Jacket

Jacket

Jacket

Jacket is a software platform designed for engineers, scientists, and analysts who want maximum application performance, minimum programming difficulty, and maximum leverage of available technical computing resources, including laptops, desktops, servers, clusters and the Cloud. The Jacket platform consists of a language processing system and runtime that automatically optimizes existing applications or new algorithms for GPU computing.

Jacket currently supports the MATLAB language as a frontend to the platform. MATLAB is the technical computing language of choice across a broad range of industries with over 1 million users. Jacket's language processing system automatically translates MATLAB code to high performance primitives required for best utilization of GPUs. Working in concert with the language processing system, Jacket's runtime system optimizes memory transfers, compiles code on-the-fly for realtime tuned performance, and launches GPU kernels efficiently for maximal performance. All GPU-specific programming details are handled by Jacket, freeing the user to focus on science, engineering, and analytics.

Jacket是一個軟件平台，為希望最大程度地提高應用性能，最低的編程難度，最大限度地利用現有的技術計算資源，包括筆記本，台式機，服務器，集群和雲設計的工程師，科學家和分析師們。Jacket平台由語言處理系統和運行時，會自動優化現有的應用程序或用於 GPU計算的新算法。

目前，Jacket支持MATLAB語言作為前端平台。MATLAB是技術計算語言，是跨越行業的選擇，而擁有超過100萬用戶。Jacket的語言處理系統自動轉換MATLAB代碼，使GPU达到最佳的高性能。Jacket的運行時系統協同的工作與語言處理系統，優化內存傳輸，實時調整性能飛編譯的代碼，並有效地啟動最大性能的GPU內核。所有GPU特定的編程細節處理由Jacket处理，使用戶集中在科學，工程和分析。

Jacket introduces new data types to MATLAB which let the user move data and computations to the GPU. Changing the data type allows the user to tap into the GPUs tremendous computational power. All standard data types are supported, including real and complex versions of singles, doubles, and integars. CUDA-capable GPUs of all types are supported. For instance, NVIDIA's Fermi-based Tesla C2050 contains 448 cores that are now fully accessible with Jacket, but those with lesser budget can also run Jacket on GeForce and Quadro GPUs!

Jacket引入了新的數據類型到MATLAB，讓用戶移動數據和計算到GPU。更改數據類型允許用戶使用GPU巨大的計算能力。Jacket支持所有的標準數據類型，包括實數和複數版本的單打，雙打和整數。Jacket也支持所有支持CUDA的GPU的類型。例如，NVIDIA的Fermi - 基於Tesla C2050包含448核心現在完全Jacket合宜的，但那些較小的預算也可以在GeForce和Quadro GPU運行Jacket！