最新一代的Intel腳位架構也就是LGA 1155,正式在今年1月初發表並同步上市
首波6系列晶片組共有兩款,分別為P67/H67晶片組,兩者有著不同的市場定位
依以往LGA 1156的命名規則來判斷,P代表高效能,沒有VGA輸出功能,常見為ATX設計
H67也可以得知為內建VGA輸出裝置,但沒有CPU超倍頻功能,較適合搭配Micro ATX規格
從一月份到現在二月底,windwithme已經分享過兩款P67的超頻測試
使用CPU為較熱門的i7-2600K與i5-2500K兩款,做為新一代PC平台的效能與超頻教學
不過Intel在1/31發佈SATA2裝置的BUG問題,長時間使用下可能會碰到HDD效能衰退的狀況
在同時Intel也聲明要回收B2晶片組的訊息,並盡快在三月後提供更新版B3晶片組來解決此問題
此舉導致整個MB/NB廠商必須盡快做出因應之道,幾乎各家MB廠都已經採取產品下架或是對已購買者提供的換貨方案。
有關這方面的後續處理訊息,希望已購買或是未來會購買的消費者,可以多了解目前6系列晶片組的狀況
回到正題,此回是小弟第一篇分享Intel H67晶片組的相關測試文章
MB品牌為ECS 精英電腦,產品型號為H67H2-M,屬於Black Series系列,為自家高階產品線
最近一年多來,可以在網路或商場看到ECS也開始致力於OC市場,希望能在消費市場逐步建立起自己的品牌形象。
H67H2-M外包裝使用亮面反光材質
這樣的外盒設計會讓產品看起來比較搶眼
內附配件
英文說明書、簡易安裝手冊、IO檔板、驅動軟體光碟與IO介面塑膠保護蓋
SATA線材、外接eSATA裝置
H67H2-M本體
主要為黑色PCB設計,上方的貼紙已經說明產品的主要特色
CPU、DDR3插槽上的針腳使用15μ黃金接點,可提供 3 倍額外的抗氧化性、耐熱性以及耐刮性。
Micro ATX規格,尺寸大小為244mm x 244mm
H67H2-M屬於較高階的H67產品,在用料與設計上有別於市場其他平價的H67
主機板左下方
1 X PCI-E Gen 2.0 X16
2 X PCI-E X1
1 X PCI
Realtek RTL8111E雙網路晶片,支援Teaming
Realtek ALC892音效晶片,支援8聲道與High Definition Audio技術
Etron USB 3.0 控制晶片,後方IO介面上兩個USB 3.0擴充裝置
主機板右下方
3 X 白色SATA,H67晶片提供,SATA2規格
2 X 灰色SATA,H67晶片提供,SATA3規格
可混合使用,並且達到支援 RAID 0, RAID 1,RAID 5及RAID 10等規格
Power、Reset按鈕,內建Debug LED
類似火燄形狀的散熱片,下方為Intel H67晶片組,為單晶片設計
主機板右上方
4 X DIMM DDR3,支援1066/1333,DDR3最高容量支援到32GB。
旁邊為24-PIN電源輸入
主機板左上方
LGA 1155 CPU安裝處
H67H2-M採用4+1+1相供電,Intel限制H67無法超頻CPU與32nm製程上已經足夠使用。
IO
D-SUB/DVI/HDMI/顯示連接埠
clr CMOS按鈕
4 X USB 2.0(紅/黑色)
2 X USB 3.0(藍色)
1 X eSATA/USB 2.0共用(紅色)
2 X RJ-45網路孔
1 X S/PDIF
搭載ECS Black Series常見的Qooltech III雙熱導管散熱模組設計
LGA 1155針腳外觀
針腳使用15μ黃金接點,俗稱三倍金用料
BIOS畫面
M.I.B X,ECS超頻技術
此頁面有調整電壓選項.DDR3 1T/2T與目前CPU時脈/DDR3容量/電壓狀況
CPU Voltage +20~320mV
DIMM -800~+630mV
IMC Voltage +15~+675mV
ICC超頻選項頁面
CPU技術設定頁面
TurboBoost、C1E節能技術在此可以選擇
DRAM時脈資訊
Intel內建GPU超頻選項
超頻時建議先將Graphics Current設定Max
時脈設定在Graphics Core Ratio Limit,預設為22
每一階代表時脈為50MHz,圖中設定34則為34 X 50 => 1700MHz
IGD Memory有三種選擇,分別是32/64/128MB
DVMT/FIXED Memory的選項也有三種,128/256MB/Maximum
其他CPU相關技術選項
除了關閉C1E之外,還要將Power Technology做關閉才可以讓CPU維持在原廠規格時脈3.3GHz運作。
PC Health Status
H67晶片組不提供CPU外頻或是倍頻上的超頻選項
基本上就是給All In One PC來使用,所以Intel在這方面比起上一代LGA 1156做出更多的限制
不過對於內建GPU超頻選項卻是H67的特點,這方面的BIOS設計在ECS H67H2-M上都有照顧到。
測試平台
CPU: Intel Core i5-2500K
MB: ECS H67H2-M
DRAM: CORSAIR CMX8GX3M2A1600C9
VGA: Integrated Graphics  
HD: Intel X25-V 40GB RAID 0
POWER: CORSAIR CX430W
Cooler: Intel 原廠散熱器
OS: Windows7 Ultimate 64bit
CPU方面使用Core i5-2500K
產品型號為Intel Core i5-2500K,K系列為2010年中出現的新代號,代表著不鎖頻的功能
時脈為3.3GHz,支援新一代Turbo Boost 2.0技術,最高可達到3.70GHz效能
實體4 Cores但沒有Hyper-Threading技術,一共可達到4執行緒,簡稱4C/4T
32nm製程,功耗95W,L3 Cache共有6MB,目前為LGA 1155中第二高階規格的CPU
由於Sandy Bridge i3系列在二月底後才會上市,此回先用較為高階的4Cores i5來做測試
待未來小弟若有入手i3-2100,會再做詳細的效能測試,本次也可以了解i5-2500K內建GPU的實際效能。
效能
CPU 99.8 X 35 => 3293.4MHz
開啟Turbo Boost、開啟C1E
DDR3 1064.4 CL7 7-7-20 1T
Hyper PI 32M X4 => 11m 54.091s
CPUMARK 99 => 568
Nuclearus Multi Core => 19330
Fritz Chess Benchmark => 20.81/9986
CrystalMark 2004R3 => 223913
CINEBENCH R11.5
CPU => 5.41 pts
CPU(Single Core) => 1.47 pts
PCMark Vantage => 15688
Windows體驗指數 - CPU 7.5
i5-2500K CPU搭配在H67上的效能與先前測試過的P67互有勝負,但之間的落差不會太大
使用H67或P67晶片組在預設值下都有很相近的效能,這方面表現還不錯
對比於先前分享過的i5-760,新款2500K在CPU預設值效能上大約有2~3成左右的增加
一方面是同時脈下,LGA 1155有比LGA 1156高出10%左右的效能
再加上i5-2500K為3.3GHz,而i5-760只有2.8GHz所造成的效能差異,提供給網友作為參考
DDR3頻寬比較
DDR3 1064.4 CL7 7-7-20 1T
ADIA64 Memory Read - 14091 MB/s
Sandra Memory Bandwidth - 14133 MB/s
以上DDR3頻寬效能測試,可以看出應該有改進上一代的問題
也就是對於內建GPU的CPU,DDR3在頻寬的話會低上許多,這是LGA 1156 i3/i5 32nm CPU的特殊狀況。
Sandy Bridge i5-2500K同樣內建GPU,但在DDR3頻寬上有很優秀的表現
只運作DDR3 1064就已經比上一代i3-540 DDR3 1333多出約36~59%的頻寬
希望未來有機會分享到i3-2100時也有如此高的頻寬,BIOS內沒有DDR3 1333選項供選擇是較為不足的地方。
溫度表現(室溫約20度)
系統待機時 - 20~29
CPU全速時 - 66~71
Intel Burn Test v2.4,Stress Level Maximum
Intel原廠銅底散熱器,個人建議使用者在安裝前再塗些散熱膏,散熱效能會比較好
由主機板監控硬體再透過軟體顯示出來的2500K溫度如以上的數據,會有些許的誤差,並非真正實際的溫度
感覺上溫度數據有些偏高,但實際靠近散熱器旁邊的風溫,沒有這樣的溫度感覺,也許是測溫軟體顯示偏高的結果。
耗電量測試
系統待機時 - 29W
CPU全速時 - 111W
開啟C1E節能技術搭配32nm製程的技術,讓擁有4Cores核心的2500K在耗電量方面低到驚人
全速時的耗電量也可以讓人接受,就算日後擴充中階等級的VGA,在PSU的需求也可能只需要400~450W而已。
H67H2-M採用H67晶片組的平台,可以讓消費者擁有更高效能的小型PC主機或是HTPC等用途定位
Sandy Bridge架構上不同於上一代LGA 1156平台,此回在Core i3~i7全系列CPU中都內建GPU
名稱為Intel HD Graphics,共有HD2000、HD3000兩個系列
其中較高階版本HD3000,目前只有2405S/2500K/2600K系列有搭載HD3000等級之GPU
以下測試為i5-2500K搭載的HD3000,一般狀況為850MHz,最高預設時脈達到1100MHz
也就是在前半部BIOS部分中有提過為22倍頻
Intel HD Graphics HD3000效能測試
預設時脈1100MHz
3DMark Vantage => P1677
StreetFighter IV Benchmark
1280 X 720 => 40.42 FPS
DEVIL MAY CRY4
1280 X 720 => 37.14/32.26/41.35/24.38 fps
FINAL FANTASY XIV
1280 X 720 => 669
H67平台在3D效能有很大的進步,比起上一代H55搭配i3-540的3D效能來說
3DMark Vantage有4倍以上的進步,StreetFighter IV Benchmark大約兩倍左右的效能
這對內建GPU的世界來說,可以說是大躍進,從先前LGA 1156的內建3D效能追上對手AMD後
Intel在這回LGA 1155 Sandy Bridge平台更以3D效能倍進的優勢而來
對於Intel想積極擴展在HTPC市場佔有率的企圖可說是不言而喻
超頻時脈1700MHz
3DMark Vantage => P2536
StreetFighter IV Benchmark
1280 X 720 => 47.49 FPS
DEVIL MAY CRY4
1280 X 720 => 48.49/43.25/57.83/33.63 fps
FINAL FANTASY XIV
1280 X 720 => 881
使用H67H2-M BIOS超頻GPU功能,在不加壓的情況下超頻到34倍頻
以上數據可以知道HD3000超頻範圍不算小,一般來說要超頻到1700~1800MHz並不會太困難
超頻後在3DMark Vantage有近50%的效能增加
其他方面如StreetFighter IV大約20%,DEVIL MAY CRY4/FINAL FANTASY XIV也有30%左右
如果在這些Game Benchmark能像有3DMark Vantage那麼高的3D效能,結果應該會讓消費者更滿意
影音效能的進步也是Sandy Bridge主要的特色之一
以下為測試Intel最新推出的Quick Sync Video技術
MediaEspresso來比較內建GPU與外接VGA兩種情形下,影片轉檔所需的時間
DVD檔案VTS_01_1.VOB,容量大小為1023MB,2500K轉成MP4檔案
Intel HD Graphics HD3000
開啟Intel Quick Sync Video(品質較佳)
6分49秒
msi N560GTX-Ti Twin Frozr II
開啟nVIDIA CUDA技術
8分09秒
可以見到在同一個檔案中,Intel Quick Sync Video比nVIDIA CUDA快上1分20秒
如果將原本使用外接VGA配合CUDA技術,換成H67自身輸出Intel內建HD3000大約可以節省37%以上的時間。
在使用者為外接VGA的狀況下,此時就無法開啟Intel Quick Sync Video技術,運用時需要注意到這一點
此外兩種轉檔技術之下,畫質的表現可能也是大家在意的一點,這方面日後小弟會再用肉眼仔細比較差異。
影片撥放測試
使用手邊購買的原版影片,藍光版本為Inception 全面啟動
以上影片為該公司版權所有。
藍光影片因為要在PowerDVD影片放映模式與Windows桌面同時截圖不易,此處直接拍攝畫面做為參考。
CPU使用率大約落在13~20%左右,對於目前高解析度的1920 X 1080的影片也可以順暢撥放
ECS H67H2-M
優點
1.屬於較為高階的H67產品,包裝精美與配件用料都還不錯
2.15μ黃金接點、Realtek RTL8111E雙網路晶片、USB 3.0/eSATA等用料
3.BIOS提供超頻GPU功能,讓Sandy Bridge的3D效能可以發揮到更高的水準
4.搭配2500K在功耗方面表現讓人滿意,在整體效能也有很高的水準
5.Qooltech III為雙熱導管散熱模組、內建除錯燈號與POWER/Reset按鈕
缺點
1.BIOS建議可以開啟DDR3時脈選項,提供1066/1333讓使用者自由選擇
2.CPU旁邊的散熱片面積有些過於龐大,若要加裝較高階的散熱器可能會有卡到的狀況
3.相比於其他一線MB廠,未來ECS在通路擴展與送修能否便利將是消費者比較在意的兩個環節
效能比 ★★★★★★★★☆☆
用料比 ★★★★★★★★★☆
規格比 ★★★★★★★★☆☆
外觀比 ★★★★★★★★★☆
性價比 ★★★★★★★★☆☆
ECS推出的H67H2-M在各方面表現都很不錯,BIOS方面若能再增加DDR3細部選項會更好
規格上對比於其他MB大廠的H67也已經稱得上相當有水準的用料
在Micro ATX中擁有雙網路、USB 3.0、eSATA等擴充性,打造出高階的小PC主機
價位方面,在美國售價約美金149元,折合台幣約4372元
以這樣等級的用料水準,屬於高階的H67設計,在市場價位上也會較高一點
加上各家MB品牌在市場上的H67價位仍然不低,ECS H67H2-M的市場價位,C/P值還算不錯。
Sandy Bridge整合型晶片組未來較為超值的主打應該會是H61,價位上會有更多的空間
有關Intel 6系列晶片組的Bug事件,ECS自家網站也在2/7發出聲明
停止對於Sandy Bridge B2版本的出貨動作,對於已經購買的消費者,未來也可以無條件做更換B3全新品的服務
這方面的處理模式與其他一線MB大廠相同,算是很有誠意的後續處理方式,希望將P67/H67設計缺陷帶來的傷害減到最低。
因為Sandy Bridge架構加入市場,也讓Micro ATX MB效能又往前邁向一步,加上在多媒體方面也有明顯地進步。
尤其是Intel Quick Sync Video的轉檔效能,讓人使用過後覺得印象十分深刻
對於喜歡小型PC主機的消費者是一大利多,不過在H61還沒上市前,使用H67是需要較高花費的選擇
本篇分享ECS H67H2-M規格與各方面效能介紹,希望能提供給有需要的消費者做為參考
如果小弟的分享文章對您有幫助,請幫我按個推,感謝

windwithme 發表在 痞客邦 留言(2) 人氣()



Intel從去年2010下半年度開始,在NoteBook領域的架構開始跨入新的世代
主要是因為Intel推出不同於以往的Core 2 Duo的核心架構
開始讓各家品牌著手於導入Core i系列的產品線,此架構已經出現在PC市場超過一兩年的時間。
讓活躍於NoteBook市場許多年的Core 2 Duo正式下台一鞠躬

windwithme 發表在 痞客邦 留言(1) 人氣()



In early January 2011, Intel announced new generation platform, LGA 1155 CPU socket.
It’s called Sandy Bridge, also named as 2nd generation Intel Core Processor.
There are three key differences with old structure. First, the CPU clock changes from 133MHz to 100MHz.

windwithme 發表在 痞客邦 留言(1) 人氣()



以往VGA市場上的經驗看來,一個世代的系列產品大約可以生存一年多到兩年左右
由於NVIDIA與ATI兩大GPU晶片製造商在這兩三年來技術競爭得越來越激烈

windwithme 發表在 痞客邦 留言(6) 人氣()



在2011年1月初Intel推出新一代的平台,CPU腳位為LGA 1155
新架構統稱為Sandy Bridge,也稱為第2代Intel Core(酷睿)處理器
此架構比起上一代LGA 1155有些許不同,首先是CPU外頻133MHz改為100MHz
第二項不同處在於CPU不超頻的狀況下,最高可支援到DDR3 2133
此外每款Sandy Bridge CPU都有內建新一代的GPU,搭配H6X晶片組會有GPU輸出的功能
本文在1/30剛完成沒多久就碰到Intel在1/31發表的召回聲明,網路上已經有很多的資料與相關討論
大意就是6系列晶片組在SATA2的裝置有部份機率會出現不穩定或找不到HDD裝置的狀況
對於已經購買的消費者,某些MB品牌是建議先安裝在兩個P67 SATA3的裝置上
如果碰到使用2個以上的HDD/SSD/DVD-ROM裝置的使用者
除非購買的P67有再加裝兩個非P67提供的SATA裝置,否則在這方面就比較難配置

windwithme 發表在 痞客邦 留言(2) 人氣()



In the past 3 years, PSU market competition is on 80Plus.
All brands make own design be more green and get better certificate from 80Plus.
As per 80Plus level, it has platinum, gold, silver and bronze. It bases on energy efficiency.
Of course, this standard is not absolutely. The excellent PSUs also need other features.
I think components are also the key factors.
CORSAIR is a well-known memory brand from USA. They have very good reputation in DRAM products.
CORSAIR PSU product is just around 3 years. Their products are positioning at high quality and stability for mid-high end.
Previously, CORSAIR PSU has HX, TX and VX series. The highest one is 80Plius Silver certificate.
In months ago, they launched new AX, GS, and CX product series.
This review is most high end AX series, the 80Plus Gold level.
It belongs to Professional Series Gold. AX has 3 products, 750/850/1200W.
I use entry AX 750W to share this latest high end PSU.
First of all, let’s see the package. The box main symbols are spec and technical feature.
7 years warranty, 80Plus Gold certificate are 2 main features.
Accessories
Product manual, cable ties, CORSAIR sticker, screws and power cord
PSU body is packed by flannelette bag
Professional Series Gold AX750W Body
The shell is black matt baking for good quality.
Hybrid Silent Fan Control technology has Fanless mode, Quiet mode and Cooling mode.
When the PSU output below 20%, the fan is complete stationary and silent.
Quiet mode is working as PSU output between 20~50%.
The fan starts to work and the noise level is 16dBA.
High quality cable collecting pack and modulized cables
12 X SATA
8 X 4 Pin
2 X Floppy
1 X ATX 24 Pin
1 X EPS/ATX12V 8/4 Pin
4 X PCI-E 6+2 Pin
The back design is cell heating holes.
Active PFC spec for 110~240VAC
Product Spec
Model, CMPSU-750AX
Simplex +12V design. Total is 62A and max is 744W
Modulized cable connectors. It can be better if it’s with clearly indicate.
Even MB 24/8 Pin cable is also moduliozed. It’s very rare design.
I remember there is only X-SERIES 750W using this.
Take a look inside.(PS:Please don’t do it by yourself, due to it will be out of warranty.)
The fan is SANYODENKI 9S1212F404. 12cm dual balls bearing with temperature controller.
A bird’s eye view
Internal heatsink is black
The capacitors are solid and Japanese made to enhance AX750W stability and durability.
Double faced PCB design
I guess CORSAIR AX750W and my previous review, Seasonic X-SERIES 750W, should be from same design team.
Many designs and components are similar. However, after checking the detail, the PCB design is some different.
CORSAIR PCB and heatsink are black and heatsink is bigger.
Also PCB some layout is changed. AX750W could be X-SERIES 750W revised version.
I use MB GIGABYTE X58A-UD3R; due to X58 is the most high end Intel platform.
It supports On/Off Charge, USB 3.0/SATA3, Cloud OC and 2oz copper PCB.
The platform is X58A-UD3R with Intel Core i7-950.
The main reason is LGA1366 is the most power consumption platform so far. It’s very suitable for PSU loading test.
System Configuration
CPU: Intel Core i7-950
MB: GIGABYTE GA-X58A-UD3R
DRAM: CORSAIR DOMINATOR-GT CMG6GX3M3A2000C8
VGA: GIGABYTE GTX460 OC
HD: Intel X25-V 40GB RAID 0
POWER: CORSAIR Gold AX750W
Cooler: Mega Shadow Deluxe Edition
OS: Windows7 Ultimate 64bit
The test below is under default setup and OC CPU to 4140MHz to compare PSU output status in middle and high power consumption modes.
Default Setup
CPU 133.2 X 24 => 3197.7MHz
DARM DDR3 1865.2 CL7 8-7-20 1T
OCCT V3.1.0
POWER SUPPLY Mode
Enter to OS desktop - 153~156W
Run OCCT Power Supply Test Items
i7-950 default is 3.07GHz. 4 physical cores and 4 HT cores full load plus GTX460 - 392~418W。
Power Angel shows this X58 default mode max power consumption is 418W
According to 80Plus website,
output 50% transfer efficiency is around 90.80%
After transferring, 110V 418W is around 380W real output.
OCCT Test Result
CPU Voltage Variation
Disable Intel C1E in default mode. CPU clock and voltage both rise to highest.
3.3V
The highest is 3.34V and mostly it’s between 3.31~3.33V during test. The voltage wave is around 0.90%
5V
Mostly it’s between 4.84~4.89V. The voltage wave is around 1.02%.
12V
The highest is 12.37V and the mostly is 12.24V. The voltage wave is 1.05%.
3V and 5V voltage wave is around 0.9~1.02%. It’s very good condition.
Even 12V is only 1.05%. You can see this PSU in high load condition is still working excellent.
Using More Accurate Meter
Idle
5V testing result is 5.03~5.05V which is higher than board meterage 4.89V.
OCCT Full Speed
The voltage is 5.06V which is higher than idle condition.
Idle
12V testing result is 12.18~12.19V which is some different with board meterage 12.37V
OCCT Full Speed
The voltage is 12.17V which is lower than idle condition around 0.01~0.02V
5V is rising to higher voltage in full speed condition.
12V only has 0.01~0.02V small wave. This meterage is much lower than mainboard measurement.
OC System Setup
CPU 180 X 23 => 4140MHz
DARM DDR3 1800 CL7 8-7-20 1T
Enter to OS Desktop - 186~190W
Run OCCT Power Supply Test Items
i7-950 OC to 4.14GHz, 4 physical cores and 4 HT cores full load plus GTX460 - 596~636W。
Power Angel shows this X58 OC mode max power consumption is 636W
According to 80Plus website, there are only 20, 50 and 100% transfer efficiency.
I decide to use transfer efficiency at 100% and it’s about 88.11%
After transferring, 110V 636W is around 560W real output.
OCCT Test Result
CPU Voltage Variation
Disable Intel C1E in default mode. CPU clock and voltage both rise to highest.  
3.3V
The highest is 3.33V and mostly it’s between 3.28~3.3V during test. The voltage wave is around 1.50%
5V
Mostly it’s between 4.87~4.76V. The voltage wave is around 2.26%.
12V
The highest is 12.3V and the mostly is between 12.18~12.24V. The voltage wave is 0.96%.
3V and 5V voltage wave is around 1.5% and 2.26%.
Even 12V is only 1.05%. You can see this PSU in high load condition is still working excellent.
3V and 5V voltage wave is around 1.50 and 2.26%.
The wave range is higher than default mode. However, it’s still much lower than acceptable 5% tolerance.
12V is even lower. It’s just only 0.96%. You can see this PSU in 75% high load condition is still working excellent.
Using More Accurate Meter
Idle
5V testing result is 5.05V which is higher than board meterage 4.87V.
OCCT Full Speed
The voltage is 5.06V which is lower than idle condition around 0.01V
Idle
The voltage is 12.17~12.18V which is different with board meterage 12.3V
OCCT Full Speed
The voltage is 12.14V which is lower than idle condition around 0.03~0.04V
You can see no matter idle or OC, 5V is raising 0.1V
There is no 5V dropping 2.26% status in OCCT software.
12V wave is only 0.03~0.04V. Even though it’s bigger than default, it’s still only 0.33% wave.
These 2 factors are much lower than mainboard meterage.
CORSAIR Professional Series Gold AX750W Conclusion
Good
1.Professional Series new packing and accessories are very good quality
2.CORSAIR provides 7 years warranty
3.Modulized design for better flexibility. The inner capacitors are Japanese made and solid caps.
4.80Plus Gold Certificate and the energy efficiency is 90.80%
5.5V is no dropping in meter test and 12V is only falling 1%. The voltage wave is very small.
6.It supports Hybrid Silent Fan Control technology which fanless below 20% output.
Weak
1.Suggesting they can add 6 X PCI-E 6+2 Pin
2.80Plus Gold Certificate and good components make the price be too high
Performance ★★★★★★★★★☆
Components ★★★★★★★★★★
Specification ★★★★★★★★★☆
Outlook     ★★★★★★★★☆☆
C/P Value   ★★★★★★★★☆☆
I used to review Seasonic X-SERIES 750W which is good components and performance Gold product.
You can see CORSAIR Professional Series Gold AX750W should be ODM by Seasonic.
Before, Seasonic ODM PSU has better components than own brand one.
AX750W inner layout is very different with X-SERIES 750W. I think it should be the revised upgrade version.
There are some special parts within my test.
First is the fanless mode in Idle condition. The PSU body temperature is also low.
During CPU/VGA full speed running, fan rpm is controlled by output rate. I don’t hear any obvious noise.
PSU body temperature is always below 40 degree C. It’s good control for noise and temperature.
CORSAIR AX750W price is around 170USD.
The key concerns of buying PSU are components, 80Plus Level, price, warranty and brand awareness.
AX750W is high end components level. I think 80Plus level doesn’t equal to components.
For the price, many non-modulized or lower components grade PSU price will be much lower.
I hope this review can be reference for the users who want to own a Gold Certificate and high end components PSU.
windwithme will introduce more Gold Certificate PSU in near future. Thanks. :)

windwithme 發表在 痞客邦 留言(0) 人氣()



近兩三年來有關於Power Supply的市場產品好像變成以競爭80Plus為主
各品牌無不希望自家的產品設計可以更節能,拿到80Plus更高的認證
經由80Plus的制度來分級,一共有白/銅/銀/金四種等級,以轉換功率的高低為區分標準。
當然這樣的節能標準並非絕對,優秀的Power Supply必須還有其他的特點
面子裡子都要兼顧到,所以個人認為用料的好壞也是很重要的一個環節
美商CORSAIR是一家歷史悠久的Memory品牌,在DRAM產品都有著良好的口碑
CORSAIR的產品線在Power Supply領域中並不算悠久,約略只有三年左右的時間
應該不難了解該品牌在產品定位上多是要求高品質與穩定為主要目標,走的是中高檔的路線
先前CORSAIR Power Supply擁有HX、TX與VX三系列的主要產品線,最高的80Plus認證等級達到銀牌。
最近幾個月又推出新產品線,主要分別為AX、GS、CX三個系列
此回的主角是三系列中認證等級最高的AX,也就是80Plus金牌的最高節能等級
屬於Professional Series Gold,目前AX一共有三種等級可以選擇,750/850/1200W
測試中以AX入門款750W來分享新規格的高階Power Supply產品
首先看到產品外包裝,主要標示產品規格與技術說明
七年保固、80Plus金牌認證為此產品的兩大特點
內附的配件
產品說明書、束線帶、CORSAIR貼紙、固定螺絲與電源線
Power Supply本體使用一個質感很好的絨布套包裹
Professional Series Gold AX750W本體
外殼使用黑色霧面烤漆,這部分質感相當不錯
Hybrid Silent Fan Control技術,以溫控的方式運用無響、靜音及散熱三種模式
在低負載20%以下時風扇並不會轉動,會形成完全沒有聲音的Power Supply
靜音模式是負載在20~50%時,風扇開始運轉,產生的聲音大小為16dBA
高質感的線材收納包與模組化線材
12 X SATA
8 X 4 Pin
2 X Floppy
1 X ATX 24 Pin
1 X EPS/ATX12V 8/4 Pin
4 X PCI-E 6+2 Pin
後方設計為市面上常見的蜂巢式散熱孔
Active PFC規格,適用於110~240VAC
產品規格表
代號為CMPSU-750AX
單路+12V設計,總共62A,最高可達744W
模組化線材接孔,此處如果有明確的安裝說明貼紙會更好
連MB所需要24/8 Pin都使用模組化方式,這方面倒是極少見的設計
先前的印象好像只有X-SERIES 750W有此設計
拆開內部一覽(註:一般來說拆開Power Supply會喪失保固,沒有必要請勿模仿)
風扇品牌使用SANYODENKI 9S1212F404,為一款12公分的雙滾珠溫控靜音風扇
鳥瞰整個內部構造
內部散熱片的顏色為黑色
內部使用電容鋁質固態電容與日系電容加強AX750W的穩定性與耐用度
雙面板PCB設計
猜測CORSAIR AX750W與小弟先前分享過的Seasonic X-SERIES 750W應屬系出同門
許多地方的設計、用料都很神似,比較兩款產品的內部PCB設計還是有些許不同
CORSAIR PCB與散熱片改用黑色,而且散熱片較多且面積也較廣
另外PCB內有些許佈線做過更動,AX750W有可能是X-SERIES 750W的改良版本
測試的主機板使用GIGABYTE X58A-UD3R,在Intel平台中屬於最高階的桌上PC平台
支援On/Off Charge功能、USB 3.0/SATA3、Cloud OC與2盎司純銅電路板等設計
以高階入門等級的X58A-UD3R搭配Intel Core i7-950為主要平台
原因是LGA 1366耗電量是目前較高的平台,適合做Power Supply負載測試
測試平台
CPU: Intel Core i7-950
MB: GIGABYTE GA-X58A-UD3R
DRAM: CORSAIR DOMINATOR-GT CMG6GX3M3A2000C8
VGA: GIGABYTE GTX460 OC
HD: Intel X25-V 40GB RAID 0
POWER: CORSAIR Gold AX750W
Cooler: Mega Shadow Deluxe Edition
OS: Windows7 Ultimate 64bit
以下測試將會以系統預設值與CPU超頻4140GHz兩種狀況的耗電量
主要目的是要比較出中度耗電與高度耗電下的電壓波動幅度
預設系統設定值
CPU 133.2 X 24 => 3197.7MHz
DARM DDR3 1865.2 CL7 8-7-20 1T
OCCT V3.1.0
POWER SUPPLY模式
進入OS系統桌面下 - 153~156W
執行OCCT中Power Supply測試項目
i7-950預設3.07GHz,四核實體加四核HT模擬滿載加上GTX460 - 392~418W。
Power Angel在這個X58預設平台的耗電量,顯示出來最高約418W
依照80Plus網站的資料,負載50%時轉換功率約90.80%
把110V 418W轉換後大約最高耗電量為380W
OCCT測試結果
CPU電壓變化
預設值下關閉Intel C1E節能技術,CPU時脈與電壓都會升到最高
3.3V
最高時3.34V,測試中大多在3.31~3.33V跳動,電壓跳動範圍在0.90%左右
5V
大約落在4.84~4.89V,電壓跳動範圍在1.02%左右
12V
最高時12.37V,測試中大多一直線落在12.24V,電壓跳動範圍在1.05%左右
3V、5V的電壓跳動範圍在0.9~1.02%左右,這兩個電壓數據的表現很好
就連12V的電壓落差也只有1.05%,對於此款Power Supply在中負載的狀況表現屬於高水準
使用比較精準的電表測量
待機時
5V測出來為5.03~5.05V,比起主機板測出來的4.89V還要高一些
OCCT全速運作時
電壓範圍5.06V,電壓數據在全速時比待機還要高一點
待機時
12V測出來為12.18~12.19V,比起主機板測出來的12.37V有些差異
OCCT全速運作時
電壓範圍12.17V,電壓在全速時降到比待機時低0.01~0.02V
5V表現在全速時反而升到比較高的電壓
12V也只有0.01~0.02V較小的波動範圍,這方面的數據都比起主機板軟體測出的電壓波動低上許多。
超頻系統設定值
CPU 180 X 23 => 4140MHz
DARM DDR3 1800 CL7 8-7-20 1T
進入OS系統桌面下 - 186~190W
執行OCCT中Power Supply測試項目
i7-950超頻4.14GHz,四核實體加四核HT模擬滿載加上GTX460 - 596~636W。
Power Angel在X58改為超頻平台的耗電量,顯示出來最高約636W
依照80Plus網站的資料,因為只有測試20/50/100%的轉換功率
最後決定採用負載100%時轉換功率,大約88.11%
環境中使用110V最高636W轉換後大約最高耗電量為560W
OCCT測試結果
CPU電壓變化
關閉Intel C1E節能技術,CPU時脈與電壓都會升到最高
3.3V
最高時3.33V,測試中大多在3.28~3.3V跳動,電壓跳動範圍在1.50%左右
5V
大約落在4.87~4.76V,電壓跳動範圍在2.26%左右
12V
最高時12.3V,測試中大多在12.18~12.24V跳動,電壓跳動範圍在0.96%左右
3V、5V的電壓跳動範圍各在1.50%與2.26%左右
這兩個電壓跳動範圍比預設值較多一些,比起一般可容許的5%電壓誤差值還要低上許多
12V的電壓落差比預設值更低,只有0.96%左右,在耗電量負載率75%的狀況下表現還是很好
使用比較精準的電表測量
待機時
5V測出來為5.05V,比起主機板測出來的4.87V還要高一些
OCCT全速運作時
電壓範圍5.06V,電壓數據在全速時比待機還要高0.1V
待機時
12V測出來為12.17~12.18V,比起主機板測出來的12.3V還是有點差異
OCCT全速運作時
電壓範圍12.14V,電壓在全速時降到比待機時低0.03~0.04V
可以得知不管在待機或是超頻時,5V的數據都是往上提升0.1V
並沒有OCCT使用軟體測試中,5V最高會掉2.26%的情況發生
12V也只有變為0.03~0.04V,雖然比預設值的範圍還要大,但只有0.33%的波動範圍
這兩方面電壓的數據都比起主機板軟體測出的電壓波動低上許多
CORSAIR Professional Series Gold AX750W使用心得總結
優點
1.Professional Series的新包裝與附件都更有質感
2.CORSAIR提供高達七年的原廠保固服務
3.模組化的最高彈性設計,標榜內部使用固態電容與日系電容搭配
4.通過80Plus金牌認證,轉換功率高達90.80%
5.5V在電表測量中電壓沒有往下掉,12V的落差也在1%以下,電壓波動範圍很低
6.同樣支援Hybrid Silent Fan Control技術,20%低負載的狀況時,散熱風扇不會運轉
缺點
1.建議可以增加到6 X PCI-E 6+2 Pin
2.80Plus金牌認證加上優秀的用料,導致價格也偏高
效能比 ★★★★★★★★★☆
用料比 ★★★★★★★★★★
規格比 ★★★★★★★★★☆
外觀比 ★★★★★★★★☆☆
性價比 ★★★★★★☆☆☆☆
先前分享過Seasonic X-SERIES 750W,是一款用料與效能都相當高的金牌產品
看得出來CORSAIR Professional Series Gold AX750W也應該是由Seasonic代工
以往由Seasonic代工的Power Supply經常會比自家的產品用料還好
AX750W在內部佈線明顯與X-SERIES 750W不同,推測應為改良後的加強版
測試中比較特殊的地方,第一印象就是在於風扇的控制
在一般待機使用下,風扇不會轉動,此外Power Supply本體的溫度也不高
測試中CPU/VGA全速的狀態下,風扇也依負載率來控制轉速,也沒有聽到明顯的噪音
Power Supply本體的溫度也壓在40度以下,對於噪音與溫度這兩方面的控制表現的不錯
CORSAIR AX750W在台灣市場可以看到的普遍價位大約台幣8000元以上,約美金274元
算是高價位的750W,也希望未來的價位可以降到只比Seasonic X-SERIES 750W高一點會更好。
挑Power Supply大約會考慮到用料品質、80Plus節能等級、市場售價、保固與個人品牌喜好度等幾點。
AX750W屬於高階用料的等級,個人認為不要把80Plus認證等級高低與用料的好壞畫上等號
當然價位方面,目前有許多非模組化或是用料比較沒有那麼優秀的Power Supply,價格上會低上許多。
希望本篇文章可以提供給想要金牌認證又想要擁有高階用料的消費者做為參考
最近小弟windwithme會再陸續介紹幾款品牌的金牌Power Supply分享,謝謝 :)

windwithme 發表在 痞客邦 留言(1) 人氣()



技嘉科技在1/7發表最新系列的高階產品線,採用上市兩年以上的Intel X58晶片組
此舉讓同時正在全力發表P67/H67新晶片組的其他品牌有著很大的不同
發表期間與美國CES消費電子展同步,顯示技嘉對於G1-Killer此系列的重視

windwithme 發表在 痞客邦 留言(0) 人氣()



About one and a half years ago, Intel released LGA 1156, and plus three chipset P55/H55/H57.
As to CPU, main products are Core i3/i5/i7, and their obvious features are:
dual-core with built-in GPU, quad-core with built-in HT, and quad-core without built-in HT respectively.
As mentioned before, the primary goal of LGA 1156 platform is to replace Intel high-intermediate-level products;
besides, LGA 1366 is still the predominant platform for the high-level ones.
The product positioning of the new LGA 1155 is similar to that of LGA 1156.
To most users, they probably feel that the replacing is too fast. When facing three kinds of LGA platform, they even feel perplexed.
If future price of LGA 1156 can be lowered appropriately,
it is also a good trend that the platform becomes the mainstream for low-intermediate-level products.
At the beginning of 2010, Intel introduced 32nm CPU, and the related products are:
dual-core—LGA 1156 i3 and i5; six-core—LGA 1366 i7 980X Extreme, representing low and high-level CPU respectively.
As for 32nm quad-core processor, Sandy Bridge LGA 1155 will be released in 2011.
This article will introduce the latest LGA 1155 platform, and analyse important features about 32nm Sandy Bridge CPU.
As for the CPU, it is Intel Core i7-2600K, and “K” means that CPU multiplier can be unlocked.
The clock is 3.4GHz (maximum: 3.80GHz), and Turbo Boost 2.0 technology is supported.
There are 4 Cores and Hyper-Threading technology, so 8 Threading is supported, also abbreviated as 4C/8T.
And, the CPU uses 32 nm manufacturing process, with 95W and 8MB L3 Cache, is the highest-level specification amongst current LGA 1155.
Back side of 2600K D1 and 2400 D2:
The motherboard is BIOSTAR TSERIES TP67XE, and this brand’s products are famous for parity and overclocking.
The colour of the package is different; this time red makes the looking of the motherboard more distinguishing feature.
Accessories:
User’s Manual, Rear I/O Panel for ATX Case, Serial ATA Cable, CorssFire/SLI Bridge, Fully Setup Driver CD.
As for colour matching, PCB is black, and PCI-E or DIMM are matched red or white; the whole appearance looks not bad.
Red matching is an eye-catching design.
More and more electronic products are matched red and black,
the colour combination (similar to the colour used in automobile racing) that stresses speed or high performance.
Lower-left of the motherboard:
2 X PCI-E 2.0 X16: support 2Way CrossFireX and SLI technology, with bandwidth X8+X8
2 X PCI-E X1
2 X PCI
LAN chipset: Realtek RTL8111DL
Codec: Realtek ALC892, with 8+2 channels and supporting Blu-ray Codec technology
Lower-right of the motherboard:
3 X red SATAII: provided by P67 chipset, support RAID 0, RAID 1, RAID 5, and RAID 10
2 X white SATAIII: provided by P67 chipset, support RAID 0, RAID 1, RAID 5, and RAID 10
Power button, reset button, and debug indicators
Upper-right of the motherboard:
4 X DIMM DDR3, supporting 1066/1333/1600/1866/2133, maximum capacity of DDR3 is 16GB. Only by overclocking can DDR3 2133 above be achieved.
1 X IDE, 24-pin power connector
Upper-left of the motherboard:
TP67XE adopts 10-phase power supply, the design that is enough for 32nm CPU. There are two 8-Pin power connectors,
which are usually seen only on high-level motherboard (such as for server).
The existence of dual 8-Pin is not aimed at high power consumption of P67,
but at LN2 users who have to consume more 12V under high clock (such as overclocking).
IO:
6 X USB 2.0 (Red/Black)
2 X USB 3.0 (Blue)
1 X eSATA/USB 2.0 (both Red)
1 X 1394
1 X RJ-45
1 X S/PDIF
In the past, I preferred not using a computer case when testing hardware.
But this time many net pals hope me to use a case so that the test result,
such as overclocking ability and temperature, will be more similar to most consumers’ condition.
The case is high-intermediate-level product
Antec DF-35, with 1 x top 140mm fan, 1 x rear white 120mm fan, and 2 x front variable-speed 120mm fans with white LED.
DF-35 has such advanced cooling system that it is helpful to keep the system steady after the computer is overclocked.
Test platform:
CPU: Intel Core i7-2600K
MB: BIOSTAR TSERIES TP67XE
DRAM: CORSAIR DOMINATOR-GT CMT4GX3M2B2133C9
VGA: GIGABYTE GTX460 OC
HD: Intel X25-V 40GB RAID 0
POWER: Antec High Current Gamer 620W
Cooler: Mega Shadow Deluxe Edition
Case: Antec DF-35
OS: Windows7 Ultimate 64bit
Whole LGA 1155 system in Antec DF-35:
BIOSTAR provides much software:
Energy-saving software has been very popular in recent years.
CPU can effectively lower power consumption according to user’s setting.
TOVERCLOCKER is BIOSTAR’s unique software
The software provides CPU, Memory information, and also shows H/W Monitor.
In TOVERCLOCKER, OC TWEAKER, which can adjust many clocks, is a useful tool to overclock.
Default performance:
CPU 100.1 X 35 => 3502.7MHz (Turbo Boost enabled, C1E disabled)
DDR3 2135.2 CL7 10-7-27 1T 1.650V
The architecture of Sandy Bridge CPU is different from that of previous Core i7. The bus speed is changed to 100MHz.
And about other frequency information, the highest ratio of 1366 CPU is 133/1866, 1156 CPU is 133/1600.
Now the default frequency of 1155 CPU can achieve 100/2133, making DDR3 clock show better performance.
Hyper PI 32M X8=> 14m 57.750s
CPUMARK 99=> 540
Nuclearus Multi Core => 22513
Fritz Chess Benchmark => 27.58/13236
CrystalMark 2004R3 => 332817
CINEBENCH R11.5
CPU => 6.83 pts
CPU(Single Core) => 1.43 pts
PCMark Vantage => 19941
3DMark Vantage
CPU SCORE => 53268
The default clock of 2006K is 3.4GHz. After Turbo Boost 2.0 is enabled, the clock can achieve 3.5~3.8GHz by CPU usage.
Hence, the test figures above are great, so the result means CPU can easily deal with many tasks under any circumstances.
And as to CPU performance under the same clock, I find that LGA 1155 is about 10% higher than LGA 1156.
As to DDR3 bandwidth, probably with more advanced Memory Controller technology, I find there is 20~30% performance improvement in much software.
Therefore, about basic performance and DDR3 bandwidth, Intel Sandy Bridge CPU with latest architecture has made great progress.
Temperature:
System standby: 33℃~37℃
[IMG]http://img257.imageshack.us/img257/6573/bp67dt1.pngCPU in full usage: 44℃~47℃
Intel Burn Test v2.4,Stress Level Maximum
Hence about the temperature of 2600K:
Standby: about 3℃~5℃lower than the temperature of 1156 45nm CPU.
Full usage: about 10℃lower than the temperature of 1156 45nm CPU.
It goes without saying that as to testing temperature, different motherboards display divergent results.
But through the comparing, we can observe the temperature difference between 32nm and 45nm of 4C8T.
Power consumption:
Standby: 84W
CPU full speed: 125W
As to power consumption, with 32nm manufacturing process technology, the power saving of this CPU is 30~50W more than that of 45nm Core i7.
So obviously Sandy Bridge CPU can save more energy, and this is also one of the reasons why consumers may prefer quad-core platform.
Please refer to my BIOS setting:
Boot logo:
Main page:
Compliency is AMI UEFI X64 2.0
O.N.E page:
The default setting of Turbo Mode is 38, 37, 36, and 35, with better CPU speeding up mode in version 2.0.
As to DRAM frequency, 1066~2133 modes are supported, so users can enjoy better DDR3 performance.
CPU Base Clock shows 10000, which means current bus speed is 100.00MHz; if it shows 10560, which means 105.60MHz.
Therefore, the item may not show an integer, such as 133, 166, 200MHz.
The setting above is DDR3 arguments, providing users to choose faster arguments; of course, making the system steady is also important.
There are 4 modes for CPU VCore Mode: SPEC Volgate/Auto/Offset Mode/Fixed Mode
CPU VCore Offset +0.010~+0.520V
CPU VCore Fixed 1.000~1.790V
DRAM Voltage 1.300~2.200V
CPU Configuration:
Some functions about CPU, such as common HT technology; you also can choose how many cores you want to activate.
PC Health Status:
With built-in monitor chipset, you can know main voltage or temperature information.
As to overclocking, this time it is not necessary for LGA 1155 to adjust too many items.
So just focus on CPU voltage, clock multiplier; and DRAM clock, arguments, voltage.
Due to the limited scope of bus speed (about 5~7%), using K series CPU would be a better choice to overclock substantially.
As to DDR3, I use US brand CORSAIR high-level DOMINATOR-GT series—the latest CMT4GX3M2B2133C9.
GT series use red and black as colour matching; in recent year, more and more motherboard and DRAM also use the two colours.
The argument specification is CL9 10-9-27 1.50V.
Compared with DDR3 voltage specification (1.650V) recommended by Intel, 1.50V only of GT 2133 can make the system steady.
AIRFLOW of Red-black version with DOMINATOR-GT makes the appearance quality nicer.
CPU 3502.8MHz DDR3 2135.2 CL7 10-7-27 1T 1.650V
ADIA64 Memory Read: 22132 MB/s
Sandra Memory Bandwidth: 28132 MB/s
MaXXMEM Memory-Copy: 21755 MB/s
CPU 5020.4MHz DDR3 2142 CL7 10-7-27 1T 1.650V
ADIA64 Memory Read: 25582 MB/s
Sandra Memory Bandwidth: 28475 MB/s
MaXXMEM Memory-Copy: 28139 MB/s
After CPU clock rises, the test figures of memory bandwidth in ADIA64 and MaXXMEM also rise obviously,
but not in Sandra Memory Bandwidth; this is the main difference amongst the three software.
And about the test figure of bandwidth, even if LGA 1155 and LGA 1156 are both dual-channel design,
the former shows about 20% greater performance than the latter does.
Although Sandy Bridge CPU only uses dual-channel design, its memory bandwidth can compete with that of LGA 1366.
ADIA64 Memory Read: 26605 MB/s
Sandra Memory Bandwidth: 29903 MB/s
MaXXMEM Memory-Copy: 29007 MB/s
Due to the limited scope of bus speed of Sandy Bridge architecture, about only 104~108MHz can be adjusted (depending upon each CPU).
The default CPU bus speed can contribute to achieve DDR3 2133 clock rate, so overclocking would not be a big problem.
If DRAM quality is good enough, performance over DDR3 2200 is not a difficult challenge.
Overclocking performance:
CPU 100.4 X 50 => 5020.2MHz(Turbo Boost disabled; C1E disabled)
DDR3 2142 CL7 10-7-27 1T 1.650V
Hyper PI 32M X8=> 12m 06.384s
CPUMARK 99=> 774
Nuclearus Multi Core => 32585
Fritz Chess Benchmark => 39.29/18859
CrystalMark 2004R3 => 427585
CINEBENCH R11.5
CPU => 9.82 pts
CPU(Single Core) => 2.05 pts
PCMark Vantage => 25062
3DMark Vantage
CPU SCORE => 65044
After 2600K is overclocked from 3.5 GHz to 5GHz, performance increases 20~30% in most software mentioned above.
When CPU is in full usage, voltage only needs 1.392V; hence, 32nm manufacturing process makes overclocking ability much better.
According to the related information I acquired on the Internet, 2600K over 5GHz is not commonly seen;
generally, the scope 2600K D1 is about 4.7~4.9GHz under air-cooling system.
The performance of Sandy Bridge is still better than 45nm 4C8T CPU (overclocking scope is 4.1~4.3GHz).
Temperature:
System standby: 43℃~46℃
CPU in full usage: 74℃~83℃
Intel Burn Test v2.4,Stress Level Maximum
If you want to overclock over 5GHz, you have to increase voltage, and note that temperature also rises much.
If you prefer low voltage and temperature for long-time using, 4.5GHz can be steady under 1.272V only.
Users can set personal setting by their preference.
Power consumption test:
System standby: 139W
CPU in full usage: 257W
Compared with power consumption of default clock rate:
Power consumption of system standby: 65% higher
Power consumption of CPU in full usage: 105% higher
It seems that the consumption increase so much, but as far as 5GHz high clock rate and high performance platform is concerned, 257W can be acceptable.
By the way, is 5GHz the maximum for 2600K under air-cooling? If fortunately, 5.1GHz can be achieved steadily.
CPU 100 X 51 => 5100.4MHz(Turbo Boost disabled; C1E disabled)
DDR3 2133.6 CL7 10-7-27 1T 1.650V
Hyper PI 32M X8=> 11m 48.382s
Fritz Chess Benchmark => 40.11/19251
As to the latest Sandy Bridge LGA 1155 CPU in 2011, 2300~2500K all use L3 Cache 6MB, and only 2600/2600K is equipped with 8MB.
K series are 2500K and 2600K, but other Sandy Bridge CPU, with limited scope of bus speed, may only have 5~8% overclocking scope.
Therefore, LGA 1155 is categorised definitely.
To the consumers who want to overclock, choosing K series is suggested, but with higher prices.
And most of the other Sandy Bridge CPUs may be suitable to the consumers who don’t overclock.
About the consideration above, I’m not sure that Intel is overconfident with Sandy Bridge performance, or that Intel wants to let LGA 1156 survive.
Originally, I thought that the reason why clock rate and power consumption of LGA 1155 platform made great progress was that 45nm manufacturing process became 32nm.
However, afterwards I find that all Sandy Bridge CPUs are built in GPU,
so it will be more convenient for users in future to use the chipset with display output function, such as H67.
Moreover, another advantage of this new platform is that it has higher performance of DDR3 and Turbo Boost 2.0 technology.
I use new generation P67 chipset in the test.
When manufacturers release P67 motherboards in the market, their prices, I estimate, will be a bit expensive than the prices of P55 motherboards.
BIOSTAR TP67XE is still featured parity, and the price is about US$160.
The motherboard supports the latest USB 3.0 and SATA3; unique BIO-Remote2 multimedia software; and THX TrueStudioPRO sound effects technology.
Furthermore, TP67XE also shows great performance of overclocking, so it can be a high C/P P67 product.
BIOSTAR now also aims for supporting multimedia software and hardware technology, making its products more competitive in the market.
I shared much overclocking experience about X58 and P55 before,
and I shall continue overclocking and analysing P67 motherboards from various brands in future.
If there is something this article doesn’t mention, I shall renew related information in the future reports.
Finally, I hope with the release of new generation products, the prices of previous generation platforms can be lowered faster.
So, consumers can choose more platforms with good quality at bargain prices.
Update
BIOSTAR TP67XE 1/3 BIOS - Internal PLL Voltage Override
2500K D2 OC 5404MHz

windwithme 發表在 痞客邦 留言(11) 人氣()



還記得時間在2008/11月左右,Intel舉辦了一場Core i7最新產品的發表會
會後兩年左右的時間,這期間內windwithme分享過15篇以上的X58測試文章
還有多款LGA 1366 Core i7 CPU的效能測試或超頻教學提供網友做為購買參考
才剛滿兩年沒過多久,2011年初Intel就隨即發表第2代的Core i系列處理器
使用新一代的腳位LGA 1155,還有搭配最新P67/H67晶片組,此架構俗稱的Sandy Bridge平台。
這場最新平台發表會訂定在2011/1/6,台北花博天使生活館做一場盛大的產品發表與說明活動
在G4入口處的建築物
一進到會場就看到許多媒體在拍攝Intel最新晶片
主持人是頗有名氣,形象也很專業的何戎
開場請來Intel許多主管級人物對於第2代Intel Core(酷睿)處理器做說明
Intel最新CPU 晶片不免要多秀一下
左右兩旁大型投影機,讓每位主講人對於新技術與新架構能更清楚地說明
3D繪圖技術也是本次的重點項目,據windwithme不負責任的消息指出
Sandy Bridge比起上一代Core i系列有2倍以上的3D效能增進
Memory的效能也更增加20%,第2代對於很多方面的效能都有明顯的增加
Turbo Boost 2.0技術,將自動超頻功能以核心數更加細分
實際以轉檔軟體CINEBENCH R11.5來測試多工下,核心時脈的變化
這部份是介紹Quick Sync Video的轉檔效能
將台北101煙火影片以第2代Intel Core CPU迅速轉檔完成後,再傳到iPhone中欣賞
三位Intel主管正式開啟第2代Intel Core CPU時代
3D眼鏡也是一項重點,發表會中有展示3D影片
看來Intel未來也將3D影像列為重要的發展目標
各大主機板廠所展示的P67/H55相關產品
Intel自家的DP67BG,在台灣市場比較難買到
小弟直到目前還沒使用過Intel公板MB,希望日後會有機會分享給網友測試心得
Intel DH67CF,尺寸為ITX規格
GIGABYTE GA-H67N--USB3(ITX)
GIGABAYTE目前最高階的產品GA-P67A-UD7
GA-H67MA-UD2H(Micro ATX)
近期在台灣也很重視高階超頻的品牌-ECS精英電腦
高階Black系列P67H2-A
ECS也會推出ITX規格的H67H2-I
BIOSTAR主打超頻與平價路線
H67 Micro ATX的產品為TH67XE
P67為TP67XE,此款小弟已經分享過相關文章,超頻能力表現優秀
ASUS SABERTOOTH P67
Maximus IV Extreme
MSI P67A-GD55
P67A-GD80
ASRock Fatal1ty P67 Professional
NoteBook方面未來也將會導入最新架構
擁有Intel Quick Sync Video的轉檔高效能、新一代的HD Graphics與Wireless Display 2.0
第2代Intel Core為32nm製程技術,也可以解決先前45nm i7的溫度與耗電量過高的問題
各大NB品牌也展示在未來搭載第2代Intel Core(酷睿)處理器的產品
MSI為遊戲推出的GT680
HP Dv6-4015tx
ASUS N435
Lenovo與GIGABYTE
新一代的Intel Core(酷睿)CPU在多媒體影音有更多的進步
未來在影片轉檔或是3D效能方面,對於NoteBook的效能將會有很大的幫助
ShowGirl部份,天使館內有四位Intel當天請來的專屬SG
右邊這位好像是常上娛樂節目的乃蓁
拿著Intel最新晶片讓參加發表會者拍攝,應該也是SG常見的任務
四位SG合照
最後是windwithme在花博會場的隨手拍攝
館內有可愛的貓型動畫
紅色好像快要爆掉的拍攝模式...
花博景觀很漂亮,感覺上如果花卉的數量可以多一些會更好
Intel在花博內舉辦發表會確實很特殊,主要是想把電腦科技結合生活與藝術中
影片3D特效也將是未來硬體的特點,會中Intel也展示3D影片搭配3D眼鏡使用,看起來更有立體感
第2代Intel Core(酷睿)處理器在多媒體影音上有許多新的技術,在此只能提出一些重點項目
當然未來要更了解新世代CPU的效能與新功能,可以參考小弟windwithme陸續發表的分享文章
下一篇Sandy測試平台為GIGABYTE P67A-UD7與Intel Core i7-2600K
但Intel Core i7-2600K要20號後小弟才會購入,希望過年時間可以來的及分享,謝謝  :)

windwithme 發表在 痞客邦 留言(6) 人氣()



大約在一年半前Intel推出LGA 1156平台,主要是以P55/H55/H57三種晶片組做為搭配
CPU這方面主要是Core i3/i5/i7三種,分別為雙核心內建GPU或四核心有無內建HT為產品區別
先前提過LGA 1156平台主要任務是取代Intel產品中高價位的市場,高階以上依然由LGA 1366來主導。
新推出的LGA 1155產品定位近似LGA 1156,對於大部份使用者來說,總覺得好像改朝換代太過於迅速。
當消費者在面對三種不同腳位的平台時,可能會有無所適從的感覺,這點應該是最讓消費者困擾的地方
如果未來LGA 1156的產品線可以適當地降價,成為中低階價位的主流平台,這樣的發展也不見得是一件壞事。
Intel在2010年初開始導入32nm製程的CPU,雙核心部分有LGA 1156的i3與i5兩系列CPU
六核心方面有LGA 1366的i7 980X Extreme,入門與高階CPU紛紛都有32nm CPU產品上市
讓人覺得在2010年內,Intel市場上遲遲沒有32nm的四核心產品發表
這個問題從2011年新平台中有了解答,也就是俗稱Sandy Bridge的LGA 1155新一代平台
本篇將介紹2011年最新的LGA 1155平台,剖析32nm Sandy Bridge CPU有哪些進步
CPU方面為Intel Core i7-2600K,K系列為半年前出現的新代號,代表有著不鎖頻的功能
時脈為3.4GHz,支援新一代Turbo Boost 2.0技術,最高可達到3.80GHz效能
實體4 Cores並有Hyper-Threading技術,一共可達到8執行緒,簡稱4C/8T
32nm製程,功耗95W,L3 Cache共有8MB,為目前LGA 1155中最高階規格的CPU
2600K D1與2400 D2兩種Revision背部比較
MB使用BIOSTAR TSERIES TP67XE,此MB品牌的產品以平價與超頻為主打
外包裝不同於以往的配色,改用全紅色系讓產品外觀看起來更有特色
內附配件
產品說明書、IO檔板、SATA線材、CorssFire/SLI橋接器與驅動軟體光碟
配色方面主要是黑色PCB,其他PCI-E或DIMM以紅白為搭配,外觀質感還不錯
紅色搭配在產品中比較搶眼,尤其目前越來越多3C都習慣以紅黑做為搭配
頗有賽車方面的味道,用顏色來強調競速或是高效能印象的目的
主機板左下方
2 X PCI-E 2.0 X16,支援2Way CrossFireX與SLI技術,頻寬為X8+X8
2 X PCI-E X1
2 X PCI
Realtek RTL8111E網路晶片
Realtek ALC892音效晶片,支援8+2聲道與Blu-Ray Codec技術
主機板右下方
3 X 紅色SATAII,P67晶片提供,SATA2規格,支援 RAID 0, RAID 1,RAID 5及RAID 10
2 X 白色SATAIII,P67晶片提供,SATA3規格,支援 RAID 0, RAID 1,RAID 5及RAID 10
Power、Reset按鈕,內建Debug LED
主機板右上方
4 X DIMM DDR3,支援1066/1333/1600/1866/2133,DDR3最高容量支援到16GB。
DDR3 2133以上的規格需要配合超頻才可以達成,旁邊為24-PIN電源輸入
主機板左上方
TP67XE採用10相供電,以32nm CPU來說已經屬於非常足夠使用的供電設計
擁有2個8Pin電源輸入,一般都是在高階Server等級的MB才會看到的設計
雙8Pin的設計並不是P67耗電量偏高的問題,而是使用針對LN2使用者在高時脈下超頻需要耗更多的12V所設計。
IO
6 X USB 2.0(紅/黑色)
2 X USB 3.0(藍色)
1 X eSATA/USB 2.0共用紅色)
1 X 1394
1 X RJ-45網路孔
1 X S/PDIF
以往windwithme測試習慣使用裸測平台,也有不少網友提過希望裝入機殼實際測試
如此一來對於超頻能力的範圍與溫度顯示會更貼近絕大多數消費者的使用環境
故此回裝入Case中做超頻測試,採用在市場中高階價位的產品-Antec DF-35
頂端140mm風扇、後方120mm白色LED風扇、前方兩個120mm附風速控制鈕的白色LED風扇。
DF-35擁有這樣良好的散熱系統配備,對於消費者超頻後長時間使用很有幫助
測試平台
CPU: Intel Core i7-2600K
MB: BIOSTAR TSERIES TP67XE
DRAM: CORSAIR DOMINATOR-GT CMT4GX3M2B2133C9
VGA: GIGABYTE GTX460 OC
HD: Intel X25-V 40GB RAID 0
POWER: Antec High Current Gamer 620W
Cooler: Mega Shadow Deluxe Edition
Case: Antec DF-35
OS: Windows7 Ultimate 64bit
將整套LGA 1155電腦配備裝入Antec DF-35中的樣貌
BIOSTAR的軟體安裝頁面,可使用的軟體種類也相當豐富
近年來在各大MB廠很熱門的節能軟體,依選擇的狀態不同,讓CPU可以有效地降低耗電量
TOVERCLOCKER為BIOSTAR獨家設計的軟體
內容有CPU、Memory的相關資訊,也可以直接看到H/W Monitor內的各種狀況
其中的OC TWEAKER為使用軟體做調整多種時脈來達到超頻的效果,較為常見的一種簡易式超頻方式。
預設效能
CPU 100.1 X 35 => 3502.7MHz(開啟Turbo Boost,關閉C1E)
DDR3 2135.2 CL7 10-7-27 1T 1.650V
Sandy Bridge CPU架構不同於以往的Core i7,外頻改為100Mhz
另外在CPU除頻方面也做了些許改變,以往1366最高比例是133/1866、1156最高是133/1600
現在1155 CPU在預設外頻可以達到100/2133的比例,讓DDR3時脈有機會可以拉到更高的效能。
Hyper PI 32M X8=> 14m 57.750s
CPUMARK 99=> 540
Nuclearus Multi Core => 22513
Fritz Chess Benchmark => 27.58/13236
CrystalMark 2004R3 => 332817
CINEBENCH R11.5
CPU => 6.83 pts
CPU(Single Core) => 1.43 pts
PCMark Vantage => 19941
3DMark Vantage
CPU SCORE => 53268
2006K預設時脈為3.4GHz,開啟Turbo Boost 2.0後,依CPU使用率不同可以達到3.5~3.8GHz
因此對於以上的測試都有著很高的測試數據,面對各種使用狀況下也幾乎都可以輕鬆勝任
有關於同時脈下的CPU效能,經個人比較後,大約比LGA 1156還要高10%左右
DDR3的頻寬方面,在很多軟體上也有20~30%的效能增進,推測可能用更先進Memory Controller技術。
Intel讓最新架構的Sandy Bridge CPU在基本效能與DDR3頻寬都有顯著地進步
溫度表現
系統待機時 - 33~37
CPU全速時 - 44~47
Intel Burn Test v2.4,Stress Level Maximum
2600K待機溫度比起1156 45nm CPU大約降低約3~5度,全速溫度大約降低10度左右
每個品牌的MB測溫都不會一模一樣,因為使用的測溫晶片也不一定相同
加上MB測出來的溫度只提供參考,但透過溫度比較還是可以大略看到4C8T在32nm與45nm的溫度差異性。
耗電量測試
系統待機時 - 84W
CPU全速時 - 125W
在耗電量方面由於有32nm的製程技術,透過Power Angle測試出來的瓦數結果
尚未使用過完全一致的平台做比較,大略推算比起45nm Core i7應該會有30~50W左右的省電效能
Sandy Bridge CPU在耗電量有顯著地降低,這方面的優勢也會拉高消費者對於四核心平台的購買慾望。
參考小弟windwithme在BIOS調出來的設定方式
開機畫面
BIOS主頁
使用AMI UEFI X64 2.0架構
O.N.E調效頁面
Turbo Mode模式預設值為38、37、36、35,新一代2.0有更精細的CPU加速模式
DRAM除頻支援1066~2133等模式,讓使用者更容易發揮該有的DDR3效能
CPU Base Clock顯示10000,表示現在外頻時脈為100.00MHz
若為10560的話,則表示時脈為105.60MHz,相較以往133、166、200MHz整數顯示法不同
上方為DDR3參數設定選項,可供使用者來選擇更快的參數,能穩定使用也是主要的前題
CPU VCore Mode共有四種,SPEC Volgate/Auto/Offset Mode/Fixed Mode
CPU VCore Offset +0.010~+0.520V
CPU VCore Fixed 1.000~1.790V
DRAM Voltage 1.300~2.200V
CPU Configuration
一些CPU相關的主要功能開啟或關閉
像是常見的HT技術,或是可以選擇只想使用幾個核心等等功能
PC Health Status
可由MB內建的監控晶片來了解目前各種主要電壓或系統內溫度狀況
超頻方面這次在LGA 1155沒有需要動到太多的選項
只要著手於CPU電壓、倍頻,DRAM時脈、參數與電壓這幾樣即可
也因為外頻的範圍只在5~7%左右可以調整,建議要大幅度超頻只能選擇K系列CPU
DDR3方面使用美國品牌CORSAIR的高階DOMINATOR-GT系列,最新推出的CMT4GX3M2B2133C9
GT系列清一色使用紅黑色澤的搭配,MB與DRAM這幾年都有越來越多這種配色的路線
參數規格CL9 10-9-27 1.50V,相較於Intel建議DDR3電壓規範為1.650V來說
新款GT 2133可以只需要使用1.50V穩定,在產品規格更跨進了一步
紅黑色版本的AIRFLOW搭配DOMINATOR-GT外觀質感更佳
CPU 3502.8MHz DDR3 2135.2 CL7 10-7-27 1T 1.650V
ADIA64 Memory Read - 22132 MB/s
Sandra Memory Bandwidth - 28132 MB/s
MaXXMEM Memory-Copy - 21755 MB/s
CPU 5020.4MHz DDR3 2142 CL7 10-7-27 1T 1.650V
ADIA64 Memory Read - 25582 MB/s
Sandra Memory Bandwidth - 28475 MB/s
MaXXMEM Memory-Copy - 28139 MB/s
CPU時脈提升之後,記憶體頻寬在ADIA64與MaXXMEM這兩款軟體有明顯拉高許多
反觀Sandra Memory Bandwidth的數據並沒有因為CPU時脈而提升,這是三款Memory測試軟體不同地方。
在頻寬的數據中,可以看到同為雙通道的架構,LGA 1155比起1156平台有高上20%左右的效能
Sandy Bridge CPU雖然只使用雙通道的設計,在記憶體頻寬上已經進步到與老大哥1366互有高下
CPU 5064.4MHz DDR3 2251 CL7 10-7-27 1T 1.650V
ADIA64 Memory Read - 26605 MB/s
Sandra Memory Bandwidth - 29903 MB/s
MaXXMEM Memory-Copy - 29007 MB/s
受限於Sandy Bridge架構在CPU外頻範圍不大,依每顆CPU體質不同,普遍只有約104~108MHz的範圍可調整。
因為CPU預設外頻最高可以達到DDR3 2133的時脈,所以超頻方面也不是太大的問題
DRAM體質夠好的話,超過DDR3 2200以上不是很困難的事,對於頻寬效能也有一定的提升
超頻效能
CPU 100.4 X 50 => 5020.2MHz(關閉Turbo Boost,關閉C1E)
DDR3 2142 CL7 10-7-27 1T 1.650V
Hyper PI 32M X8=> 12m 06.384s
CPUMARK 99=> 774
Nuclearus Multi Core => 32585
Fritz Chess Benchmark => 39.29/18859
CrystalMark 2004R3 => 427585
CINEBENCH R11.5
CPU => 9.82 pts
CPU(Single Core) => 2.05 pts
PCMark Vantage => 25062
3DMark Vantage
CPU SCORE => 65044
將2600K從3.5GHz超頻到5GHz之後,在上述大多數軟體都有20~30%的效能增長
對於效能的提升有很大的幫助,全速時的電壓也只需要1.392V,可見得32nm製程讓超頻範圍也提升不少
依目前小弟在網路上許多資料得知,1.4V可以上5GHz的2600K D1並不常見,一般的空冷穩定範圍大約在4.7~4.9GHz左右。
Sandy Bridge這樣的表現還是比45nm的4C8T CPU超頻範圍落在4.1~4.3GHz的表現還好
溫度表現
系統待機時 - 43~46
CPU全速時 - 74~83
Intel Burn Test v2.4,Stress Level Maximum
超頻到5GHz除了電壓要拉高之外,溫度方面也提高不少,這方面是需要使用者拿捏的地方
如果要使用低電壓加上低溫長時間使用的話,小弟試過4.5GHz穩定也只需要1.272V
使用者到時可以依自己對於效能或是溫度方面的需求再做其他的電壓設定
耗電量測試
系統待機時 - 139W
CPU全速時 - 257W
對比預設時脈的耗電量,待機時大約高出65%,全速時高出105%,看似提高不少的耗電量
不過以此5GHz高時脈與高效能的平台來說,只有257W的全速耗電量,已經算是不太耗電的數據
5GHz就是2600K空冷穩定的最高極限了嗎?!
經windwithme花一段時間調校後發現當然不是...只要手氣夠好的話,2600K D1 5.1GHz穩定也是有機會達成的!
感覺上買CPU手氣好才是最重要的,序號批號只提供參考,小弟也有多次參考網友分享,衝去買體質好的CPU批號。
買近十顆同批號的還是無法達到網友的最佳標準...後來只能安慰自己運氣才是王道
CPU 100 X 51 => 5100.4MHz(關閉Turbo Boost,關閉C1E)
DDR3 2133.6 CL7 10-7-27 1T 1.650V
Hyper PI 32M X8=> 11m 48.382s
Fritz Chess Benchmark => 40.11/19251
對於2011年最新LGA 1155的Sandy Bridge CPU產品線
2300~2500K都是使用L3 Cache 6MB,只有2600/2600K兩款為8MB
不鎖頻的K系列為2500K與2600K兩款,其他Sandy Bridge CPU受限於外頻的範圍,可能只有5~8%的超頻範圍
1155的市場區分相當地清楚,想要超頻的消費者,建議要選購K系列的CPU產品,但是價位也高上一些
其他大多數的Sandy Bridge CPU可能較適合大多數不超頻的消費者來使用
有關於這方面的思維,讓人搞不清楚是Intel對於Sandy Bridge效能太有信心,或是想讓LGA 1156可以繼續生存下去。
原本以為在LGA 1155平台上只是看到45nm製程進化到32nm,對應到時脈、耗電量都能有效地進步
實際接觸後發現Sandy Bridge CPU也開始全都內建GPU的設計,讓使用者未來方便搭配H67等有輸出功能的晶片組。
另外Turbo Boost 2.0與更高效能的DDR3技術,都是這個新平台上的優勢
測試中是以新一代的P67晶片組,預估各廠在P67 MB剛上市時,會比自家同級的P55 MB還貴上一些
BIOSTAR TP67XE依然是走平價路線,大約市場價位會落在美金160元左右
同時支援最新的USB 3.0與SATA 3技術,加上自家研發最新的BIO-Remote2多媒體影音功能
還有THX TrueStudioPRO音效技術,在超頻方面也有很出色的表現,TP67XE不失為高C/P的P67產品
BIOSTAR也開始致力於多媒體影音技術的軟硬體支援,讓自家MB產品擁有更多的市場優勢
先前做過很多篇X58與P55的超頻教學分享,對於P67 MB未來也會有很多品牌的超頻測試與分析
本篇文章還是有對於其他項目沒有詳細提到的地方,windwithme在日後的分析報導也會依續補上
最後希望藉由新一代產品的推出,可以加速上一代的平台降價,讓消費者可以有更多超值平台的選擇,也是一個不錯的方向 :)

windwithme 發表在 痞客邦 留言(4) 人氣()



由消費者自行選擇各零件來DIY電腦的風氣在市場上已經行之有年
DIY的優勢是在於自由度高,比起品牌電腦的不同在於由單一品牌來負責後續服務與維修

windwithme 發表在 痞客邦 留言(0) 人氣()

Blog Stats
⚠️

成人內容提醒

本部落格內容僅限年滿十八歲者瀏覽。
若您未滿十八歲,請立即離開。

已滿十八歲者,亦請勿將內容提供給未成年人士。