NVIDIA GT300規格終曝光 流處理器、顯存驚人

火野亂太郎

終於，NVIDIA DX11圖形芯片GT300的規格參數浮出水面了，不出所料地強大，很多地方都超出了預期。

GT300，或稱G300，內部代號Fermi(費米)，按照歷史傳統又稱NV70，最高端型號GF100——不管叫什麼，規格如下：

－ 製造工藝：台積電40nm
－ 晶體管：30億個
－ 流處理器：512個(32×16)
－ 一級緩存：1MB
－ 二級統一緩存：768KB
－ 顯存位寬：384-bit
－ 顯存容量：1.5/3/6GB GDDR5
－ 浮點精度：半速IEEE754雙精度

GT200擁有240個流處理器，而GT300一舉增加到了512個，並分成16組SM(Streaming Multiprocessor)，每組32個流處理器。

GT300集成了六個64-bit顯存控制器，合計規格384-bit，低於GT200 512-bit，但搭配速度更快的GDDR5顯存，容量上消費級桌面卡GeForce (380?)主要是1.5GB，專業卡Quadro和Tesla上則會配備驚人的3/6GB GDDR5。

GT300還改變了GPU功能執行方式，單精度模式下每個時鐘循環均可執行512個積和熔加運算(Fused Multiply-Add/FMA)，雙精度下則是256個。

IEEE浮點算法格式方面，NVIDIA過去支持的是IEEE754-1985，現在GT300則升級到了最新的IEEE754-2008。雖然具體的浮點運算能力還不清楚，但非常有希望擺脫近年來始終不如對手的尷尬。

除此之外，GT300架構原生支持C(CUDA)、C++、DirectCompute、DirectX 11、Fortran、OpenCL、OpenGL 3.1/3.2等一系列技術標準，尤其是C++，這是GPU歷史上第一次能夠順利運行這種代碼，再加上C、Fortran就可以看出NVIDIA在通用計算方面的雄心壯志了。

NVIDIA CEO黃仁勳將在GPU技術大會上首次公開展示GT300顯卡原型，時間是太平洋時間30日下午一點，北京時間1日凌晨五點。讓我們拭目以待。



GF100規格





http://www.nvidia.com/page/home.html

Fermi是為GPGPU而生，其規格大家都知道了不再重複，只挑選一些重要特性！

• 512 Cuda core，雙精度提升最高達8x，各方面性能全面提升；
• 並行數據緩存，業界首款支持緩存分級的GPU
• 全新Giga Thread調度系統，改進的並行數據管理效率，雙向數據傳輸，智能管理上萬條進程
• 首款支持ECC內存的GPU

The Next Generation CUDA Architecture, Code Named Fermi
The Soul of a Supercomputer in the Body of a GPU

512 CUDA Cores               

• Optimized performance and accuracy with up to 8X faster double precision
• Versatile accelerators for a wide variety of applications

NVIDIA Parallel DataCache               

• First GPU architecture to support a true cache hierarchy in combination with on-chip shared memory
• Improves bandwidth and reduces latency through L1 cache』s configurable shared memory
• Fast, coherent data sharing across the GPU through unified L2 cache

NVIDIA GigaThread™ Engine               

• Increased efficiency with concurrent kernel execution
• Dedicated, bi-directional data transfer engines
• Intelligently manage tens of thousands of threads

              
ECC Support               

• First GPU architecture to support ECC
• Detects and corrects errors before system is affected
• Protects register files, shared memories, L1 and L2 cache, and DRAM

另外，NV為並行計算還推出了基於MS virsual studio 工作環境的Nexus套件。

NVIDIA Introduces Nexus, The Industry』s First Integrated GPU/CPU Environment For Developers Working With Microsoft Visual Studio

SANTA CLARA, Calif. —Sept 30, 2009—NVIDIA Corp. today introduced NVIDIAR Nexus, the industry』s first development environment for massively parallel computing that is integrated into Microsoft Visual Studio, the world』s most popular development environment for Windows-based solutions and Web applications and services.

「NVIDIA Nexus is going to improve programmer productivity immediately,」 said Tarek El Dokor at Edge 3 Technologies. 「An integrated GPU and CPU development solution is something Edge 3 has needed for a long time. The fact that it』s integrated into the Visual Studio development environment drastically reduces the learning curve.」

NVIDIA Nexus radically improves productivity by enabling developers of GPU computing applications to use the popular Microsoft Visual Studio-based tools and workflow in a transparent manner, without having to create a separate version of the application that incorporates diagnostic software calls. NVIDIA Nexus also includes the ability to run the code remotely on a different computer. Nexus includes advanced tools for simultaneously analyzing efficiency, performance, and speed of both the graphics processing unit (GPU) and central processing unit (CPU) to give developers immediate insight into how co-processing affects their applications.

Nexus is composed of three components:

    * The Nexus Debugger is a source code debugger for GPU source code, such as CUDA C, HLSL and DirectCompute. It supports source breakpoints, data breakpoints and direct GPU memory inspection. All debugging is performed directly on the hardware.
    * The Nexus Analyzer is a system-wide performance tool for viewing GPU events (kernels, API calls, memory transfers) and CPU events (core allocation, threads and process events and waits)—all on a single, correlated timeline.
    * The Nexus Graphics Inspector provides developers the ability to debug and profile frames rendered using APIs such as Direct3D. Developers can use the Graphics Inspector™ to scrub through draw calls, look at any textures, vertex buffers, and API state in the entire frame.

The NVIDIA Nexus supports Windows 7 and Windows Vista operating systems and full integration within Visual Studio (2008 SP1 standard edition or later).

A BETA version of NVIDIA Nexus is scheduled to be available on Oct. 15. For more information on NVIDIA Nexus or to register as a developer, please visit: www.nvidia.com/nexus. Developers can register for the BETA program in person at the GPU Tech Conference, being held this week in San Jose, Calif. Both standard and professional versions of NVIDIA Nexus will be available upon final release.

老黃很酷呦

諜照嘍 誰說沒實物卡 純得瑟？？？看老黃的表情


























GF100 (Saw-zall GTX)
40nm DX11 Cuda3
~590mm^2 ~3.2 billion transistors
24.5 x 24.5mm2 die
1536mb .4ns samsung/hynix 5gbps
~233gb/sec bandwidth
700c / 1750s / 1250m
512 MIMD / 128 tmu / 64 rop
195watt TDP
launch nov 25th, major retail christmas/jan.
$549 - $599

[ 本帖最後由 火野亂太郎 於 09-10-1 02:37 PM 編輯 ]

天地無雙

haha 老黃

雲一0一

散熱個樣唔得wo
都唔知幾時先推出市場..

美人魚暗之惡魔

原帖由 雲一0一 於 09-10-1 04:29 PM 發表
散熱個樣唔得wo
都唔知幾時先推出市場..

應該都要年底或明年年頭
點解個價錢都有埋的=.=
我係第度睇冇價錢

monstering

原帖由 雲一0一 於 09-10-1 04:29 PM 發表
散熱個樣唔得wo
都唔知幾時先推出市場..

launch nov 25th, major retail christmas/jan.
$549 - $599

牙峻＊

勁貴..- -..

♀坤♂

下年先出掛

Ｘ～鏗

1個8PIN @@?

美人魚暗之惡魔

原帖由 Ｘ～鏗 於 09-10-1 10:14 PM 發表
1個8PIN @@?

聽講係8+6
但2個插頭的位置都唔同
一個上面一個側面

痛心

ati 應該悲哀
岩岩單卡 張5890 打win 左285
佢轉頭又出gt3系=.=

火野亂太郎

原帖由 痛心 於 09-10-2 08:06 AM 發表
ati 應該悲哀
岩岩單卡 張5890 打win 左285
佢轉頭又出gt3系=.=

咩5890打win285-_-"??你係咪打錯字-_-"

美人魚暗之惡魔

原帖由 火野亂太郎 於 09-10-2 04:41 PM 發表

咩5890打win285-_-"??你係咪打錯字-_-"

佢應該想話5850
5890應該係打295

★╮鬼儿

個價0k啵..

大鐵

3千尾~4千頭都可以入得貨

火野亂太郎

本來諗住等5870x2 但係而家n仔又出d咁既野...睇黎又排有排忍-_-

痛心

大家冇睇過評測?
5890 係效能上 點夠gtx 295

同埋GTX295 係雙卡
點同 5890 單卡果D打