How about the strength of the four core A7? MT6589 Dual Core A9 | Ainongji - 上海419贵族宝贝-上海后花园1314-上海品茶网-阿拉爱上海

PS: This article was transferred from Baidu , Author

As we all know, 6589 has received unprecedented popularity in the bar due to its excellent power consumption control, memory occupation and the price of conscience. Naturally, there are many 6589 blowers and 6589 fans
Yujian has made a comparison between 6589 and 4210 and 8960 before, and the final result is clear to everyone. 6589 is only half the weight of the two high-end dual cores, and even can't compete in many places
What about the performance of 6589 and low-end dual core SoC? I happen to have two low-end dual cores: Loser 865 and S899t
Can Aunt De 4430, who has entered her twilight years, and U8500, who was once elected as the "weakest dual core", fight against 6589?

Explain the test platform
BOC Motorola ME865, OMAP4430@1GHz 。 From 3K in January 2012. The system version is the official 4.1.2
Lenovo S899t, A9500 (U8500 baseband free version) @ 1GHz. From October 1, 2012. The system version is official 4.0.3
ZTE U956， [email protected] 。 From July 900, 2013. The system version is noble MIOS V5 based on 4.1.2
These three machines also just represent the average level of the flagship 1000 yuan smart machines in 2012 and 2013
Because it is mainly about the comparison of SoC, the screen, Wi Fi signal, call and endurance heating have not been tested. This comparison only includes three items: CPU GPU and DSP

First, the CPU part:
CPU mainly measures the score of video soft solution and Sunspider
I think it is more practical to test these two than other scoring software
In terms of software solution, we use the CLANNAD OP in the Yujian video package and the AB OP in the 8800 video package
These two videos are also popular high parameter videos with ref=16, and the software used is BS Player 1.9
Originally, I wanted to test the 720P soft solution (after all, MIUI can't be hard solved at station B), but after a few random tests, they were all full frames, so I didn't go further
CLANNAD OP:

From top to bottom, they are S899t, ME865 and U956. The reason for the black screen of the last two videos is that the last few seconds of the soft solution are black

Angel Beats! OP:

Same order
To summarize:
It can be seen that 6589 is more than 20% ahead of 4430 and 8500 even after deducting the frequency advantage,
This is different from when Yu Jian took 6589 and 8960/4210
If the 4430 is the standard A9 dual core, the U8500 is just a little weaker than the A9

Then Sunspider 1.0, the first entertainment score of the popular 78, measured the CPU's single thread performance:
The browser used is the latest version of Chrome Beta
Originally, I ran three self built browsers, but the self built browser I thought was too junk (3000+ms), so I didn't put it up

Summary:

The gap between U8500 and the other two SoCs has been widened since the beginning
In addition, the A9 at 1GHz is not better than the A7 at 1.2GHz... it should have an advantage in frequency

That's all for the CPU part. I think it's quite scientific
Next is the GPU part of the bar, which was argued to be no good. I felt a bit shocked after the test
In the GPU part, we will first benchmark Antutu 3D Benchmark, and then some small games
After all, the low-frequency 544MP of 6589, the high-frequency 540 of 4430, and the single core Mali-400 of U8500 are unlikely to be able to compete with large games... So I choose 3D games that do not have too much pressure on GPU, Temple 2 and the Wizard of Magic, as well as Rapide 2, which is relatively popular in the bar recently.
Among them, Temple 2 and the Wizard of Devil's Realm have selected the medium and high quality, while the torrent speedboat 2 is the highest quality (one can only drive to Medium, is it a special effect specially prepared for a manufacturer)

First, the benchmark test, the latest version of Antu 3D Benchmark, has two scores of corresponding resolution and 1024 × 2048 Off screen
Although the entertainment rabbit has always been criticized, the 3D score run independently from the hodgepodge is still relatively meaningful

It doesn't need to tell which machine it is
The summary is as follows:
There is no doubt that the score of U956 is directly separated from the other two dual cores,
3D scores are about 865 points ahead, 23% Offscreen scores are even close to 100% ahead
The 3D score of the entertainment rabbit is directly related to the theoretical performance of the GPU, but it is unknown whether the actual game experience is like this

In the actual game, there are two temples:

This is medium quality

Next comes the Wizard of the Devil Kingdom:

Medium quality

High picture quality
The sequence is S899t, then U956
Why is there no ME865?
This is really a bit painful. Because I was very excited when measuring the frame number, I didn't react until I fell down, so I didn't take a screenshot at all
These days, I found that the data is not convincing. I went to the temple again and planned to test it again. However, Opensource, which has already painted the magic, has lost about 10% of the temple's score, so the screenshot cannot be cut
So the data of ME865 can only be provided by the average number of frames measured in real time when I was competing with others on Weibo for 6589 3D performance. If you think it is not convincing, then it is not convincing. After all, there is a torrent speedboat 2, which is perfectly cut off by three models

The summary is as follows:
Temple 2 with medium special effects basically has no pressure, they are all full frames or close to full frames
The Temple 2 with not too high special effects is different from the magic realm wizards with two special effects. The average frame begins to be compressed to about 50
However, I think the optimization of the magic realm wizard is better than that of the temple 2. The magic realm wizard is obviously smoother than the temple 2 in the same frame number
Here comes the unscientific point: the frame number of ME865 is higher than that of U956
Theoretically, this is impossible. The SGX544MP on the U956 is higher than the SGX540 on the Loser 865 in both theoretical performance and frequency
So I began to try to find the reason from the resolution: 720P has 92W pixels, while qHD has only 52W pixels
In this way, 720P has a bigger disadvantage than qHD in terms of pixels, which may smooth the huge gap between 540 and 544
However, it is not ruled out that there is something wrong with my test steps, so I hope you can test 4430 and 6589
By the way, the single core Mary on the U8500 is dead

The game also has a torrent speedboat 2. If the data of the two temples is not complete or the credibility is questionable, the torrent speedboat 2 test is at least a blackpoint free version of three truncated pictures

All three are of the highest image quality available. Why is there a Medium? How should I know?
Game frames:
I chose the first level, the first map, from the beginning to the end
Why the first map? Because I don't want to play this game at all. If the game wasn't so popular in the bar, I wouldn't test it!

From top to bottom, the S899t, ME865, U956 and U956 are still cut a second or two late
Summary:

Why does the U956 still fail the ME865? I'm still a little confused, but it's probably the resolution
However, despite the low frame count, due to the nature of this game (NC speedboats bouncing around), I don't feel very stuck when playing
However, the S899t is an exception. The average frame is only one frame lower than that of the U956, but the actual experience is not good. It may be garbage optimization

Finally, the DSP part
DSP, I choose the Royal Sword video bag that you like to see
And the h.264 ref test package of LittlePox

Due to the reason that three models are jammed together, the picture is relatively small. Please try to open the big picture
The U956's format support is really poor. It seems that it can only support a few formats such as mkv, mp4, avi (it is said that 4.2 has magic oil). It seems that because of the MTK (or MIUI
865 won't say anything. It was tested bad many years ago
It is worth mentioning that the S899t, the original black residue U8500 on the Sony machine, has made a big turnaround here. Except for a few videos, it can basically be hard understood, so it can be said that there is a video god oil
However, I would like to say that many of the Yujian video packages tested are exotic codes (such as Can I Have This Dance and FSN OP), as well as many wmv and rmvb video formats that have basically retired from the era. Therefore, this table is good for pure entertainment, and often does not represent the actual use of the machine

The following is part h.264 S899t。 The U8500's support for h.264 is really strong, and it directly swallowed the AB OP with ref=16 (but the frame number is not too high). Lenovo's video magic oil once again played its role

In the era of 4.0, the dog demon can only solve 720P to ref=91080P to ref=4. After 4.1, it has slightly improved. 720P can reach ref=11, and 1080P can also eat ref=7. But it is still weak by comparison
In addition, the dog demon was The lock of hl4.0 is finally solved after 4.1. It can be played to 5.1 with a player like MX+

The h.264 of the U956 is quite good. The 720P cuts ref=161080P and also supports ref=14, which means that the parameters can be played as long as they are not too BT
By the way, the S899t supports sys hw. Of course, the U956 also supports sys hw after it is upgraded to 4.2. However, the dog demon does not exist at any time
In general, in the current era of h.264, 6589's video capability is commendable, especially 4.2 and sys hw. Together with low power consumption, it should become a powerful tool for many people

In the end, we conclude:
1. 6589 has a certain advantage over A9 dual core in terms of CPU, but it has little advantage due to Neon
2. The SGX544MP of 6589 with 720P does not lead the previous generation A9 dual core, and the game party suggests to join qHD
3. The h.264 hard solution of 6589 is quite good. Non game party and video party can enter the 720P version, which is also quite pleasant to watch