204:
Some units are used more than others, and in an effort to increase the processor's entire performance, they attempted to find a "sweet spot" in the number of units needed for optimum efficiency without the need for excess silicon. In this architecture the name pixel pipeline lost its meaning as pixel
150:
processing is becoming more important, this number still holds some weight. Best example of this is the X1600 XT. This card has a 3 to 1 ratio of pixel shader processors/texture mapping units. As a result, the X1600 XT achieves lower performance when compared to other GPUs of the same era and class
196:
A pipeline isn't formally accepted as a technical term. There are different pipelines within a graphics processor as there are separate functions being performed at any given time. Historically, it has been referred to as a pixel processor that is attached to a dedicated TMU. A Geforce 3 had four
174:
units. It is the TMU's job to apply texture operations to pixels. The number of texture units in a graphics processor is used when comparing two different cards for texturing performance. It is reasonable to assume that the card with more TMUs will be faster at processing texture information. In
200:
The ATI Radeon 9700 was first to break this mould, by placing a number of vertex shader engines independent of the pixel shaders. The R300 GPU used in the Radeon 9700 had four global vertex shaders, but split the rest of the rendering pipeline in half (it was, so to speak, dual core) each half,
151:(such as nVidia's 7600GT) . In the mid range, texture mapping can still very much be a bottleneck. However, at the high end, the X1900 XTX has this same 3 to 1 ratio, but does just fine because screen resolutions top out and it has more than enough texture mapping power to handle any display.
91:(ROPs). For example, in AMD's Cypress GPU, each shader pipeline (of which there are 20) has four TMUs, giving the GPU 80 TMUs. This is done by chip designers to closely couple shaders and the texture engines they will be working with.
175:
modern GPUs TMUs contain
Texture Address Units(TA) and Texture Filtering Units(TF). Texture Address Units map texels to pixels and can perform texture addressing modes. Texture Filtering Units optionally perform hardware based
208:
The vertex shader had long been decoupled, starting with the R300, but the pixel shader was not so easily done, as it required colour data (e.g. texture samples) to work with, and hence needed to be closely coupled to a TMU.
212:
Said coupling remains to this day, where the shader engine, made of units able to run either vertex or pixel data, is tightly coupled to a TMU but has a crossbar dispatcher between its output and the bank of ROPs.
75:
The TMU came about due to the compute demands of sampling and transforming a flat image (as the texture map) to the correct angle and perspective it would need to be in 3D space. The compute operation is a large
116:
take a texture and 'map' it to a piece of geometry. That is, they wrap the texture around the geometry and produce textured pixels which can then be written to the screen. Textures can be an actual image, a
143:. This is called texture mapping and is accomplished by texture mapping units (TMUs) on the videocard. Texture fill rate is a measure of the speed with which a particular card can perform texture mapping.
998:
351:
467:
848:
403:
677:
993:
381:
295:
344:
430:
1073:
629:
193:
A pipeline is the graphics card's architecture, which provides a generally accurate idea of the computing power of a graphics processor.
602:
641:
624:
337:
239:, when more than one sample is merged into one pixel. All data rendered has to travel through the ROP in order to be written to the
636:
558:
197:
pixel pipelines, each of which had two TMUs. The rest of the pipeline handled things like depth and blending operations.
832:
597:
1068:
1047:
964:
901:
386:
1013:
707:
360:
32:
324:
266:, the texture mapping unit provides a convenient means of approximating arbitrary functions with data tables.
170:
Textures need to be addressed and filtered. This job is done by TMUs that work in conjunction with pixel and
822:
1008:
969:
780:
712:
976:
954:
933:
728:
651:
619:
235:. Its job is to control the sampling of pixels (each pixel is a dimensionless point), so it controls
299:
246:
Therefore, the ROP is where the GPU's output is assembled into a bitmapped image ready for display.
1035:
981:
875:
672:
523:
1025:
682:
585:
232:
226:
88:
63:, whereas when first introduced it was implemented as a separate processor, e.g. as seen on the
1018:
817:
692:
580:
188:
176:
60:
20:
1030:
1003:
938:
916:
880:
812:
100:
986:
911:
646:
590:
457:
222:
165:
77:
52:
80:, which CPUs of the time (early Pentiums) could not cope with at acceptable performance.
827:
553:
462:
275:
1062:
863:
607:
503:
482:
263:
171:
56:
40:
36:
870:
445:
423:
418:
259:
236:
231:
The Render Output
Pipeline is an inherited term, and more often referred to as the
147:
140:
122:
109:
44:
1040:
858:
733:
533:
472:
240:
928:
896:
853:
743:
508:
440:
113:
959:
518:
396:
329:
262:
in 1,2, or 3 dimensions may be used to store arbitrary data. By providing
906:
391:
134:
118:
105:
48:
800:
667:
612:
528:
513:
477:
408:
64:
697:
687:
573:
568:
498:
452:
435:
413:
84:
923:
805:
790:
773:
768:
763:
758:
753:
748:
738:
255:
201:
called a quad, had four pixel shaders, four TMUs and four ROPs.
16:
Component of a GPU which alters bitmap images for use as textures
795:
785:
563:
112:, and the textures that cover that geometry. Texture units in a
333:
325:"The Differential Geometry of Texture-Mapping and Shading"
139:
To render a 3D scene, textures are mapped over the top of
947:
889:
841:
721:
660:
546:
491:
374:
367:
243:, from there it can be transmitted to the display.
205:processors were no longer attached to single TMUs.
47:to be placed onto an arbitrary plane of a given
345:
8:
59:it is implemented as a discrete stage in a
371:
352:
338:
330:
287:
125:for advanced surface lighting effects.
296:"Dev Hardware - The Graphics Pipeline"
108:are generally composed of two things:
7:
678:Input–output memory management unit
51:as a texture, in a process called
14:
35:(GPUs). They are able to rotate,
217:Render output pipelines (ROPs)
83:In 2013, TMUs are part of the
1:
160:Texture mapping units (TMUs)
1074:Geometry in computer vision
31:) is a component in modern
1090:
220:
186:
163:
132:
98:
33:graphics processing units
708:Video display controller
361:Graphics processing unit
89:Render Output Pipelines
87:and decoupled from the
823:Shared graphics memory
71:Background and history
1009:Hardware acceleration
713:Video processing unit
934:Performance per watt
703:Texture mapping unit
652:Unified shader model
25:texture mapping unit
876:Integrated graphics
1026:Parallel computing
902:Display resolution
683:Render output unit
673:Geometry processor
233:render output unit
227:Render output unit
1069:Graphics hardware
1056:
1055:
871:External graphics
854:Discrete graphics
818:Memory controller
581:Graphics pipeline
542:
541:
189:Graphics pipeline
177:texture filtering
129:Texture fill rate
61:graphics pipeline
21:computer graphics
1081:
1031:Vector processor
1014:Image processing
1004:Graphics library
939:Transistor count
881:System on a chip
813:Memory bandwidth
693:Stream processor
372:
354:
347:
340:
331:
312:
311:
309:
307:
298:. Archived from
292:
101:Digital geometry
1089:
1088:
1084:
1083:
1082:
1080:
1079:
1078:
1059:
1058:
1057:
1052:
943:
885:
837:
717:
656:
647:Tiled rendering
538:
487:
458:InfiniteReality
363:
358:
321:
316:
315:
305:
303:
294:
293:
289:
284:
272:
252:
229:
223:Raster graphics
221:Main articles:
219:
191:
185:
168:
166:Texture mapping
162:
157:
137:
131:
103:
97:
85:shader pipeline
78:matrix multiply
73:
67:graphics card.
53:texture mapping
17:
12:
11:
5:
1087:
1085:
1077:
1076:
1071:
1061:
1060:
1054:
1053:
1051:
1050:
1045:
1044:
1043:
1033:
1028:
1023:
1022:
1021:
1011:
1006:
1001:
996:
991:
990:
989:
984:
974:
973:
972:
967:
962:
951:
949:
945:
944:
942:
941:
936:
931:
926:
921:
920:
919:
914:
904:
899:
893:
891:
887:
886:
884:
883:
878:
873:
868:
867:
866:
861:
851:
845:
843:
839:
838:
836:
835:
830:
828:Texture memory
825:
820:
815:
810:
809:
808:
803:
798:
793:
788:
778:
777:
776:
771:
766:
761:
756:
751:
746:
736:
731:
725:
723:
719:
718:
716:
715:
710:
705:
700:
695:
690:
685:
680:
675:
670:
664:
662:
658:
657:
655:
654:
649:
644:
639:
634:
633:
632:
622:
617:
616:
615:
605:
600:
595:
594:
593:
588:
578:
577:
576:
571:
566:
556:
554:Compute kernel
550:
548:
544:
543:
540:
539:
537:
536:
531:
526:
521:
516:
511:
506:
501:
495:
493:
489:
488:
486:
485:
480:
475:
470:
465:
460:
455:
450:
449:
448:
443:
438:
428:
427:
426:
421:
416:
411:
401:
400:
399:
394:
389:
378:
376:
369:
365:
364:
359:
357:
356:
349:
342:
334:
328:
327:
320:
319:External links
317:
314:
313:
286:
285:
283:
280:
279:
278:
276:Execution unit
271:
268:
251:
248:
218:
215:
187:Main article:
184:
181:
164:Main article:
161:
158:
156:
153:
141:polygon meshes
133:Main article:
130:
127:
99:Main article:
96:
93:
72:
69:
57:graphics cards
15:
13:
10:
9:
6:
4:
3:
2:
1086:
1075:
1072:
1070:
1067:
1066:
1064:
1049:
1046:
1042:
1039:
1038:
1037:
1034:
1032:
1029:
1027:
1024:
1020:
1017:
1016:
1015:
1012:
1010:
1007:
1005:
1002:
1000:
997:
995:
992:
988:
985:
983:
980:
979:
978:
975:
971:
968:
966:
963:
961:
958:
957:
956:
953:
952:
950:
946:
940:
937:
935:
932:
930:
927:
925:
922:
918:
915:
913:
910:
909:
908:
905:
903:
900:
898:
895:
894:
892:
888:
882:
879:
877:
874:
872:
869:
865:
862:
860:
857:
856:
855:
852:
850:
847:
846:
844:
840:
834:
831:
829:
826:
824:
821:
819:
816:
814:
811:
807:
804:
802:
799:
797:
794:
792:
789:
787:
784:
783:
782:
779:
775:
772:
770:
767:
765:
762:
760:
757:
755:
752:
750:
747:
745:
742:
741:
740:
737:
735:
732:
730:
727:
726:
724:
720:
714:
711:
709:
706:
704:
701:
699:
696:
694:
691:
689:
686:
684:
681:
679:
676:
674:
671:
669:
666:
665:
663:
659:
653:
650:
648:
645:
643:
640:
638:
635:
631:
628:
627:
626:
623:
621:
618:
614:
611:
610:
609:
608:Rasterisation
606:
604:
601:
599:
598:HDR rendering
596:
592:
589:
587:
584:
583:
582:
579:
575:
572:
570:
567:
565:
562:
561:
560:
557:
555:
552:
551:
549:
545:
535:
532:
530:
527:
525:
522:
520:
517:
515:
512:
510:
507:
505:
504:Apple silicon
502:
500:
497:
496:
494:
490:
484:
483:Apple silicon
481:
479:
476:
474:
471:
469:
466:
464:
461:
459:
456:
454:
451:
447:
444:
442:
439:
437:
434:
433:
432:
429:
425:
422:
420:
417:
415:
412:
410:
407:
406:
405:
402:
398:
395:
393:
390:
388:
385:
384:
383:
380:
379:
377:
373:
370:
366:
362:
355:
350:
348:
343:
341:
336:
335:
332:
326:
323:
322:
318:
302:on 2006-05-28
301:
297:
291:
288:
281:
277:
274:
273:
269:
267:
265:
264:interpolation
261:
257:
249:
247:
244:
242:
238:
234:
228:
224:
216:
214:
210:
206:
202:
198:
194:
190:
182:
180:
178:
173:
172:vertex shader
167:
159:
154:
152:
149:
144:
142:
136:
128:
126:
124:
120:
115:
111:
107:
102:
94:
92:
90:
86:
81:
79:
70:
68:
66:
62:
58:
54:
50:
46:
42:
38:
34:
30:
26:
22:
1036:Video coding
702:
637:Tessellation
547:Architecture
304:. Retrieved
300:the original
290:
260:texture maps
253:
250:Use in GPGPU
245:
237:antialiasing
230:
211:
207:
203:
199:
195:
192:
169:
148:pixel shader
145:
138:
104:
82:
74:
55:. In modern
45:bitmap image
28:
24:
18:
1019:Compression
890:Performance
842:Form factor
734:Framebuffer
698:Tensor unit
688:Shader unit
620:Ray-tracing
559:Fabrication
534:Intel 2700G
468:3dfx Voodoo
463:NEC µPD7220
241:framebuffer
123:normal maps
110:3D geometry
1063:Categories
929:Frame rate
897:Clock rate
859:Clustering
661:Components
441:Radeon Pro
282:References
121:, or even
114:video card
960:Scrolling
864:Switching
519:VideoCore
183:Pipelines
106:3D scenes
907:Fillrate
586:Geometry
446:Instinct
270:See also
135:Fillrate
119:lightmap
95:Geometry
49:3D model
987:Texture
917:Texel/s
912:Pixel/s
849:IP core
801:HBM-PIM
668:Blitter
642:T&L
613:Shading
529:Imageon
524:Vivante
514:PowerVR
478:Glaze3D
409:GeForce
375:Desktop
155:Details
146:Though
65:Voodoo2
41:distort
965:Sprite
924:FLOP/s
722:Memory
591:Vertex
574:MOSFET
569:FinFET
499:Adreno
492:Mobile
453:Matrox
436:Radeon
414:Quadro
404:Nvidia
306:11 May
39:, and
37:resize
1041:Codec
999:GPGPU
806:HBM3E
791:HBM2E
774:GDDR7
769:GDDR6
764:GDDR5
759:GDDR4
754:GDDR3
749:GDDR2
739:SGRAM
424:Tegra
419:Tesla
382:Intel
256:GPGPU
1048:VLIW
994:ASIC
970:Tile
948:Misc
833:VRAM
796:HBM3
786:HBM2
744:GDDR
630:SIMT
625:SIMD
564:CMOS
509:Mali
308:2006
225:and
23:, a
781:HBM
729:DMA
603:MAC
431:AMD
397:Arc
368:GPU
254:In
29:TMU
19:In
1065::
982:GI
977:3D
955:2D
473:S3
392:Xe
387:GT
258:,
179:.
43:a
353:e
346:t
339:v
310:.
27:(
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.