GeForce 600 series
Introduced the Kepler architecture with a focus on performance per watt.
The GeForce 600 series, a line of graphics processing units from Nvidia, debuted in 2012 and marked the arrival of the Kepler architecture. This series includes GPUs based on both the older Fermi design and the newer Kepler design. Kepler itself is named after the German mathematician and astronomer Johannes Kepler.
Nvidia’s primary aim with Kepler was to improve performance per watt, rather than just raw performance as with the previous Fermi architecture. The key method was a unified clock: by dropping the separate shader clock used in earlier GPUs, efficiency rose. According to Nvidia, two Kepler cores use about 90% of the power of one Fermi core, and the lower clock speed cuts power consumption in that area by 50%. However, achieving similar performance levels requires more cores.
Kepler also introduced bindless textures. Previously, the CPU had to load each texture and bind it to one of 128 fixed slots before the GPU could use it. This limited the number of textures available at once and added unnecessary CPU work. With bindless textures, the GPU can access any texture already in memory, removing both the slot limit and the binding performance penalty.
The memory clock was increased to 6 GHz, requiring a new memory controller and bus. While still below the 7 GHz theoretical limit of GDDR5, this was a significant jump from the 4 GHz speed of Fermi’s memory controller.
Key features of Kepler-based members of the 600 series include: PCI Express 3.0, DisplayPort 1.2, HDMI 1.4a with 4K×2K output, PureVideo VP5 hardware video acceleration (up to 4K×2K H.264 decode), a hardware H.264 encoding block (NVENC), support for up to four independent 2D displays or three stereoscopic/3D displays (NV Surround), the new SMX streaming multiprocessor, a new instruction scheduler, bindless textures, CUDA Compute Capability 3.0, GPU Boost, TXAA, and fabrication by TSMC on a 28 nm process.
The SMX architecture uses a single unified core clock instead of a double-pumped shader clock, which is the main source of power efficiency. Because two Kepler CUDA cores consume 90% of the power of one Fermi core, the SMX needs extra processing units to execute a full warp per cycle. To stay competitive, Kepler doubled CUDA cores per array from 16 to 32, increased CUDA core arrays from three to six, and doubled load/store and SFU groups.
Quick Facts
- Codename
- GK10x
- Architecture
- Fermi · Kepler
- Created
- March 22, 2012
- Model
- GeForce series
- Transistors
- 292M 40 nm (GF119)
Facts from the source article.
Lore & Background
The GeForce 600 series debuted in 2012, with initial members being entry-level laptop GPUs sourced from the older Fermi architecture. On March 22, 2012, Nvidia unveiled the GTX 680 for desktop PCs and the GeForce GT 640M, GT 650M, and GTX 660M for notebook PCs. The GTX 690 was announced on April 29, 2012, as the first dual-GPU Kepler product, followed by the GTX 670 on May 10, 2012.
Kepler is named after the German mathematician, astronomer, and astrologer Johannes Kepler. The architecture introduced a unified clock, abandoning the shader clock of previous designs to increase efficiency. Two Kepler cores use about 90% of the power of one Fermi core, and the reduction in clock speed delivers a 50% reduction in power consumption in that area. Kepler also introduced bindless textures, which allow the GPU to access any texture loaded into memory, removing the limitations of a fixed-size binding table.
Key features of Kepler-based members include PCI Express 3.0, DisplayPort 1.2, HDMI 1.4a 4K x 2K output, Purevideo VP5 hardware video acceleration (up to 4K x 2K H.264 decode), hardware H.264 encoding (NVENC), support for up to 4 independent 2D displays or 3 stereoscopic displays, a new Streaming Multiprocessor Architecture (SMX), a new instruction scheduler, GPU Boost, TXAA, and CUDA Compute Capability 3.0.
Reader's Guide
The GeForce 600 series is significant as the vehicle for Nvidia's Kepler architecture, which marked a strategic shift from the raw performance focus of Fermi to a priority on performance per watt. This was achieved primarily through a unified clock design and a new Streaming Multiprocessor Architecture (SMX) that doubled CUDA cores per array while reducing power consumption. The series also introduced several lasting features: bindless textures, which improved texture handling efficiency; GPU Boost, a dynamic clock adjustment mechanism analogous to CPU turbo boosting; and TXAA, an anti-aliasing method exclusive to Kepler GPUs designed to reduce temporal aliasing. The inclusion of NVENC provided a fixed-function hardware video encoding block, and driver features such as Adaptive VSync and later Dynamic Super Resolution (DSR) expanded the user experience. The series also saw the first dual-GPU Kepler product, the GTX 690. While Kepler-based cards supported Direct3D 11 and Direct3D 12, not all features of the latter API were provided. The GeForce 600 series was succeeded by the GeForce 700 series.
Did You Know?
- The GeForce 600 series introduced the Kepler architecture, named after the German mathematician and astronomer Johannes Kepler.
- Two Kepler CUDA cores consume about 90% of the power of one Fermi CUDA core.
- GPU Boost dynamically increases clock speed within TDP limits, with a default power target of 170W.
- TXAA, exclusive to Kepler GPUs, is an anti-aliasing method designed to reduce shimmering and temporal aliasing in games.
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