ZeniThink ZT-180 Specifications Faking, Not ARMv7 A8-Cortex!

[xaueious]

I am a ROM cook for the Apad iRobot, with contacts to Shenzhen. I also know how to read Chinese. This is how I uncovered the story.

In case you don't want to read 20 pages (which is filled with spam posts from scammers, by the way), here's the summary:

What happened
The Zenithink tablet was originally advertised as a 10" tablet with an A8 Cortex CPU. The A8 Cortex CPU would have allowed for Adobe Flash playback if the device was ever upgraded to Android 2.2. Previous CPU versions are not only incapable of using Adobe Flash, they are also 30-40% slower and less efficient overall. It was revealed that the CPU specification was in fact faked, and the CPU is the older and slower ARM11 variant.

What we have done here is uncovered faked specifications and an attempt to cover up the situation, both evidenced by the firmware implementation to artificially boost benchmark scores (which inherently leads to instability), and by the manufacturer's passive stance on the false specifications being spread.

What's the big deal
Online retailers were selling the device for as much as $300US, which would have been a fair price provided that the device had a real A8 Cortex CPU. However, since this is an ARM11 device, $300US is a rip off pricing. A good reference for a device with similar specifications is the Augen Gentouch, which is around $150, but with a 7" screen and warranty support. The device retails for the equivalent of $155US (1050RMB/Yuan) in China. Assuming $50 shipping, you should be paying no more than $200 for the device to be shipped.
* * *
Technical Details Below
Part 1
It has been confirmed in China that the Zenithink ZT180 / Wiipad does NOT have ARMv7 A8 Cortex CPUs. Also Flatpad, Orphan M16 are all the same thing.

append: Even though Zenithink claims that they never made any official statements about the tablet being A8 Cortex, their silence on the topic when their product was being misrepresented internationally is no better than an active role in false advertising.

There's no way resellers didn't have a clue about this with their sample products. I find it hard to believe that they didn't suspect anything. Instead, they jacked their prices to obscene numbers, cashing in on the false advertising and misleading customers.

In addition to hardware quality control issues, the CPU has been suspected to be an ARM 11 variant, with the Android OS running with the highly unstable JIT Dalvik engine (stable only in Android 2.2).

This also means this device is NOT flash capable.

Original source:
Unofficial User Discussion Forums, Refund
Mirror:
1 - Fullscreen

虎翼商城作为国内比较早的专业销售山寨笔记本电脑，山寨平板的直销网站。

目前针对卓尼斯CE版、android 2.1版的软硬件问题：

包括CPU质疑造假、电池、WIFI模块、系统经常死机、软件不兼容等问题；由于作为销售商同


样的不知情，为维护商城信誉和品质，我们已经在2010年7月31日停止销售此款机型。


前期购机的用户凭借收据或者淘宝交易记录凭证可以在商城退款退货（非本站购机客户请自行联系购买商家）。


针对此款不成熟的产品对广大客户造成的影响深表歉意。

This is a large retailer of Shanzhai products announcing on the Zenithink forums, declaring a full refund for all Zenithink ZT180 products. Hardware issues include allegations of CPU specification faking, system instability, wifi module failures, software incompatibilities.

I did some digging for myself and found benchmarks of the ZT180:
JT's Digital World » Zenithink 10 inches Cortex-A8 Ipad Clone ZT180

Linpack:

MFLOPS : 9.878
Time: 0.55 Seconds
Norm Res: 3.36
Precision : 2.220446049250313E-16

Also, Softweg CPU info

unknown p7901a p7901a/p7901a/p7901a/:2.1-update1/ERE27/eng.root.20100704.171303:eng/test-keys
Processor ARMv6-compatible processor rev 5 (v6l) Linux BogoMIPS 1005.97

The combination of the CPU benchmark and the Linpack score means one thing: it is an ARM 11 (ARM v6) CPU with JIT enabled.

* * *
Part 2
Here's more info if you guys aren't going to do more digging for yourself.
¸æËùÓб»×¿Äá˹ÆÛÆ*µÄÓû§µÄ¹«¿ªÐÅ!ÎÒÃÇÒªÇóÍË»õ!ÍË»õ!ÍË»õ!ÍË»õ! - MIDÖ÷ÂÛ̳ - iMP3ËæÉíÓ°Òô MP3|MP4|MP3ÂÛ̳|MP4ÂÛ̳|MP3ÏÂÔØ|MP4ÏÂÔØ|¶ú»úÂÛ̳| - Powered by Discuz!

Google translate really sucks, so I'll just translate bits of it.

This is an 11 page thread on imp3.net of China buyers of the Flatpad/ZT180/Wiipad asking for full refunds on this scam.

Zenithink is a Shanzhai laptop company with a focus on the Anyka AK88 chipset. From2009, then started selling Windows CE microtablets for 3xx Yuan, and reaped good profits. Since they wanted to enter the higher-end MID (mobile internet device) product sector, but did not have chipset manufacturing capabilities, they resorted to Remark chips, purchasing Shanghai Tsinghuaic IX2X0 series CPUs, and proceeded to stamp their own logos on the CPUs to be shipped. However, these chips, did not contain A8 Cortex, but ARM11 with a coreclock of 800MHz and supporting 1080p hardware decoding. Right now the chipset only supports MP4 1080p hardware decoding, and does not actually support any other formats! Since ARM Cortex A8 processors process data at an efficiency of double of ARM11, this is clearly an act of trickery and deception in advertising.

* * *
Part 3
For any shops who are sneaking in here to dispute the fact that their tablets have fake specifications:

JIT ARTIFICALLY INFLATES BENCHMARK SCORES.
An ARM11 CPU with JIT results in scores 2-3 times higher in Dalvik based benchmarks. The i5700, an ARM11 800MHz CPU, reaches up to 8MFLOPs on Linpack with JIT enabled. With the CPU at 1GHz, it falls right into the range of 9-10 MFLOPs seen in every Zenithink benchmark.
This makes it very likely that it is an ARM 11 based CPU

Typical clock-per-clock benchmarks for A8Cortex are at least 20-30% higher compared to ARM11 cores, for both OMAP 3430 and Snapdragon CPUs. THe Samsung Galaxy S CPU, the most efficient A8Cortex on the market, has a Linpack of 8.5 MFLOPs. A benchmark score at 1GHz at 9-10 MFLOPs without JIT enabled is IMPOSSIBLE for an A8Cortex CPU or any CPU on the market today. The Nexus One scores a maximum of 7 MFLOPs at 1 GHz without JIT on Android 2.1. 9MFLOPs would indicate that the CPU has to be at least 20-30% more efficient (impossible for the same architecture), or has a clockspeed faster by that same percentage.
From this it is obvious that it is NOT A8 Cortex.

* * *
Part 4
Update 9amET, Aug 2
On the same thread where a poster has issued an official update for Android 2.1 on Zenithink tablets, the same poster, who claims to be a representative from Zenithink, claims that the ZT180 was never listed as an A8 Cortex processor.

Clearly playing games here.

This is not a verified poster, but we'll continue to get more information on this.

你好，对于cpu不是ARM V7 而是ARM V6 架构 的质疑，如何看待

卓尼斯公司从来没有过官方的书面或者官方网站上公布过所用的芯片是ARM V7的架构。

* * *
Part 5
On the Orphan M16, I have yet to see what is inside it, but I have no reason to believe in them or Chinagrabber whatsoever.

I can ask my Shanzhai contacts about Orphan, but from what I've seen of these guys I wouldn't trust them at all.

* * *
Part 6
Update Aug 2 5pm ET

I was reading xda developers and stumbled upon more specifications:
Zenithink ZT-180 1GHz 10" tablet - Page 2 - Android Forums

Specs update.
CPU - IMAPX200 r5 (over 1000 bogoMIPS) But this is odd, because it is identified as ARMv6TEJ - This may be closely related to the Shouying PID1110. I'm confused by this as it would indicate that we have an ARM11 and not a CortexA8 as described. Can anyone confirm the CortexA8 is there?

GPU - Vivante GC600 :: Vivante Corporation I Products/Mobile Visual Reality :: nice specs

Current driver is OpenGL 1.1 so that's why some OpenGL apps are failing.

Frame buffer and crossbar combiner is not supported, but DOT3 and cube maps are there.

Confirmed 256MB RAM, but only about 90MB is available after boot.

uses a zenithink 2 axis accelerometer.

* * *
Part 7

* * *

Update Aug 2 5pm ET

I was reading xda developers and stumbled upon more specifications:
Zenithink ZT-180 1GHz 10" tablet - Page 2 - Android Forums

Specs update.
CPU - IMAPX200 r5 (over 1000 bogoMIPS) But this is odd, because it is identified as ARMv6TEJ - This may be closely related to the Shouying PID1110. I'm confused by this as it would indicate that we have an ARM11 and not a CortexA8 as described. Can anyone confirm the CortexA8 is there?

GPU - Vivante GC600 :: Vivante Corporation I Products/Mobile Visual Reality :: nice specs

Current driver is OpenGL 1.1 so that's why some OpenGL apps are failing.

Frame buffer and crossbar combiner is not supported, but DOT3 and cube maps are there.

Confirmed 256MB RAM, but only about 90MB is available after boot.

uses a zenithink 2 axis accelerometer.

* * *

Update Aug3, 8am

I have received a system dump and a kernel log (dmesg) for the device.
http://www.androidtablets.net/forum...ions-faking-not-armv7-a8-cortex.html#post1909

Continued discussion on unofficial Zenithink user forums have indicated that Zenithink is now standing by their position that they never said that their device was A8 Cortex. As I have stated earlier, they did not give any statements on the false information in the Shanzhaiben article despite knowing that there was misinformation spreading.

If there was any question that this was not a switch-and-bait scheme, there is none now.

 

[TechNinja]

Thanks for the info! *bump*

I'm glad I don't make hasty purchases. This was going to be my first tablet too!

 

[mindwave]

I too had them high on my shopping list. in another thread I added this comment, but maybe someone here will reply, I'm wondering if WIISTECH (sic?) knew something about this. over the weekend i went to their site looking for their 10.2 android and couldnt find hide nor hair.

now maybe the page was down for updating i dont know i havent been back, but still, interesting

especially since flac comp was going to be my method.

 

[xaueious]

I have found more evidence that Android is running with JIT, and is ARMv6.
System dump from http://forum.xda-developers.com/showpost.php?p=7476741&postcount=27

dmesg on pastebin
<5>Linux version 2.6.32.9 (roo - Anonymous - EuHGKsiH - Pastebin.com

Edit: guys, readelf output of libdvm.so, towards the end

No version information found in this file.
Attribute Section: aeabi
File Attributes
  Tag_CPU_name: "6"
  Tag_CPU_arch: v6
  Tag_ARM_ISA_use: Yes
  Tag_THUMB_ISA_use: Thumb-1
  Tag_VFP_arch: VFPv2
  Tag_ABI_PCS_wchar_t: 4
  Tag_ABI_FP_denormal: Needed
  Tag_ABI_FP_exceptions: Needed
  Tag_ABI_FP_number_model: IEEE 754
  Tag_ABI_align8_needed: Yes
  Tag_ABI_align8_preserved: Yes, except leaf SP
  Tag_ABI_enum_size: int
  Tag_ABI_HardFP_use: SP and DP

We also have evidence that the file has at least been compiled with JIT, although it is still not clear where they have enabled it. Here's the readelf -a of libdvm.so
ELF Header: Magic: 7f 45 - Anonymous - i3hGiAy8 - Pastebin.com

I'm no expert, but I can see references to JIT in the libdvm readelf. I have an Acer Liquid, compiled for without JIT, and these lines referring to JIT are there in my build.

Here:

    22: 000674c1   156 FUNC    GLOBAL DEFAULT    7 dvmJitToPatchPredictedCha
...
   100: 00067225   116 FUNC    GLOBAL DEFAULT    7 dvmJitUnchainAll
...
   103: 00061c41    96 FUNC    GLOBAL DEFAULT    7 dvmJitf2l
...
   232: 00061b05    24 FUNC    GLOBAL DEFAULT    7 dvmJitStopTranslationRequ
...
   269: 00014de8     0 NOTYPE  GLOBAL DEFAULT    7 dvmJitToTraceSelect
...
   320: 00061ca1   260 FUNC    GLOBAL DEFAULT    7 dvmJitLookupAndAdd
...
   539: 00062299   200 FUNC    GLOBAL DEFAULT    7 dvmJitStartup
...
   677: 0003aa0d    10 FUNC    GLOBAL DEFAULT    7 dvmJdwpNetFree
...
   772: 00067121   132 FUNC    GLOBAL DEFAULT    7 dvmJitChain

I might have missed a few.

These lines are also not present in my compiled AOSP build of Dalvik, without the WITH_JIT flag

Here's an example corresponding to source code, you can see that in Jit.c ([platform/dalvik.git] / vm / interp / Jit.c)
android.git.kernel.org Git - platform/dalvik.git/blob - vm/interp/Jit.c

dvmJitStartup is only defined '#ifdef WITH_JIT'

  16 #ifdef WITH_JIT

  17 

  18 /*

  19  * Target independent portion of Android's Jit

  20  */

  21 

  22 #include "Dalvik.h"

  23 #include "Jit.h"

  24 

  25 

  26 #include "dexdump/OpCodeNames.h"

  27 #include <unistd.h>

  28 #include <pthread.h>

  29 #include <sys/time.h>

  30 #include <signal.h>

  31 #include "compiler/Compiler.h"

  32 #include "compiler/CompilerUtility.h"

  33 #include "compiler/CompilerIR.h"

  34 #include <errno.h>

  35 

  36 int dvmJitStartup(void)

This proves that libdvm.so has been compiled with the flag 'WITH_JIT'. I have no doubt in my mind that they have compiled this to use it in running systems on purpose.

 

[stunno]

Oh Dear! I've bought this one 10.2" Android 2.1 Apad 1 GHz 256 RAM Tablet PC HD UMPC on eBay (end time 30-Jul-10 04:46:21 BST) in the last couple of days.

Have I bought a pile of cr@p?

 

[xaueious]

It took a while, but I finally got an admin on Slatedroid who happens to be developing for the device to confirm that this is indeed running JIT.

It depends on how much you paid. 181 Pounds is okay for what it is. This really affects members who bought it for an outrageous $285USD from third-party resellers such as Flatcomputing and Orphan

 

[bigscot]

It took a while, but I finally got an admin on Slatedroid who happens to be developing for the device to confirm that this is indeed running JIT.

It depends on how much you paid. 181 Pounds is okay for what it is. This really affects members who bought it for an outrageous $285USD from third-party resellers such as Flatcomputing and Orphan

Thanks for the info about the JIT. How ever the people that payed the 285 made out slightly on the deal over the ones that payed in pounds, if they live in the us that is.

Let me google that for you

 

[iregentle]

This thread was very informative and another big reason to be careful.

I bought:

G01 10.2'' 10'' Inch Notebook Tablet PC,Tablet laptop,Mini PC,Touch screen, MID, UMPC, with ZT-180 1GHz CPU, 256MB Memory,Wifi from Sinoband Co. Ltd.'s store.

I'm out $38 on a return from the United States after buying a 10" android touchscreen from SINOBAND CO. LTD. I received the product with a non functional touchscreen. I hooked up a mouse and downloaded a android benchmark program. The processor reported 300mhz, had android 1.5, and the sold product stated it had 1 Ghz. I was the victim of a bait and switch on top of a non functional screen, so I had to send it back. Why should I pay return shipping? Additionally on the return they are refusing to receive it?

Bullet Attempted Delivery Abroad, July 31, 2010, 12:00 am, CHINA PEOPLES REP
Bullet Attempted Delivery Abroad, July 30, 2010, 1:27 pm, CHINA PEOPLES REP
Bullet Attempted Delivery Abroad, July 29, 2010, 2:12 pm, CHINA PEOPLES REP
Bullet Arrived Abroad, July 29, 2010, 8:25 am, CHINA PEOPLES REP
Bullet International Dispatch, July 25, 2010, 4:00 pm, ISC CHICAGO IL (USPS)

Priority Mail International # CJ167155545US (Sequence Number 1 of 1)
Shipped to: PENG JINWEI
SINOBAND CO. LTD.
2-202
YINGSHUI SHANZHUANG 2ND DISTRICT
MINZHI ST. BAOAN DISTRICT
SHENZHEN, GUANGDONG
CHINA 518131

Return information from messages with seller:

Sinoband Co. Ltd.

RETURN ADDRESS:
Peng Jinwei
2-202,Yingshui Shanzhuang 2nd District
Minzhi St. Baoan District
Shenzhen,518131
China
Tel:13312908250

Sinoband made early adoption a dangerous proposition.

 

[bigscot]

WOW that was fast, It took them less the an hours to get my order canceled even though it was after business hours. If they ever get this cpu thing figured out i may still get one. They seem like an honest company that hat shafted by suppliers.

 

[xaueious]

Thanks for the info about the JIT. How ever the people that payed the 285 made out slightly on the deal over the ones that payed in pounds, if they live in the us that is.

Let me google that for you

My mistake . I meant 180US. Sorry


To iregentle, you got a Rockchip 10". ARMv5 CPUs report 50% MFLOPs typically

 

[Laharl]

Don't fall for cheap salesmanship I was contacted by Merimobiles and made aware of the cpu problem and they were very nice to me and when they found out they offered a refund or a discount!!!!!!!!!!!

And THIS WAS THE DAY THE CPU WAS FOUND OUT TO BE MISLABLED!

So I went ahead and got the Epad from them they were honest and upfront with me on the item and it was not a hassle!!!!!

"Hi as you may or may not already know from the news circulating on the internet, The Zenithink processor is clocked at 800Mhz and is a ARMv6 CPU. Please understand the original specs that we got from the Zenithink manufacturer is 1GHz processor, although after dissasembling the device as noted on many blogs, news websites and forum, we found out like many others that in reality the processor is a ARMv6 clocked at 800Mhz."

This is what they told me after they found out and they did it ASAP!!
From what I heard of the reviews the tablet is still a great tablet its just not a 1ghz Product.

ZT-180 800MHz processor 10" tablet
CPU is ARM 800MHz
Price is $219.99 plus shipping which still beats out the flatpad and alot of other places for the Zenithink tablet
So Get it HERE!!!!
https://www.merimobiles.com/ProductDetails.asp?ProductCode=MERI0512

So have no fears and order from Merimobiles!

 

[xaueious]

They are still wrong. It actually does seem to be running at a legitimate 1GHz. The key is that it's ARM11 and not Cortex A8.

Merimobiles is still expensive. You should not be paying more than $200 US for this, shipped.

I still don't recommend this tablet. The Android build is no good on this.

 

[Laharl]

Well I only paid $220.00 for the tablet Shipped so I didn't do to bad just 20 over 200 so Im ok with it for now and I read at XDA that they just updated the RoM for it and its supposed to be running alot better now.

 

[HMNSL8]

The units are great for developers serious about getting a head start on larger format Android development. but yes, they are not really mainstream ready. They are able to be flashed with custom/ updated ROMs.

HMN-10001

We have them in hand and have run specs on them, etc. Note: THESE particular units do not have the Market installed and are really stripped down. There is supposed to be updated Firmware this week.

If you are interested we have two open box units available for $200 each cash. They are located in the Seattle area.

We have some pics also:
http://www.facebook.com/album.php?aid=11833&id=126524480724118

 

[HMNSL8]

If anyone gets their hands on a more stable Firmware update w/ the Market, etc can you pass it along to the community?