Jon Peddie Research Professional Graphics CGW Webinar.

Slides:



Advertisements
Similar presentations
Computer Graphics Prof. Muhammad Saeed. 2 Hardware ( Graphic Cards ) II Hardware II Computer Graphics 1 August 2012.
Advertisements

© 1998, Progress Software Corporation 1 Migration of a 4GL and Relational Database to Unicode Tex Texin International Product Manager.
NexSentry Imaging Solutions
February 2010 CELSIUS H700 Workstation power on-the-go Copyright 2010 FUJITSU TECHNOLOGY SOLUTIONS.
Copyright © 2006 by The McGraw-Hill Companies, Inc. All rights reserved. McGraw-Hill Technology Education Introduction to Computer Administration Introduction.
NVIDIA® NVS Professional Business Graphics
COMPUTER GRAPHICS CS 482 – FALL 2014 NOVEMBER 10, 2014 GRAPHICS HARDWARE GRAPHICS PROCESSING UNITS PARALLELISM.
GPU and PC System Architecture UC Santa Cruz BSoE – March 2009 John Tynefield / NVIDIA Corporation.
Understanding the graphics pipeline Lecture 2 Original Slides by: Suresh Venkatasubramanian Updates by Joseph Kider.
Graphics Pipeline.
Case Study: T-Mobile Retail Personal Coverage Check Kiosk Joe Wong, Integral GIS Sean Alexis, T-Mobile April 18, 2007.
Click Here to Begin. Objectives Purchasing a PC can be a difficult process full of complex questions. This Computer Based Training Module will walk you.
NVIDIA Confidential. Adobe Video Market Opportunity SAM 220K Premiere Pro Forecast / 1 st year 20% Professional 75% Quadro Attach (upgrade) Quadro Market.
GPU Virtualization on VMware’s Hosted I/O Architecture Micah Dowty Jeremy Sugerman USENIX WIOV
Rage Fury MAXX™. The Answer to today’s 3D dilemma High performance AND High quality AND Universal application acceleration.
GRAPHICS AND COMPUTING GPUS Jehan-François Pâris
NVIDIA ® Quadro ® Mobile Workstation Solutions Intel ® Core i5 and i7 “Calpella” Based Platforms October, 2010.
OS Case Study: The Xbox 360  Instructor: Rob Nash  Readings: See citations in the slides.
Programming with CUDA, WS09 Waqar Saleem, Jens Müller Programming with CUDA and Parallel Algorithms Waqar Saleem Jens Müller.
A Crash Course on Programmable Graphics Hardware Li-Yi Wei 2005 at Tsinghua University, Beijing.
1 ITCS 6/8010 CUDA Programming, UNC-Charlotte, B. Wilkinson, Jan 19, 2011 Emergence of GPU systems and clusters for general purpose High Performance Computing.
3D Graphics Processor Architecture Victor Moya. PhD Project Research on architecture improvements for future Graphic Processor Units (GPUs). Research.
Evolution of the Programmable Graphics Pipeline Patrick Cozzi University of Pennsylvania CIS Spring 2011.
Status – Week 283 Victor Moya. 3D Graphics Pipeline Akeley & Hanrahan course. Akeley & Hanrahan course. Fixed vs Programmable. Fixed vs Programmable.
Presentation by David Fong
GPU Graphics Processing Unit. Graphics Pipeline Scene Transformations Lighting & Shading ViewingTransformations Rasterization GPUs evolved as hardware.
NVIDIA Confidential. Product Description World’s most popular 3D content creation tool Used across Design, Games and VFX markets Over +300k 3ds Max licenses,
Realtime 3D Computer Graphics Computer Graphics Computer Graphics Software & Hardware Rendering Software & Hardware Rendering 3D APIs 3D APIs Pixel & Vertex.
Under the Hood: 3D Pipeline. Motherboard & Chipset PCI Express x16.
High Performance in Broad Reach Games Chas. Boyd
Interactive Visualization of Volumetric Data on Consumer PC Hardware: Introduction Daniel Weiskopf Graphics Hardware Trends Faster development than Moore’s.
CSU0021 Computer Graphics © Chun-Fa Chang CSU0021 Computer Graphics September 10, 2014.
Dell Precision T1600 Sales Aid  What’s New? The Dell Precision T1600 is the next generation of our single-socket, entry-level T1500 workstation. Designed.
NVIDIA Quadro for Mac by PNY Accelerate Your Entire Workflow.
GPU Programming Robert Hero Quick Overview (The Old Way) Graphics cards process Triangles Graphics cards process Triangles Quads.
CHAPTER 4 Window Creation and Control © 2008 Cengage Learning EMEA.
Graphics Hardware and Graphics in Video Games COMP136: Introduction to Computer Graphics.
Computer Graphics Graphics Hardware
Nick Sims Like a motherboard, a graphics card is a printed circuit board that houses a processor and RAM.
The Graphics Rendering Pipeline 3D SCENE Collection of 3D primitives IMAGE Array of pixels Primitives: Basic geometric structures (points, lines, triangles,
NVIDIA GPUs Power Adobe® Creative Suite® 6 Production Premium Best Performance - Broadest GPU Support ® Premiere Pro CS6 After Effects CS6 SpeedGrade.
1 Introduction to Computer Graphics SEN Introduction to OpenGL Graphics Applications.
NVIDIA Quadro Plex Scalable Visualization Solutions July 2009.
Emergence of GPU systems and clusters for general purpose high performance computing ITCS 4145/5145 April 3, 2012 © Barry Wilkinson.
Stream Processing Main References: “Comparing Reyes and OpenGL on a Stream Architecture”, 2002 “Polygon Rendering on a Stream Architecture”, 2000 Department.
Siemens NX Overview.
Computer Graphics 3 Lecture 6: Other Hardware-Based Extensions Benjamin Mora 1 University of Wales Swansea Dr. Benjamin Mora.
Introduction Why are virtual machines interesting?
1 Introduction to Computer Graphics with WebGL Ed Angel Professor Emeritus of Computer Science Founding Director, Arts, Research, Technology and Science.
GAM666 – Introduction To Game Programming ● Programmer's perspective of Game Industry ● Introduction to Windows Programming ● 2D animation using DirectX.
From Turing Machine to Global Illumination Chun-Fa Chang National Taiwan Normal University.
GPU Computing for GIS James Mower Department of Geography and Planning University at Albany.
Avaya - Proprietary. Use pursuant to your signed agreement or Avaya policy. 1 Selling Avaya Scopia XT5000 with IP Office.
Postmortem: Deferred Shading in Tabula Rasa Rusty Koonce NCsoft September 15, 2008.
Image Fusion In Real-time, on a PC. Goals Interactive display of volume data in 3D –Allow more than one data set –Allow fusion of different modalities.
Computer Graphics Graphics Hardware
Emergence of GPU systems for general purpose high performance computing ITCS 4145/5145 July 12, 2012 © Barry Wilkinson CUDAIntro.ppt.
Petri Nordlund Chief Architect Bitboys Oy
Petri Nordlund Chief Architect Bitboys Oy
Computer Graphics Lecture 32
Graphics Processing Unit
Introduction to OpenGL
From Turing Machine to Global Illumination
sv7: Blazing Visualization on a Commodity Cluster
Introduction to Computer Graphics with WebGL
Graphics Processing Unit
Computer Graphics Graphics Hardware
RADEON™ 9700 Architecture and 3D Performance
Introduction to OpenGL
Competitive Analysis: Apple Inc.
Presentation transcript:

Jon Peddie Research Professional Graphics CGW Webinar

Jon Peddie Research Evolution of Professional Graphics Yesterdays Landscape Pro graphics distinguished from consumer/corporate graphics by most every metric –Vendor, chips, boards, bus, memory, video I/O, OS, middleware/APIs, usage, performance, price Excusive domain of Traditional Proprietary Workstation (TPW) vendors –Sgi, Sun, HP, DEC and IBM drove the innovation –Proprietary UNIX/VMS were the only OSes

Jon Peddie Research Evolution of Professional Graphics Todays Landscape Hardware migration from in-house to IHV –IHVs are vertical: chips and AIBs –TPW vendors no longer build graphics chips Enable IHV hardware with drivers for proprietary Unix A few unique high-end board configurations Gaming is driving innovation –E.g. programmable shaders, floating-point precision –Even $1K+ multi-board monsters: Alienware and Nvidias SLI Much harder to distinguish from consumer AIBs –Cost premium has dropped considerably

Jon Peddie Research Graphics Hardware Differentiation Professional vs. Consumer Brand Reliability –ISV certification Customer support Breadth of driver support –OpenGL ICDs –64-bit Linux and Windows drivers Performance and price can be a low priority –Legacy requirements can sustain lagging hardware

Jon Peddie Research Graphics Hardware Differentiation Professional vs. Consumer Remaining GPU differences artificial and/or minor –(Virtually) no difference in raw die –Nvidia and ATI lead with same GPU/VPUs from consumer line relatively minor driver, package and or board-exposed features Board-level differences significant at high-end only –Value varies by application –Physical memory DCC and vis-sim have never-ending appetite for textures –Display support optimized for pro applications Framelock, genlock, interface type (e.g. SDI) Number and datarate of video interfaces Ultra-high resolution (e.g. dual dual-link for up to 9 Mpixel displays)

Jon Peddie Research 2003 Professional Graphics Hardware Market Nearly 2.1 million professional graphics AIBs sold Almost $1B in revenue Legacy in-house graphics from TPW vendors small but significant –Only 5% of units shipped, but 16% of revenue –Incremental opportunity for IHVs Units are in the low-end, but revenue is in the mid-range

Jon Peddie Research JPR Pro Graphics AIB Classes Class ASP Range minmax 2D$400 Entry-3D$350 Mid-range$350$950 High-end$950$1,500 Ultra High-end$1,500

Jon Peddie Research Vendor Profile 5% (units) and 11% (revenue) share in 2003 –but 26% and 44% unit share in high and ultra-high Pro Gfx flagship: Realizm What sets Realizm apart –Exclusive focus on professional apps –Chip-level scalability –16-bit FP format in frame buffer –Virtual, paged video memory Where 3DLabs is going –Fighting hard to keep high-end dominance Largest physical memory, Multi-chip AIBs, Genlock / framelock –Realizm trickle-down to mid-range and low-end?

Jon Peddie Research Vendor Profile 17% (units) and 15% (revenue) share in 2003 –Unbranded presence in 2D applications Pro graphics flagship: Fire GL 7100 What sets Fire GL apart –A strong mid-range focus (31%) –Subjective edge in quality and quality/performance –Perf/W has won mobile and embedded sockets ATI dominant in mobile workstations (67%) Where ATI is going –Best positioned to ride growth in mobile workstation –Can it (should it) ignore high end of market?

Jon Peddie Research Vendor Profile JPR estimates* 9% (units) and 11% (revenue) share in 2003 –But ~17% in 2D segment Slanted heavily toward direct sales Not directly targeting power renderers –Appeal on basis of image quality and specific, niche features Where is Matrox going? –Road ahead looks difficult in keeping pace on GPUs Last major introduction, Parhelia, was out in May 2002 Move to programmable shaders and floating-point requires overhaul –Some key competitive advantages going away More 2D competition from Nvidia, ATI and maybe soon IGPs Fewer areas of differentiation, e.g. super-high res (9 Mpixel) –OEM presence declining –Continued focus on custom-fit solutions for large customers * Matrox is private and does not disclose financials

Jon Peddie Research Vendor Profile 67% (units) and 47% (revenue) share in 2003 Pro gfx flagship: Quadro FX 4000 (NV40 GPU) What sets Nvidia apart –Breadth of offerings, entry to ultra-high end –Shader Model 3.0 vs. 2.0 –SLI: Board level scalablity –Custom offerings for DCC, vis-sim Where Nvidia is going –Trying to take share in existing segments From 3Dlabs in the high/ultra-high end –Think margin, not units –Sales synergy From ATI in the mobile space (MXM and Axiom) –Getting GPUs into new segments, like render farms

Jon Peddie Research Pro Graphics Technology Trends Final stage of migration to fully programmable architecture –Richer, cleaner programming: large code, predication, branching –Changing how graphics hardware vendors will compete Leveraging parallelism –Todays flagship GPUs: 6 vertex and 16 pixel pipelines (ATI/Nvidia) –Chip-level (3DLabs) and Board-level (Nvidia) scalability Continuing to annex upstream processing –Physics, kinematics, simulation, animation, tessellation Vehicle for general purpose computing (GPGPU), –Why Intels biggest threat may someday be not AMD but Nvidia Floating-point precision GDDR3 memory

Jon Peddie Research PCI Express for Graphics Serial, point-to-point, packets –More a network interconnect than a traditional I/O bus Variable number of lanes –Graphics design center: 16-lane More bandwidth, but remember: –Directionally constrained: 4GB/s up, 4GB/s down –In-band command, control and packet overhead reduces bw Just in time to carry the load –Most apps on most hardware today not constrained by AGP 8X –Some may be … it all depends HD video editing Hybrid CPU/GPU render for DCC Src: PCI-SIG

Jon Peddie Research PCI Express Graphics AIBs Form factor derived from PCI Power budgets –10W: ×1 cards (<= 6.6 length) –25W: ×1 cards (> 7.0 length), ×4 cards, ×8 cards, ×16 low-profile graphics and ×16 server I/O –75W: full-height graphics cards –High-end Graphics Spec will allow auxiliary power for up to 150W Src: PCI-SIG

Jon Peddie Research PCI Express Connectors Up-plugging allowed OEMs encouraged to support wider connectors –Link width not determined by connector or interface, negotiated at config time –More end-user flexibility –Allows dual high-bw ( AGP 8X) graphics AIBs Src: PCI-SIG

Jon Peddie Research GPU Interfaces to PCI Express To bridge or not to bridge Initial plans spurred some mud-slinging –ATI planned all native PCIe interfaces –Nvidia indicated plans to bridge with on-board HSI (AGP 16X) –3DLabs Realizm depends on configuration In the end, it will most likely be a non-issue –Speedup of full-speed PCIe interface is exception and debatable –ATI will likely bridge back to AGP –HSI does not preclude native PCIe – NV45 is out already –3DLabs likely to fill in low-end PCIe offerings, too Dell should ship Nvidia and ATI PCIe AIBs July, 3DLabs later this quarter

Jon Peddie Research Pro Graphics Market Trend Forecast Strong growth in Mobile Workstations Final phase in transition to all-IHV graphics AIBs configured for specific applications –Genlock and SDI for DCC studio apps –Framelock for vis-sim and wall-display applications IGPs for pro graphics? Never say never. –What about Grantsdale for 2D workstation apps? –Why Nvidia/ATI/3DLabss biggest competitor may someday be Intel GPUs to final frame rendering?

Jon Peddie Research Nvidias Application-specific AIB Configurations Nvidia Quadro FX 3000G I/O Nvidia Quadro FX 4000 SDI I/O

Jon Peddie Research GPUs in the Render Farm? Graphics hardware is absent in the render farm ISVs/IHVs looking to final-frame speedup as well –Enablers Primary: advent of programmable hardware shaders with compilers Secondary: FP color precision, more flexible programming (larger code, predication, branching) –Nvidia Gelato, Mental Images Mental Ray 3.3 Vendors would welcome 10Ks of incremental professional GPUs Not a slam-dunk –Global illumination, raycasting techniques (e.g raytracing and volume rendering) dont map very well (at least not yet) to GPUs

Jon Peddie Research Technology Forecast Impact of Longhorn Image quality –Gamma, sRGB, 32-bit FP, Text enhancements Virtualization to support Avalon, Presentation Manager –Virtual memory, mostly under OS/driver interaction –GPU: Hyper Threading-like context management Pixel rates will be especially stressed –Lots of temporary textures, surfaces to be warped, composited, blended Dual, cascaded vertex shaders Moving to (optional) programmable hardware tessellation Security & stability –simpler drivers, hang prevention OpenGL ICDs should be upgraded for Longhorn (but not required)

Jon Peddie Research Windows Graphics Foundation Longhorn and Beyond Src: Microsoft, WinHEC 2004

Jon Peddie Research Backup Slides

Jon Peddie Research Hardware Differentiation vs. Consumer Disappearing Historical DifferentiatorFuture differentiator? OpenGL vs. DirectXMinor, esp. with OGL2 Anti-aliased points / linesNo. Can be rendered with shaded polygons; even enhanced (e.g. miter) Rendering performanceVery little… performance driven by games Color fidelityVery little … internal FP32, stored as FP16 (display?); attention to sRGB and gamma becoming pervasive AntialiasingVery little, ># samples at high end (highest end is software) Display resolutionYes, at the high end with 9 MPix 2 nd order rendering features: e.g. two-sided lighting, user clip planes Very little … implemented with shader Multi-displayLittle … even IGPs going dual-display Genlock, FramelockYes, at high end Multi-chip implementationsYes, at high end Overlay planesYes, but trivial to implement StereoYes, but relatively easy to implement