
-GPU can handle more 3D graphic intense games better (major +)
-power consumption somewhat better over 3.7v RK2918 devices (minor +)

I will point out that:
GC800 (RK2918) - 4 texture units, clocked around 575MHz, single core GPU
Mali 400MP (A10) - 8 texture units, clocked either 300 (or 250?) MHz, dual core GPU; ie, 2 X 300MHz GPUs.

What happens is, for simple games (less or simpler textures), the RK2918 will handle them slightly faster & smoother because graphics going to 1 GPU core. For A10, graphics go to 2 GPU cores, which slows things down a little but for complex textures, having 2 GPUs with an extra 4 texture units better handles the graphics.

Note: Having less texture units on the GC800 is what negatively impacts its performance.
If you go into Quadrant bench you can check GPU info. ie, Wopad i7: max texture units:4, max texture size:8192, max lights:8

http://reviews.imp3....artq/t15/43.jpg

From above link of GLBenchmark comparison. Notice how:
#1. 8726-M3 & A10 have very close GPU performance? Difference is probably because of clock speed of the GPUs. 8726 uses 250MHz Mali400s & A10 uses 300MHz Mali400s though I have not been able to confirm this info.
#2. GC800 just barely beats out Mali400 (A10) in the standard test (ie, less textures). ie, scores are very close. 19.1 vs 18.8 FPS
#3. Mali400 hits GC800 hard in high quality (PRO) test. (ie, lots more textures to deal with). ie, 19.6 vs 10.5 FPS.
#4. SGX540 does somewhat bad in the S5PC110 (A8) but fairly good in OMAP4 (dual A9s).

So, in terms of better gaming performance on Chinese brand tablets. I would say choose A10 over RK2918.

nice :-)

Thanks for adding some numbers to the above review.

I would also suspect that aMlogic is using a slower clocked gpu. They probably have to because of the 65nm production process vs the smaller better 55nm process used in the A10.

I am interested in the new oMap processor.Ti has been using powerVR chipsets for a long time. I think we are going to see a shift away from chips like the Tegra and PowerVR that are more proprietary and don't adhere to industry standards. This will hopefully help to finally counteract the inherent fragmentation that does exist with android devices.

My 2 Cents,

Roman

[Roman2025]

rowwor, on 24 March 2012 - 01:20 PM, said:

Ok, my first thought, you did a great job for this long whiting.

Thanks! , it was a work over the course of a couple of days and my hand is now a bit cramped lol

[Nickos-V]

Roman2025, on 25 March 2012 - 04:16 PM, said:

Thanks for adding some numbers to the above review.

Yes, many of the Chinese chip makers are switching over to Mali 400 GPU which is a very good GPU compared to the other ones that were being using.

Little worried though looking @ that Adreno 220 GPU score for high end GLBenchmark tests. It seems to suffer same problem as RK2918 but have not confirmed if this score is right. GLBenchmark website is getting updated today and not working right. The bad news is that Google with their Nexus tablet are very likely getting Snapdragon w/Adreno GPU. Which maybe bad for high end 3D gaming if those results are true. Adreno 220 scores very high in NenaMark bench but that only does a specific benchmark. ie, probably more simple 3D bench.

That is why I dislike using GPU benches because many just check a certain game scene and not representative of how good the GPU really is.

I did find a good GPU benchmark which runs different tests to see how good the GPU really performs. It's called An3dBench. Also, there is an XL version which is supposed to have more intense game tests but I prefer the standard version myself.

Wopad I7 (RK2918) An3dBench @ 800x480
Fillrate ST/MT: 21.35/21.38 MP/s
High object count: 18.01 fps
Multiple lights: 55.27 fps
High polygon count: 30.5 fps
Keyframe animation: 55.52 fps
Game level: 39.95 fps
TOTAL SCORE: 5404

On the i7 (RK2918), you can see that the high object count is very weak, the high polygon count is kinda weak and game level is almost good but not great. Also, this test was performed @ 800x480. Meaning @ at higher res., the RK2918 would start to stutter on really intense 3D games. ie, 3D racing and first person shooter with high detail graphics.

Edit: removed Ten3 An3dBench because was run @ 455x320. Way too low for fair comparison to the i7.

This post has been edited by Nickos-V: 26 March 2012 - 08:20 PM

[Nickos-V]

Seems the Ten3 ran An3dBench @ much lower resolution. No wonder it did so well. Wish I could have seen the results @ 1024x768. Will have to edit my previous post.

GLBenchmark probably the best choice for testing out mobile GPUs.

[Roman2025]

Thanks for the updates. I always appreciate your input as you seem to have the inside scoop on chipset/hardware stuff .

[ralp]

thanks good point. i was thinking about Gemei G9 and Cube U9GT-2. i think i will buy G9 after reading this article

[Dansplans]

Can I throw a wrench in here? A Soc that is starting to appear in tablets as we speak.

A new chip in town marries a dual core Cortex A5 (low power consumption) with a Mali400. According to benchmarks I've read the dual A5 beats both A8 and A9, and of course you have already sung the praises of the Mali400.

Don't get me wrong, I love my A10 devices, but if this new Soc delivers, it could alter the landscape much as the A10 did...

[Roman2025]

Dansplans, on 28 March 2012 - 09:41 PM, said:

Can I throw a wrench in here? A Soc that is starting to appear in tablets as we speak.

A new chip in town marries a dual core Cortex A5 (low power consumption) with a Mali400. According to benchmarks I've read the dual A5 beats both A8 and A9, and of course you have already sung the praises of the Mali400.

Don't get me wrong, I love my A10 devices, but if this new Soc delivers, it could alter the landscape much as the A10 did...

I have done some reading on the A5. It was originally designed by Arm to be a cheaper alternative to the Cortex-A8 and/or (I think...) a new co-processor base design for things like DSPs and PMU and all those other acronyms that represent different co-processors. (I think) many co-processor chips are actually currently designed around Arm 11.

However, some A5s in testing have been shown to exceed A8 chipsets and nearly match A9 chipsets. With that being said...

The Amlogic Blah Blah AM3 is an A9 (granted it is single core) and the A10 and probably RK2918 (both cortex A8's) are both better choices. Just because the chip is built around a good reference design like Cortex-A9 doesn't immediately make it a winner. The NEC EV2 Dual-Core A9 chipset also comes to mind. It was decent (actually quite speedy) but buggy and clocked quite slowly @ 533 MHz per core. The thing that probably hurt it the most though was that it had an under-powered PowerVR-530 GPU... all of which is part of the chipset design...

So ultimately it will come down to the actual specific chipsets and how the perform when review time comes around . If the companies that are embracing the A5 reference designs play their cards right though, I think we could see some very interesting chips hitting the budget market in the near future.

Cheers,

Roman

This post has been edited by Roman2025: 30 March 2012 - 01:41 AM

[Roman2025]

ralp, on 28 March 2012 - 12:43 PM, said:

thanks good point. i was thinking about Gemei G9 and Cube U9GT-2. i think i will buy G9 after reading this article

Be careful ordering the Gemei G9... I read many spec sheets that say it only has a 3800 mAh battery... That is going to be absolutely horrid battery life... like 3 - 4 hours max... . Those are most likely "knock off" devices ... "fakes"...

The real Gemei G9 should have a 7000mAh battery. Don't settle for anyone selling anything at all different.

-Roman

[Joan Segura]

Hello,

as the only person in the world doubting between Onda vi40 or Cube U9GT2 I'd like to thank your detailed post. And now the winner is... I don't know hahahahaaaa.

Thanks and best regards,
Joan.

[GAB]

Nickos-V, on 24 March 2012 - 03:37 PM, said:

Roman, great job with your write up. I agree with the conclusion. AllWinner A10 is the better chip and people should be buying tablets with that. Buyers should pass on RK2918 tablets.

A10 benefits over RK2918 (both have same CPU performance; ie, Linpack scores)
-somewhat more responsive because of multiple co-processors (minor +)
-GPU can handle more 3D graphic intense games better (major +)
-power consumption somewhat better over 3.7v RK2918 devices (minor +)

I will point out that:
GC800 (RK2918) - 4 texture units, clocked around 575MHz, single core GPU
Mali 400MP (A10) - 8 texture units, clocked either 300 (or 250?) MHz, dual core GPU; ie, 2 X 300MHz GPUs.

What happens is, for simple games (less or simpler textures), the RK2918 will handle them slightly faster & smoother because graphics going to 1 GPU core. For A10, graphics go to 2 GPU cores, which slows things down a little but for complex textures, having 2 GPUs with an extra 4 texture units better handles the graphics.

Note: Having less texture units on the GC800 is what negatively impacts its performance.
If you go into Quadrant bench you can check GPU info. ie, Wopad i7: max texture units:4, max texture size:8192, max lights:8

http://reviews.imp3....artq/t15/43.jpg

From above link of GLBenchmark comparison. Notice how:
#1. 8726-M3 & A10 have very close GPU performance? Difference is probably because of clock speed of the GPUs. 8726 uses 250MHz Mali400s & A10 uses 300MHz Mali400s though I have not been able to confirm this info.
#2. GC800 just barely beats out Mali400 (A10) in the standard test (ie, less textures). ie, scores are very close. 19.1 vs 18.8 FPS
#3. Mali400 hits GC800 hard in high quality (PRO) test. (ie, lots more textures to deal with). ie, 19.6 vs 10.5 FPS.
#4. SGX540 does somewhat bad in the S5PC110 (A8) but fairly good in OMAP4 (dual A9s).

So, in terms of better gaming performance on Chinese brand tablets. I would say choose A10 over RK2918.

I have MOMO15 (A10/512DDR3/10") and it has ICS 4.0.3:
1. I get Egypt standrad 26FPS and high 25FPS.
2.In quadrant standard i get next result:
total:1757
CPU:1408
Memory:1300
I/O:3634
2D:885
3D:1559
3. the unit is NOT stable and it crash at least 2-3 per day!
4. The result using ANDROID 2.3.4 was giving 2D result of less than to 200
and unit was less stable than ICS 4.0.3 and some games like airAtack HD looks much better.
5. The units battery can last for at least 6-7 hours.

The unit is more stable in ICS 4.0.3 than 2.3.4.
Preformance in ICS is better mainly in 2D.
GPU looks to work better in ICS.

THE MAIN PROBLEM IS STABILTY!
So even that in terms of preformance is better the stabilty in my point of view is
not less important so i don't give the crown to A10 at this point.

This post has been edited by GAB: 08 April 2012 - 03:31 PM

[Nickos-V]

GAB, on 08 April 2012 - 03:14 PM, said:

THE MAIN PROBLEM IS STABILTY!
So even that in terms of preformance is better the stabilty in my point of view is
not less important so i don't give the crown to A10 at this point.

My (rk2918) GLBenchmark results can be found here:
http://www.slatedroi...323#entry345323

I agree that stability is very important. Surprised your tablet is crashing on you. What is max CPU clock? Is it 1 or 1.2 GHz? If 1.2 GHz, can you use some app to limit it to 1 GHz. These CPUs were not meant to run any faster than 1 GHz.

If CPU clock has 1 GHz for max speed then either OS issue or defective unit. Since your tablet seems to crash on Android 2.3 & 4.0 then appears to point towards hardware issue. Something you will have to follow up in the MOMO forums. ie, if this only happens to you then maybe defective tablet.

This post has been edited by Nickos-V: 08 April 2012 - 04:31 PM

[GAB]

Nickos-V, on 08 April 2012 - 04:30 PM, said:

My (rk2918) GLBenchmark results can be found here:
http://www.slatedroi...323#entry345323

I agree that stability is very important. Surprised your tablet is crashing on you. What is max CPU clock? Is it 1 or 1.2 GHz? If 1.2 GHz, can you use some app to limit it to 1 GHz. These CPUs were not meant to run any faster than 1 GHz.

If CPU clock has 1 GHz for max speed then either OS issue or defective unit. Since your tablet seems to crash on Android 2.3 & 4.0 then appears to point towards hardware issue. Something you will have to follow up in the MOMO forums. ie, if this only happens to you then maybe defective tablet.

my CPU is set to 1.2Ghz reducing the CPU speed affect screen touch screen.

ICS 4.0.3 is more stable than 2.3.4. i think that 512RAM for 10" is problematic...

[Roman2025]

GAB, on 08 April 2012 - 03:14 PM, said:

I have MOMO15 (A10/512DDR3/10") and it has ICS 4.0.3:
1. I get Egypt standrad 26FPS and high 25FPS.
2.In quadrant standard i get next result:
total:1757
CPU:1408
Memory:1300
I/O:3634
2D:885
3D:1559
3. the unit is NOT stable and it crash at least 2-3 per day!
4. The result using ANDROID 2.3.4 was giving 2D result of less than to 200
and unit was less stable than ICS 4.0.3 and some games like airAtack HD looks much better.
5. The units battery can last for at least 6-7 hours.

The unit is more stable in ICS 4.0.3 than 2.3.4.
Preformance in ICS is better mainly in 2D.
GPU looks to work better in ICS.

THE MAIN PROBLEM IS STABILTY!
So even that in terms of preformance is better the stabilty in my point of view is
not less important so i don't give the crown to A10 at this point.

RK2918 is a more stable SoC... I will give you that. But the kind of problems you are describing with your A10 device are not common across all A10 tablets... not at all. My Elf is on Elvish .3 firmware right now and solid as a rock and it runs great.

Also, the GC800 GPU in the Rockchip RK2918 is not nearly as well supported for gaming as the Mali-400 which leads to artifacting and jacked up textures and bumpmap and normal map screw ups in games like Shadow Gun... Not too mention much less performance.

Yes.. the crown does go to the A10... I have owned/do own/sold many A10 devices and I own and have sold several different RK2918 powered devices. The A10 is the hands-down winner in "bang for your buck" as long as you get a stable model. Which is all I sell... Hence "Top Notch Tablets"... we try to avoid the crap that doesn't work well. Even we make mistakes though. The Novo 7 Advanced had stability issues and we dealt with a lot of returns and replacements as a result. But the ones that were stable, ran excellently well and the only complaint I had was the poor WiFi.

You are trying to run ICS on a 512 Mb device too... that in large part is your issue. ICS eats up resources like crazy. My Elf Tablet usually only has a little over 500mb available of memory and it is a 1 Gb device.... where do you think that leaves the Momo 15 (and my N7 Advanced when I try to run ICS on it...)?

It is a shame ployer didn't provide stable GB firmware for the Momo 15. I have had some QC issues with the Momo 9 as well... this is another reason why we are looking towards releasing our own brand:

Quantum: http://www.slatedroi...series-roadmap/

The first two devices will be based around the A10 and we are going to try to get an early RK30xx device on the market later this summer.

Anyhow, benchmarks, real-world use, pricing, etc. all lead to the A10 being the superior chipset and clearly the better buy. As usual though, in the world of China Tabs, you have to be very careful about what model you get.

Kind Regards,

Roman

This post has been edited by Roman2025: 14 April 2012 - 12:56 PM

[Roman2025]

Had someone else PM me about how the S5PV210/110 SoC stacks up in this comparison. I haven't personally owned a device powered by this chipset but I have done a whole lot of research around it.... here is my 2-cents...

---------------------------------
It was the top of the heap for a long time. However... it can't process video nearly as well as either the RK2918 or A10 platform.

It is great for gaming though but the A10 can keep up. The RK2918 cannot.

It is more power hungry than both the RK218 and A10.

My guess is that android performance might be a bit smoother on the S5PV210/110 but I haven't ever owned a device to compare.

The fact that it eats through battery, can't smoothly decode HD video (not nearly as well as the A10 or RK2918) and the fact that the A10 is generally cheaper with as good or in some cases/benchmarks better gaming performance makes me lean toward the A10 platform.

------------------------------

With that being said, it is probably one of the most compatible chipsets as far as having a wider range of apps and games that will work well on it. It is also very well supported by the community. If you get an S5PV110 with a good size battery, it is still a good device. You just aren't going to be watching a high quality bluray rip of Lord of the Rings on it pumped out over HDMI to your 50" plasma... The A10 can handle this and in 3D if you really want something to show off... (that would be dual simultaneous 1080p decoding...)

So it really does come down to "needs" and "wants" when comparing the S5PV210/110 with the A10.

To be clear... my understanding is that the S5PV210/110 can handle 720p video in most formats without too much issue but that is chokes on most 1080p video.

Cheers,

Roman

This post has been edited by Roman2025: 14 April 2012 - 12:54 PM

[Nickos-V]

In terms of gaming performance and power use; the A10 is the clear winner!!! =)

The S5PV210 uses more power. Most common Samsung chip found in Chinese tablets. S5PC110 should do kinda better power wise but probably not as good as the A10.

I agree with Roman that the AllWinner A10 is the best Cortex A8 chip out there. Forget the rest. Either go for A10 tablet today or wait for Chinese dual core: 8726MX or RK3066 with tablets in next 2-5 months.

GLBenchmark 2.1 results at 800x480 to prove this point:
                        Nexus S    Novo 7    I7
                        S5PC110    A10     RK2918
Egypt High (FPS):        27.1      30.8      15
Pro High (FPS):          33.0      49.0      18

http://www.glbenchma...stgroup=overall
http://www.glbenchma...nchmark=glpro21

You can clearly see that the A10 chip with Mali400MP beats the other ARM chips in the high-end 3D benchmarks.

This post has been edited by Nickos-V: 15 April 2012 - 03:52 PM