![]() |
||
Elite Member
![]() ![]() ![]() ![]() ![]() 加入日期: Mar 2002 您的住址: 台北市
文章: 4,505
|
引用:
有點像是NPC的升級模組吧 右下角被模糊的那個,是說sample在Feb 04出來的嗎? 然後量產在Apirl 04?
__________________
收到…那我就不去上課了…… GeForce3新舊Driver測試(有差嗎?) 9800SE軟改簡單教學(Catalyst+RivaTuner)。私人收藏 Magnifier.tw 2003-10-26撼訊9800XT Preview2003-11-039700Pro/9800Pro/9800XT大集合! 2004-05-28ATi強力接棒者--Radeon X800Pro !2004-06-30DX9.0c帶給了GeForceFX什麼影響?(Far Cry) 2005-01-15X800XL現身─ATI中/高階新戰將!2005-02-17剪不斷、理還亂─『CPU&顯示卡』前世今生徹底解析! 2005-06-30[G70系列之1]7800GTX真的非『最高階』CPU不可嗎? more...... |
||||||||
![]() |
![]() |
New Member
加入日期: Apr 2004
文章: 2
|
The reason NVIDIA sucks is not because of their video card, it is what they did after. They discredited 3Dmark2003 for being honest benchmark. 3Dmark2003 revealed NV3x performed miserably in DX9. NVIDIA called them a liar and purposely inflated their 3Dmark scores. Publicly, they vigorously promoted how FP32 is light years ahead of FP24, how their hardware is the one and only IEEE 754 complaint video card, how useless synthetically benchmarks like 3Dmark2003 are, because they do not represent real performances NV3X posses under REAL DX9 games. “You won't be able to experience the full power of our NV3x hardware until the DX9 games ship” Sadly, NV fans agreed them. Making all sorts of excuses for them, the most absurd of which is Microsoft and ATI scammed them by leaving key details out. I guess ATI got scammed by Microsoft and NVIDIA upon the release of DX7 and DX8 as well. So when 3Dmark scores are low, it is because ATI paid futuremark and 3Dmark2003 is a dishonest useless benchmark. When the scores match that of ATI's, it is the true performance of NV3x series. Double standards, anyone? Every DX9 or partial DX9 game shipped to date performed better on R3x0, sometimes by a close 50%, and yet the facts above doesn't strike you a little bit odd? Why is it 5950 could score 6500 marks just like 9800 XT, but I can play the game at 1024x768 with all details high and the 5950 could only manage at 800x600? The fact that 3Dmark2003 revealed that 2 years back?
|
||
![]() |
![]() |
*停權中*
加入日期: Jun 2002 您的住址: 天天看WAFER被切
文章: 197
|
引用:
已弟工作的地方來說 封裝過5600 5700 以及5900 當然良率的確很糟糕 以5600來說 大約只有一半 49X顆DIE裡面只有約22X的GOOD DIE 5700來說 大約剩1/3 5900不用講了 上次碰到一片最離譜的 一片WAFER上496顆DIE裡面只有49顆GOOD DIE 大致來說5900的良率大約只有在1/6~1/8之間而已 |
|
![]() |
![]() |
Senior Member
![]() ![]() ![]() 加入日期: Nov 2002
文章: 1,294
|
引用:
1/6-1/8的良率....真的是太糟糕了 假如平均一片wafer只有70顆能用,一片代工兩千美金 單顆未封測的die就要30美金了 |
|
![]() |
![]() |
*停權中*
加入日期: Feb 2003 您的住址: 台灣
文章: 4,036
|
回覆: 發燒友們!我們的新玩具來了!!GF 6800定價499美金
引用:
題外話 為何一堆人陷於PCI-E跟AGP8X的陷阱裡?? NV40用AGP8X代表什麼??這代表頻寬根本還沒餵飽 那花更多的錢用 NV40核心的PCI-E顯卡根本就沒必要 完全不用可惜 難道大家忘了Ti4200-4x跟Ti4200-8X的故事了嗎?? 大家一窩蜂去期待PCI-E 保證受傷害 |
|
![]() |
![]() |
*停權中*
加入日期: Sep 2003
文章: 212
|
引用:
小弟有個疑問... Bad DIE 最後用途是??? (當垃圾丟,還是回收再利用或者是....) |
|
![]() |
![]() |
Basic Member
加入日期: Jan 2004 您的住址: 台北
文章: 15
|
個人還是認為R420跟NV40都是破2XXM晶體管的顯示晶片,這次ATI如果重演X800 XT--16 pipes, X800 Pro--12 pipes,X800 SE--- 8 pipes,這樣低獲利的方式肯定的是未來幾季的財報不怎麼好看,雖然我們都很喜歡ATI這樣做
X800 Pro改卡變X800 XT -----Dream On
__________________
![]() |
![]() |
![]() |
Basic Member
加入日期: Jan 2004 您的住址: 台北
文章: 15
|
引用:
ATI的做法是因為在設計就有類似fuse的設計,部分功能不能工作的管線可以關閉,如R9500/R9800SE或是將來的X800系列個產品線(用管線數分級),到現在還是懷疑R420與NV40的管線設計是傳統的還是像NV35那樣有彈性
__________________
![]() |
|
![]() |
![]() |
Advance Member
![]() ![]() 加入日期: Jan 2002
文章: 437
|
回覆: 回覆: 發燒友們!我們的新玩具來了!!GF 6800定價499美金
引用:
NV40用AGP8X只能代表他打算鞏固舊有AGP市場 不代表他用不到PCI-E 你的硬碟也不曾跑超過66MB/s吧 但是你還在用DMA66嗎? 除非他用橋接晶片接PCI-E,不然效能多少都會有點差異 只是這個差異的大小還是未知數而已 |
|
![]() |
![]() |
*停權中*
加入日期: Apr 2001 您的住址: 香港
文章: 1,441
|
希望NV45有AGP8X的版本
![]() |
![]() |
![]() |