Английская Википедия:GeForce 900 series

Материал из Онлайн справочника
Перейти к навигацииПерейти к поиску

Шаблон:Short description Шаблон:Redirect Шаблон:For Шаблон:Use mdy dates Шаблон:Infobox GPU The GeForce 900 series is a family of graphics processing units developed by Nvidia, succeeding the GeForce 700 series and serving as the high-end introduction to the Maxwell microarchitecture, named after James Clerk Maxwell. They are produced with TSMC's 28 nm process.

With Maxwell, the successor to Kepler, Nvidia expected three major outcomes: improved graphics capabilities, simplified programming, and better energy efficiency compared to the GeForce 700 series and GeForce 600 series.[1]

Maxwell was announced in September 2010,[2] with the first Maxwell-based GeForce consumer-class products released in early 2014.[3]

Architecture

Шаблон:Main

Шаблон:AnchorFirst generation Maxwell (GM10x)

First generation Maxwell GM107/GM108 were released as GeForce GTX 745, GTX 750/750 Ti and GTX 850M/860M (GM107) and GT 830M/840M (GM108). These new chips provide few consumer-facing additional features; Nvidia instead focused on power efficiency. Nvidia increased the amount of L2 cache from 256 KiB on GK107 to 2 MiB on GM107, reducing the memory bandwidth needed. Accordingly, Nvidia cut the memory bus from 192 bit on GK106 to 128 bit on GM107, further saving power.[4] Nvidia also changed the streaming multiprocessor design from that of Kepler (SMX), naming it SMM. The structure of the warp scheduler is inherited from Kepler, which allows each scheduler to issue up to two instructions that are independent from each other and are in order from the same warp. The layout of SMM units is partitioned so that each of the 4 warp schedulers in an SMM controls 1 set of 32 FP32 CUDA cores, 1 set of 8 load/store units, and 1 set of 8 special function units. This is in contrast to Kepler, where each SMX has 4 schedulers that schedule to a shared pool of 6 sets of 32 FP32 CUDA cores, 2 sets of 16 load/store units, and 2 sets of 16 special function units.[5] These units are connected by a crossbar that uses power to allow the resources to be shared.[5] This crossbar is removed in Maxwell.[5] Texture units and FP64 CUDA cores are still shared.[4] SMM allows for a finer-grain allocation of resources than SMX, saving power when the workload isn't optimal for shared resources. Nvidia claims a 128 CUDA core SMM has 86% of the performance of a 192 CUDA core SMX.[4] Also, each Graphics Processing Cluster, or GPC, contains up to 4 SMX units in Kepler, and up to 5 SMM units in first generation Maxwell.[4]

GM107 supports CUDA Compute Capability 5.0 compared to 3.5 on GK110/GK208 GPUs and 3.0 on GK10x GPUs. Dynamic Parallelism and HyperQ, two features in GK110/GK208 GPUs, are also supported across the entire Maxwell product line.

Maxwell provides native shared memory atomic operations for 32-bit integers and native shared memory 32-bit and 64-bit compare-and-swap (CAS), which can be used to implement other atomic functions.

While it was once thought that Maxwell used tile-based immediate mode rasterization,[6] Nvidia corrected this at GDC 2017 saying Maxwell instead uses Tile Caching.[7]

NVENC

Шаблон:Main Maxwell-based GPUs also contain the NVENC SIP block introduced with Kepler. Nvidia's video encoder, NVENC, is 1.5 to 2 times faster than on Kepler-based GPUs meaning it can encode video at 6 to 8 times playback speed.[4]

PureVideo

Шаблон:Main Nvidia also claims an 8 to 10 times performance increase in PureVideo Feature Set E video decoding due to the video decoder cache paired with increases in memory efficiency. However, H.265 is not supported for full hardware decoding, relying on a mix of hardware and software decoding.[4] When decoding video, a new low power state "GC5" is used on Maxwell GPUs to conserve power.[4]

Second generation Maxwell (GM20x)

Second generation Maxwell introduced several new technologies: Dynamic Super Resolution,[8] Third Generation Delta Color Compression,[9] Multi-Pixel Programming Sampling,[10] Nvidia VXGI (Real-Time-Voxel-Global Illumination),[11] VR Direct,[12][13][14] Multi-Projection Acceleration,[9] and Multi-Frame Sampled Anti-Aliasing (MFAA)[15] (however support for Coverage-Sampling Anti-Aliasing (CSAA) was removed).[16] HDMI 2.0 support was also added.[17][18]

Second generation Maxwell also changed the ROP to memory controller ratio from 8:1 to 16:1.[19] However, some of the ROPs are generally idle in the GTX 970 because there are not enough enabled SMMs to give them work to do and therefore reduces its maximum fill rate.[20]

Second generation upgraded NVENC which supports HEVC encoding and adds support for H.264 encoding resolutions at 1440p/60FPS & 4K/60FPS compared to NVENC on Maxwell first generation GM10x GPUs which only supported H.264 1080p/60FPS encoding.[14]

Maxwell GM206 GPU supports full fixed function HEVC hardware decoding.[21][22]

Advertising controversy

GTX 970 hardware specifications

Issues with the GeForce GTX 970's specifications were first brought up by users when they found out that the cards, while featuring 4 GB of memory, rarely accessed memory over the 3.5 GB boundary. Further testing and investigation eventually led to Nvidia issuing a statement that the card's initially announced specifications had been altered without notice before the card was made commercially available, and that the card took a performance hit once memory over the 3.5 GB limit were put into use.[23][24][25]

The card's back-end hardware specifications, initially announced as being identical to those of the GeForce GTX 980, differed in the amount of L2 cache (1.75 MB versus 2 MB in the GeForce GTX 980) and the number of ROPs (56 versus 64 in the 980). Additionally, it was revealed that the card was designed to access its memory as a 3.5 GB section, plus a 0.5 GB one, access to the latter being 7 times slower than the first one.[26] The company then went on to promise a specific driver modification in order to alleviate the performance issues produced by the cutbacks suffered by the card.[27] However, Nvidia later clarified that the promise had been a miscommunication and there would be no specific driver update for the GTX 970.[28] Nvidia claimed that it would assist customers who wanted refunds in obtaining them.[29] On February 26, 2015, Nvidia CEO Jen-Hsun Huang went on record in Nvidia's official blog to apologize for the incident.[30] In February 2015 a class-action lawsuit alleging false advertising was filed against Nvidia and Gigabyte Technology in the U.S. District Court for Northern California.[31][32]

Nvidia revealed that it is able to disable individual units, each containing 256KB of L2 cache and 8 ROPs, without disabling whole memory controllers.[33] This comes at the cost of dividing the memory bus into high speed and low speed segments that cannot be accessed at the same time unless one segment is reading while the other segment is writing because the L2/ROP unit managing both of the GDDR5 controllers shares the read return channel and the write data bus between the two GDDR5 controllers and itself.[33] This is used in the GeForce GTX 970, which therefore can be described as having 3.5 GB in its high speed segment on a 224-bit bus and 0.5 GB in a low speed segment on a 32-bit bus.[33]

On July 27, 2016, Nvidia agreed to a preliminary settlement of the U.S. class action lawsuit,[31] offering a $30 refund on GTX 970 purchases. The agreed upon refund represents the portion of the cost of the storage and performance capabilities the consumers assumed they were obtaining when they purchased the card.[34]

Async compute support

Файл:GTXTITANX.jpg
GTX TITAN X

While the Maxwell series was marketed as fully DirectX 12 compliant,[35][36][37] Oxide Games, developer of Ashes of the Singularity, uncovered that Maxwell-based cards do not perform well when async compute is utilized.[38][39][40][36]

It appears that while this core feature is in fact exposed by the driver,[41] Nvidia partially implemented it through a driver-based shim, coming at a high performance cost.[40] Unlike AMD's competing GCN-based graphics cards which include a full implementation of hardware-based asynchronous compute,[42][43] Nvidia planned to rely on the driver to implement a software queue and a software distributor to forward asynchronous tasks to the hardware schedulers, capable of distributing the workload to the correct units.[44] Asynchronous compute on Maxwell therefore requires that both a game and the GPU driver be specifically coded for asynchronous compute on Maxwell in order to enable this capability.[45] The 3DMark Time Spy benchmark shows no noticeable performance difference between asynchronous compute being enabled or disabled.[45] Asynchronous compute is disabled by the driver for Maxwell.[45]

Oxide claims that this led to Nvidia pressuring them not to include the asynchronous compute feature in their benchmark at all, so that the 900 series would not be at a disadvantage against AMD's products which implement asynchronous compute in hardware.[39]

Maxwell requires that the GPU be statically partitioned for asynchronous compute to allow tasks to run concurrently.[46] Each partition is assigned to a hardware queue. If any of the queues that are assigned to a partition empty out or are unable to submit work for any reason (e.g. a task in the queue must be delayed until a hazard is resolved), the partition and all of the resources in that partition reserved for that queue will idle.[46] Asynchronous compute therefore could easily hurt performance on Maxwell if it is not coded to work with Maxwell's static scheduler.[46] Furthermore, graphics tasks saturate Nvidia GPUs much more easily than they do to AMD's GCN-based GPUs which are much more heavily weighted towards compute, so Nvidia GPUs have fewer scheduling holes that could be filled by asynchronous compute than AMD's.[46] For these reasons, the driver forces a Maxwell GPU to place all tasks into one queue and execute each task in serial, and give each task the undivided resources of the GPU no matter whether or not each task can saturate the GPU or not.[46]

Products

GeForce 900M (9xxM) series

Some implementations may use different specifications.

Model Launch Code name Fab (nm) Transistors (million) Die size (mm2) Bus interface Core configШаблон:Efn Clock speeds Fillrate Memory API support (version) Processing power (GFLOPS) TDP (watts) SLI supportШаблон:Efn
Base core clock (MHz) Boost core clock (MHz) Memory (MT/s) Pixel (GP/s)Шаблон:Efn Texture (GT/s)Шаблон:Efn Size (MiB) Bandwidth (GB/s) Type Bus width (bit) DirectX OpenGL OpenCL Vulkan Single precisionШаблон:Efn Double precisionШаблон:Efn
GeForce 910M[47][48][49] Aug 18, 2015 GF117Шаблон:Efn 28 585 116 PCIe 3.0 x8 96:16:8 775 1550 1800 3.1 12.4 1024 14.4 GDDR3 64 12.0 (11_0)[50][51] 4.6 1.1 Шаблон:N/a 297.6 1/12 of SP 33 No
March 15, 2015 GK208 Шаблон:Unk 87 384:16:8 575 575 5.13 9.2 2048 1.2 1.1 441.6 18.4
GeForce 920M[52][53][54] March 13, 2015 GF117Шаблон:Efn 585 116 96:16:8 775 1550 3.1 12.4 1024 1.1 Шаблон:N/a 297.6 1/12 of SP
GK208 Шаблон:Unk 87 384:32:16 954 954 7.6 30.5 2048 1.2 1.1 732.7 22.9
GeForce 920MX[55][56] March 2016 GM108Шаблон:Efn 1870 148 256:24:8 1072 1176 8.58 25.7 2048 DDR3 GDDR5 549 1/32 of SP 16
GeForce 930M[57][58] March 13, 2015 384:24:8 928 941 7.4 22.3 2048 DDR3 712.7 22.3 33
GeForce 930MX[59][60] March 1, 2016 Шаблон:Unk Шаблон:Unk PCIe 3.0 x8 384:24:8 952 1020 2000 Шаблон:Unk Шаблон:Unk 2048 Шаблон:Unk DDR3 GDDR5 Шаблон:Unk Шаблон:Unk Шаблон:Unk Шаблон:Unk Шаблон:Unk Шаблон:Unk
GeForce 940M[61][62][63] March 13, 2015 GM107 1870 148 PCIe 3.0 x16 640:40:16 1029 1100 2002 16.5 41.2 2048 16 - 80.2 GDDR5 DDR3 128 1.2 1.1 1317 41.1 75 No
GM108Шаблон:Efn Шаблон:Unk Шаблон:Unk PCIe 3.0 x8 384:24:8 8.2 24.7 64 790.3 24.7 33
GeForce 940MX[64][65] March 10, 2016 1870 148 384:24:8 1122 1242 8.98 26.93 2048
4096 		16.02 (DDR3)
40.1 (GDDR5) 		861.7 Шаблон:Unk 23
GeForce 945M[66][67][68] 2015 GM107 Шаблон:Dunno 640:40:16 1029 1085 rowspan="2" Шаблон:Dunno 16.46 41.2 rowspan="2" Шаблон:Dunno rowspan="2" Шаблон:Dunno DDR3 GDDR5 128 1,317.1 rowspan="2" Шаблон:Dunno 75 rowspan="2" Шаблон:Dunno
GM108Шаблон:Efn Шаблон:Dunno Шаблон:Dunno PCIe 3.0 x8 384:24:8 1122 1242 8.98 26.93 64 861.7 23
GeForce GT 945A[69][70] March 13, 2015 Шаблон:Unk Шаблон:Unk 384:24:8 1072 1176 1800 8.58 25.73 2048 14.4 DDR3 Шаблон:Unk Шаблон:Unk Шаблон:Unk 33 Шаблон:Unk
GeForce GTX 950M[71][72] March 13, 2015 GM107 1870 148 PCIe 3.0 x16 640:40:16 914 1085 5012 14.6 36.6 2048(GDDR5)
4096(DDR3) 		80(GDDR5)
32(DDR3) 		DDR3 GDDR5 128 1.2[73] 1.1 1170 36.56 75 No
GeForce GTX 960M[74][75] 640:40:16 1029 1085 16.5 41.2 2048
4096 		80 GDDR5 1317 41.16 65
GeForce GTX 965M[76][77] January 5, 2015 GM204 5200 398 1024:64:32 924 950 5000 30.2 60.4 12.0 (12_1)[50][51] 1945 60.78 60 Yes
GeForce GTX 970M[78] October 7, 2014 1280:80:48 924 993 5012 37.0 73.9 3072
6144 		120 192 2365 73.9 75
GeForce GTX 980M[79] 1536:96:64 1038 1127 49.8 99.6 4096
8192 		160 256[80] 3189 99.6 100

Шаблон:Notelist

Chipset table

GeForce 900 (9xx) series

Шаблон:See also

Model Launch Code name Process Transistors (billion) Die size (mm2) Core configШаблон:Efn Bus interface L2 Cache
(MB) 		Clock Speeds Memory FillrateШаблон:Efn Processing power (GFLOPS)Шаблон:EfnШаблон:Efn TDP (Watts) SLI support Release price (USD)
Base (MHz) Boost (MHz) Memory (MT/s) Size (GB) Bandwidth (GB/s) Bus type Bus width (bit) Pixel (GP/s)Шаблон:Efn Texture (GT/s)Шаблон:Efn Single precision Double precision MSRP
GeForce GT 945A[81][82][83] February, 2016 GM108 TSMC
28HP 		Шаблон:Unk Шаблон:Unk 512:24:8 (4) PCIe 3.0 x8 ? 1072 1176 1800 1 / 2 14.4 DDR3 / GDDR5 64 8.5
9.4 		25.7
28.2 		1,097.7
1,204.2 		34.3
37.6 		33 rowspan="1" Шаблон:No Шаблон:Okay
GeForce GTX 950[84] August 20, 2015 GM206-250 2.94 227 768:48:32 (6) PCIe 3.0 x16 1 1024 1188 6600 2 105.7 GDDR5 128 32.7
38.0 		49.1
57.0 		1,572.8
1,824.7 		49.1
57.0 		90 (75Шаблон:Efn) 2-way SLI $159
GeForce GTX 950 (OEM)[85] Шаблон:Unk GM206 1024:64:32 (8) 935 Шаблон:Unk 5000 80.0 29.9
  		59.8
  		1,914.9
, 		59.8
  		Шаблон:Unk Шаблон:Okay
GeForce GTX 960[86] January 22, 2015 GM206-300 1127 1178 7000 2
4Шаблон:Efn 		112.1 36.0
37.6 		72.1
75.3 		2,308.0
2,412.5 		72.1
75.3 		120 $199
GeForce GTX 960 (OEM)[87] Шаблон:Unk GM204 5.2 398 1280:80:48 (10) 924 Шаблон:Unk 5000 3 120.0 192 44.3
  		73.9
  		2,365.4
, 		73.9
  		Шаблон:Unk Шаблон:Okay
GeForce GTX 970[88] September 18, 2014 GM204-200 1664:104:56 (13) 1.75 1050 1178 7000 3.5 +
0.5Шаблон:Efn 		196.3 +
28.0Шаблон:Efn 		224 +
32Шаблон:Efn 		58.8
65.9 		109.2
122.5 		3,494.4
3,920.3 		109.2
122.5 		145 4-way SLI $329
GeForce GTX 980[89] September 18, 2014 GM204-400 2048:128:64 (16) 2 1126 1216 4 224.3 256 72.0
77.8 		144.1
155.6 		4,612.0
4,980.7 		144.1
155.6 		165 $549
GeForce GTX 980 Ti[90] June 1, 2015 GM200-310 8 601 2816:176:96 (22) 3 1000 1075 6 336.5 384 96.0
103.2 		176.0
189.2 		5,632.0
6,054.4 		176.0
189.2 		250 $649
GeForce GTX TITAN X[91] March 17, 2015 GM200-400 3072:192:96 (24) 12 192.0
206.4 		6,144.0
6,604.8 		192.0
206.4 		$999

Шаблон:Notelist

Discontinued support

Шаблон:Cite web

32-bit drivers for 32-bit operating systems were discontinued after the release of driver 391.35 in March 2018.[92]

Notebook GPUs based on the Kepler architecture moved to legacy support in April 2019 and stopped receiving critical security updates after April 2020.[93][94] The Nvidia GeForce 910M and 920M from the 9xxM GPU family are affected by this change.

Nvidia announced that after release of the 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.[95]

See also

References

Шаблон:Reflist

External links

Шаблон:NVIDIA

Партнерские ресурсы
Криптовалюты
Магазины
Хостинг
Разное

  1. ↑ Шаблон:Cite web
  2. ↑ Шаблон:Cite web
  3. ↑ Шаблон:Cite web
  4. ↑ 4,0 4,1 4,2 4,3 4,4 4,5 4,6 Шаблон:Cite news
  5. ↑ 5,0 5,1 5,2 Шаблон:Cite web
  6. ↑ Шаблон:Cite news
  7. ↑ Шаблон:Cite news
  8. ↑ Шаблон:Cite web
  9. ↑ 9,0 9,1 Шаблон:Cite web
  10. ↑ Шаблон:Cite web
  11. ↑ Шаблон:Cite web
  12. ↑ Шаблон:Cite web
  13. ↑ Шаблон:Cite web
  14. ↑ 14,0 14,1 Шаблон:Cite web
  15. ↑ Шаблон:Cite web
  16. ↑ Шаблон:Cite web
  17. ↑ Шаблон:Cite web
  18. ↑ Шаблон:Cite web
  19. ↑ Шаблон:Cite web
  20. ↑ Шаблон:Cite web
  21. ↑ Шаблон:Cite web
  22. ↑ Шаблон:Cite web
  23. ↑ Шаблон:Cite web
  24. ↑ Шаблон:Cite web
  25. ↑ Шаблон:Cite web
  26. ↑ Шаблон:Cite web
  27. ↑ Шаблон:Cite web
  28. ↑ Шаблон:Cite web
  29. ↑ Шаблон:Cite web
  30. ↑ Шаблон:Cite web
  31. ↑ 31,0 31,1 Шаблон:Cite news
  32. ↑ Шаблон:Cite news
  33. ↑ 33,0 33,1 33,2 Шаблон:Cite web
  34. ↑ Шаблон:Cite news
  35. ↑ Ошибка цитирования Неверный тег <ref>; для сносок dx12_1 не указан текст
  36. ↑ 36,0 36,1 http://international.download.nvidia.com/geforce-com/international/images/nvidia-geforce-gtx-980-ti/nvidia-geforce-gtx-980-ti-directx-12-advanced-api-support.png Шаблон:Bare URL image
  37. ↑ Шаблон:Cite web
  38. ↑ Шаблон:Cite web
  39. ↑ 39,0 39,1 Шаблон:Cite web
  40. ↑ 40,0 40,1 Шаблон:Cite web
  41. ↑ Шаблон:Cite web
  42. ↑ Шаблон:Cite web
  43. ↑ Шаблон:Cite web
  44. ↑ Шаблон:Cite web
  45. ↑ 45,0 45,1 45,2 Шаблон:Cite news
  46. ↑ 46,0 46,1 46,2 46,3 46,4 Шаблон:Cite news
  47. ↑ Шаблон:Cite web
  48. ↑ Шаблон:Cite web
  49. ↑ Шаблон:Cite web
  50. ↑ 50,0 50,1 Ошибка цитирования Неверный тег <ref>; для сносок D3D11.3 не указан текст
  51. ↑ 51,0 51,1 Ошибка цитирования Неверный тег <ref>; для сносок 11.3And12.0RenderingFeaturesAreEquivalent не указан текст
  52. ↑ Шаблон:Cite web
  53. ↑ Шаблон:Cite web
  54. ↑ Шаблон:Cite web
  55. ↑ Шаблон:Cite web
  56. ↑ Шаблон:Cite web
  57. ↑ Шаблон:Cite web
  58. ↑ Шаблон:Cite web
  59. ↑ Шаблон:Cite web
  60. ↑ Шаблон:Cite web
  61. ↑ Шаблон:Cite web
  62. ↑ Шаблон:Cite web
  63. ↑ Шаблон:Cite web
  64. ↑ Шаблон:Cite web
  65. ↑ Шаблон:Cite web
  66. ↑ Шаблон:Cite web
  67. ↑ Шаблон:Cite web
  68. ↑ Шаблон:Cite web
  69. ↑ NVIDIA™ GeForceGT 945A (1GB GDDR5) user-selectable by application via NVIDIA Control Panel http://store.hp.com/us/en/ContentView?catalogId=10051&langId=-1&storeId=10151&eSpotName=Sprout-Pro#!
  70. ↑ Шаблон:Cite web
  71. ↑ Шаблон:Cite web
  72. ↑ Шаблон:Cite web
  73. ↑ Шаблон:Cite web
  74. ↑ Шаблон:Cite web
  75. ↑ Шаблон:Cite web
  76. ↑ Шаблон:Cite web
  77. ↑ Шаблон:Cite web
  78. ↑ Шаблон:Cite web
  79. ↑ Шаблон:Cite web
  80. ↑ Шаблон:Cite web
  81. ↑ Шаблон:Cite web
  82. ↑ Шаблон:Cite web
  83. ↑ Шаблон:Cite webШаблон:Dead linkШаблон:Cbignore
  84. ↑ Шаблон:Cite web
  85. ↑ Шаблон:Cite web
  86. ↑ Шаблон:Cite web
  87. ↑ Шаблон:Cite web
  88. ↑ Шаблон:Cite web
  89. ↑ Шаблон:Cite web
  90. ↑ Шаблон:Cite web
  91. ↑ Шаблон:Cite web
  92. ↑ Шаблон:Cite web
  93. ↑ Шаблон:Cite news
  94. ↑ Шаблон:Cite web
  95. ↑ Шаблон:Cite web
Источник — http://wikihandbk.com/ruwiki/index.php?title=Английская_Википедия:GeForce_900_series&oldid=14869744