NVIDIA Quadro PNY Roadshow Januar, Februar 2011 Lutz Eigenfeld 1
GeForce Quadro
Amaze
Design
Tesla Explore
Tegra Anywhere
2
fermi
Quadro
Exponentially Better
fermi 3
Quadro
4
Professional Graphics Inflection Points
Fixed Function 3D Pipelines
Programmable Shaders
Computational Visualization 5
1st and 2nd Era 1,4 1,2
1 0,8 0,6
TeraFlops
0,4
Triangles (Billions) 0,2 0 1999
2006
2007
2008
2010
6
3rd Era 1,4
Triangles 1,2
1
TeraFlops
0,8 0,6 0,4 0,2 0 1999
2006
2007
2008
2010
Quadro 6000 7
Quadro
Tesla
GeForce
Performance
Viewperf (Geomean)
13x
1x
1x
Graphics
OpenGL
4.0
4.0
4.0
DX
11
11
11
Quad Buffered OGL Stereo
Yes
-
-
ECC
Yes
Yes
-
Fast Double Precision
Yes
Yes
-
WS ISV Certs
Yes
-
-
Tesla ISV Certs
Yes
Yes
-
G-Sync
Yes
-
-
SDI
Yes
-
-
SLI Multi-OS
Yes
-
-
3D Vision Pro
Yes
-
-
Performance Drivers (3dsMax; Autocad)
Yes
-
-
Enterprise Sales and Support
Yes
Yes
-
NVIDIA
NVIDIA
~ 40 Various AIC
Compute Software
Critical Adders
Support
Manufacturer
8
Brand Transition Introducing NVIDIA NVS NVIDIA Quadro FX
NVIDIA Quadro NVS
Quadro = Professional Visualization Graphics
NVS = Professional Business Graphics
9
QuadroPlex 7000
Quadro 6000
Quadro 5000
Quadro 4000
6 GB GDDR5 448 CUDA Cores ECC Stereo G-Sync Compatible HD SDI Compatible
12 GB GDDR5 Total 896 Cores ECC Stereo Quad DVI-DL G-Sync Incl. DHIC Compatible: 1,792 Cores 24GB 8 DVI-DL
2.5 GB GDDR5 352 CUDA Cores ECC Stereo G-Sync Compatible HD SDI Compatible
2 GB GDDR5 256 CUDA Cores Optional Stereo HD SDI Compatible
10
Quadro Fermi Family Product Segment
Target Audience
Key Additional Relevant Features
Quadro Solution
4D Seismic Analysis 4D Medical Imaging
+ 6GB GPU Memory + 448 CUDA Parallel Cores
Quadro 6000
Digital Special Effects Product Styling
+ G-Sync + SLI Frame Rendering + Mosaic Advanced Features + ECC Memory
Quadro 5000
High End MCAD Digital Effects Broadcast
24% Better performance than Quadro 2000 + SDI + Native Stereo + Fast Double Precision +Dual Copy Engines
Quadro 4000
Midrange CAD Midrange DCC
+42% better Performance than Quadro 600 +SLIMOS
Quadro 2000
Entry
Volume CAD Volume DCC
+44% better performance than FX 380 + Mosaic Technology
Quadro 600
Entry
Volume CAD Volume DCC Productivity Apps
Ultra High-End
High-End
High-End
Mid-Range
Quadro FX 380 11
Model
Quadro 4000
Quadro 5000
Quadro 6000
QuadroPlex 7000
CUDA Cores
256
352
448
896
Memory Size
2.0GB GDDR5
2.5GB GDDR5
6.0GB GDDR5
12GB GDDR5
Memory Bandwidth
89.6 GB/s (256-bit)
120 GB/s (320-bit)
144 GB/s (384-bit)
144 GB/s (384-bit)
Power
142 W (SSA)
152 W (DSA)
225 W (DSA)
System (3U/Deskside)
Outputs
DP(2) + DL-DVI + ST (optional)
DP (2) + DL-DVI + ST
DP(2) + DL-DVI + ST
DL-DVI (4) + ST (2)
Fast Double Precision
Yes
Yes
Yes
Yes
ECC
No
Yes
Yes
Yes
SLI FR and FSAA
No
Yes*
Yes*
Yes
SLI Mosaic Mode
No
Yes
Yes
Yes
SLI Multi OS
Yes
Yes
Yes
No
G-Sync Compatible
No
Yes
Yes
Standard
SDI Compatible
Yes
Yes
Yes
N/A
3D Vision Pro
Yes
Yes
Yes
Yes
12
Quadro 2000: Ultimate price-performance
•
Quadro 2000
CUDA Cores
192
Memory Size
1.0GB GDDR5
Memory Bandwidth
41.6 GB/s (256-bit)
Power
62 W
Form Factor
4.376” x 7” Single Slot
Display IO
DVI-I (1) + DP (2)
SLI Multi OS
Yes
3D Vision Pro
Yes (USB)
DirectX11
Yes
OpenGL 4.1
Yes
Delivers best in class performance across leading CAD and DCC applications
• •
Model
Certified on a broad range of leading applications
Integral part of a sophisticated ecosystem that accelerates the professional workflow 13
Quadro 600: Ultimate Value Model
Quadro 600
CUDA Cores
96
Memory Size
1.0GB DDR3
Memory Bandwidth
25.6 GB/s (256-bit)
Power
40W
Form Factor
•
•
4.376” x 6.6” Full Height Bracket 2.731” x 6.6” Low Profile Bracket
Display IO
DVI-I (1) + DP (1)
3D Vision Pro
Yes (USB)
DX11
Yes
OpenGL 4.1
Yes
Delivers best in class performance per watt
Most advanced feature set in the professional entry segment •
Flexible form factor 14
NVIDIA Quadro 4000 for Mac GPU: 256-core CUDA Parallel Computing Processor Fast Double Precision Form Factor: Full ATX; 4.376 (H)” x 9.50 (L)” Single Slot Frame Buffer: 2 GB GDDR5; 256 bit
Connectors: (1) DisplayPort; (1) DL DVI-I; (1) 3D Stereo DisplayPort to mini-DisplayPort adapter included
6-pin aux power connector Total Board Power: 142W Availability: November 2010 Through NVIDIA Quadro channel partners and APPLE.COM 15
NVIDIA Business Graphics Line-up - 2011 NVIDIA Quadro NVS 450 (ATX profile PCIe x16)
NVIDIA Quadro NVS 420 (Low Profile PCIe x16 & x1)
New!
NVIDIA NVS 300 Low Profile (PCIe x16 and x1)
NVIDIA Quadro NVS 295 Low Profile (PCIe x16 and x1) NVIDIA Quadro NVS 290 Low Profile (PCIe x16 and x1)
Quad DisplayPort Quad DVI-D (Single/Dual-Link) Quad VGA Quad DisplayPort Quad DVI-D (Single-Link)
Dual DVI-I(Single Link) Dual VGA Dual Display Port Dual DisplayPort Dual DVI-D (Single/Dual Link) Dual VGA Dual DVI-I(Single Link) Dual VGA
16
Dual NVS Generation Comparison Product Features
NVS 290
NVS 295
NVS 300
Gen 1
Gen 2
Gen 2
Memory Size & Type
256MB DDR2
256MB GDDR3
512 MB DDR3
Memory Bandwidth
6.4 GB/s
11.2 GB/s
12.6 GB/s
Power
21 Watts
23 Watts
17.5 Watts
Supported Displays (Native)
SL-DVI-I VGA
Display Port
SL-DVI-I VGA Display Port (with audio)
1920x1200
2560x1600
2560x1600
HDMI Support
No
Yes
Yes
Audio Support
No
Yes (HDMI thru SPDIF only)
Yes (DP & HDMI over PCIE)
PCIE Generation
Max Digital Display Resolution
17
Teaching CUDA 18
Tesla Data Center & Workstation GPU Solutions
Tesla M-series GPUs M2070 M2050 M1060
Integrated CPU-GPU Servers & Blades
Tesla S-series 1U Systems S2050 S1070
Tesla C-series GPUs C2070 C2050 C1060
OEM CPU Server + Tesla S-series 1U
Workstations 2 to 4 Tesla GPUs 19
NVIDIA Tesla GPU Computing Products Server Module
1U Systems
Workstation Boards
Tesla M2070 / Tesla M2050
Tesla M1060
Tesla S2050
Tesla S1070
Tesla C2070 / Tesla C2050
Tesla C1060
GPUs
1 T20 GPU
1 T10 GPU
4 T20 GPUs
4 T10 GPUs
1 T20 GPU
1 T10 GPU
Single Precision
1030 GFlops
933 GFlops
4120 GFlops
3732 GFlops
1030 Gflops
933 GFlops
Double Precision
515 Gflops
78 GFlops
2060 GFlops
312 GFlops
515 Gflops
78 GFlops
Memory
6 GB / 3 GB
4 GB
12 GB (S2050)
16 GB 4 GB / GPU
6 GB / 3 GB
4 GB
Mem BW
148.4 GB/s
102 GB/s
148.4 GB/s
102 GB/s
144 GB/s
102 GB/s
20
Tesla S20xx 1U GPU Systems NextIO vCORE™
S2050
S20xx uses same accessories as S1070
Express 2070
Processors
4 Tesla 20-series GPUs
Number of Cores
1792 per 1U (448 / GPU)
Single precision performance
4120 Gigaflops per 1U 1030 Gigaflops / GPU
Double precision performance
2060 Gigaflops per 1U 515 Gigaflops / GPU
GPU Memory
12 GB per 1U 3 GB / GPU 2.625 GB with ECC on
24 GB per 1U 6 GB / GPU 5.25 GB with ECC on
Memory Interface
GDDR5
System I/O
2x PCIe x16 Gen2 Host interface cards (HICs) OR 2x PCIe x8 Gen2 Host interface cards (HICs) OR 1x PCIe x16 Gen2 Dual Host interface cards (DHIC)
Power
1200 W (max)
Available
June 2010
Dezember 2010
21
The World’s Fastest Supercomputer Tianhe-1A 2.507 Petaflop
7168 Tesla M2050 GPUs National Supercomputing Center in Tianjin 22
Professional Applications are 3D Stereo • Mission critical feature • Until now display technology had lagged
• Introducing NVIDIA 3D Vision Pro for Professionals
NVIDIA 3D Vision Pro Highest Quality 3D Stereo Solution • 120 Hz Active Shutter solution • Quadro 3-Pin Stereo – direct to GPU
Designed for Multi User Collaborative Environments • RF communication for coverage up to 100ft • Does not require “line of sight” Advanced Host and Application Management • Explicit coupling, Battery Levels, Operation • Accelerometer and Compass API (TBA) Available October 2010 • Via NVIDIA Channel Partners
24
Consumer vs. Professional Stereo Consumer/GeForce
Professional/Quadro
Yes
Yes
Yes, limited multidisplay configurations
Yes
Quad Buffered OpenGL Support
No
Yes
Supported Stereo Display Types
3D Vision Certified
3D Vision Certified Left/Right Display Interlaced Passive Active CRTs & projectors
3D Vision Only
All
Surround Gaming 3x1 Only
Up to 16
Stereo must be primary
Yes
G-Sync support for multiple displays and systems
No
Yes
Stereo Connector for direct GPU control of glasses
No
Yes
Consumer 3D Vision SW
DirectX Stereo Support
Supported 3D Glasses Multiple Stereo Displays Mixed Stereo and Regular Display
25
Danke f端r Ihre Aufmerksamkeit
26
ISV Pipeline http://www.nvidia.com/tesla
NVIDIA Confidential – Internal Only
Increasing Number of Professional CUDA Applications Available Now
Tools
Libraries
Future
CUDA C/C++
PGI Accelerators
Platform LSF Cluster Manager
TauCUDA Perf Tools
Parallel Nsight Vis Studio IDE
TotalView Debugger
PGI CUDA Fortran
CAPS HMPP
Bright Cluster Manager
Allinea DDT Debugger
ParaTools VampirTrace
AccelerEyes Jacket (MATLAB)
CUDA FFT CUDA BLAS
EMPhotonics CULAPACK
Thrust C++ Template Lib
NVIDIA NPP Perf Primitives
MAGMA (LAPACK)
NVIDIA Video Libraries
Headwave Suite
OpenGeoSolutions OpenSEIS
GeoStar Seismic Software Suite
Acceleware RTM Solver
StoneRidge RTM
ffA SVI Pro
VSG Open Inventor
Seismic City RTM
Tsunami RTM
AMBER
NAMD
HOOMD
TeraChem
BigDFT ABINT
Acellera ACEMD
GROMACS
LAMMPS
VMD
GAMESS
CP2K
Schrodinger Core Hopping
CUDA-BLASTP
MUMmerGPU
CUDA-MEME
PIPER Docking
CUDA SW++ (SmithWaterman)
GPU-HMMR
CUDA-EC
HEX Protein Docking
ACUSIM AcuSolve 1.8
Autodesk Moldflow 2010
Prometch Particleworks
Remcom XFdtd 7.0
Wolfram Mathematica
Paradigm GeoDepth RTM
Panorama Tech
Oil & Gas Paradigm SKUA DL-POLY
Bio-Chemistry
BioInformatics
CAE
Available
Announced
OpenEye ROCS
FluiDyna OpenFOAM
LSTC LS-DYNA 971
Metacomp CFD++
MSC.Software Marc 2010.2
Increasing Number of Professional CUDA Applications Available Now
Video
Rendering
Adobe Premier Pro CS5
ARRI Various Apps
GenArts Sapphire Plugins
TDVision TDVCodec
Black Magic Da Vinci Resolve
MainConcept CUDA Encoder
Elemental Video Transcode
Fraunhofer JPEG2000
Cinnafilm Pixel Strings
Assimilate SCRATCH
Bunkspeed Shot (iray)
Refractive SW Octane
Random Control Arion
ILM Plume
Works Zebra Zeany
Cebas finalRender
mental images iray (OEM)
NVIDIA OptiX (SDK)
Caustic Graphics OpenRL (SDK)
Weta Digital PantaRay
Lightworks Artisan
Chaos Group V-Ray GPU
NAG RNG
Numerix Counterparty Risk
SciComp SciFinance
RMS Risk Mgt Solutions
Aquimin AlphaVision
Hanweck Volera Options Analytics
Murex MACS
Agilent EMPro 2010
CST MICROWAVE STUDIO
Agilent ADS SPICE Simulator
Acceleware FDTD Solver
Synopsys Sentaraus TCAD
SPEAG SEMCAD X
Gauda Optical Proximity Correct
Acceleware EM Solution
Siemens Foursight 4D Ultrasound
Digisens DigiHCT Medical Imaging
Schrodinger Core Hopping
Useful Progress Medical Imaging
MotionDSP Ikena Video
Manifold GIS
Dalsa Sapera (Machine Vision)
Digital Anarchy Beauty Box Photo
Finance
EDA
Other
Available
Future
Announced
The Foundry Kronos