GeForce 400 series
Nvidia's Fermi architecture introduced Direct3D 11 and high-performance computing features.
The GeForce 400 series is a line of graphics processing units from Nvidia that introduced the Fermi microarchitecture. Originally planned for a November 2009 launch, the series was delayed and released on March 26, 2010, with retail availability beginning in April 2010. Its main rival was ATI's Radeon HD 5000 series.
The Fermi architecture, named after physicist Enrico Fermi, was Nvidia’s first major GPU redesign since the Tesla microarchitecture used in the G80. The initial chip, GF100, contained 512 stream processors arranged in 16 groups of 32, with 3.0 billion transistors manufactured by TSMC on a 40 nm process. It was Nvidia’s first chip to support OpenGL 4.0 and Direct3D 11. No fully enabled GF100 GPU was ever sold. The GTX 480 had one streaming multiprocessor disabled, the GTX 470 had two streaming multiprocessors and one memory controller disabled, and the GTX 465 had five streaming multiprocessors and two memory controllers disabled. Consumer GeForce cards used 256 MB of GDDR5 memory per enabled memory controller, resulting in total capacities of 1.5 GB, 1.25 GB, or 1.0 GB. The Tesla C2050 had 512 MB on each of six controllers, and the Tesla C2070 had 1024 MB per controller; both Tesla cards had 14 active groups of stream processors.
High-performance Tesla chips included optional ECC memory and could perform one double-precision floating-point operation per core per cycle. Consumer GeForce cards were artificially driver-restricted to one double-precision operation per four cycles. With support for Visual Studio and C++, Nvidia targeted professional, commercial, and high-performance computing markets.
The amount of on-chip SRAM per ALU decreased compared to the previous G200 generation, despite an increase in L2 cache from 256 kB per 240 ALUs to 768 kB per 512 ALUs. Fermi had 32,768 registers per 32 ALUs (versus 16,384 per 8 ALUs), 48 kB of shared memory per 32 ALUs (versus 16 kB per 8 ALUs), and 16 kB of cache per 32 ALUs (versus 8 kB constant cache per 8 ALUs plus 24 kB texture cache per 24 ALUs). Detailed register counts are listed in the CUDA Compute Capability Comparison Table.
Quick Facts
- Codename
- GF10x
- Architecture
- Fermi
- Created
- April 12, 2010
- Model
- GeForce series
- Transistors
- 260M 40 nm (GT218 - GeForce GT 405 only)
- Entry
- GT 405 / GT 420 / GT 430
Facts from the source article.
Lore & Background
Nvidia described the Fermi microarchitecture as the next major step in its line of GPUs following the Tesla microarchitecture used since the G80. The GF100, the first Fermi-architecture product, is large: 512 stream processors, in sixteen groups of 32, and 3.0 billion transistors, manufactured by TSMC in a 40 nm process. It is Nvidia's first chip to support OpenGL 4.0 and Direct3D 11. No products with a fully enabled GF100 GPU were ever sold. The GTX 480 had one streaming multiprocessor disabled. The GTX 470 had two streaming multiprocessors and one memory controller disabled. The GTX 465 had five streaming multiprocessors and two memory controllers disabled. Consumer GeForce cards came with 256MB attached to each of the enabled GDDR5 memory controllers, for a total of 1.5, 1.25 or 1.0GB; the Tesla C2050 had 512MB on each of six controllers, and the Tesla C2070 had 1024MB per controller. Both the Tesla cards had fourteen active groups of stream processors.
Many users reported high temperatures and power consumption while receiving correspondingly poor performance improvement in the GeForce 400 series Fermi GPUs when compared to rival competitor AMD's Radeon HD 5000 series - leading AMD to create and release a promotional video 'The Misunderstanding' to poke fun at the issue. In the video, a police unit is seen commencing a raid on a house with a large thermal profile, indicating a grow operation. However, upon entering the home it is apparent that the source of the high temperature is a Fermi GPU. It became a common joke that one could fry an egg on a Fermi GPU at full load.
Reader's Guide
The GeForce 400 series marked Nvidia's transition to the Fermi microarchitecture, which introduced support for Direct3D 11 and OpenGL 4.0, as well as features targeting high-performance computing such as ECC memory and double-precision floating-point support on Tesla variants. However, the series was notable for its high power consumption and thermal output, which drew criticism and mockery from competitor AMD. The GF100 chip was never fully enabled in any consumer product; the GTX 480, 470, and 465 had varying numbers of disabled streaming multiprocessors and memory controllers. The series also included the GeForce 405, an OEM-only card based on the older Tesla core, making it the only GeForce 400 card not based on Fermi. Nvidia later released the GF110 chip with the GTX 580, a redesigned GF100 that used significantly less power and allowed all 16 SMs to be enabled. The Fermi architecture's legacy includes its role in establishing Nvidia's compute capabilities, with GF100 cards providing Compute Capability 2.0 and GF104/106/108 cards providing Compute Capability 2.1. Nvidia announced in April 2018 that Fermi would move to legacy driver support status and be maintained until January 2019.
Did You Know?
- The GeForce 400 series introduced the Fermi microarchitecture, named after Italian physicist Enrico Fermi.
- The GeForce 405, an OEM-only card, is based on the GT218 (Tesla) core and is the only GeForce 400 card not based on Fermi.
- AMD released a promotional video titled 'The Misunderstanding' poking fun at the high temperatures of Fermi GPUs.
More in Nvidia and AMD Graphics Processors 1-24
Spotted an error? Know more?
Reader corrections go straight into our review queue. Suggest an edit · How this site is sourced