GeForce 400 Series

Nvidia GeForce 400 Series
Codename(s)	GF100 / GF104 / GF106 / GF108 (Fermi)
Release date	April 12, 2010
Entry-level GPU	GT 430, GT 440
Mid-range GPU	GTS 450
High-end GPU	GTX 460, GTX 465, GTX 470, GTX 480
Direct3D and shader version	Direct3D 11 Shader Model 5.0
OpenGL version	OpenGL 4.2
OpenCL version	OpenCL 1.1
Predecessor	GeForce 300 Series
Successor	GeForce 500 Series

The GeForce 400 Series is the 11th generation of Nvidia's GeForce graphics processing units. The series was originally slated for production in November 2009,^[1] but, after a number of delays, launched on March 26, 2010 with availability following in April 2010.

Overview

Nvidia has given the architecture an internal name of Fermi, after the Italian physicist Enrico Fermi, a key developer of the nuclear reactor, who also gave his name to the Fermi acceleration mechanism in astrophysics. Nvidia claims that the Fermi architecture is the next major step in its line of GPUs following 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. At launch, no product was available with all the stream processors active: the GTX 480 has one group disabled, the GTX 470 has two groups and one memory controller disabled, and the GTX 465 has five groups and two memory controllers disabled. Consumer GeForce cards come 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 has 512MB on each of six controllers, and the Tesla C2070 has 1024MB per controller. Both the Tesla cards have fourteen active groups of stream processors.

In the more expensive "Tesla" configurations, the chip features optional ECC protection on the memory, and can perform one double-precision floating-point operation per cycle per core; the consumer GeForce cards are artificially driver restricted to one DP operation per four cycles. With these features, combined with support for Visual Studio and C++, Nvidia hopes to appeal to the High-Performance Computer users who might presently be using Tesla systems.^[2]

History

NVIDIA GeForce GTX 480

Fermi core chip on GTX 480 card

On 30 September 2009, Nvidia released a white paper describing the architecture:^[3] the chip features 16 'Streaming Multiprocessors' each with 32 'CUDA Cores' capable of one single-precision operation per cycle or one double-precision operation every other cycle, a 40-bit virtual address space which allows the host's memory to be mapped into the chip's address space, meaning that there is only one kind of pointer and making C++ support significantly easier, and a 384-bit wide GDDR5 memory interface. As with the G80 and GT200, threads are scheduled in 'warps', sets of 32 threads each running on a single shader core. While the GT200 had 16 KB 'shared memory' associated with each shader cluster, and required data to be read through the texturing units if a cache was needed, GF100 has 64 KB of memory associated with each cluster, which can be used either as a 48 KB cache plus 16 KB of shared memory, or as a 16 KB cache plus 48 KB of shared memory, along with a 768 KB L2 cache shared by all 16 clusters.

The white paper describes the chip much more as a general purpose processor for workloads encompassing tens of thousands of threads - reminiscent of the Tera MTA architecture, though without that machine's support for very efficient random memory access - than as a graphics processor.

Pre-launch statements

At a press event on January 7, 2010 at CES Jen-Hsun Huang said that the GF100 products were in production but did not give a shipping date.^[4]

On January 18, 2010, Nvidia released the GF100 graphics architecture details through a white paper.^[5]

On February 2, 2010, Nvidia tweeted the official titles of the GF100 (Fermi) retail cards, the GeForce GTX 480 and the GeForce GTX 470.^[6]

February 22, 2010: According to Nvidia's twitter update, the Fermi based Geforce GTX 400 series will be "unveiled" at the PAX East 2010,^[7] in a later update Nvidia released the launch date of March 26, 2010 for the GTX 470 and GTX 480 to clear up confusion over the PAX announcement.^[8]

March 4, 2010: Tom Petersen at NVIDIA describes how the performance of GeForce GTX 480 compares to the Radeon HD 5870 in a single test case.^[9]

March 26, 2010: The complete architecture along with the GTX 470 and 480 were officially launched at PAX EAST.

April 7, 2010: Limited product availability started to show.^[10][11]

April 12, 2010: Official release date for most manufacturers, bar EVGA, which released its cards on April 7, 2010.

Current limitations and trade-offs

The quantity of on-board SRAM per ALU actually decreased proportionally compared to the previous G200 generation, despite the increase of the L2 cache from 256kB per 240 ALUs to 768kB per 512 ALUs, since Fermi has only 32768 registers per 32 ALUs (vs. 16384 per 8 ALUs), only 48kB of shared memory per 32 ALUs (vs. 16kB per 8 ALUs), and only 16kB of cache per 32 ALUs (vs. 8kB constant cache per 8 ALUs + 24kB texture cache per 24 ALUs). Parameters such as the number of registers can be found in the CUDA Compute Capability Comparison Table in the reference manual.^[12]

Products

• ¹ SPs - Shader Processors - Unified Shaders (Vertex shader / Geometry shader / Pixel shader) : TMUs - Texture mapping units : Render Output unit
• ² Each Streaming Multiprocessor(SM) in the GPU of GF100 architecture contains 32 SPs and 4 SFUs. Each Streaming Multiprocessor(SM) in the GPU of GF104/106/108 architecture contains 48 SPs and 8 SFUs. Each SP can fulfil one single precision fused multiply–add (FMA) operation per cycle. Each SFU can fulfil four SF operations per cycle. One FMA operation counts for two floating point operations. So the theoretical single precision peak performance, with shader count [n] and shader frequency [f, GHz], can be estimated by the following, FLOPS_sp ≈ f × n × 2 (FMA). Total Processing Power: for GF100 FLOPS_sp ≈ f × m ×(32 SPs × 2(FMA) + 4 × 4 SFUs) and for GF104/106/108 FLOPS_sp ≈ f × m × (48 SPs × 2(FMA) + 4 × 8 SFUs) or for GF100 FLOPS_sp ≈ f × n × 2.5 and for GF104/106/108 FLOPS_sp ≈ f × n × 8 / 3.^[13]

SP - Shader Processor (Unified Shader, CUDA Core), SFU - Special Function Unit, SM - Streaming Multiprocessor.

• ³ Each SM in the GF100 contains 4 texture filtering units for every texture address unit. The complete GF100 die contains 64 texture address units and 256 texture filtering units^[14] Each SM in the GF104/106/108 architecture contains 8 texture filtering units for every texture address unit. The complete GF104 die contains 64 texture address units and 512 texture filtering units, the complete GF106 die contains 32 texture address units and 256 texture filtering units and the complete GF108 die contains 16 texture address units and 128 texture filtering units.^[15]

Model	Year	Code name	Fab (nm)	Transistors (million)	Die size (mm2)	Die Count	Bus interface	Memory (MiB)	SM count	Config core 1,3	Clock rate			Fillrate		Memory configuration			API support (version)			GFLOPs (FMA)2	TDP (watts)	Release price (USD)
											Core (MHz)	Shader (MHz)	Memory (MHz)	Pixel (GP/s)	Texture (GT/s)	Bandwidth (GB/s)	DRAM type	Bus width (bit)	DirectX	OpenGL	OpenCL
GeForce 405 (OEM)	September 16, 2011	?	40	?	?	1	PCIe 2.0 x16	512 1024	1	?	790	1402	1580	?	?	12.6	GDDR3	64	11	4.2	1.1	?	25	OEM
GeForce GT 420 (OEM)	September 3, 2010	GF108	40	585	116	1	PCIe 2.0 x16	2048	1	48:8:4	700	1400	1800	2.8	5.6	28.8	GDDR3	128	11	4.2	1.1	134.4	50	OEM
GeForce GT 430 (OEM)	October 11, 2010	GF108	40	585	116	1	PCIe 2.0 x16	2048	2	96:16:4	700	1400	1600 1800	2.8	11.2	25.6 28.8	GDDR3	128	11	4.2	1.1	268.8	60	OEM
GeForce GT 430	October 11, 2010	GF108	40	585	116	1	PCIe 2.0 x16	1024	2	96:16:4	700	1400	1800	2.8	11.2	28.8	GDDR3	128	11	4.2	1.1	268.8	49	$79
GeForce GT 440	February 1, 2011	GF108	40	585	116	1	PCIe 2.0 x16	512 1024 2048	2	96:16:4	810	1620	1800 3200	3.24	12.96	28.8 51.2	GDDR3 GDDR5	128	11	4.2	1.1	311.04	65	$79
GeForce GT 440 (OEM)	October 11, 2010	GF106	40	1170	238	1	PCIe 2.0 x16	1536 3072	3	144:24:24	594	1189	1800	14.26	14.26	43.2	GDDR3	192	11	4.2	1.1	342.43	56	OEM
GeForce GTS 450 (OEM)	October 11, 2010	GF106	40	1170	238	1	PCIe 2.0 x16	1536	3	144:24:24	790	1580	1804	18.96	18.96	86	GDDR5	192	11	4.2	1.1	455.04	106	OEM
GeForce GTS 450	September 13, 2010	GF106	40	1170	238	1	PCIe 2.0 x16	512 1024	4	192:32:16	783	1566	1804	12.53	25.06	57.73	GDDR5	128	11	4.2	1.1	601.34	106	$129
GeForce GTX 460 SE	November 15, 2010	GF104	40	1950	332	1	PCIe 2.0 x16	1024	6	288:48:32	650	1300	1700	20.8	31.2	108.8	GDDR5	256	11	4.2	1.1	748.8	150	$160?-$180?
GeForce GTX 460 (OEM)	October 11, 2010	GF104	40	1950	332	1	PCIe 2.0 x16	1024	7	336:56:32	650	1300	3400	20.8	36.4	108.8	GDDR5	256	11	4.2	1.1	873.6	150	OEM
GeForce GTX 460	July 12, 2010	GF104	40	1950	332	1	PCIe 2.0 x16	768	7	336:56:24	675	1350	3600	16.2	37.8	86.4	GDDR5	192	11	4.2	1.1	907.2	150	$199
								1024		336:56:32				21.6		115.2		256					160	$229
GeForce GTX 465	May 31, 2010	GF100	40	3200	529	1	PCIe 2.0 x16	1024	11	352:44:32	607	1215	3206	19.42	26.71	102.6	GDDR5	256	11	4.2	1.1	855.36	200	$279
GeForce GTX 470	March 26, 2010	GF100	40	3200	529	1	PCIe 2.0 x16	1280	14	448:56:40	607	1215	3348	24.28	34	133.9	GDDR5	320	11	4.2	1.1	1088.64	215	$349
GeForce GTX 480	March 26, 2010	GF100	40	3200	529	1	PCIe 2.0 x16	1536	15	480:60:48	700	1401	3696	33.60	42	177.4	GDDR5	384	11	4.2	1.1	1344.96	250	$499

As of November 8, 2010, Nvidia released the GF110 chip, along with the GTX580 (480's replacement). It is a redesigned GF100 chip, which uses significantly less power. This allows Nvidia to enable all 16 SMs (all 16 cores), which was previously impossible on the GF100 "NVIDIA GeForce GTX 580". http://www.nvidia.com/object/product-geforce-gtx-580-us.html. . Various features of the GF100 architecture remain available only on the more expensive Tesla series of cards.^[16] For consumer products, double precision performance has been limited to a quarter of that of the "full" Fermi architecture. Error checking and correcting memory (ECC) is also disabled on consumer cards.^[17] The GF100 cards provide Compute Capability 2.0, while the GF104/106/108 cards provide Compute Capability 2.1.

Chipset table

Main article: Comparison of Nvidia graphics processing units

See also

• Comparison table of GeForce 400 Series
• GeForce 8 Series
• GeForce 9 Series
• GeForce 200 Series
• GeForce 300 Series
• GeForce 500 Series
• Nvidia Quadro
• Nvidia Tesla

References

1. ^ "OFFICIAL: NVIDIA says GT300 on schedule for Q4 2009, yields are fine - Bright Side Of News*". Brightsideofnews.com. http://www.brightsideofnews.com/news/2009/9/25/nvidia-says-gt300-on-schedule-for-q4-20092c-yields-are-fine.aspx. Retrieved 2010-09-20. 
2. ^ "Next Generation CUDA Architecture, Code Named Fermi". Nvidia.com. http://www.nvidia.com/object/fermi_architecture.html. Retrieved 2010-09-20. 
3. ^ http://www.nvidia.com/content/PDF/fermi_white_papers/NVIDIA_Fermi_Compute_Architecture_Whitepaper.pdf
4. ^ "Live from NVIDIA's CES press event". Engadget. 2010-01-07. http://www.engadget.com/2010/01/07/live-from-nvidias-ces-press-event/. Retrieved 2010-09-20. 
5. ^ "The next generation of NVIDIA GeForce GPU". Nvidia.com. http://www.nvidia.com/object/gf100.html. Retrieved 2010-09-20. 
6. ^ "Twitter / NVIDIA GeForce: Fun Fact of the Week: GeFo ...". http://twitter.com/NVIDIAGeForce/status/8523402386. 
7. ^ "Twitter / NVIDIA GeForce: The wait is almost over! T ...". http://twitter.com/NVIDIAGeForce/status/9483846060. 
8. ^ "Twitter / NVIDIA GeForce: Apologies for the confusion ...". http://twitter.com/NVIDIAGeForce/status/9503345711. 
9. ^ "YouTube / GTX 480 Unigine and 3D Vision Surround Demo (GF100)". http://www.youtube.com/watch?v=vpdPSZB8A8E. 
10. ^ "NVIDIA’s GeForce GTX 400 Series Shows Up "Early"". http://www.anandtech.com/show/3642/nvidias-geforce-gtx-400-series-shows-up-early. 
11. ^ "http://www.evga.com/forums/tm.aspx?m=284103". http://www.evga.com/forums/tm.aspx?m=284103. 
12. ^ Compute Capability Comparison Table in "Page 147-148, Appendix G.1, CUDA 3.1 official reference manual". http://developer.download.nvidia.com/compute/cuda/3_1/toolkit/docs/NVIDIA_CUDA_C_ProgrammingGuide_3.1.pdf. . Page 97 in Appendix A lists the older NVIDIA GPUs and shows all G200 series to be compute capability 1.3, while Fermi-based cards have compute capability 2.x (page 14, Section 2.5).
13. ^ siliconmadness.com (2010). "Nvidia Announces Tesla 20 Series". http://www.siliconmadness.com/2009/11/nvidia-announces-tesla-20-series.html. 
14. ^ NVIDIA's GeForce GTX 480 and GTX 470: 6 Months Late, Was It Worth the Wait?
15. ^ NVIDIA’s GeForce GTX 460: The $200 King
16. ^ "Statement by NVIDIA on their General CUDA GPU Computing Discussion forum". http://forums.nvidia.com/index.php?showtopic=165055. 
17. ^ "NVIDIA Tesla C2xxx webpage". http://www.nvidia.com/object/product_tesla_C2050_C2070_us.html. , note from the description one may infer that on Teslas, ECC may be switched on and off using 1/8 of existing on-board memory, unlike standard ECC memory modules which requires 1/8 extra memory chips (that is, one extra chip to be mounted on the printed circuit board for every 8).

External links

• The Next Generation of Nvidia GeForce
• Fermi architecture
• GTX 400 Overview
• GeForce GTX 480
• GeForce GTX 470
• GeForce GTX 465
• GeForce GTX 460
• GeForce GTS 450
• GeForce GT 440
• GeForce GT 430
• GeForce GTX 485M
• GeForce GTX 480M
• GeForce GTX 470M
• GeForce GTX 460M
• GeForce GT 445M
• GeForce GT 435M
• GeForce GT 425M
• GeForce GT 420M
• GeForce GT 415M
• GeForce 410M
• GeForce 405
• Nvidia Parallel Nsight
• David Kanter (September 30, 2009). "Inside Fermi: Nvidia's HPC Push". realworldtech.com. http://www.realworldtech.com/page.cfm?ArticleID=RWT093009110932. Retrieved December 16, 2010.