AFDS 2012 研讨会技术专题讲座安排

Edison

[url=]BRKFST[/url]	Breakfast & Networking	Special Event
[url=]CC-4022[/url]	Characterizing and Evaluating the Memcached Key-Value Store Application on AMD Systems	Presentation
[url=]CC-4211[/url]	Accelerating The Cloud with Heterogeneous Computing	Presentation
[url=]CC-4218[/url]	AMD Cloud Workload Analysis and Tuning with SWAT	Presentation
[url=]CC-4228[/url]	Heterogeneous Computing in Healthcare Big Data	Presentation
[url=]CC-4257[/url]	GPU Acceleration of Interactive Large Scale Data Analytics Utilizing The Aparapi Framework	Presentation
[url=]CC-4265[/url]	Power Estimation for Web Server Applications on AMD CPUs	Presentation
[url=]CC-4289[/url]	Accelerate Computing with FPGAs Using OpenCL	Presentation
[url=]CC-4327[/url]	Big Data in the Era of Heterogeneous Computing	Panel
[url=]CC-4328[/url]	Virtualization & The Future Of The Hyperscale Data Center	Presentation
[url=]CC-4329[/url]	Scale or Fail: Give Your App the Speed It Needs in the Cloud!	Presentation
[url=]CC-4332[/url]	Next Generation Citrix XenServer Architecture For Scalable Clouds	Presentation
[url=]CC-4334[/url]	IOMMUv2: The Ins and Outs of the Heterogeneous GPU Use	Presentation
[url=]CC-4344[/url]	Hadoop and GPU Compute	Presentation
[url=]CE-4060[/url]	Next Generation Human Machine Interface	Presentation
[url=]CE-4161[/url]	Microsoft Surface: A Gateway to NUI	Presentation
[url=]CE-4251[/url]	Actually Building Stuff: How Small APUs Enable Big Products	Presentation
[url=]CE-4256[/url]	Running Android Apps on Windows	Presentation
[url=]CE-4263[/url]	Vision and Presence: Better Presentations Through Technology	Presentation
[url=]CE-4294[/url]	Interaction Continuum : Touch Through Gesture	Presentation
[url=]CE-4295[/url]	Gesture Based Interaction Using Heterogeneous Computing	Presentation
[url=]CE-4303[/url]	Heterogeneous Collaboration	Presentation
[url=]CE-4311[/url]	Content Creation and the Mobile Revolution	Presentation
[url=]CE-4318[/url]	Motion Interface - The Next Frontier	Presentation
[url=]CE-4324[/url]	Wireless Display - Solution and SDK	Presentation
[url=]CE-4343[/url]	Motion and Orientation Sensors Moves Up to the Next Level: Sensor Fusion and Sensor Hub Solutions Expand New Applications and User Experience	Presentation
[url=]CE-4348[/url]	Leveraging GPU to Drive Cutting Edge Augmented Reality Experiences	Presentation
[url=]CE-4564[/url]	Moving Content Around Your Home and Connected Devices	Presentation
[url=]CG-4133[/url]	GPU Accelerated Rigid Body and Soft Body Game Physics	Presentation
[url=]CG-4134[/url]	The Technology behind AMD's "Leo Demo"	Presentation
[url=]CG-4168[/url]	Global Illumination Using Ray-Bundle Tracing	Presentation
[url=]CG-4192[/url]	OpenCL Optimizations on ImageMagick: Convert, Edit, and Compose Images	Presentation
[url=]CG-4242[/url]	Game Development Optimization and Debugging using AMD's GPU PerfStudio 2	Presentation
[url=]CG-4244[/url]	Tessellation Implementation and Optimization Using OpenCL	Presentation
[url=]CG-4272[/url]	New OpenGL Extensions Made Possible by the Southern Islands GPU Family	Presentation
[url=]CG-4278[/url]	GPGPU Algorithms:  How Heterogenous Systems Architecture Can Be Leveraged to Optimize Such Algorithms in Video Games	Presentation
[url=]CG-4280[/url]	Smart 3D Content Creation: Leveraging the Cloud	Presentation
[url=]CG-4292[/url]	Compute in the Future of Gaming	Presentation
[url=]CG-4293[/url]	GPU Compute in Games, Present and Future	Presentation
[url=]CG-4304[/url]	One Size Doesn't Fit All: Building High-performance Game Content Authoring Tools with Fabric Engine	Presentation
[url=]CG-4306[/url]	The Future of PC Gaming	Panel
[url=]CG-4404[/url]	Flash: The Next Generation	Presentation
[url=]HC-3940[/url]	OpenCL+APU+GPU+Matrix Multiplication	Presentation
[url=]HC-4153[/url]	Real-time Molecular Dynamics Simulation and Visualization	Presentation
[url=]HC-4159[/url]	MSC.Nastran on AMD GPGPU Processors	Presentation
[url=]HC-4160[/url]	Octree for Particle Simulations on Heterogenous Computers	Presentation
[url=]HC-4162[/url]	Fast GEMM Implementation on Heterogeneous CPU-GPU Systems	Presentation
[url=]HC-4163[/url]	Efficient Clustering for Amorphous “Big Data” Streams	Presentation
[url=]HC-4167[/url]	Accelerating Sparse Linear Solvers on Heterogeneous Devices	Presentation
[url=]HC-4200[/url]	OpenCL and APPML with NX Nastran	Presentation
[url=]HC-4220[/url]	Curve Based Monte Carlo Financial Simulations on APUs, FPGAs and GPUs	Presentation
[url=]HC-4246[/url]	Accelerating Medical Imaging Applications Using OpenCL and APUs	Presentation
[url=]HC-4264[/url]	APPML: A High-Performance OpenCL Math Library	Presentation
[url=]HC-4279[/url]	A Proof-of-Concept C++ AMP Implementation of the BiConjugate Gradient Stabilized Method on AMD CPUs, GPUs and APUs	Presentation
[url=]HC-4345[/url]	GPGPU-based Programmable Ultrasound System Architecture	Presentation
[url=]HC-4364[/url]	The Path to Exascale - Early Experiences with Heterogenous Compute	Presentation
[url=]HC-4500[/url]	The MulticoreWare Application Development and Deployment Environment for HSA	Presentation
[url=]HC-4680[/url]	Aparapi: OpenCL GPU and Multi-Core CPU Heterogeneous Computing for Java	Presentation
[url=]HC-4682[/url]	AMD Developer Tools and Math Libraries for Heterogeneous Compute and Multicore Environments – Use Cases & Best Practices	Presentation
[url=]HC-HOL01[/url]	Self-Paced Hands On Lab	Hands On Lab
[url=]KEYTHU[/url]	Keynote: Thursday	Keynote
[url=]KEYTUE[/url]	Keynote: Tuesday	Keynote
[url=]KEYWED[/url]	Keynote: Wednesday	Keynote
[url=]LUNCH1[/url]	Lunch & Networking	Special Event
[url=]LUNCH2[/url]	Lunch and Departures	Special Event
[url=]MM-4101[/url]	Video Transcoding in a Heterogeneous System	Presentation
[url=]MM-4170[/url]	OpenCL Enabled Face Detection Plug-in for IrfanView	Presentation
[url=]MM-4172[/url]	Optimizing VLC Using OpenCL and Other Fusion Capabilities	Presentation
[url=]MM-4196[/url]	Building Real-Time Professional Visualization Solutions with OpenCL	Presentation
[url=]MM-4203[/url]	OpenCL Acceleration of x264	Presentation
[url=]MM-4234[/url]	Optimizing Media Creation with Heterogenous Computing Architecture	Presentation
[url=]MM-4241[/url]	Accelerated Vision and Image Processing Using CUVILib	Presentation
[url=]MM-4248[/url]	Feedback from the Trenches	Presentation
[url=]MM-4268[/url]	VPL Overview: Performant and Extensible OpenCL-based Video Processing Framework	Presentation
[url=]MM-4281[/url]	GPU Accelerated Face Recognition in Photo and Video	Presentation
[url=]MM-4308[/url]	Accelerating SURF on HSA	Presentation
[url=]MM-4310[/url]	How a Joint Effort Between Software and Hardware Solutions Drives to UX Excellence	Presentation
[url=]MM-4346[/url]	Beyond Channel-Based Audio	Presentation
[url=]MM-4403[/url]	Accelerate a Fully Functional Photo Editing Software with GPU	Presentation
[url=]MM-4523[/url]	Media Players in the Heterogeneous Computing Era: Baofeng’s Vision and Actions	Presentation
[url=]PG-3680[/url]	AMD Fusion and Holographic Rendering	Presentation
[url=]PG-4000[/url]	Solid Modeling on the Fusion	Presentation
[url=]PG-4149[/url]	Implementing OpenCL Support in GEGL and GIMP	Presentation
[url=]PG-4190[/url]	Biomedical Data Analysis on Heterogeneous Platforms	Presentation
[url=]PG-4191[/url]	GPU Enablement in Adobe Photoshop	Presentation
[url=]PG-4202[/url]	EDEM Software Increases Simulation and Visualization Performance Using AMD Technology	Presentation
[url=]PG-4206[/url]	Using Fusion Shared Memory for Huge 3D Model and Highly Accurate Rendering	Presentation
[url=]PG-4210[/url]	Scalable Mixed Heterogeneous Clusters for the Real-Time High-Fidelity Driving Simulator System	Presentation
[url=]PG-4213[/url]	Assessing the Relevance of APU for High Performance Scientific Computing	Presentation
[url=]PG-4230[/url]	On Heterogeneous Parallel Computations: GPU vs APU	Presentation
[url=]PG-4253[/url]	Direct OpenCL Graphics Rendering	Presentation
[url=]PG-4266[/url]	Heterogeneous Kd-tree Construction on an APU	Presentation
[url=]PG-4271[/url]	Advanced Flow Control in GCN	Presentation
[url=]PG-4326[/url]	Using DirectGMA for Peer to Peer Data Transfers	Presentation
[url=]PG-4338[/url]	Blender Cycles and Tiles. Enhancing User Experience	Presentation
[url=]PG-4660[/url]	GPU Accelerated Video Editing	Presentation
[url=]PL-3660[/url]	OpenCL C++	Presentation
[url=]PL-4040[/url]	C++ Memory Model and Processor Architectures	Presentation
[url=]PL-4130[/url]	Can GPGPU Programming be Liberated from the Data-Parallel Bottleneck	Presentation
[url=]PL-4151[/url]	PTask: Operating System Abstractions To Manage GPUs as Compute Devices	Presentation
[url=]PL-4152[/url]	Chapel: Multiresolution Parallel Programming	Presentation
[url=]PL-4156[/url]	GMAC-2: Easy and Efficient Programming for OpenCL	Presentation
[url=]PL-4158[/url]	Run Fast Anywhere: Managed Platform GPGPU with OpenCL	Presentation
[url=]PL-4174[/url]	Dataflow and Transactional Memory to Simplify Programming Heterogeneous Devices	Presentation
[url=]PL-4175[/url]	Performance Portability in the OP2 Unstructured Mesh Library	Presentation
[url=]PL-4189[/url]	OpenVL: A High-Level Developer Interface to Computer Vision	Presentation
[url=]PL-4201[/url]	Realities of Write-Once, Run-Anywhere Kernels: A Case Study	Presentation
[url=]PL-4212[/url]	McVM: A Just-in-Time Compiler for MATLAB Language for Heterogeneous Systems	Presentation
[url=]PL-4238[/url]	Malloc in OpenCL Kernels: Why and How?	Presentation
[url=]PL-4245[/url]	Bolt: A C++ Template Library for HSA	Presentation
[url=]PL-4297[/url]	An Overview of HSAIL	Presentation
[url=]PL-4312[/url]	Accelerating OpenCV on AMD GPUs with OpenCL	Presentation
[url=]PT-3601[/url]	Harnessing GPU Compute with C++ AMP: Part 1	Presentation
[url=]PT-3602[/url]	Harnessing GPU Compute with C++ AMP: Part 2	Presentation
[url=]PT-3900[/url]	OpenCL Meta-Programming for BLAS	Presentation
[url=]PT-3980[/url]	SimpleCL - Rapid OpenCL Development Tool	Presentation
[url=]PT-4169[/url]	Task Manager (TM) - A Parallel Building Library for Heterogeneous Computing	Presentation
[url=]PT-4173[/url]	Parallel Path Analyzer (PPA) - A Visualized System Dependency Analyzer for Heterogeneous Computing	Presentation
[url=]PT-4229[/url]	Quickly Optimize OpenCL™ Applications with SlotMaximizer	Presentation
[url=]PT-4232[/url]	MXPA – The Multicore Cross-Platform Architecture for Performance-Portable Computing	Presentation
[url=]PT-4233[/url]	CU2CL: An Automated CUDA-to-OpenCL Source-to-Source Translator	Presentation
[url=]PT-4249[/url]	Practical Implementations of OpenCL ALTELL Security Library	Presentation
[url=]PT-4255[/url]	clMAGMA: Heterogeneous High-Performance Linear Algebra with OpenCL	Presentation
[url=]PT-4291[/url]	Productive OpenCL GPU Libraries	Presentation
[url=]PT-4330[/url]	Introduction to AMD Developer Tools	Presentation
[url=]PT-4331[/url]	Advanced OpenCL™ Debugging and Profiling Using AMD Tools	Presentation
[url=]PT-4365[/url]	Using SWARM for Load Balancing on Heterogeneous Hardware	Panel
[url=]PT-4380[/url]	Programming Heterogeneous Many-cores Using Directives	Presentation
[url=]SE-3840[/url]	Harnessing GPUs for Password Recovery and Computer Forensics	Presentation
[url=]SE-4166[/url]	Securing the Cloud	Presentation
[url=]SE-4171[/url]	Crypto++ Optimizations - Using Heterogeneous Computing to Improve Encryption / Decryption Performance	Presentation
[url=]SE-4176[/url]	Using GPUs to Accelerate and Isolate Runtime Monitors	Presentation
[url=]SE-4226[/url]	The Impact of Heterogeneous Computing Solutions On Cloud Computing, Consumerization of IT, and Persistent Security Threats.	Presentation
[url=]SE-4239[/url]	Big Data Protection & Compression on GPUs	Presentation
[url=]SE-4262[/url]	Leverage AMD Platforms for Cryptographic Primitives	Presentation
[url=]SE-4333[/url]	Trusted Execution Environment: The New Solution for Higher and More Flexible Security	Presentation
[url=]WELCOME[/url]	Welcome Reception	Special Event
[url=]WT-4236[/url]	Performance Evaluation of AMD-APARAPI Using Real World Applications	Presentation
[url=]WT-4273[/url]	GPU Acceleration in Chrome	Presentation
[url=]WT-4402[/url]	New Browser Technology and Its Impact on the China Market	Presentation
[url=]WT-4540[/url]	Fabric Engine: High-Performance Computing for Dynamic Languages	Presentation

sourpotato

不错。mark一记

mooncocoon

开始说C++了，这是好事。

gtx5

有没有GCN2.0

BDFMK2

现在的Fusion还是不太给力啊，不知道完全融和版会有什么惊喜

骨刺

BDFMK2 发表于 2012-4-20 22:21
现在的Fusion还是不太给力啊，不知道完全融和版会有什么惊喜

高温高热的火炉

mooncocoon

说到融合，在4月初的HD7970“技术说明会”上，AMD称“HSA的最终目的是让CPU能够使用GPU中丰富的浮点运算资源，并接管GPU的线程和任务管理工作，以便令其能够更好更专注的完成浮点运算和并行吞吐任务。”
“为了达到目的，我们会让GPU对CPU开放并改进它的设计，让它能够更好的跟CPU结合在一起并各司其职。”
然后，它们强化了HD7970的ACE，并在说出刚刚那句话之后不到1分钟的时间里又说出了“就算是融合在一起，GPU也同样需要有强大的线程管理能力”这句话。

所以……所以我不知道未来的Fusion会是个什么样子了，你们谁能理解这不到一分钟前后所发生的事情的话，请给我答案……

BDFMK2

mooncocoon 发表于 2012-4-20 23:01
说到融合，在4月初的HD7970“技术说明会”上，AMD称“HSA的最终目的是让CPU能够使用GPU中丰富的浮点运算资源 ...

“HSA的最终目的是让CPU能够使用GPU中丰富的浮点运算资源，并接管GPU的线程和任务管理工作，以便令其能够更好更专注的完成浮点运算和并行吞吐任务。”

“就算是融合在一起，GPU也同样需要有强大的线程管理能力”

我觉得这两句话不冲突呀。莫非想让shader ALU取代SFU的工作？流水线长度和延迟看似不可能！如果线程调度器从ACE变成X86，在shader ALU规模不变的情况下，性能和灵活性能提高多少呢？

rSKip

我觉得AMD要实现的Fusion是“CPU/GPU - API - 程序”关键就是完美地架设好这层API 物理上两者还是分离的
CPU跑高串行度的线程/进程 GPU跑高并行度的 对程序员来说是有一些核心可用 CPU和GPU区别不是特别大 芯片间任务分配、同步等靠API来打理 不用人工雕琢。

mooncocoon

“HSA的最终目的是让XXXXX，并接管GPU的线程和任务管理工作”
“就算是融合在一起，GPU也同样需要有强大的线程管理能力”
CPU接管之后GPU也同样需要强大的XXX，那要CPU接管做什么？

这是很现实的问题——如果要以前者为发展方向，一级仲裁机制对GPU来说就将逐渐变得多余，GPU只需要专注于下级的线程派放能力即可，这可以腾出更多的晶体管资源来强化吞吐以及缓冲，对性能功耗比有极大的促进。而后者的发展方向则完全相反，ACE必须进一步得到强化，并且一步步的将更多过去通过CPU来实现的仲裁过程以硬件的形式固化进GPU，就像Fermi那个德行一样，这是对GPU灵活性的提升，同时也是挑战原教旨GPGPU的一种做法。

这不是单纯的快慢的问题，这是一个制定计划的连贯性以及执行性的问题。一边是计划着CPU要接手，一边却又是GPU跃跃欲试并且不让接手，两个南辕北辙的东西同时出现在当前AMD的发展路线上，并且能够共存在一分钟不到的同段描述里，难道这是一个日式的“第一，第一，第一以及第一”的计划列表么

NORAWITHMYCALL

AMD现在的情况就是造桥中，2边都在往中间靠拢，就是缺少最后一块最关键的箱梁。