廣告刊登

 
... ... ...
... ...
... ... ...



 
標題: Surprise! Atom N280 was beaten by N270
odinchong
版主
Rank: 7Rank: 7Rank: 7


UID 4093
精華 7
積分 1897
帖子 1452
閱讀權限 100
註冊 2008-7-28
狀態 離線
發表於 2009-6-16 14:48  資料  個人空間  短消息  加為好友 
Click 入香港 Apple Store 選購 iPhone 4、iPad、新款 iMac 及 2010 全新 Mac Mini!

Surprise! Atom N280 was beaten by N270

As we know, the new Atom N280 is a little bit more powerful than its predecessor N270 under various benchmark tests. However, after running a series of tests between them, Odin found the results rather surprising – N280 may not be fast enough to beat N270 off, or put it in other words, N280 MAY EVEN BE BEATEN by its one year old precessor. (Disclaimer: all the test results performed below may be biased due to errors of sampling and test methods.)

眾所周知,Atom N280在跑分時會比它的舊版本N270好一點點。但是,Odin做了一系列的測試後,得出一個令人驚訝的結果:對,不但N280不足以擊倒N270,甚至可以說會被它的舊版本擊倒。(注意:不同的測試樣本及測試結果可能會影響結果的準確性。)

Based on the most common usage scenarios of netbooks, Odin has focused on 4 kinds of tests: word processing, 2D graphics (vectors), single-task compressing, and multi-task compressing. Besides, I have tested 2 more other processors, Intel CULV platform’s SU3500 and AMD Yukon platform’s MV-40, to act as convincing reference points in our speed tests.

Odin根據netbook的用處而將測試分開為四大部份,分別是:文字處理、平面繪圖(向量圖)、檔案壓縮、及多工作業下的壓縮。另外,筆者特意找來兩個被公認為比Atom更強的平台,確保測試中較快的一個系統,就是效能較強的系統。這兩個參照樣本是Intel CULV的SU3500以及AMD Yukon的MV-40。

In our first test, Odin opened a word document (in Simplified Chinese, with file size 5.97MB and 3.2 million characters/32 milion including spaces) in Microsoft Word 2003 and asked Word to calculate the total number of characters in the document, in order to find out how long it takes the system to finish this job.

首先,Odin利用MS Word來計算一個文件檔案(簡體中文,5.97MB、300萬字/連空白共3,200萬字元),然後計算系統完成計算所需的時間。



The test results are valid because of the faster results obtained in the SU3500 and MV-40 systems. Nevertheless, the N280 system is not much faster than the N270 one when dealing with such a large document. Perhaps tests on more larger documents may prove that N280 is observably better than N270, but to be honest, how many of us edit encyclopedias on netbooks every day?

SU3500及MV-40的佳績可確認測試是有效的。但是,N280在如此大型的文件上進行計算時,沒有做出比N270較佳的成績。Odin不排除N280在應付更嚴竣的測試下會做出比N270更佳的成績,但是,我們真的需要在Netbook上編輯百科全書嗎?

In our second test, we will perform some 2D graphic processing using the CorelDraw X4 suite. Odin tried to convert a JPEG (1.17MB, 2440x1448 pixels) file to a 256-color high quality vector file using the CorelDraw’s default settings (photos will be downsized automatically) in order to measure the time it takes to perform such task.

其次,平面繪圖測試將以Coreldraw X4 suite為主。Odin會利用Coreldraw的預設設定(照片會被自動縮小)去把一個1.17M、2440x1448像素的JPEG檔案轉為256色高品質向量圖。



The results are unbelievable – Atom N270 has beaten all of the other processors! The surprising results may be casued by the difference of Operating Systems (Windows XP/Windows Vista), but with independent HD3410 GPU equipped, the Yukon platform performs even worser than the integrated GMA4500MHD platform (Someone has told me that the poor performance of HD3410 in this test may be due to the power-saving mechanisms of the ATI chip). Although the results are questionable, given the fact that the CULV and Yukon platform can easily beat Atom N280 off, our test is not absolutely invalid – that is, N280 may still be the worst system in our 2D graphic tests and it took almost a double amount of time than that of N270's.

結果實在難以置信!Atom N270居然將所有對手都擊敗!這個驚人的結果可能來自作業系統的不同所致(Windows XP/Windows Vista),但Yukon獨立顯示核心HD3410亦比較弱的GMA4500MHD要慢(據知可能是因為節能系統被自動啟動所致)。無論如何,即使測試結果亦需相榷,但CULV及Yukon平台都比N280要優勢這一點,某程度上也反映了測試結果不能完全反映事實。對,N280很可能是在Odin的平面測試中最慢的一個平台,而且,它還差不多可能需要用比N270多一倍的時間!

The single-task compression test is quite simple: Odin tried to compress a 100MB video file to RAR format with WINRAR default settings to see how long it takes.

單工壓縮測試就非常簡單:Odin用WINRAR的預設設定,把一個100MB的影像檔案壓縮為RAR檔,然後看看它需要多少時間。



Again, faster results of SU3500 and MV-40 proved that the test results valid – although MV-40 is far faster than SU3500 (SU3500 is better when it tested in benchmark software). However, N280 is a little faster than N270 – yes, a little bit at all.

SU3500及MV-40較佳的成績確認了這個測試某程度上是準確的──雖然MV-40居然遠比SU3500快得多(SU3500在跑分上比MV-40高出不少)。無論如何,在這回合裡,N280是比N270要快,但就只是僅僅一點點而已。

The multi-tasking test is similar to the compression test, but here Odin plays a 1080p video file using KMplayer (without hardware acceleration) and at the same time browsing 5 Adobe Flash-based webpages on Firefox simultaneously. This is to ensure that the compressing process is under 100% of CPU usage.

最後的多工作業測試與單工壓縮測試相似,但Odin同時用KMplayer播放1080p影片(硬解加速功能關閉),並用Firefox一起瀏覽五個flash-based的網頁──確保整個壓縮過程在100%CPU使用率下進行。



The results are surprising, too. N280 did yield a really poor result and much worse than its predecessor, N270. While SU3500 and MV-40 yield much better results, I think the test results are valid even though the benchmark-optimized SU3500 processor has been beaten by MV-40 in real life use cases, too.

結果仍然驚人──N280大幅落後於它的上一代,N270。由於SU3500及MV-40都取得更佳的成績,所以Odin相信這個結果是可靠的──雖然跑分能力很強的SU3500仍然輸給MV-40。

The consequence of the tests was unpredictable: the one-year-old N270 may still be a little bit slower than N280 in several scenarios, but I did find that N280 can perform much worser than its predecessor on vector drawing and multi-tasking. (Also unpredictable is that the SU3500 CPU, while performs extremely well in various benchmarks, was beaten by AMD’s MV-40 in 3 of these tests.)

結果是我們意料不到的:N270也許在一些範疇上輕微地不及N280,但在向量圖繪畫及多工作業上卻大敗。(另一個更想不到的是:很善於跑分的SU3500,卻連續在三個項目上敗給最大對手MV-40)



Let us come back to our discussions on netbook performance—what makes these tests interesting to look at is that the 2 of the weaknesses found in N280 fall into the netbook weakness areas that Intel emphasized.  Mysteriously, we found that the earlier N270 CPU does perform very well on these areas while N280 doesn’t. Hey Intel, could you explain why this is the case?

讓我們回到Netbook──最有趣的一點是:N280的兩個明顯弱點往往都是Intel所著重的範疇──對,是Intel強調Netbook做得不好的範疇。但是,我們卻發現N270做得不錯,但N280卻做得一塌胡塗,原因何在?Intel可以解釋一下嗎?

For your reference, cnet.com has found some similar results on the multi-tasking issues.

www.cnet.com也發現N280的多工作業能力比N270要差的成績,大家可以參考一下。


Link: Eee PC 1008HA Review
Link: N120 Review

[ 本帖最後由 odinchong 於 2009-6-17 10:05 編輯 ]

頂部
Louis
版主
Rank: 7Rank: 7Rank: 7



UID 235
精華 0
積分 672
帖子 588
閱讀權限 100
註冊 2008-1-1
狀態 離線
發表於 2009-6-16 18:48  資料  個人空間  短消息  加為好友 
Click 入香港 Apple Store 選購 iPhone 4、iPad、新款 iMac 及 2010 全新 Mac Mini!

勵害 ! 勵害 !

頂部
chen_junru
普通會員
Rank: 2



UID 12752
精華 0
積分 117
帖子 89
閱讀權限 20
註冊 2009-5-21
狀態 離線
發表於 2009-6-16 20:21  資料  個人空間  短消息  加為好友 
Click 入香港 Apple Store 選購 iPhone 4、iPad、新款 iMac 及 2010 全新 Mac Mini!

如果同样是用945平台,那么n280不可能比n270强,我觉得有必要把n280+gl40的成绩贴出来。

gl40用的是667的内存频率,而945是533,这样的话,就没有完全满足n280高内存带宽的特征。

也就是说主板的内存带宽是个瓶颈。

頂部
peter852
普通會員
Rank: 2



UID 1288
精華 0
積分 51
帖子 21
閱讀權限 20
註冊 2008-3-14
狀態 離線
發表於 2009-6-16 20:35  資料  個人空間  短消息  加為好友  添加 peter852 為MSN好友 通過MSN和 peter852 交談
Click 入香港 Apple Store 選購 iPhone 4、iPad、新款 iMac 及 2010 全新 Mac Mini!
等ion netbook la

頂部
chen_junru
普通會員
Rank: 2



UID 12752
精華 0
積分 117
帖子 89
閱讀權限 20
註冊 2009-5-21
狀態 離線
發表於 2009-6-16 21:04  資料  個人空間  短消息  加為好友 
Click 入香港 Apple Store 選購 iPhone 4、iPad、新款 iMac 及 2010 全新 Mac Mini!
贴一张图,用945主板的n280,内存带宽只有266×2,而cpu的FSB却有667。
ion的内存带宽是800,不但足以匹配533的N270,相信搭配667的N280也是绰绰有余。


頂部
odinchong
版主
Rank: 7Rank: 7Rank: 7


UID 4093
精華 7
積分 1897
帖子 1452
閱讀權限 100
註冊 2008-7-28
狀態 離線
發表於 2009-6-16 21:53  資料  個人空間  短消息  加為好友 
Click 入香港 Apple Store 選購 iPhone 4、iPad、新款 iMac 及 2010 全新 Mac Mini!


QUOTE:
原帖由 chen_junru 於 2009-6-16 20:21 發表
如果同样是用945平台,那么n280不可能比n270强,我觉得有必要把n280+gl40的成绩贴出来。

gl40用的是667的内存频率,而945是533,这样的话,就没有完全满足n280高内存带宽的特征。

也就是说主板的内存带宽是个瓶颈。

完全同意你的說法。不過我起初天真地以為1.66GHz的N280總會比1.60GHz的N270快一點點。
想不到只是快一點點,但多工及繪圖卻大幅減弱了。

頂部
UMPC
超級版主
Rank: 8Rank: 8
微腦狂人



UID 5
精華 1
積分 4256
帖子 3866
閱讀權限 150
註冊 2007-8-14
狀態 離線
發表於 2009-6-16 23:07  資料  個人空間  短消息  加為好友 
Click 入香港 Apple Store 選購 iPhone 4、iPad、新款 iMac 及 2010 全新 Mac Mini!
I would rather expect the comparison between MSi U100 and U100+  or Eee PC 1000H and 1000HE.
N120 and NC10 are quite different in hardware build and especially the motherboard may be totally different. if someone can test the above pairs will be much better.
anyway, thanks odin for the professional testing
http://www.umpcfever.com/news/?postid=2603





微腦狂熱份子

UMPCFever Twitter 啟動
頂部
odinchong
版主
Rank: 7Rank: 7Rank: 7


UID 4093
精華 7
積分 1897
帖子 1452
閱讀權限 100
註冊 2008-7-28
狀態 離線
發表於 2009-6-16 23:27  資料  個人空間  短消息  加為好友 
Click 入香港 Apple Store 選購 iPhone 4、iPad、新款 iMac 及 2010 全新 Mac Mini!


QUOTE:
原帖由 UMPC 於 2009-6-16 23:07 發表
I would rather expect the comparison between MSi U100 and U100+  or Eee PC 1000H and 1000HE.
N120 and NC10 are quite different in hardware build and especially the motherboard may be totally dif ...

Basically, the most important factors of affecting system performance are:
1. CPU
2. Northbridge Chipset (Memory and FSB Controller)
3. Southbridge Chipset (I/O and Storage Controller)
4. GPU
5. Harddisk
6. RAM

Motherboard does essential because of its built-in North/southbrdge chipset.  Apart from the chipset, voltage does effect on your system -- but mostly on system's stability than performance -- if your GHz is fixed

However, HDD play a much impportant role in performance.  Nevertheless, if both of the system are 5400rpm SATA tradition HDD, and both of them haven't AHCI/NCQ/RAID enable, the quality differences between them may not cause a significant changes in performance.

Thus, I don't think the these differences may cause critical bias in my test, but Eee PC's do. Actually, 1000H's HDD is NCQ disable but 1000HE's is enable -- it will cause a bigger difference in performance.

[ 本帖最後由 odinchong 於 2009-6-18 21:00 編輯 ]

頂部
UMPC
超級版主
Rank: 8Rank: 8
微腦狂人



UID 5
精華 1
積分 4256
帖子 3866
閱讀權限 150
註冊 2007-8-14
狀態 離線
發表於 2009-6-16 23:42  資料  個人空間  短消息  加為好友 
Click 入香港 Apple Store 選購 iPhone 4、iPad、新款 iMac 及 2010 全新 Mac Mini!


QUOTE:
原帖由 odinchong 於 2009-6-16 23:27 發表


Basically, the most important factors of affecting system performance are:
1. CPU
2. Northbridge Chipset (Memory and FSB Controller)
3. Southbridge Chipset (I/O and Storage Controller)
4. ...

哈哈, 唔夠你熟. 都係你專業d





微腦狂熱份子

UMPCFever Twitter 啟動
頂部
odinchong
版主
Rank: 7Rank: 7Rank: 7


UID 4093
精華 7
積分 1897
帖子 1452
閱讀權限 100
註冊 2008-7-28
狀態 離線
發表於 2009-6-17 10:00  資料  個人空間  短消息  加為好友 
Click 入香港 Apple Store 選購 iPhone 4、iPad、新款 iMac 及 2010 全新 Mac Mini!
Apart from hardware, I would rather to say that OS may play a importatn role in the test -- especially the multi-tasking test.  Actually my NC10 is quite clean but all of the sample machine are factory default setting -- which install too much rubbish, but it is quite hard to format their machine in such a short testing period.

Anyway, N280 is not such a power processor as we think.  chen_junru has point out the critical factor that N280 cannot run in 667 FSB under 945GSE, so the main difference between N280 and N270 is 0.06 MHz and is there any improvement with this 0.06 MHz?

頂部
 



當前時區 GMT+8, 現在時間是 2010-9-3 16:32

TOP
本論壇支付平台由支付寶提供
攜手打造安全誠信的交易社區 Powered by Discuz! 5.5.0  © 2001-2007 Comsenz Inc.
Processed in 0.034046 second(s), 9 queries , Gzip enabled  
清除 Cookies - 聯繫我們 - 數碼縱橫 - Archiver

Copyright c 2007 Digital XY. | All Rights Reserved