[Nickos-V]

sikke, on 07 February 2013 - 01:33 PM, said:

offtopic but heres rk3188 quadcore results

http://blog.t-kstore...he-pipo-max-m9/

Thanks for the link!!! Exactly what I wanted to see. I started a new thread for it to discuss further.

Later,

[Tablet_Collector]

Hi Nickos-V,

I have added some results for my Ramos W42 from here and put together a spreadsheet showing all the results (Momo12, W42, Chuwi V10 and M9).

Note: To make the comparison on a level playing field all tablets have 1280 x 800 resolution and, where possible, the same version of Antutu (v3.1.1) is used.

You need to view this in a browser to see the correct formatting (it looks bad from Tapatalk)...

Antutu 3.1.1                                                                                           Antutu 3.0.3
           All 1280 x 800                                                                  FW 10/1/13
           Ployer Momo12   Ramos W42     Chuwi V10      Pipo M9    @ 1800 Mhz    Venus7      Venus7      Ampe A10
              RK3066      Exynos 4412     All A31        RK3188      RK3188     ATM7029     ATM7029      FS i.MX6
RAM             1810          3095          1536          3217        3601        1762        726          1997
CPU Int         1761          4291          2889          4973        5567        2622        1510         3146
CPU Float       1829          3169          1827          3596        4025        2458        334          2288
2D              1187          1144          1034          1415        1584        921         803          662
3D              3697          2438          4528          3988        4464        2347        1453         1685
Database        555           420           425           545          610        330         315          470
SD write         67           150           150            75          84          58          52          137
SD read         204           172           203           204          228        101         106          204
CPU Freq      1608 x 2      1400 x 4      1008 x 4      1608 x 4    1800 x 4    1200 x 4    1200 x 4    1200 x 4?

Totals         11,110        14,879        12,592        18,013      20,164      10,599      5,299        10,589

The most noticeable thing for me is the poor 3D results from the 4412 compared to the RK3066 which is not far behind the RK3188. Surely this shows that there is a driver/firmware issue with the Exynos in the W42? I wonder if the Ramos W30 fares any better? Will see if I can dig up any Antutu results.

Jez

This post has been edited by Tablet_Collector: 13 February 2013 - 08:49 AM
Reason for edit: Replaced Hero II with 2 x Venus results

[Roman2025]

I am surprised the 3D GPU results weren't any better on the RK3188? I was under the impression that even though it has a Mali-400 MP4 GPU that it was clocked significantly higher than the RK3066 GPU.

Also, what is up with the 1.6 Ghz Core clock? Were we not promised 1.8 - 2.0 Ghz for the RK3188

Meh...

Anyone have Allwinner A31 Quad numbers to compare against just for reference? I am really curious now how it stacks up. I am not expecting it to fare well but then again I expected much more out of the RK3188.

Regards,

Roman

[Nickos-V]

Roman2025, on 08 February 2013 - 02:05 PM, said:

I am surprised the 3D GPU results weren't any better on the RK3188? I was under the impression that even though it has a Mali-400 MP4 GPU that it was clocked significantly higher than the RK3066 GPU.

Also, what is up with the 1.6 Ghz Core clock? Were we not promised 1.8 - 2.0 Ghz for the RK3188

Anyone have Allwinner A31 Quad numbers to compare against just for reference?

Yes, I and others were surprised to see that 533 MHz Mali-400MP4 GPUs were not much better than one found in RK3066. ie. RK3188 has slightly better 3D performance. 4412 3D scores (in quadrant & Antutu) are not good though and hopefully firmware update will improve that.

They all lied about CPU frequencies this time around.
AllWinner A31 I saw 1.4 GHz - ended up 1 GHz
Actions ATM7029 said 1.5 GHz Cortex A9 - ended up 1.2 GHz Cortex A5
RK3188 said 1.8 GHz - ended up with 1.6 GHz

Quad 1.6 GHz A9 is really great but I get tired of the lying from the Chinese. Rockchip also gives out the smaller lies compared to their competitors.

Antutu 3.1 results for Chuwi V10 Quad (A31 with 1280x800). Maybe Jez will add them in to compare easier?
http://mp3.zol.com.cn/352/3526740.html

A31 has the strongest 3D score of them all. Also, really strong SD write like 4412. Biggest problem is Cortex A7 @ 1 GHz is holding it back. Cortex A7 @ 1.5 (or 1.4) GHz would have given much better RAM & CPU scores. A 1.5 or 1.4 GHz A31 would have given a very strong ARM chip and really great.

This post has been edited by Nickos-V: 08 February 2013 - 04:27 PM

[Tablet_Collector]

Nickos-V/Roman,

I agree the RK3188 results aren't quite as stellar as I was expecting either. Thanks for the Chuwi V10 Antutu results link Nickos, as per your suggestion I have added the A31 figures for the Chuwi V10 to the above table for comparison. I have also added the Antutu total scores.

I have also added theoretical figures for the RK3188 if they ever do manage to clock it at 1.8 Ghz.
Where Theoretical result = (1800/1608) x Pipo M9 result

Might be fun to add in Actions ATM7029 and Freescale iMX6 too. EDIT: Now added these too. I could only find Antutu 3.0.3 results for these so if anyone can find Antutu 3.1.1 then please let me know. I know there is some debate over whether ATM7029 firmware has a "cheat" for Antutu so please treat the results carefully.

I think it would be useful to add rankings for various categories e.g. 2D, 3D, RAM etc. What do you folks think?

Cheers
Jez

This post has been edited by Tablet_Collector: 10 February 2013 - 05:36 AM

[janesh]

Hi Jezzer, i found here a antutu 2.9.4 benchmark of the Ramos W30, and here another, W30HD compared with Nexus 10.

This post has been edited by janesh: 10 February 2013 - 10:32 AM

[Nickos-V]

Tablet_Collector, on 09 February 2013 - 07:08 PM, said:

Might be fun to add in Actions ATM7029 and Freescale iMX6 too. EDIT: Now added these too. I could only find Antutu 3.0.3 results for these so if anyone can find Antutu 3.1.1 then please let me know. I know there is some debate over whether ATM7029 firmware has a "cheat" for Antutu so please treat the results carefully.

Yes, ATM7029 score dropped with 3.1 to around 6500. Then Ainol released new firmware and score is now 10500 to 11500. That is why I don't trust Antutu that much. Other benchmarks like Vellamo stay consistent. Quadrant probably better one to test with too but Antutu has now become the popular benching app. Used to be Quadrant about 1.5 years ago.

This person has both Hero Quad & Venus but did not give detailed Antutu 3.1 results. I will post and ask him to include those.
http://www.slatedroi...ge__pid__594514

You can rank each category if you like to. Why not. People should keep in mind that certain issues may not show up though. Like Youtube stuttering which affects RK3066. I am wondering if it will affect RK3188 too.

Some benchmarks are better than others and good to test with multiple ones when possible. That way will know for sure if one of the benches is being manipulated.

Taken from Mc-Bub Blog - "Review of Ainol Novo 10 Hero II Quad Core Tablet PC 10.1 inch IPS Android 4.1"


Notice how Hero II beats X7 in CF-Bench but in Sunspider & V8 does poorly. All of those tests are CPU intense. How can X7 score badly in CF-Bench but excellent in Sunspider & V8??? Makes little sense to me. ie. I can see how CF-Bench would score the X7 lower because it uses dual core 1 GHz but why it performs way better in those other two tests is a real mystery. X7 even outscores W30 in those two benches. How can a dual 1 GHz A9 chip beat out a quad 1.4 GHz A9 chip??? The 1.4 GHz A9 should have won either way whether the tests were single or multi-threaded! ie. CF-Bench seems to give more accurate & realistic results. Somehow Sunspider & V8 were manipulated by OMAP4 firmware to show really great performance for that chip?

This post has been edited by Nickos-V: 10 February 2013 - 11:25 AM

[Tablet_Collector]

janesh, on 10 February 2013 - 08:47 AM, said:

Hi Jezzer, i found here a antutu 2.9.4 benchmark of the Ramos W30, and here another, W30HD compared with Nexus 10.

Thanks janesh the W30 scores are similar to the W42 but not quite as good. Maybe because it is Antutu 2.9.4 vs 3.1.1.

That W30HD vs Nexus 10 was interesting. The Nexus 10 does better overall but then you would expect that at double the price. For me the biggest weakness with the Nexus 7 and 10 is the lack of a TF card slot. Anyway better stop before I go too far off-topic

This post has been edited by Tablet_Collector: 11 February 2013 - 09:38 AM

[Tablet_Collector]

Nickos-V, on 10 February 2013 - 11:17 AM, said:

{snip lots of useful information}

Thanks for that Nickos-V. Just shows how certain benchmarks can be manipulated.

Just for comparison I ran CF-Bench v1.2 on my Ramos W42...



Pretty good result, actually scoring better than the Galaxy S3 overall

Hey janesh, how does the Momo11 Quad do on CFBench? I imagine it would be even better.

This post has been edited by Tablet_Collector: 11 February 2013 - 09:42 AM

[janesh]

Slightly better Jez, Ployer momo11 quad score one time 14206, after that between 14150 and 14200.
My Ramos W42 one run, score 13960.
I ran geekbench2 on both devices, Ramos W42, 1880, ployer momo11 quad, 1967, which are very good scores.
Overall, momo11 quad feels slightly faster and both devices, no lag, browser lag or force closes.
When my wife, ipad user, does not grumble using that ployer and especially the Ramos W42, it should be a very good device.

[vvuv]

Nickos-V, on 08 February 2013 - 02:50 PM, said:

Yes, I and others were surprised to see that 533 MHz Mali-400MP4 GPUs were not much better than one found in RK3066. ie. RK3188 has slightly better 3D performance. 4412 3D scores (in quadrant & Antutu) are not good though and hopefully firmware update will improve that.

They all lied about CPU frequencies this time around.
AllWinner A31 I saw 1.4 GHz - ended up 1 GHz
Actions ATM7029 said 1.5 GHz Cortex A9 - ended up 1.2 GHz Cortex A5
RK3188 said 1.8 GHz - ended up with 1.6 GHz

Quad 1.6 GHz A9 is really great but I get tired of the lying from the Chinese. Rockchip also gives out the smaller lies compared to their competitors.

Antutu 3.1 results for Chuwi V10 Quad (A31 with 1280x800). Maybe Jez will add them in to compare easier?
http://mp3.zol.com.cn/352/3526740.html

A31 has the strongest 3D score of them all. Also, really strong SD write like 4412. Biggest problem is Cortex A7 @ 1 GHz is holding it back. Cortex A7 @ 1.5 (or 1.4) GHz would have given much better RAM & CPU scores. A 1.5 or 1.4 GHz A31 would have given a very strong ARM chip and really great.

Not exactly about Rockchip. When the first RK3066 tablets were released (as well as engineering samples) they were initially clocked at 1.2-1.4ghz. Now the newest firmwares increased their clock speed to 1.6ghz (1608mhz) per core.

Same thing as RK3066 will probably be for the RK3188. This PiPo M9 is an engineering sample, The next firmware would probably clock the RK3188 to 1.8ghz.

[Nickos-V]

vvuv, on 11 February 2013 - 05:36 PM, said:

Not exactly about Rockchip. When the first RK3066 tablets were released (as well as engineering samples) they were initially clocked at 1.2-1.4ghz. Now the newest firmwares increased their clock speed to 1.6ghz (1608mhz) per core.

Good point. Have to wait and see.

My U30GT came at 1.6 GHz and I was in the 1st batch. But, Yuandao clocked their N101 at 1.4 GHz and later upped it to 1.6 GHz. So, very possible that tablet makers start with 1.6 GHz and raise the clock to 1.8 GHz. Time will tell for sure. 1.6 GHz quad is still great. Exynos 4412 only has 1.4 GHz speed in comparison. Though 1.8 GHz will be even better. =)

[Nickos-V]

Jez, you should replace the Hero II 3.0 score with Venus 3.1 score to give fair comparison.
http://www.slatedroi...post__p__597362

The Venus is 1280x800 7" tablet by Ainol with ATM7029.


Edit 1:
Your CF-Bench score is slightly better than S3. (6.5%) Same as W30 score in the table I posted above.

Quote

Hey janesh, how does the Momo11 Quad do on CFBench? I imagine it would be even better.

CF-Bench is a CPU benchmark. So, same CPUs should score very close to each other.

Quote

Slightly better Jez, Ployer momo11 quad score one time 14206, after that between 14150 and 14200. My Ramos W42 one run, score 13960.

Performance differences are either from motherboard design, quality of components or firmware optimization between the tablet makers.

This post has been edited by Nickos-V: 12 February 2013 - 06:46 PM

[rui no onna]

Here are the scores for my Ainol Novo7 Venus (1280x800, ATM7029 similar to Ainol Novo 10 Hero II):

AnTuTu 3.1.1
RAM: 726
CPU Int: 1510
CPU Float: 334
2D: 803
3D: 1453
Database: 315
SD write: 52
SD read: 106
CPU Freq: 1200 x 4

Overall: 5,299

This is with 2013.01.10 firmware (before they tweaked it to get high scores in AnTuTu again).

[Tablet_Collector]

janesh, on 11 February 2013 - 01:47 PM, said:

Slightly better Jez, Ployer momo11 quad score one time 14206, after that between 14150 and 14200.
My Ramos W42 one run, score 13960.
I ran geekbench2 on both devices, Ramos W42, 1880, ployer momo11 quad, 1967, which are very good scores.
Overall, momo11 quad feels slightly faster and both devices, no lag, browser lag or force closes.
When my wife, ipad user, does not grumble using that ployer and especially the Ramos W42, it should be a very good device.

Thanks janesh I will give geekbench a try.

Nickos-V, on 12 February 2013 - 06:31 PM, said:

Jez, you should replace the Hero II 3.0 score with Venus 3.1 score to give fair comparison.
http://www.slatedroi...post__p__597362

Done.

Quote

The Venus is 1280x800 7" tablet by Ainol with ATM7029.

Edit 1:
Your CF-Bench score is slightly better than S3. (6.5%) Same as W30 score in the table I posted above.

CF-Bench is a CPU benchmark. So, same CPUs should score very close to each other.

Performance differences are either from motherboard design, quality of components or firmware optimization between the tablet makers.

Noted thanks.

rui no onna, on 12 February 2013 - 10:08 PM, said:

Here are the scores for my Ainol Novo7 Venus (1280x800, ATM7029 similar to Ainol Novo 10 Hero II):

AnTuTu 3.1.1
RAM: 726
CPU Int: 1510
CPU Float: 334
2D: 803
3D: 1453
Database: 315
SD write: 52
SD read: 106
CPU Freq: 1200 x 4

Overall: 5,299

This is with 2013.01.10 firmware (before they tweaked it to get high scores in AnTuTu again).

Thanks I have also included these results in the table.

Jez

[Roman2025]

This gets ranked by me as one of the more interesting threads we have had in a while

Thanks for all the performance results...

Here are my conclusions thus far... (probably obvious)

1. If you have an RK3066 there is pretty much NO reason to upgrade to any of the other chipsets represented here. In fact I would advise against it as the RK3066 platform has the "maturity" factor now and you don't have to worry as much about game-killing early design/motherboard/firmware flaws in existing products.

2. If you are buying a new device, the RK3066 can be hopefully found at a lower price in the near future and would be well worth it. If you are wanting high-end (in China tab terms) you should definitely go RK3188 and if you want another value option vs the RK3066 you can try to find an Allwinner A31 which may actually be good on power consumption due to the Cortex-A7 cores making it an attractive option in that regard.

3. Avoid the ridiculous excuses for a processor that are the ATM7029 and i.Mx6 Quadcore. The i.Mx6 gets marks for being on of the first quad China chips available but that is nostalgic/novelty value only. The ATM7029 gets no marks whatsoever.

4. The RK3066 was the first China Chip to legitimately compete with the chips used in major western brands (thought I would throw this thought in there). Argue that if you want but honestly I don't even think previous generation A10/RK2918 were truly competitive. The RK3066 could be sold in stores today and outpace much of what is sitting on the shelf with it.

Cheers,

Roman

PS - And oh yeah.... I want my bloody "theoretical" RK3188 running at 2.0 Ghz..

This post has been edited by Roman2025: 13 February 2013 - 11:20 AM

[BamaPanda]

Roman ...Since you discussing processors a bit here, how would you place the new 4412 Exynos?
They are appearing in quite a few of what many consider to be quality tabs, such as Ployer and Ramos. I know there is some concern about battery size in some of these, but I would bet that is more about meeting a price point than anything else.
Somewhat like Onda giving us a high end Retina screen with a lower cost CPU ...just a result of economy.
Thoughts?

[rui no onna]

Roman2025, on 13 February 2013 - 11:17 AM, said:

3. Avoid the ridiculous excuses for a processor that are the ATM7029 and i.Mx6 Quadcore. The i.Mx6 gets marks for being on of the first quad China chips available but that is nostalgic/novelty value only. The ATM7029 gets no marks whatsoever.

The ATM7029 isn't that awful. ATM7029 tablets I've seen actually seem to be cheaper than RK3066 ones. While I wouldn't buy an ATM7029-based tablet for gaming, it's been pretty decent for web browsing, reading, etc and battery life seems to be quite good. I remember getting around 8+ hours battery life with it while reading. Unfortunately, I broke the screen on my Venus so now I have to wait until after the Chinese New Year celebration to buy a replacement. The device has been on standby since Saturday (2/9) and the battery level doesn't seem to have budged since then. It was at 68% when I broke the screen, will see how long until the device runs out of battery.

However, the GPU on the ATM7029 is pretty bad. I guess that's the reason Ainol went with the Allwinner A31 for the retina display Novo9 Spark rather than stick to the ATM7029. *sigh* Can't wait for these RK3188 tablets to be released.

[janesh]

Roman2025, on 13 February 2013 - 11:17 AM, said:

1. If you have an RK3066 there is pretty much NO reason to upgrade to any of the other chipsets represented here. In fact I would advise against it as the RK3066 platform has the "maturity" factor now and you don't have to worry as much about game-killing early design/motherboard/firmware flaws in existing products.

In my view and experience and in contrast with conclusion 1, there ARE several reasons to upgrade to an exynos 4412 chipset in a chinatab, if you have an RK3066 device.
After testing ten or more different tablets with RK3066, i have to say my twoo Exynos 4412 tablets are much faster, no lag, no browser lag, and no ".....is not responding, will you wait or close" ( not one time after about twoo weeks of use) also much better video playback.
Boot up is faster, after boot loading your mail is much faster (i am using five gmail accounts, one of them with a lot of google alerts every hour.) and after boot up you can immediately use your browser when mail come in.
Using a RK3066 device you better have to wait after boot, until your email is loaded, and then using your browser or what ever. The last advantage of Exynos 4412 is battery life, in proportion with RK3066, is much better for sure.

Do not misunderstand me, the RK3066 is a very good chipset, but Exynos 4412 is even better.
Perhaps one drawback, disadvantage. Exynos 4412 is still running ICS, i wonder if, and when, JB will come, but again i have to say, i do not notice this, compared with the lot of JB devices i tested.

Conclusion 2,3 and 4, i totaly agree, what a bad processor that ATM7029, A31 the same performance as RK3066, and not, or less advancement. My devices with that last chipsets, i sold them both very fast as you can see in my fingerprint.

At last i have to see if and how much RK3188 is better as Exynos 4412, for me thats doubtful.

This post has been edited by janesh: 13 February 2013 - 02:50 PM

[Roman2025]

janesh, on 13 February 2013 - 02:28 PM, said:

In my view and experience and in contrast with conclusion 1, there ARE several reasons to upgrade to an exynos 4412 chipset in a chinatab, if you have an RK3066 device.
After testing ten or more different tablets with RK3066, i have to say my twoo Exynos 4412 tablets are much faster, no lag, no browser lag, and no ".....is not responding, will you wait or close" ( not one time after about twoo weeks of use) also much better video playback.
Boot up is faster, after boot loading your mail is much faster (i am using five gmail accounts, one of them with a lot of google alerts every hour.) and after boot up you can immediately use your browser when mail come in.
Using a RK3066 device you better have to wait after boot, until your email is loaded, and then using your browser or what ever. The last advantage of Exynos 4412 is battery life, in proportion with RK3066, is much better for sure.

Do not misunderstand me, the RK3066 is a very good chipset, but Exynos 4412 is even better.
Perhaps one drawback, disadvantage. Exynos 4412 is still running ICS, i wonder if, and when, JB will come, but again i have to say, i do not notice this, compared with the lot of JB devices i tested.

Conclusion 2,3 and 4, i totaly agree, what a bad processor that ATM7029, A31 the same performance as RK3066, and not, or less advancement. My devices with that last chipsets, i sold them both very fast as you can see in my fingerprint.

At last i have to see if and how much RK3188 is better as Exynos 4412, for me thats doubtful.

Excellent response! Real world use is what it is all about (trumps benchmarks imho) and I would guess the 4412 does so much better at the things you are mentioning due to memory write speeds being about double that of the RK3066 platform. Memory timing was my argument for the RK3066 vs the Tegra III SoC. It makes a huge difference in real world use.

I also manage several accounts for google and while it does boot to the OS quickly, I can't actually start "using" my tablet until about 1 - 2 minutes or so in because Wifi kicks on and starts updating all my accounts.

This issue was magnified by about 5x on the Allwinner A10 device I had before. You ended up waiting 5 minutes before you could use the unit if I hadn't booted in a day or so because of all the updating. It also hurt having only a single-core so the processor was completely tied up during updates.

SO... What I have started doing with my RK3066 unit is never turning my tablet off and setting wifi to turn on when "plugged in" so it normally doesn't go for more than an hour or so at most without a Wifi connection which has alleviated that bottleneck for me most of time. That however is a "work around" and not a solution. The faster memory write speeds on the 4412 indeed make it a more appealing processor in that regard and even though the other benchmarks post a bit lower, it easily has enough processing power for 98% of things to be just as smooth as any other more powerful chipset.

Memory read/write speeds are not as technically "glamorous" as "3D performance" but they are absolutely vital if you want a good tablet experience.

All that being said, it seems to me that the 4412 based units tend to be more expensive then comparable RK3066 devices. So the question is, is it worth the extra $40 - $50?

I think if the difference is $25 or less then it is obvious (as long as all other tablet features remain more or less equal).

My 2 cents but I don't have any hands on with the 4412 platform.

Nice addition to the thread!

Cheers all,

Roman