Friday, June 22, 2018
Search
  
Submit your own News for
inclusion in our Site.
Click here...
Breaking News
FCC to Seek for Flexible Use of C-band and 6GHz Airwaves
AMD Presents Modular Routing Design for Chiplet-based Systems
Software Business Continues to Work For BlackBerry
Apple Turns to the U.S. Patent Office to Invalidate Qualcomm Patents
Samsung Patents Bezel-less, Notch-free Smartphone Design
China is Home to Most Smartphone Vendors
VidCon 2018: Youtube Announces Memberships, Merchandise as Alternatives to Ads
Chatting With Google Assistant Gets More Natural
Active Discussions
Which of these DVD media are the best, most durable?
How to back up a PS2 DL game
Copy a protected DVD?
roxio issues with xp pro
Help make DVDInfoPro better with dvdinfomantis!!!
menu making
Optiarc AD-7260S review
cdrw trouble
 Home > News > Graphics Cards > NVIDIA ...
Last 7 Days News : SU MO TU WE TH FR SA All News

Tuesday, January 19, 2010
NVIDIA Details GF100 GPU


Nvidia provided some details about the highly anticipated GF100 GPU, based on the new 'Fermi' architecture.

Scheduled for release sometime in April, the GF100 is comprised of four main units, Nvidia calls "Graphics Processing Clusters" or (GPC). Each GPC includes 4 "Streaming Multiprocessor cores" (SP), resulting to a total number of 16 SP.

Each GPC also has a 'Raster Engine'. Nvidia claims that the GF100 has also dedicated hardware tessellators, or what Nvidia calls 'polymorph engines', appearing one per SM. Breaking this down further, each of the SMs comprise of 32 CUDA cores, 16 or 48KB of shared memory and 16 or 48KB of L1 cache and 4 texture units.



Basic features:

3 billion transistors
512 cores CUDA cores
16 Streaming Multiprocessors (SM)
16 Polymorph engines
4 Raster Engine Units 768KB L2 cache
64 Texture units
48 ROPs
384-bit memory interface (GDDR5)

A closer look at the following diagram could be helpful to understand the GF 100's main architecture:



As you can see the GF100 looks more like a GPGPU chip rather than a GPU.

What about clock speeds, performance or power requirements? Nvidia has not provided any details. At CES2010, Nvidia showed off the tessellation capabilities pf the GF100 system using the Unigine Heaven benchmark. The company claimed that the GF100 beat the ATI 5870 by 60 percent. However, the company did not ran any synthetic benchmarks and also chose not to compare it with a comparably priced ATI HD5970 Hemlock card - which also supports tessellation.



Previous
Next
Samsung Unveils New Digital Cameras        All News        Apple's Invites Journalists to 'Special Event'
NVIDIA GeForce 196.21 WHQL Drivers Released     Graphics Cards News      Nvidia's Tesla Bio Workbench Enables Scientists to Achieve New Breakthroughs

Get RSS feed Easy Print E-Mail this Message

Related News
NVIDIA CEO Springs Special Titan V GPUs on Elite AI Researchers
Nvidia Uses AI to Produce High-quality, 240fps Slow-motion Video From 30fps Source
Computex: NVIDIA Announces $1,300 Isaac Platform for Building Robots
Hot Chips Symposium: Nvidia's Next Generation GPU, Intel Xeon Cascade Lake Processors and More
NVIDIA Introduces HGX-2, Fusing HPC and AI Computing
AMD Increased GPU Market Share in Q1
Nvidia Releases the 3GB GeForce GTX 1050 to Attract More PC Gamers
Data Center and Gaming Business Boost Nvidia's Revenue
Nvidia Kills the GeForce Partner Program
Nvidia Climbs Top 10 List of Chip Sellers
Nvidia Boosts GPU Performance in training Neural Networks, Partners With Arm Partner to Bring Deep Learning to IoT
NVIDIA Quadro GV100 GPU Announced for the Age of Deep Learning

Most Popular News
 
Home | News | All News | Reviews | Articles | Guides | Download | Expert Area | Forum | Site Info
Site best viewed at 1024x768+ - CDRINFO.COM 1998-2018 - All rights reserved -
Privacy policy - Contact Us .