賓尼._.

其待超頻能力

ccw

引用:

原帖由 dom 於 2011-9-7 19:41 發表
http://www.xbitlabs.com/news/cpu ... tml#discussion_last

OH HOLY SHXX .......
Things become UGLY ..... ...

This is such a holy......
Very wise of me to give it up so early.

dom

引用:

原帖由 ccw 於 2011/9/7 19:49 發表

This is such a holy......
Very wise of me to give it up so early.

Well have fun with your new rig

ccw

引用:

原帖由 dom 於 2011-9-7 22:11 發表


Well have fun with your new rig

Still waiting for my real rig
Damn you Tyan, build C202 and C204 and now a rubbish Q67 but not a C206!
I really think that Intel or someone is trying to hinder the companies from building a usable C206 board for E3-12x5 CPUs....

Tommi_Vercetti

引用:

原帖由 賓尼._. 於 2011-9-7 19:48 發表
其待超頻能力

只留意其外頻設定能否回復正常做法，唔好下下走去Cherry Pick倍頻先有得超

Henry

引用:

原帖由 Tommi_Vercetti 於 2011-9-8 00:03 發表


只留意其外頻設定能否回復正常做法，唔好下下走去Cherry Pick倍頻先有得超

Multi+Bus.

dom

http://www.overclock.net/tankguy ... pre-order-info.html
Looks like Bulldozer is not competitive ...........
Cheaper than 2600K = FX FAILED to Beat Sandy Bridge

Good luck with AMD .......facing Ivy Bridge in 2012 will repeat K8 Vs Core 2 Duo .....

Tommi_Vercetti

引用:

原帖由 Henry 於 2011-9-8 00:35 發表

Multi+Bus.

問題係外頻組合選擇太小，而外頻又超唔起，等於又要cherrypick倍頻

Henry

引用:

原帖由 Tommi_Vercetti 於 2011-9-8 03:09 發表


問題係外頻組合選擇太小，而外頻又超唔起，等於又要cherrypick倍頻

1: 外頻有得1MHz上
2: RAM基本上有1:6-8(9-10有可能,未知)

Bus 100MHz,對RAM除頻係行133MHz (1.33x)
假設RAM行DDR3 2000(1000MHz)
1:6 -> 1:7 = 166MHz - 142MHz = 24MHz (Bus 18MHz)
1:7 -> 1:8 = 142MHz - 125MHz = 17MHz (Bus 13MHz)
100上125(25%),行DMI應該上到啦掛

RAM行DDR3 2400(1200MHz)
1:8 = 150MHz (Bus 120MHz)
Bus先上20%,比DDR3 2000行1:6仲要鬆動.
1:6 = 200MHz (Bus 150MHz)
如果Bus可以上到50%,RAM既比例一樣能夠配合.

假設CPU行4G, DDR3 2000同DDR3 2400個倍頻差:
@DDR3 2400 (1:8): 4000/125 = 32x
@DDR3 2000 (1:6): 4000/120 = 33x(34x),先差1-2倍.

適當利用FSB極限同RAM除頻係關鍵,況且1155個Bus得果少少調都係咁超RAM.
假設CPU行5GHz,行100MHz就50x,125MHz就40x,150MHz就33x(34x).
上25MHz已經有10x既差距,應該能夠彌補上唔到倍頻既問題.

選擇太小: 而家仲有得果100/125/133/150/166幾種配搭?咩年代?

[ 本帖最後由 Henry 於 2011-9-8 04:04 編輯 ]

qcmadness

引用:

原帖由 Henry 於 2011-9-8 04:02 發表

1: 外頻有得1MHz上
2: RAM基本上有1:6-8(9-10有可能,未知)

Bus 100MHz,對RAM除頻係行133MHz (1.33x)
假設RAM行DDR3 2000(1000MHz)
1:6 -> 1:7 = 166MHz - 142MHz = 24MHz (Bus 18MHz)
1:7 -> 1:8 = 142MHz - 125MH ...

最主要唔係外頻係咪有得1MHz咁上
而係base clock個clock係咪同pci-e tie埋一齊

tie埋一齊就110MHz到頂

qcmadness

可惜, base clock tie pci-e clokc係大勢所趨

intel / amd明年ship的platform, 80%以上 (Socket 1155 / Socket FM1 / FM2) 都會係default base clock tie with pci-e clock

dom

http://www.amd.com/us/press-rele ... ssors-2011sep7.aspx
http://blogs.amd.com/work/2011/09/07/the-start-of-a-new-era/
Well Bulldozer is Shipping (Server)

[ 本帖最後由 dom 於 2011-9-8 09:57 編輯 ]

Henry

引用:

原帖由 qcmadness 於 2011-9-8 08:41 發表

最主要唔係外頻係咪有得1MHz咁上
而係base clock個clock係咪同pci-e tie埋一齊

tie埋一齊就110MHz到頂

X79應該未係喎.
如果係既話,好似1155咁用內置clockgen就得啦,省時省力,反正都超唔到.

qcmadness

引用:

原帖由 Henry 於 2011-9-8 14:14 發表

X79應該未係喎.
如果係既話,好似1155咁用內置clockgen就得啦,省時省力,反正都超唔到.

x79都係PCH, 你話呢

Henry

引用:

原帖由 qcmadness 於 2011-9-8 18:05 發表

x79都係PCH, 你話呢

睇定D先,板都未出,BIOS都仲係有Bug咁樣.
Bus上到20-25%基本上已經有足夠超RAM空間之餘,亦能夠抵銷倍頻限制.
如果Bus對PCI-E都有得除頻基本上無障礙.