601:
593:
585:
506:
838:
40:
860:
GP104: This GPU is used in the GeForce GTX 1070, GTX 1070 Ti, GTX 1080, and some GTX 1060 6 GB's. The GTX 1070 has 15/20 and the GTX 1070 Ti has 19/20 of its SMs enabled; both utilize GDDR5 memory. The GTX 1080 is a fully unlocked chip and uses GDDR5X memory. Some GTX 1060 6 GB's use GP104
731:
Instruction-level preemption. In graphics tasks, the driver restricts preemption to the pixel-level, because pixel tasks typically finish quickly and the overhead costs of doing pixel-level preemption are lower than instruction-level preemption (which is expensive). Compute tasks get thread-level or
809:
On GP104, 1 SM combines 128 single-precision ALUs, 4 double-precision ALUs (providing a 32:1 ratio), and one half-precision ALU which contains a vector of two half-precision floats which can execute the same instruction on both floats, providing a 64:1 ratio if the same instruction is used on both
670:
Dynamic load balancing scheduling system. This allows the scheduler to dynamically adjust the amount of the GPU assigned to multiple tasks, ensuring that the GPU remains saturated with work except when there is no more work that can safely be distributed to distribute. Nvidia therefore has safely
758:. An SM encompasses 128 single-precision ALUs ("CUDA cores") on GP104 chips and 64 single-precision ALUs on GP100 chips. While all CU versions consist of 64 shader processors (i.e. 4 SIMD Vector Units, each 16 lanes wide), Nvidia experimented with very different numbers of CUDA cores:
661:) floating-point operations (colloquially "half precision") can be executed at twice the rate of 32-bit floating-point operations ("single precision") and 64-bit floating-point operations (colloquially "double precision") executed at half the rate of 32-bit floating point operations.
632:
contained 128 CUDA cores per SM; Kepler had 192, Fermi 32 and Tesla 8. The GP100 SM is partitioned into two processing blocks, each having 32 single-precision CUDA cores, an instruction buffer, a warp scheduler, 2 texture mapping units and 2 dispatch
801:
providing a 2:1 ratio of single- to double-precision throughput. The GP100 uses more flexible FP32 cores that are able to process one single-precision or two half-precision numbers in a two-element vector. This is intended to better serve
600:
612:
would be the Pascal microarchitecture; announced on May 6, 2016, and released on May 27 of the same year. The Tesla P100 (GP100 chip) has a different version of the Pascal architecture compared to the GTX GPUs (GP104 chip). The
732:
instruction-level preemption, because they can take longer times to finish and there are no guarantees on when a compute task finishes. Therefore the driver enables the expensive instruction-level preemption for these tasks.
1442:
Each of those SMs also contains 32 FP64 CUDA cores - giving us the 1/2 rate for FP64 - and new to the Pascal architecture is the ability to pack 2 FP16 operations inside a single FP32 CUDA core under the right
654:— a high-bandwidth bus between the CPU and GPU, and between multiple GPUs. Allows much higher transfer speeds than those achievable by using PCI Express; estimated to provide between 80 and 200 GB/s.
592:
584:
648:
Unified memory — a memory architecture where the CPU and GPU can access both main system memory and memory on the graphics card with the help of a technology called "Page
Migration Engine".
1090:
The theoretical double-precision processing power of a Pascal GPU is 1/2 of the single precision performance on Nvidia GP100, and 1/32 of Nvidia GP102, GP104, GP106, GP107 & GP108.
2651:
847:
535:, starting with the GeForce GTX 1080 and GTX 1070 (both using the GP104 GPU), which were released on May 27, 2016, and June 10, 2016, respectively. Pascal was manufactured using
1130:
1093:
The theoretical half-precision processing power of a Pascal GPU is 2× of the single precision performance on GP100 and 1/64 on GP104, GP106, GP107 & GP108.
531:
architecture. The architecture was first introduced in April 2016 with the release of the Tesla P100 (GP100) on April 5, 2016, and is primarily used in the
44:
NVIDIA GeForce GTX 1080 Ti of the GeForce 10-line of graphics-cards, was the final major iteration featuring the Pascal microarchitecture (GP102-350-K1-A1).
716:
691:
Simultaneous Multi-Projection - generating multiple projections of a single geometry stream, as it enters the SMP engine from upstream shader stages.
2088:
2067:
1135:
1246:
798:
767:
867:
GP107: This GPU is used in the GeForce GTX 1050 and 1050 Ti. It is also used in the Quadro P1000, Quadro P600, Quadro P620 & Quadro P400.
861:
with 10/20 SMs enabled and GDDR5X memory. It is also used in the Quadro P5000, Quadro P4000, Quadro P3200 (mobile applications) and Tesla P4.
2988:
658:
857:
GP102: This GPU is used in the Titan Xp, Titan X Pascal and the GeForce GTX 1080 Ti. It is also used in the Quadro P6000 & Tesla P40.
719:
2.2 support for 4K DRM protected content playback and streaming (Maxwell GM200 and GM204 lack HDCP 2.2 support, GM206 supports HDCP 2.2).
2983:
1979:
1497:
1627:
1600:
1461:
1370:
1424:
645:
2 — some cards feature 16 GiB HBM2 in four stacks with a total bus width of 4096 bits and a memory bandwidth of 720 GB/s.
2581:
2040:
822:
1087:
is computed as 2 × operations per FMA instruction per CUDA core per cycle × number of CUDA cores × core clock speed (in GHz).
2910:
2511:
1860:
1327:
1190:
2816:
2586:
829:. It has been moved from the shader module to the TPC to allow one Polymorph engine to feed multiple SMs within the TPC.
1297:
628:
In Pascal, a SM (streaming multiprocessor) consists of between 64-128 CUDA cores, depending on if it is GP100 or GP104.
394:
373:
2551:
2526:
2516:
788:
629:
618:
609:
528:
470:
1768:
2576:
2571:
2566:
2546:
2521:
1949:
1118:
1106:
781:
484:
302:
297:
784:, 1 SM combines 192 single-precision (FP32) shader processors and 64 double-precision (FP64) units (on GK110 GPUs)
2591:
2561:
2541:
2536:
2531:
2464:
2376:
2246:
1811:
1546:
1211:
1102:
774:
763:
492:
452:
746:
A chip is partitioned into
Graphics Processor Clusters (GPCs). For the GP104 chips, a GPC encompasses 5 SMs.
2789:
2695:
2482:
755:
2870:
2494:
2289:
1229:
1701:
2935:
1628:"The NVIDIA GeForce GTX 1080 & GTX 1070 Founders Editions Review: Kicking Off the FinFET Generation"
1601:"The NVIDIA GeForce GTX 1080 & GTX 1070 Founders Editions Review: Kicking Off the FinFET Generation"
1498:"The NVIDIA GeForce GTX 1080 & GTX 1070 Founders Editions Review: Kicking Off the FinFET Generation"
1462:"The NVIDIA GeForce GTX 1080 & GTX 1070 Founders Editions Review: Kicking Off the FinFET Generation"
642:
389:
368:
242:
864:
GP106: This GPU is used in the GeForce GTX 1060 with GDDR5 memory. It is also used in the Quadro P2000.
850:
applications such as FP64 double precision compute and deep learning training that uses FP16. It uses
713:
Feature Set H hardware video decoding HEVC Main10 (10-bit), Main12 (12-bit) and VP9 hardware decoding.
2197:
2033:
2418:
2264:
1683:
92:
1898:
2674:
2639:
2269:
2259:
2254:
2235:
2230:
2225:
2220:
2215:
826:
569:
1723:
707:
Enhanced SLI Interface — SLI interface with higher bandwidth compared to the previous versions.
2860:
2740:
2308:
2303:
2298:
2279:
2274:
2178:
1275:
573:
532:
520:
151:
1571:
2920:
2915:
2755:
2440:
2406:
2387:
2359:
2210:
2205:
2188:
2183:
2173:
2168:
2146:
2141:
1837:
1654:
1523:
1355:
1114:
803:
710:
2730:
2712:
2458:
2026:
1745:
1110:
505:
1396:"NVIDIA's Next-Gen Pascal GPU Architecture to Provide 10X Speedup for Deep Learning Apps"
17:
2797:
2688:
2683:
2452:
1212:"Accelerating The Real-Time Ray Tracing Ecosystem: DXR For GeForce RTX and GeForce GTX"
1164:
2977:
2940:
2774:
2735:
2700:
2446:
2156:
562:
561:
The architecture is named after the 17th century French mathematician and physicist,
540:
118:
112:
2945:
2865:
2855:
2769:
2659:
2470:
2353:
2347:
2160:
1145:
1140:
722:
551:
183:
1793:
1395:
837:
797:
On GP100, 1 SM combines 64 single-precision (FP32) shader processors and also 32
664:
More registers — twice the amount of registers per CUDA core compared to
Maxwell.
2745:
2625:
2136:
694:
688:— new memory standard supporting 10Gbit/s data rates, updated memory controller.
442:
258:
252:
237:
232:
2005:
2875:
2664:
2488:
2412:
1305:
2121:
1429:
678:
Architectural improvements of the GP104 architecture include the following:
624:
Architectural improvements of the GP100 architecture include the following:
1547:"Nvidia GeForce GTX 1080 Simultaneous Multi-Projection & Async Compute"
931:
Texture (graphics or compute) or read-only data (compute only) cache per SM
576:
cards, a feature reserved to the Turing-based RTX series up to that point.
2116:
2111:
1980:"GeForce GTX 970: Correcting The Specs & Exploring Memory Allocation"
290:
285:
2925:
2808:
2764:
2062:
572:
on Pascal-based cards starting with the GTX 1060 6 GB, and in the
547:
278:
273:
1425:"NVIDIA Announces Tesla P100 Accelerator - Pascal GP100 Power for HPC"
1371:"NVLink, Pascal and Stacked Memory: Feeding the Appetite for Big Data"
39:
2634:
2476:
2049:
1662:
1084:
1083:
The theoretical single-precision processing power of a Pascal GPU in
685:
651:
614:
543:
524:
350:
345:
326:
321:
314:
309:
167:
71:
870:
GP108: This GPU is used in the GeForce GT 1010 and GeForce GT 1030.
2930:
2905:
2779:
2707:
2423:
2398:
2392:
851:
836:
591:
583:
504:
429:
2900:
2382:
698:
636:
596:
Die shot of the GP102 GPU found inside GeForce GTX 1080 Ti cards
536:
447:
338:
333:
87:
2837:
2612:
2329:
2060:
2022:
1230:"Ray Tracing Comes to Nvidia GTX GPUs: Here's How to Enable It"
2102:
2097:
517:
399:
266:
1950:"Nvidia GeForce GTX 1080, le premier GPU 16nm en test !"
1861:"NVIDIA Launches Tesla K20 & K20X: GK110 Arrives At Last"
1455:
1453:
1451:
821:
The
Polymorph Engine version 4.0 is the unit responsible for
791:, 1 SM combines 128 single-precision (FP32) shader processors
777:, 1 SM combines 32 single-precision (FP32) shader processors
875:
Comparison table of some Kepler, Maxwell, and Pascal chips
568:
In April 2019, Nvidia enabled a software implementation of
846:
GP100: Nvidia's Tesla P100 GPU accelerator is targeted at
2018:
1973:
1971:
588:
Die shot of the GP100 GPU used in Nvidia Tesla P100 cards
1892:
1890:
1888:
1886:
1884:
1882:
1684:"NVIDIA TITAN Xp Graphics Card with Pascal Architecture"
957:
Programmer-selectable shared memory/L1 partitions per SM
1769:"Nvidia GeForce GTX 1060 Gets GDDR5X in Fifth Makeover"
1491:
1489:
1487:
1485:
1483:
608:
In March 2014, Nvidia announced that the successor to
1594:
1592:
604:
Die shot of the GP106 GPU found inside GTX 1060 cards
1812:"NVIDIA GeForce GTX 1060 to be released on July 7th"
1418:
1416:
1191:"Samsung to Optical-Shrink NVIDIA "Pascal" to 14 nm"
2893:
2848:
2806:
2788:
2754:
2723:
2673:
2650:
2623:
2504:
2433:
2369:
2340:
2288:
2244:
2196:
2155:
2087:
2080:
1655:"Inside Pascal: NVIDIA's Newest Computing Platform"
1356:"Inside Pascal: NVIDIA's Newest Computing Platform"
1241:
1239:
754:A "Streaming Multiprocessor" is analogous to AMD's
476:
466:
461:
435:
425:
408:
382:
361:
356:
344:
332:
320:
308:
296:
284:
272:
264:
251:
225:
209:
197:
192:
176:
160:
144:
139:
131:
104:
77:
67:
49:
1724:"New Quadro Graphics Built on Pascal Architecture"
1101:The Pascal architecture was succeeded in 2017 by
509:Painting of Blaise Pascal, eponym of architecture
1131:List of eponyms of Nvidia GPU microarchitectures
963:48 KiB shared memory + 16 KiB L1 cache (default)
960:48 KiB shared memory + 16 KiB L1 cache (default)
671:enabled asynchronous compute in Pascal's driver.
1943:
1941:
1939:
1937:
1935:
1933:
1931:
1929:
1927:
1925:
1923:
1921:
1919:
1917:
1915:
1746:"Accelerating Data Center Workloads with GPUs"
1276:"NVIDIA Updates GPU Roadmap; Announces Pascal"
1165:"NVIDIA 7nm Next-Gen-GPUs To Be Built By TSMC"
674:Instruction-level and thread-level preemption.
2034:
8:
32:
2845:
2834:
2620:
2609:
2337:
2326:
2084:
2077:
2057:
2041:
2027:
2019:
1702:"NVIDIA TITAN X Graphics Card with Pascal"
704:Fourth generation Delta Color Compression.
1794:"NVIDIA GeForce 10 Series Graphics Cards"
825:. It corresponds functionally with AMD's
1136:List of Nvidia graphics processing units
873:
599:
1156:
994:16 KiB shared memory + 48 KiB L1 cache
986:32 KiB shared memory + 32 KiB L1 cache
854:. Quadro GP100 also uses the GP100 GPU.
991:16 KiB shared memory + 48 KiB L1 cache
983:32 KiB shared memory + 32 KiB L1 cache
31:
1304:. Devblogs.nvidia.com. Archived from
1121:in the consumer and business market.
999:Unified L1 cache/texture cache per SM
7:
725:HEVC Main10 10bit hardware encoding.
25:
1859:Smith, Ryan (November 12, 2012).
1335:International.download.nvidia.com
1254:International.download.nvidia.com
750:Streaming Multiprocessor "Pascal"
1978:Smith, Ryan (January 26, 2015).
1948:Triolet, Damien (May 24, 2016).
1545:Carbotte, Kevin (May 17, 2016).
38:
1274:Gupta, Sumit (March 21, 2014).
27:GPU microarchitecture by Nvidia
2491:(framebuffer in system memory)
1767:Zhiye Liu (October 22, 2018).
1653:Harris, Mark (April 5, 2016).
1369:Denis Foley (March 25, 2014).
1025:Dedicated shared memory per SM
905:Dedicated texture cache per SM
1:
1626:Smith, Ryan (July 20, 2016).
1599:Smith, Ryan (July 20, 2016).
1496:Smith, Ryan (July 20, 2016).
1460:Smith, Ryan (July 20, 2016).
1423:Smith, Ryan (April 5, 2015).
2006:"NVIDIA Turing Release Date"
1897:Nvidia (September 1, 2015).
682:CUDA Compute Capability 6.1.
2989:Graphics microarchitectures
3005:
1899:"CUDA C Programming Guide"
742:Graphics Processor Cluster
527:, as the successor to the
2984:Nvidia microarchitectures
2844:
2833:
2696:RSX 'Reality Synthesizer'
2619:
2608:
2336:
2331:Software and technologies
2325:
2076:
2056:
1838:"GTX 1060 Graphics Cards"
1247:"NVIDIA GeForce GTX 1080"
977:
974:
971:
968:
965:
956:
37:
1572:"Nvidia Pascal HDCP 2.2"
1524:"GTX 1080 Graphics Card"
1400:The Official NVIDIA Blog
1117:markets, and in 2018 by
161:Professional/workstation
18:Pascal microarchitecture
2483:Scalable Link Interface
2449:(variable refresh rate)
2341:Multimedia acceleration
841:GTX 1080 Ti PCB and die
799:double-precision (FP64)
794:On Pascal, it depends:
768:single-precision (FP32)
639:Compute Capability 6.0.
2505:GPU microarchitectures
2497:(live video upscaling)
2495:Video Super Resolution
2415:(ray tracing platform)
2245:Unified shaders &
1956:(in French). p. 2
842:
605:
597:
589:
516:is the codename for a
510:
339:Compute Capability 6.0
2936:Mellanox Technologies
1302:NVIDIA Developer Zone
840:
643:High Bandwidth Memory
603:
595:
587:
508:
152:GeForce GTX 10 series
55:; 8 years ago
2081:Fixed pixel pipeline
1576:Nvidia Hardware Page
817:Polymorph-Engine 4.0
426:Encoder(s) supported
2419:Nvidia System Tools
2012:. February 2, 2021.
1328:"NVIDIA Tesla P100"
876:
827:Geometric Processor
667:More shared memory.
546:process, and later
265:Supported Graphics
105:Fabrication process
34:
2426:(video decode API)
2379:(shading language)
2292:& tensor cores
1278:. Blogs.nvidia.com
874:
843:
766:, 1 SM combines 8
606:
598:
590:
570:DirectX Raytracing
511:
2971:
2970:
2967:
2966:
2963:
2962:
2861:Chris Malachowsky
2829:
2828:
2825:
2824:
2741:Shield Android TV
2604:
2603:
2600:
2599:
2395:(ray tracing API)
2321:
2320:
2317:
2316:
2131:
2130:
2071:
1308:on March 26, 2014
1298:"Parallel Forall"
1232:. April 11, 2019.
1076:
1075:
1051:L2 cache per chip
770:shader processors
533:GeForce 10 series
521:microarchitecture
503:
502:
279:DirectX 12 (12.1)
177:Server/datacenter
53:May 27, 2016
16:(Redirected from
2996:
2921:Cumulus Networks
2916:Bright Computing
2901:3dfx Interactive
2846:
2835:
2716:
2704:
2692:
2621:
2610:
2441:Nvidia 3D Vision
2407:Nvidia Omniverse
2388:Nvidia GameWorks
2362:(video decoding)
2356:(video decoding)
2350:(video encoding)
2338:
2327:
2085:
2078:
2065:
2058:
2043:
2036:
2029:
2020:
2014:
2013:
2002:
1996:
1995:
1993:
1991:
1975:
1966:
1965:
1963:
1961:
1945:
1910:
1909:
1907:
1905:
1894:
1877:
1876:
1874:
1872:
1856:
1850:
1849:
1847:
1845:
1834:
1828:
1827:
1825:
1823:
1808:
1802:
1801:
1790:
1784:
1783:
1781:
1779:
1764:
1758:
1757:
1755:
1753:
1742:
1736:
1735:
1733:
1731:
1720:
1714:
1713:
1711:
1709:
1698:
1692:
1691:
1680:
1674:
1673:
1671:
1669:
1650:
1644:
1643:
1641:
1639:
1623:
1617:
1616:
1614:
1612:
1596:
1587:
1586:
1584:
1582:
1568:
1562:
1561:
1559:
1557:
1551:Tomshardware.com
1542:
1536:
1535:
1533:
1531:
1520:
1514:
1513:
1511:
1509:
1493:
1478:
1477:
1475:
1473:
1457:
1446:
1445:
1439:
1437:
1420:
1411:
1410:
1408:
1406:
1392:
1386:
1385:
1383:
1381:
1366:
1360:
1359:
1358:. April 5, 2016.
1352:
1346:
1345:
1343:
1341:
1332:
1324:
1318:
1317:
1315:
1313:
1294:
1288:
1287:
1285:
1283:
1271:
1265:
1264:
1262:
1260:
1251:
1243:
1234:
1233:
1226:
1220:
1219:
1208:
1202:
1201:
1199:
1197:
1187:
1181:
1180:
1178:
1176:
1161:
1115:self-driving car
877:
804:machine learning
617:in GP104 have a
558:FinFET process.
555:
496:
488:
443:DisplayPort 1.4a
303:Shader Model 6.7
220:
216:
204:
122:
63:
61:
56:
42:
35:
21:
3004:
3003:
2999:
2998:
2997:
2995:
2994:
2993:
2974:
2973:
2972:
2959:
2889:
2883:Ranga Jayaraman
2880:Debora Shoquist
2840:
2821:
2802:
2784:
2750:
2731:Shield Portable
2719:
2713:Nintendo Switch
2710:
2698:
2686:
2669:
2646:
2615:
2596:
2500:
2473:(module/socket)
2467:(module/socket)
2461:(multi-monitor)
2459:Nvidia Surround
2455:(GPU switching)
2429:
2365:
2332:
2313:
2284:
2240:
2198:Unified shaders
2192:
2151:
2127:
2126:
2107:
2072:
2052:
2047:
2017:
2004:
2003:
1999:
1989:
1987:
1977:
1976:
1969:
1959:
1957:
1947:
1946:
1913:
1903:
1901:
1896:
1895:
1880:
1870:
1868:
1858:
1857:
1853:
1843:
1841:
1836:
1835:
1831:
1821:
1819:
1818:. June 29, 2016
1810:
1809:
1805:
1792:
1791:
1787:
1777:
1775:
1766:
1765:
1761:
1751:
1749:
1744:
1743:
1739:
1729:
1727:
1722:
1721:
1717:
1707:
1705:
1700:
1699:
1695:
1682:
1681:
1677:
1667:
1665:
1659:Parallel Forall
1652:
1651:
1647:
1637:
1635:
1625:
1624:
1620:
1610:
1608:
1598:
1597:
1590:
1580:
1578:
1570:
1569:
1565:
1555:
1553:
1544:
1543:
1539:
1529:
1527:
1522:
1521:
1517:
1507:
1505:
1495:
1494:
1481:
1471:
1469:
1459:
1458:
1449:
1435:
1433:
1422:
1421:
1414:
1404:
1402:
1394:
1393:
1389:
1379:
1377:
1368:
1367:
1363:
1354:
1353:
1349:
1339:
1337:
1330:
1326:
1325:
1321:
1311:
1309:
1296:
1295:
1291:
1281:
1279:
1273:
1272:
1268:
1258:
1256:
1249:
1245:
1244:
1237:
1228:
1227:
1223:
1210:
1209:
1205:
1195:
1193:
1189:
1188:
1184:
1174:
1172:
1171:. June 24, 2018
1163:
1162:
1158:
1154:
1127:
1111:cloud computing
1099:
1081:
891:GM204 (GTX 980)
888:GM204 (GTX 970)
835:
819:
752:
744:
739:
582:
553:
499:
491:
483:
457:
436:Display outputs
421:
409:Color bit-depth
404:
378:
247:
218:
214:
202:
188:
172:
156:
127:
120:
100:
97:
78:Manufactured by
59:
57:
54:
45:
28:
23:
22:
15:
12:
11:
5:
3002:
3000:
2992:
2991:
2986:
2976:
2975:
2969:
2968:
2965:
2964:
2961:
2960:
2958:
2957:
2954:
2951:
2948:
2943:
2938:
2933:
2928:
2923:
2918:
2913:
2908:
2903:
2897:
2895:
2891:
2890:
2888:
2887:
2886:Jonah M. Alben
2884:
2881:
2878:
2873:
2868:
2863:
2858:
2856:Jen-Hsun Huang
2852:
2850:
2842:
2841:
2838:
2831:
2830:
2827:
2826:
2823:
2822:
2820:
2819:
2813:
2811:
2804:
2803:
2801:
2800:
2798:Project Denver
2794:
2792:
2786:
2785:
2783:
2782:
2777:
2772:
2767:
2761:
2759:
2752:
2751:
2749:
2748:
2743:
2738:
2733:
2727:
2725:
2721:
2720:
2718:
2717:
2705:
2693:
2680:
2678:
2671:
2670:
2668:
2667:
2662:
2656:
2654:
2648:
2647:
2645:
2644:
2643:
2642:
2631:
2629:
2617:
2616:
2614:Other products
2613:
2606:
2605:
2602:
2601:
2598:
2597:
2595:
2594:
2589:
2584:
2579:
2574:
2569:
2564:
2559:
2554:
2549:
2544:
2539:
2534:
2529:
2524:
2519:
2514:
2508:
2506:
2502:
2501:
2499:
2498:
2492:
2486:
2480:
2474:
2468:
2462:
2456:
2453:Nvidia Optimus
2450:
2444:
2437:
2435:
2431:
2430:
2428:
2427:
2421:
2416:
2410:
2404:
2403:
2402:
2396:
2385:
2380:
2373:
2371:
2367:
2366:
2364:
2363:
2357:
2351:
2344:
2342:
2334:
2333:
2330:
2323:
2322:
2319:
2318:
2315:
2314:
2312:
2311:
2306:
2301:
2295:
2293:
2286:
2285:
2283:
2282:
2277:
2272:
2267:
2262:
2257:
2251:
2249:
2242:
2241:
2239:
2238:
2233:
2228:
2223:
2218:
2213:
2208:
2202:
2200:
2194:
2193:
2191:
2186:
2181:
2176:
2171:
2166:
2164:
2153:
2152:
2150:
2149:
2144:
2139:
2132:
2129:
2128:
2125:
2124:
2119:
2114:
2108:
2106:
2105:
2100:
2094:
2093:
2091:
2082:
2074:
2073:
2061:
2054:
2053:
2048:
2046:
2045:
2038:
2031:
2023:
2016:
2015:
1997:
1967:
1911:
1878:
1851:
1829:
1816:VideoCardz.com
1803:
1785:
1773:Tom's Hardware
1759:
1737:
1715:
1693:
1675:
1645:
1618:
1588:
1563:
1537:
1515:
1479:
1447:
1412:
1387:
1361:
1347:
1319:
1289:
1266:
1235:
1221:
1203:
1182:
1155:
1153:
1150:
1149:
1148:
1143:
1138:
1133:
1126:
1123:
1098:
1095:
1080:
1077:
1074:
1073:
1070:
1067:
1064:
1061:
1058:
1055:
1052:
1048:
1047:
1044:
1041:
1038:
1035:
1032:
1029:
1026:
1022:
1021:
1018:
1015:
1012:
1009:
1006:
1003:
1000:
996:
995:
992:
988:
987:
984:
980:
979:
976:
973:
970:
967:
964:
961:
958:
954:
953:
950:
947:
944:
941:
938:
935:
932:
928:
927:
924:
921:
918:
915:
912:
909:
906:
902:
901:
898:
895:
892:
889:
886:
883:
880:
872:
871:
868:
865:
862:
858:
855:
834:
831:
818:
815:
814:
813:
812:
811:
807:
792:
785:
778:
771:
751:
748:
743:
740:
738:
735:
734:
733:
729:
728:GPU Boost 3.0.
726:
720:
714:
708:
705:
702:
692:
689:
683:
676:
675:
672:
668:
665:
662:
655:
649:
646:
640:
634:
621:-like design.
581:
578:
501:
500:
498:
497:
495:(professional)
489:
480:
478:
474:
473:
468:
464:
463:
459:
458:
456:
455:
450:
445:
439:
437:
433:
432:
427:
423:
422:
420:
419:
416:
412:
410:
406:
405:
403:
402:
397:
392:
386:
384:
380:
379:
377:
376:
371:
365:
363:
359:
358:
354:
353:
348:
342:
341:
336:
330:
329:
324:
318:
317:
312:
306:
305:
300:
294:
293:
288:
282:
281:
276:
270:
269:
262:
261:
256:
249:
248:
246:
245:
240:
235:
229:
227:
226:Memory support
223:
222:
211:
207:
206:
199:
195:
194:
193:Specifications
190:
189:
187:
186:
180:
178:
174:
173:
171:
170:
164:
162:
158:
157:
155:
154:
148:
146:
142:
141:
140:Product Series
137:
136:
133:
129:
128:
126:
125:
115:
108:
106:
102:
101:
99:
98:
96:
95:
90:
84:
81:
79:
75:
74:
69:
65:
64:
51:
47:
46:
43:
26:
24:
14:
13:
10:
9:
6:
4:
3:
2:
3001:
2990:
2987:
2985:
2982:
2981:
2979:
2955:
2952:
2949:
2947:
2944:
2942:
2941:Mental Images
2939:
2937:
2934:
2932:
2929:
2927:
2924:
2922:
2919:
2917:
2914:
2912:
2909:
2907:
2904:
2902:
2899:
2898:
2896:
2892:
2885:
2882:
2879:
2877:
2874:
2872:
2869:
2867:
2864:
2862:
2859:
2857:
2854:
2853:
2851:
2847:
2843:
2836:
2832:
2818:
2815:
2814:
2812:
2810:
2805:
2799:
2796:
2795:
2793:
2791:
2787:
2781:
2778:
2776:
2773:
2771:
2768:
2766:
2763:
2762:
2760:
2757:
2753:
2747:
2744:
2742:
2739:
2737:
2736:Shield Tablet
2734:
2732:
2729:
2728:
2726:
2724:Nvidia Shield
2722:
2714:
2709:
2706:
2702:
2701:PlayStation 3
2697:
2694:
2690:
2685:
2682:
2681:
2679:
2676:
2672:
2666:
2663:
2661:
2658:
2657:
2655:
2653:
2649:
2641:
2638:
2637:
2636:
2635:Nvidia Quadro
2633:
2632:
2630:
2627:
2622:
2618:
2611:
2607:
2593:
2590:
2588:
2585:
2583:
2580:
2578:
2575:
2573:
2570:
2568:
2565:
2563:
2560:
2558:
2555:
2553:
2550:
2548:
2545:
2543:
2540:
2538:
2535:
2533:
2530:
2528:
2525:
2523:
2520:
2518:
2515:
2513:
2510:
2509:
2507:
2503:
2496:
2493:
2490:
2487:
2484:
2481:
2478:
2475:
2472:
2469:
2466:
2463:
2460:
2457:
2454:
2451:
2448:
2447:Nvidia G-Sync
2445:
2442:
2439:
2438:
2436:
2432:
2425:
2422:
2420:
2417:
2414:
2411:
2409:(3D graphics)
2408:
2405:
2401:(physics SDK)
2400:
2397:
2394:
2391:
2390:
2389:
2386:
2384:
2381:
2378:
2375:
2374:
2372:
2368:
2361:
2358:
2355:
2352:
2349:
2346:
2345:
2343:
2339:
2335:
2328:
2324:
2310:
2307:
2305:
2302:
2300:
2297:
2296:
2294:
2291:
2287:
2281:
2278:
2276:
2273:
2271:
2268:
2266:
2263:
2261:
2258:
2256:
2253:
2252:
2250:
2248:
2243:
2237:
2234:
2232:
2229:
2227:
2224:
2222:
2219:
2217:
2214:
2212:
2209:
2207:
2204:
2203:
2201:
2199:
2195:
2190:
2187:
2185:
2182:
2180:
2177:
2175:
2172:
2170:
2167:
2165:
2162:
2158:
2154:
2148:
2145:
2143:
2140:
2138:
2134:
2133:
2123:
2120:
2118:
2115:
2113:
2110:
2109:
2104:
2101:
2099:
2096:
2095:
2092:
2090:
2086:
2083:
2079:
2075:
2069:
2064:
2059:
2055:
2051:
2044:
2039:
2037:
2032:
2030:
2025:
2024:
2021:
2011:
2007:
2001:
1998:
1985:
1981:
1974:
1972:
1968:
1955:
1951:
1944:
1942:
1940:
1938:
1936:
1934:
1932:
1930:
1928:
1926:
1924:
1922:
1920:
1918:
1916:
1912:
1900:
1893:
1891:
1889:
1887:
1885:
1883:
1879:
1866:
1862:
1855:
1852:
1844:September 15,
1839:
1833:
1830:
1822:September 15,
1817:
1813:
1807:
1804:
1799:
1795:
1789:
1786:
1774:
1770:
1763:
1760:
1752:September 15,
1747:
1741:
1738:
1730:September 15,
1725:
1719:
1716:
1708:September 15,
1703:
1697:
1694:
1689:
1685:
1679:
1676:
1664:
1660:
1656:
1649:
1646:
1633:
1629:
1622:
1619:
1606:
1602:
1595:
1593:
1589:
1577:
1573:
1567:
1564:
1556:September 15,
1552:
1548:
1541:
1538:
1530:September 15,
1525:
1519:
1516:
1503:
1499:
1492:
1490:
1488:
1486:
1484:
1480:
1467:
1463:
1456:
1454:
1452:
1448:
1444:
1443:circumstances
1432:
1431:
1426:
1419:
1417:
1413:
1401:
1397:
1391:
1388:
1376:
1372:
1365:
1362:
1357:
1351:
1348:
1340:September 15,
1336:
1329:
1323:
1320:
1307:
1303:
1299:
1293:
1290:
1277:
1270:
1267:
1259:September 15,
1255:
1248:
1242:
1240:
1236:
1231:
1225:
1222:
1217:
1213:
1207:
1204:
1192:
1186:
1183:
1170:
1166:
1160:
1157:
1151:
1147:
1144:
1142:
1139:
1137:
1134:
1132:
1129:
1128:
1124:
1122:
1120:
1116:
1112:
1108:
1104:
1096:
1094:
1091:
1088:
1086:
1078:
1071:
1068:
1065:
1062:
1059:
1056:
1053:
1050:
1049:
1045:
1042:
1039:
1036:
1033:
1030:
1027:
1024:
1023:
1019:
1016:
1013:
1010:
1007:
1004:
1001:
998:
997:
993:
990:
989:
985:
982:
981:
962:
959:
955:
951:
948:
945:
942:
939:
936:
933:
930:
929:
925:
922:
919:
916:
913:
910:
907:
904:
903:
899:
896:
893:
890:
887:
884:
881:
879:
878:
869:
866:
863:
859:
856:
853:
849:
845:
844:
839:
832:
830:
828:
824:
816:
808:
805:
800:
796:
795:
793:
790:
786:
783:
779:
776:
772:
769:
765:
761:
760:
759:
757:
749:
747:
741:
736:
730:
727:
724:
721:
718:
715:
712:
709:
706:
703:
700:
696:
693:
690:
687:
684:
681:
680:
679:
673:
669:
666:
663:
660:
656:
653:
650:
647:
644:
641:
638:
635:
631:
627:
626:
625:
622:
620:
616:
611:
602:
594:
586:
579:
577:
575:
571:
566:
564:
563:Blaise Pascal
559:
557:
549:
545:
542:
538:
534:
530:
526:
523:developed by
522:
519:
515:
507:
494:
490:
486:
482:
481:
479:
475:
472:
469:
465:
460:
454:
451:
449:
446:
444:
441:
440:
438:
434:
431:
428:
424:
417:
414:
413:
411:
407:
401:
398:
396:
393:
391:
388:
387:
385:
383:Decode codecs
381:
375:
372:
370:
367:
366:
364:
362:Encode codecs
360:
355:
352:
349:
347:
343:
340:
337:
335:
331:
328:
325:
323:
319:
316:
313:
311:
307:
304:
301:
299:
295:
292:
291:Direct3D 12.0
289:
287:
283:
280:
277:
275:
271:
268:
263:
260:
257:
254:
250:
244:
241:
239:
236:
234:
231:
230:
228:
224:
212:
208:
200:
196:
191:
185:
182:
181:
179:
175:
169:
166:
165:
163:
159:
153:
150:
149:
147:
143:
138:
134:
130:
124:
116:
114:
110:
109:
107:
103:
94:
91:
89:
86:
85:
83:
82:
80:
76:
73:
70:
66:
52:
48:
41:
36:
30:
19:
2946:PortalPlayer
2894:Acquisitions
2866:Curtis Priem
2758:and embedded
2708:Tegra NX-SoC
2660:Nvidia Tesla
2582:Ada Lovelace
2556:
2434:Technologies
2068:List of GPUs
2009:
2000:
1988:. Retrieved
1983:
1958:. Retrieved
1953:
1902:. Retrieved
1869:. Retrieved
1864:
1854:
1842:. Retrieved
1832:
1820:. Retrieved
1815:
1806:
1797:
1788:
1776:. Retrieved
1772:
1762:
1750:. Retrieved
1740:
1728:. Retrieved
1718:
1706:. Retrieved
1696:
1687:
1678:
1666:. Retrieved
1658:
1648:
1636:. Retrieved
1631:
1621:
1609:. Retrieved
1604:
1579:. Retrieved
1575:
1566:
1554:. Retrieved
1550:
1540:
1528:. Retrieved
1518:
1506:. Retrieved
1504:. p. 10
1501:
1470:. Retrieved
1465:
1441:
1434:. Retrieved
1428:
1403:. Retrieved
1399:
1390:
1378:. Retrieved
1374:
1364:
1350:
1338:. Retrieved
1334:
1322:
1310:. Retrieved
1306:the original
1301:
1292:
1280:. Retrieved
1269:
1257:. Retrieved
1253:
1224:
1215:
1206:
1194:. Retrieved
1185:
1173:. Retrieved
1168:
1159:
1146:Nvidia NVENC
1141:Nvidia NVDEC
1100:
1092:
1089:
1082:
823:Tessellation
820:
756:Compute Unit
753:
745:
677:
623:
615:shader units
607:
567:
560:
513:
512:
357:Media Engine
298:Shader Model
29:
2746:GeForce Now
2640:Quadro Plex
2626:Workstation
2485:(multi-GPU)
2443:(stereo 3D)
2290:Ray tracing
2255:GeForce 600
2137:GeForce 256
2089:Pre-GeForce
1986:. p. 1
1954:Hardware.fr
1867:. p. 3
1778:February 2,
1634:. p. 4
1607:. p. 5
1468:. p. 9
1079:Performance
852:HBM2 memory
695:DisplayPort
467:Predecessor
205:KB (per SM)
132:Codename(s)
68:Designed by
2978:Categories
2876:Bill Dally
2871:David Kirk
2849:Key people
2677:components
2512:Fahrenheit
2489:TurboCache
2479:(protocol)
2413:Nvidia RTX
2299:GeForce 20
1375:nvidia.com
1196:August 13,
1152:References
487:(consumer)
351:Vulkan 1.3
327:OpenGL 4.6
315:OpenCL 3.0
60:2016-05-27
2807:Computer
2624:Graphics
2587:Blackwell
2360:PureVideo
2206:GeForce 8
2169:GeForce 3
2010:Techradar
1984:AnandTech
1865:AnandTech
1840:. GeForce
1704:. GeForce
1632:AnandTech
1605:AnandTech
1526:. GeForce
1502:AnandTech
1466:AnandTech
1430:AnandTech
1405:March 23,
1312:March 25,
1282:March 25,
1097:Successor
1072:4096 KiB
810:elements.
711:PureVideo
574:16 series
477:Successor
448:HDMI 2.0b
2809:chipsets
2370:Software
2117:RIVA TNT
2112:RIVA 128
1990:July 24,
1960:July 24,
1904:July 24,
1871:July 24,
1748:. NVIDIA
1726:. NVIDIA
1638:July 21,
1611:July 21,
1508:July 21,
1472:July 21,
1169:Wccftech
1125:See also
1069:2048 KiB
1066:3072 KiB
1063:2048 KiB
1060:1792 KiB
1057:1536 KiB
737:Overview
657:16-bit (
286:Direct3D
259:PCIe 3.0
210:L2 cache
198:L1 cache
184:Tesla P4
168:Quadro P
117:Samsung
50:Launched
2926:DeepMap
2839:Company
2765:GoForce
2675:Console
2552:Maxwell
2527:Rankine
2517:Celsius
2163:shaders
2063:GeForce
1668:June 3,
1436:May 27,
1380:July 7,
1175:July 6,
1105:in the
1054:512 KiB
1046:64 KiB
1020:24 KiB
789:Maxwell
630:Maxwell
619:Maxwell
610:Maxwell
580:Details
548:Samsung
529:Maxwell
471:Maxwell
462:History
274:DirectX
255:support
145:Desktop
93:Samsung
58: (
2956:Stexar
2953:MediaQ
2950:Exluna
2817:nForce
2775:Jetson
2577:Hopper
2572:Ampere
2567:Turing
2557:Pascal
2547:Kepler
2522:Kelvin
2477:NVLink
2157:Vertex
2135:
2050:Nvidia
1798:NVIDIA
1688:NVIDIA
1663:Nvidia
1581:May 8,
1216:NVIDIA
1119:Turing
1113:, and
1085:GFLOPS
1043:96 KiB
1040:96 KiB
1037:96 KiB
1034:96 KiB
1017:48 KiB
1014:48 KiB
1011:48 KiB
1008:48 KiB
937:48 KiB
908:48 KiB
900:GP100
806:tasks.
782:Kepler
686:GDDR5X
652:NVLink
633:units.
554:
544:FinFET
525:Nvidia
514:Pascal
485:Turing
418:10-bit
346:Vulkan
322:OpenGL
310:OpenCL
238:GDDR5X
219:
215:
203:
121:
72:Nvidia
33:Pascal
2931:Icera
2906:Ageia
2780:Tegra
2770:Drive
2652:GPGPU
2628:cards
2592:Rubin
2562:Volta
2542:Fermi
2537:Tesla
2532:Curie
2424:VDPAU
2399:PhysX
2393:OptiX
2354:NVDEC
2348:NVENC
2161:pixel
1331:(PDF)
1250:(PDF)
1103:Volta
897:GP104
894:GM200
885:GK110
882:GK104
848:GPGPU
833:Chips
775:Fermi
764:Tesla
723:NVENC
701:2.0b.
697:1.4,
541:16 nm
493:Volta
430:NVENC
415:8-bit
395:H.265
390:H.264
374:H.265
369:H.264
233:GDDR5
135:GP10x
111:TSMC
2790:CPUs
2756:SoCs
2689:Xbox
2684:NV2A
2383:CUDA
2265:800M
2247:NUMA
2174:4 Ti
2159:and
2147:4 MX
2122:TNT2
1992:2016
1962:2016
1906:2016
1873:2016
1846:2016
1824:2016
1780:2024
1754:2016
1732:2016
1710:2016
1670:2016
1640:2016
1613:2016
1583:2016
1558:2016
1532:2016
1510:2016
1474:2016
1438:2016
1407:2015
1382:2014
1342:2016
1314:2014
1284:2014
1261:2016
1198:2016
1177:2019
717:HDCP
699:HDMI
659:FP16
637:CUDA
537:TSMC
334:CUDA
267:APIs
253:PCIe
243:HBM2
217:KB—4
113:16FF
88:TSMC
2911:ULi
2665:DGX
2471:SXM
2465:MXM
2270:900
2260:700
2236:500
2231:400
2226:300
2221:200
2216:100
2103:NV2
2098:NV1
1107:HPC
787:On
780:On
773:On
762:On
565:.
550:'s
539:'s
518:GPU
453:DVI
400:VP9
213:256
2980::
2377:Cg
2309:40
2304:30
2280:16
2275:10
2179:FX
2008:.
1982:.
1970:^
1952:.
1914:^
1881:^
1863:.
1814:.
1796:.
1771:.
1686:.
1661:.
1657:.
1630:.
1603:.
1591:^
1574:.
1549:.
1500:.
1482:^
1464:.
1450:^
1440:.
1427:.
1415:^
1398:.
1373:.
1333:.
1300:.
1252:.
1238:^
1214:.
1167:.
1109:,
978:—
952:—
926:—
556:nm
552:14
221:MB
201:24
123:nm
119:14
2715:)
2711:(
2703:)
2699:(
2691:)
2687:(
2211:9
2189:7
2184:6
2142:2
2070:)
2066:(
2042:e
2035:t
2028:v
1994:.
1964:.
1908:.
1875:.
1848:.
1826:.
1800:.
1782:.
1756:.
1734:.
1712:.
1690:.
1672:.
1642:.
1615:.
1585:.
1560:.
1534:.
1512:.
1476:.
1409:.
1384:.
1344:.
1316:.
1286:.
1263:.
1218:.
1200:.
1179:.
1031:—
1028:—
1005:—
1002:—
975:—
972:—
969:—
966:—
949:—
946:—
943:—
940:—
934:—
923:—
920:—
917:—
914:—
911:—
62:)
20:)
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.
