GeForce 600 series

For GeForce cards with a model number of 6XX0, see GeForce 6 series.

GeForce 600 series

Release date	March 22, 2012 (March 22, 2012)
Codename	GK10x
Architecture	Fermi Kepler
Models	GeForce series GeForce GT series GeForce GTX series
Transistors	292M 40 nm (GF119) 585M 40 nm (GF108) 1.170B 40 nm (GF116) 1.950B 40 nm (GF114) 1.270B 28 nm (GK107) 1.020B 28 nm (GK208) 2.540B 28 nm (GK106) 3.540B 28 nm (GK104)
Cards
Entry-level	GT 605 GT 610 GT 620 GT 630 GT 640
Mid-range	GTX 650 GTX 650 Ti GTX 650 Ti Boost GTX 660
High-end	GTX 660 Ti GTX 670 GTX 680
Enthusiast	GTX 690
API support
DirectX	Direct3D 12.0 (feature level 11_0) Shader Model 6.5
OpenCL	OpenCL 3.0
OpenGL	OpenGL 4.6
Vulkan	Vulkan 1.2 SPIR-V
History
Predecessor	GeForce 500 series
Successor	GeForce 800M series GeForce 700 series
Support status
Fermi cards unsupported Security updates for Kepler until September 2024

The GeForce 600 series is a series of graphics processing units developed by Nvidia, first released in 2012. It served as the introduction of the Kepler architecture. It is succeeded by the GeForce 700 series.

Overview

Where the goal of the previous architecture, Fermi, was to increase raw performance (particularly for compute and tessellation), Nvidia's goal with the Kepler architecture was to increase performance per watt, while still striving for overall performance increases. The primary way Nvidia achieved this goal was through the use of a unified clock. By abandoning the shader clock found in their previous GPU designs, efficiency is increased, even though it requires more cores to achieve similar levels of performance. This is not only because the cores are more power efficient (two Kepler cores using about 90% of the power of one Fermi core, according to Nvidia's numbers), but also because the reduction in clock speed delivers a 50% reduction in power consumption in that area.

Kepler also introduced a new form of texture handling known as bindless textures. Previously, textures needed to be bound by the CPU to a particular slot in a fixed-size table before the GPU could reference them. This led to two limitations: one was that because the table was fixed in size, there could only be as many textures in use at one time as could fit in this table (128). The second was that the CPU was doing unnecessary work: it had to load each texture, and also bind each texture loaded in memory to a slot in the binding table. With bindless textures, both limitations are removed. The GPU can access any texture loaded into memory, increasing the number of available textures and removing the performance penalty of binding.

Finally, with Kepler, Nvidia was able to increase the memory clock to 6 GHz. To accomplish this, Nvidia needed to design an entirely new memory controller and bus. While still shy of the theoretical 7 GHz limitation of GDDR5, this is well above the 4 GHz speed of the memory controller for Fermi.

Kepler is named after the German mathematician, astronomer, and astrologer Johannes Kepler.

Architecture

Main articles: Fermi (microarchitecture) and Kepler (microarchitecture)

The GeForce 600 series contains products from both the older Fermi and newer Kepler generations of Nvidia GPUs. Kepler based members of the 600 series add the following standard features to the GeForce family:

• PCI Express 3.0 interface
• DisplayPort 1.2
• HDMI 1.4a 4K x 2K video output
• Purevideo VP5 hardware video acceleration (up to 4K x 2K H.264 decode)
• Hardware H.264 encoding acceleration block (NVENC)
• Support for up to 4 independent 2D displays, or 3 stereoscopic/3D displays (NV Surround)
• Next Generation Streaming Multiprocessor (SMX)
• A New Instruction Scheduler
• Bindless Textures
• CUDA Compute Capability 3.0
• GPU Boost
• TXAA
• Manufactured by TSMC on a 28 nm process

Streaming Multiprocessor Architecture (SMX)

The Kepler architecture employs a new Streaming Multiprocessor Architecture called SMX. The SMX are the key method for Kepler's power efficiency as the whole GPU uses a single "Core Clock" rather than the double-pump "Shader Clock". The SMX usage of a single unified clock increases the GPU power efficiency due to the fact that two Kepler CUDA Cores consume 90% power of one Fermi CUDA Core. Consequently, the SMX needs additional processing units to execute a whole warp per cycle. Kepler also needed to increase raw GPU performance as to remain competitive. As a result, it doubled the CUDA Cores from 16 to 32 per CUDA array, 3 CUDA Cores Array to 6 CUDA Cores Array, 1 load/store and 1 SFU group to 2 load/store and 2 SFU group. The GPU processing resources are also double. From 2 warp schedulers to 4 warp schedulers, 4 dispatch unit became 8 and the register file doubled to 64K entries as to increase performance. With the doubling of GPU processing units and resources increasing the usage of die spaces, The capability of the PolyMorph Engine aren't double but enhanced, making it capable of spurring out a polygon in 2 cycles instead of 4. With Kepler, Nvidia not only worked on power efficiency but also on area efficiency. Therefore, Nvidia opted to use eight dedicated FP64 CUDA cores in a SMX as to save die space, while still offering FP64 capabilities since all Kepler CUDA cores are not FP64 capable. With the improvement Nvidia made on Kepler, the results include an increase in GPU graphic performance while downplaying FP64 performance.

A new instruction scheduler

Additional die areas are acquired by replacing the complex hardware scheduler with a simple software scheduler. With software scheduling, warps scheduling was moved to Nvidia's compiler and as the GPU math pipeline now has a fixed latency, it now include the utilization of instruction-level parallelism and superscalar execution in addition to thread-level parallelism. As instructions are statically scheduled, scheduling inside a warp becomes redundant since the latency of the math pipeline is already known. This resulted an increase in die area space and power efficiency.

GPU Boost

GPU Boost is a new feature which is roughly analogous to turbo boosting of a CPU. The GPU is always guaranteed to run at a minimum clock speed, referred to as the "base clock". This clock speed is set to the level which will ensure that the GPU stays within TDP specifications, even at maximum loads. When loads are lower, however, there is room for the clock speed to be increased without exceeding the TDP. In these scenarios, GPU Boost will gradually increase the clock speed in steps, until the GPU reaches a predefined power target (which is 170W by default). By taking this approach, the GPU will ramp its clock up or down dynamically, so that it is providing the maximum amount of speed possible while remaining within TDP specifications.

The power target, as well as the size of the clock increase steps that the GPU will take, are both adjustable via third-party utilities and provide a means of overclocking Kepler-based cards.

Microsoft DirectX support

Both Fermi and Kepler based cards support Direct3D 11, both also support Direct3D 12, though not all features provided by the API.

TXAA

Exclusive to Kepler GPUs, TXAA is a new anti-aliasing method from Nvidia that is designed for direct implementation into game engines. TXAA is based on the MSAA technique and custom resolve filters. Its design addresses a key problem in games known as shimmering or temporal aliasing; TXAA resolves that by smoothing out the scene in motion, making sure that any in-game scene is being cleared of any aliasing and shimmering.

NVENC

Main article: Nvidia NVENC

NVENC is Nvidia's SIP block that performs video encoding, in a way similar to Intel's Quick Sync Video and AMD's VCE. NVENC is a power-efficient fixed-function pipeline that is able to take codecs, decode, preprocess, and encode H.264-based content. NVENC specification input formats are limited to H.264 output. But still, NVENC, through its limited format, can perform encoding in resolutions up to 4096×4096.

Like Intel's Quick Sync, NVENC is currently exposed through a proprietary API, though Nvidia does have plans to provide NVENC usage through CUDA.

New driver features

In the R300 drivers, released alongside the GTX 680, Nvidia introduced a new feature called Adaptive VSync. This feature is intended to combat the limitation of v-sync that, when the framerate drops below 60 FPS, there is stuttering as the v-sync rate is reduced to 30 FPS, then down to further factors of 60 if needed. However, when the framerate is below 60 FPS, there is no need for v-sync as the monitor will be able to display the frames as they are ready. To address this issue (while still maintaining the advantages of v-sync with respect to screen tearing), Adaptive VSync can be turned on in the driver control panel. It will enable VSync if the framerate is at or above 60 FPS, while disabling it if the framerate lowers. Nvidia claims that this will result in a smoother overall display.

While the feature debuted alongside the GTX 680, this feature is available to users of older Nvidia cards who install the updated drivers.

Dynamic Super Resolution (DSR) was added to Fermi and Kepler GPUs with an October 2014 release of Nvidia drivers. This feature aims at increasing the quality of displayed picture, by rendering the scenery at a higher and more detailed resolution (upscaling), and scaling it down to match the monitor's native resolution (downsampling).

History

In September 2010, Nvidia first announced Kepler.

In early 2012, details of the first members of the 600 series parts emerged. These initial members were entry-level laptop GPUs sourced from the older Fermi architecture.

On March 22, 2012, Nvidia unveiled the 600 series GPU: the GTX 680 for desktop PCs and the GeForce GT 640M, GT 650M, and GTX 660M for notebook/laptop PCs.

On April 29, 2012, the GTX 690 was announced as the first dual-GPU Kepler product.

On May 10, 2012, the GTX 670 was officially announced.

On June 4, 2012, the GTX 680M was officially announced.

On August 16, 2012, the GTX 660 Ti was officially announced.

On September 13, 2012, the GTX 660 and GTX 650 were officially announced.

On October 9, 2012, the GTX 650 Ti was officially announced.

On March 26, 2013, the GTX 650 Ti BOOST was officially announced.

Products

GeForce 600 (6xx) series

EVGA GeForce GTX 650 Ti

• ¹ SPs – Shader Processors – Unified Shaders : Texture mapping units : Render output units
• ² The GeForce 605 (OEM) card is a rebranded GeForce 510.
• ³ The GeForce GT 610 card is a rebranded GeForce GT 520.
• ⁴ The GeForce GT 620 (OEM) card is a rebranded GeForce GT 520.
• ⁵ The GeForce GT 620 card is a rebranded GeForce GT 530.
• ⁶ This revision of GeForce GT 630 (DDR3) card is a rebranded GeForce GT 440 (DDR3).
• ⁷ The GeForce GT 630 (GDDR5) card is a rebranded GeForce GT 440 (GDDR5).
• ⁸ The GeForce GT 640 (OEM) card is a rebranded GeForce GT 545 (DDR3).
• ⁹ The GeForce GT 645 (OEM) card is a rebranded GeForce GTX 560 SE.

GeForce 600M (6xxM) series

The GeForce 600M series for notebooks architecture. The processing power is obtained by multiplying shader clock speed, the number of cores and how many instructions the cores are capable of performing per cycle.

• ¹ Unified Shaders: Texture mapping units: Render output units

Model	Launch	Code Name	Fab (nm)	Bus interface	Core Configuration1	Clock Speed			Fillrate		Memory				API Support (version)				Processing Power2 (GFLOPS)	TDP (Watts)	Notes
						Core (MHz)	Shader (MHz)	Memory (MT/s)	Pixel (GP/s)	Texture (GT/s)	Size (MiB)	Bandwidth (GB/s)	DRAM Type	Bus Width (bit)	DirectX	OpenGL	OpenCL	Vulkan
GeForce 610M	Dec 2011	GF119 (N13M-GE)	40	PCIe 2.0 x16	48:8:4	450	900	1800	3.6	7.2	1024 2048	14.4	DDR3	64	12.0 (11_0)	4.6	1.1	—	142.08	12	OEM. Rebadged GT 520MX
GeForce GT 620M	Apr 2012	GF117 (N13M-GS)	28		96:16:4	625	1250	1800	2.5	10		14.4 28.8		64 128					240	15	OEM. Die-Shrink GF108
GeForce GT 625M	October 2012	GF117 (N13M-GS)										14.4		64
GeForce GT 630M	Apr 2012	GF108 (N13P-GL) GF117	40 28			660 800	1320 1600	1800 4000	2.6 3.2	10.7 12.8		28.8 32.0	DDR3 GDDR5	128 64					258.0 307.2	33	GF108: OEM. Rebadged GT 540M GF117: OEM Die-Shrink GF108
GeForce GT 635M	Apr 2012	GF106 (N12E-GE2) GF116	40		144:24:24	675	1350	1800	16.2	16.2	2048 1536	28.8 43.2	DDR3	128 192					289.2 388.8	35	GF106: OEM. Rebadged GT 555M GF116: 144 Unified Shaders
GeForce GT 640M LE	March 22, 2012	GF108 GK107 (N13P-LP)	40 28	PCIe 2.0 x16 PCIe 3.0 x16	96:16:4 384:32:16	762 500	1524 500	3130 1800	3 8	12.2 16	1024 2048	50.2 28.8	GDDR5 DDR3	128			1.1 1.2	N/A ?	292.6 384	32 20	GF108: Fermi GK107: Kepler architecture
GeForce GT 640M	March 22, 2012	GK107 (N13P-GS)	28	PCIe 3.0 x16	384:32:16	625	625	1800 4000	10	20		28.8 64.0	DDR3 GDDR5				1.2	1.1	480	32	Kepler architecture
GeForce GT 645M	October 2012	GK107 (N13P-GS)				710	710	1800 4000	11.36	22.72									545
GeForce GT 650M	March 22, 2012	GK107 (N13P-GT)				835 745 900*	950 835 900*	1800 4000 5000*	15.2 13.4 14.4*	30.4 26.7 28.8*	1024 2048 *	28.8 64.0 80.0*	DDR3 GDDR5 GDDR5*						729.6 641.3 691.2*	45	Kepler architecture
GeForce GTX 660M	March 22, 2012	GK107 (N13E-GE)				835	950	5000	15.2	30.4	2048	80.0	GDDR5						729.6	50	Kepler architecture
GeForce GTX 670M	April 2012	GF114 (N13E-GS1-LP)	40	PCIe 2.0 x16	336:56:24	598	1196	3000	14.35	33.5	1536 3072	72.0		192			1.1	—	803.6	75	OEM. Rebadged GTX 570M
GeForce GTX 670MX	October 2012	GK106 (N13E-GR)	28	PCIe 3.0 x16	960:80:24	600	600	2800	14.4	48.0		67.2					1.2	1.1	1152		Kepler architecture
GeForce GTX 675M	April 2012	GF114 (N13E-GS1)	40	PCIe 2.0 x16	384:64:32	620	1240	3000	19.8	39.7	2048	96.0		256			1.1	?	952.3	100	OEM. Rebadged GTX 580M
GeForce GTX 675MX	October 2012	GK106 (N13E-GSR)	28	PCIe 3.0 x16	960:80:32	600	600	3600	19.2	48.0	4096	115.2					1.2	1.1	1152		Kepler architecture
GeForce GTX 680M	June 4, 2012	GK104 (N13E-GTX)			1344:112:32	720	720	3600	23	80.6									1935.4
GeForce GTX 680MX	October 23, 2012	GK104			1536:128:32			5000		92.2		160							2234.3	100+
Model	Launch	Code Name	Fab (nm)	Bus interface	Core Configuration1	Clock Speed			Fillrate		Memory				API Support (version)				Processing Power2 (GFLOPS)	TDP (Watts)	Notes
						Core (MHz)	Shader (MHz)	Memory (MT/s)	Pixel (GP/s)	Texture (GT/s)	Size (MiB)	Bandwidth (GB/s)	DRAM Type	Bus Width (bit)	DirectX	OpenGL	OpenCL	Vulkan

(*)-Apple MacBook Pro Retina 2012 with 512MB or 1024MB GDDR5 configuration.

Chipset table

GeForce 600 (6xx) series

See also: Comparison table of GeForce 600 series

Model	Launch	Code name	Fab (nm)	Transistors (million)	Die size (mm2)	Bus interface	SM count	Core config	Clock rate					Fillrate		Memory configuration				Supported API version				Processing power (GFLOPS)		TDP (Watts)	Release Price (USD)
									Core (MHz)	Average Boost (MHz)	Max Boost (MHz)	Shader (MHz)	Memory (MHz)	Pixel (GP/s)	Texture (GT/s)	Size (MB)	Bandwidth (GB/s)	DRAM type	Bus width (bit)	Vulkan	Direct3D	OpenGL	OpenCL	Single precision	Double precision
GeForce 605	April 3, 2012	GF119	TSMC 40 nm	292	79	PCIe 2.0 x16	1	48:8:4	523	—	—	1046	898 (1796)	2.09	4.2	512 1024	14.4	DDR3	64	—	12	4.6	1.2	100.4	Un­known	25	OEM
GeForce GT 610	May 15, 2012	GF119-300-A1				PCIe 2.0 x16, PCIe x1, PCI		48:8:4	810	—	—	1620	1000 1800	3.24	6.5	512 1024 2048	8 14.4							155.5	Un­known	29	Retail
GeForce GT 620	April 3, 2012	GF119				PCIe 2.0 x16		48:8:4		—	—		898 (1796)		6.5	512 1024	14.4							155.5	Un­known	30	OEM
	May 15, 2012	GF108-100-KB-A1		585	116		2	96:16:4	700	—	—	1400	1000–1800	2.8	11.2	1024 2048	8–14.4							268.8	Un­known	49	Retail
GeForce GT 625	February 19, 2013	GF119		292	79		1	48:8:4	810	—	—	1620	898 (1796)	3.24	6.5	512 1024	14.4							155.5	Un­known	30	OEM
GeForce GT 630	April 24, 2012	GK107	TSMC 28 nm	1300	118	PCIe 3.0 x16		192:16:16	875	—	—	875	891 (1782)	14	14	1024 2048	28.5		128	1.2				336	14	50
	May 15, 2012	GF108-400-A1	TSMC 40 nm	585	116	PCIe 2.0 x16	2	96:16:4	700	—	—	1620	1600–1800	2.8	11.2	1024 2048 4096	25.6–28.8			—				311	Un­known	49	Retail
		GF108						96:16:4	810	—	—	1620	800 (3200)	3.2	13	1024	51.2	GDDR5						311	Un­known	65
	May 29, 2013	GK208-301-A1	TSMC 28 nm	1020	79	PCIe 2.0 x8	1	384:16:8	902	—	—	902	900 (1800)	7.22	14.44	1024 2048	14.4	DDR3	64	1.2				692.7	Un­known	25
GeForce GT 635	February 19, 2013	GK208				PCIe 3.0 x8		384:16:8	967	—	—	967	1001 (2002)	7.74	15.5		16							742.7	Un­known	35	OEM
GeForce GT 640	April 24, 2012	GF116	TSMC 40 nm	1170	238	PCIe 2.0 x16	3	144:24:24	720	—	—	1440	891 (1782)	17.3	17.3	1536 3072	42.8		192	—				414.7	Un­known	75
		GK107	TSMC 28 nm	1300	118	PCIe 3.0 x16	2	384:32:16	797	—	—	797	891 (1782)	12.8	25.5	1024 2048	28.5		128	1.2				612.1	25.50	50
	June 5, 2012								900	—	—	900	891 (1782)	14.4	28.8	2048 4096	28.5							691.2	28.8	65	$100
	April 24, 2012								950	—	—	950	1250 (5000)	15.2	30.4	1024 2048	80	GDDR5						729.6	30.40	75	OEM
	May 29, 2013	GK208-400-A1	TSMC 28 nm	1020	79	PCIe 2.0 x8		384:16:8	1046	—	—	1046	1252 (5008)	8.37	16.7	1024	40.1		64					803.3	Un­known	49
GeForce GT 645	April 24, 2012	GF114-400-A1	TSMC 40 nm	1950	332	PCIe 2.0 x16	6	288:48:24	776	—	—	1552	1914	18.6	37.3		91.9		192	—				894	Un­known	140	OEM
GeForce GTX 645	April 22, 2013	GK106	TSMC 28 nm	2540	221	PCIe 3.0 x16	3	576:48:16	823.5	888.5	—	823	1000 (4000)	14.16	39.5		64		128	1.2				948.1	39.53	64
GeForce GTX 650	September 13, 2012	GK107-450-A2		1300	118		2	384:32:16	1058	—	—	1058	1250 (5000)	16.9	33.8	1024 2048	80							812.54	33.86		$110
	November 27, 2013	GK-106-400-A1		2540	221					—	65															?
GeForce GTX 650 Ti	October 9, 2012	GK106-220-A1					4	768:64:16	928	—	—	928	1350 (5400)	14.8	59.4		86.4							1425.41	59.39	110	$150 (130)
GeForce GTX 650 Ti Boost	March 26, 2013	GK106-240-A1						768:64:24	980	1032	—	980	1502 (6008)	23.5	62.7		144.2		192					1505.28	62.72	134	$170 (150)
GeForce GTX 660	September 13, 2012	GK106-400-A1					5	960:80:24			1084		1502 (6008)	23.5	78.4	1536+512 3072	96.1+48.1 144.2		128+64 192					1881.6	78.40	140	$230 (180)
	August 22, 2012	GK104-200-KD-A2		3540	294		6	1152:96:24 1152:96:32	823.5	888.5	899	823	1450 (5800)	19.8	79	1536 2048 3072	134 186		192 256					2108.6	79.06	130	OEM
GeForce GTX 660 Ti	August 16, 2012	GK104-300-KD-A2					7	1344:112:24	915	980	1058	915	1502 (6008)	22.0	102.5	2048	96.1+48.1 144.2		128+64 192					2459.52	102.48	150	$300
GeForce GTX 670	May 10, 2012	GK104-325-A2						1344:112:32			1084		1502 (6008)	29.3	102.5	2048 4096	192.256		256					2459.52	102.48	170	$400
GeForce GTX 680	March 22, 2012	GK104-400-A2					8	1536:128:32	1006	1058	1110	1006	1502 (6008)	32.2	128.8		192.256							3090.43	128.77	195	$500
GeForce GTX 690	April 29, 2012	2x GK104-355-A2		2x 3540	2x 294		2x 8	2x 1536:128:32	915	1019	1058	915	1502 (6008)	2x 29.28	2x 117.12	2x 2048	2x 192.256		2x 256					2x 2810.88	2x 117.12	300	$1000
Model	Launch	Code name	Fab (nm)	Transistors (million)	Die size (mm2)	Bus interface	SM count	Core config	Clock rate					Fillrate		Memory configuration				Supported API version				Processing power (GFLOPS)		TDP (Watts)	Release Price (USD)
									Core (MHz)	Average Boost (MHz)	Max Boost (MHz)	Shader (MHz)	Memory (MHz)	Pixel (GP/s)	Texture (GT/s)	Size (MB)	Bandwidth (GB/s)	DRAM type	Bus width (bit)	Vulkan	Direct3D	OpenGL	OpenCL	Single precision	Double precision

Discontinued support

Nvidia stopped releasing 32-bit drivers for 32-bit operating systems after the last Release 390 driver, 391.35, was released in March 2018.

Kepler notebook GPUs moved to legacy support in April 2019 and stopped receiving critical security updates in April 2020. Several notebook Geforce 6xxM GPUs were affected by this change, the remaining ones being low-end Fermi GPUs already out of support since January 2019.

Nvidia announced that after Release 470 drivers, it would transition driver support for the Windows 7 and Windows 8.1 operating systems to legacy status and continue to provide critical security updates for these operating systems through September 2024.

Nvidia announced that all remaining Kepler desktop GPUs would transition to legacy support from September 2021 onwards and be supported for critical security updates through September 2024. All remaining GeForce 6xx GPUs would be affected by this change.

See also

• List of Nvidia graphics processing units

• GeForce 400 series
• GeForce 500 series
• GeForce 700 series
• GeForce 800M series
• GeForce 900 series
• GeForce 10 series
• Nvidia Quadro
• Nvidia Tesla

External links

• Introducing the GeForce GTX 680 GPU
• Introducing The GeForce GTX 670 GPU
• Meet Your New Weapon: The GeForce GTX 660 Ti. Borderlands 2 Included.
• Kepler For Every Gamer: Meet The New GeForce GTX 660 & 650
• Kepler Whitepaper
• Introducing The GeForce GTX 680M Mobile GPU
• GeForce 600M Notebooks: Powerful and Efficient
• GeForce GTX 690
• GeForce GTX 680
• GeForce GTX 670
• GeForce GTX 660 Ti
• GeForce GTX 660
• GeForce GTX 650 Ti BOOST
• GeForce GTX 650 Ti
• GeForce GTX 650
• GeForce GT 640
• GeForce GTX 680MX
• GeForce GTX 680M
• GeForce GTX 675MX
• GeForce GTX 670MX
• GeForce GTX 660M
• GeForce GT 650M
• GeForce GT 645M
• GeForce GT 640M
• A New Dawn
• Nvidia Nsight
• techPowerUp! GPU Database