要酸就酸吧
今天的GTX680對上7970
不就是6950對上560ti

同樣是同價位 前者比後者低耗、高能

消費者只用選最符合自己的產品
而不是馬上就酸另一個陣營 :rolleyes:

甚至是還有把500美元當成中階產品
還不如等nvidia把200美元的產品拿出來再說吧
主要的消費群關心的是這一塊

有純CUDA測試程式嗎??

呃!那個多3倍的Unit並不是以前純CUDA單元哦!
這樣您有稍稍釋懷嗎?:)
如果Nv放生0penCL改變CUDA架構是有機會,不過目前還沒有動靜。
至於那些轉檔程式老實說無法發揮CUDA的完全效能,和純科學運算差距很大
在轉檔上的效能進步來自軟体架構改進會多一些…

WHQL-Certified GTX 680 Drivers Limit Sandy Bridge-E To PCI-E 2.0 Speeds

WHQL-Certified GTX 680 Drivers Limit Sandy Bridge-E To PCI-E 2.0 Speeds
Posted 03/23/2012 at 10:34am | by Brad Chacos

If you've dropped the dough on a spiffy new enthusiast-level Sandy Bridge-E processor, you may decide to drop a corresponding level of dough on a spiffy new enthusiast-level Nvidia HTX 680 graphics card. (If so, we salute your Maximum-ness.) There's just one little caveat you should be aware of, however; Nvidia's initial WHQL drivers for the GTX 680 nerfs data transfer levels to slower PCI-E 2.0 speeds, rather than the blazing fast enthusiast-level PCI-E 3.0 x16 speeds the X79 chipset is capable of.

Hopefully it's just a short-term bump in the road due to software validation issues rather than a long-term hassle. Nvidia owns up to the slowed speeds on its specification page for the graphics card:

GeForce GTX 680 supports PCI Express 3.0. The Intel X79/SNB-E PCI Express 2.0 platform is only currently supported up to 5GT/s (PCIE 2.0) bus speeds even though some motherboard manufacturers have enabled higher 8GT/s speeds.

There is some good news, however: TechPowerUp reports that the pre-launch drivers shipped on the disc that comes with the GTX 680 enable PCI-E 3.0 on compatible Sandy Bridge-E systems. Basically, if you're a SB-E owner who feels the need for that extra data transfer speed, you can stick with the included pre-launch drivers for now and hold off on any new drivers updates until Nvidia comes up with a fix.

我知道不同世代的CUDA數量不能對等比較

我的意思是，CUDA數量已經增加為3倍
"假如"實際上還無法勝過前代僅有1/3數量的GTX580表現(假設來說)
這樣不代表CUDA架構有大改嗎？
(對照從G80以來的NV 1D Scalar SP設計來說，是否在SP規模上有所縮解？)

另外，轉檔輸出的比較
純粹是指同一段影片，在9800GTX、GTX280、GTX460上面開啟CUDA加速
輸出所需的時間比較(例如以TMPGEnc最新版來說)
雖然說轉檔程式可能無法發揮CUDA的全部能力，但只改變顯示卡這個變因
應該可以比較出三張卡在某些限制情況下的CUDA效能吧？

你買了嗎?
等你的外接開箱文

你買了嗎?
等你的外接開箱文


:D :D :D :D :D

No...我在等非公版...公版趕快出清吧...
:laugh: :D

你要買OC版?
我要等SOC與DCII之類的...

三風扇的OC就行了...