View source for RIVA 128

{{Short description|Graphics processing unit developed by Nvidia}}
{{Infobox GPU
| name = Nvidia RIVA 128
| image = Elsa Victory Erazor-AGP-4 - Nvidia RIVA 128-7094.jpg
| codename = NV3
| created = {{start date|1997|08|25}}<ref>{{Cite press release|title=NVIDIA Announces Major OEM Design Wins For The RIVA 128 3D Multimedia Accelerator|url=http://www.nvidia.com/news/pr/pr082597FR.html|author=Nvidia Corporation|author-link=Nvidia|date=August 25, 1997|archive-url=https://web.archive.org/web/19980613043410/http://www.nvidia.com/news/pr/pr082597FR.html|archive-date=June 13, 1998|url-status=dead|access-date=April 21, 2020}}</ref>
| entry = 
| midrange = 
| highend = RIVA 128, RIVA 128 ZX
| d3dversion = Direct3D 5.0
| openglversion = 1.0
| predecessor = [[NV1]]
| successor = [[RIVA TNT]]
| support status = Unsupported
| manufacturer = [[STMicroelectronics|SGS-Thomson]], [[TSMC]]
| designfirm = [[Nvidia]]
}}
The '''RIVA 128''', or "'''NV3'''", was a consumer [[graphics processing unit]] created in 1997 by [[Nvidia]]. It was the first nVidia product to integrate 3D acceleration in addition to traditional 2D and video acceleration. Its name is an acronym for '''''R'''eal-time '''I'''nteractive '''V'''ideo and '''A'''nimation accelerator''.<ref name=NVRIVA128brochure>[https://web.archive.org/web/19970706013810/http://www.nvidia.com/product/riva128/index.html RIVA 128 Brochure], Nvidia, accessed October 9, 2007.</ref>

The RIVA 128 followed Nvidia's less successful "[[NV1]]" accelerator and was the first product to gain Nvidia widespread recognition. It was also a major change in Nvidia's technological direction.

[[File:Diamond Viper V330 AGP @ nVIDIA RIVA 128.jpg|thumbnail|Diamond Viper V330 4Mb @ RIVA 128 GPU]]

[[File:ASUS RIVA 128ZX.png|thumb|An ASUS RIVA 128ZX AGP]]

==History==

By 1996, Nvidia was in poor shape financially.  It had initially pursued a completely different type of rendering technology called quadratic texture mapping with its first product, the NV1.<ref name="Witt">{{cite magazine |last1=Witt |first1=Stephen |title=How Jensen Huang's Nvidia Is Powering the A.I. Revolution |url=https://www.newyorker.com/magazine/2023/12/04/how-jensen-huangs-nvidia-is-powering-the-ai-revolution |access-date=December 5, 2023 |magazine=The New Yorker |date=November 27, 2023 |archive-date=November 27, 2023 |archive-url=https://archive.today/20231127154917/https://www.newyorker.com/magazine/2023/12/04/how-jensen-huangs-nvidia-is-powering-the-ai-revolution |url-status=live }}</ref>  Then the company had spent a year trying to use its inferior technology to build the graphics chip for [[Sega]]'s [[Dreamcast]] video game console.<ref name="wsj cohen 2024">{{Cite news |last=Cohen |first=Ben |date=May 18, 2024 |title=The 84-Year-Old Man Who Saved Nvidia |url=https://www.wsj.com/business/nvidia-stock-jensen-huang-sega-irimajiri-chips-ai-906247db |work=The Wall Street Journal |url-access=subscription |archive-url=https://archive.today/20240518101129/https://www.wsj.com/business/nvidia-stock-jensen-huang-sega-irimajiri-chips-ai-906247db |archive-date=May 18, 2024 |url-status=live}}</ref> Sega finally had to pull the plug on Nvidia's project and switch vendors, but was convinced to keep Nvidia alive with a $5 million investment.<ref name="wsj cohen 2024" />

Nvidia then laid off half of its 100 employees and focused its remaining resources on developing the RIVA 128.<ref name="Witt" /> By the time the RIVA 128 was released in August 1997, Nvidia was down to one month of payroll.<ref name="Witt" />  This extremely desperate situation resulted in what remains "the unofficial company motto" today: "Our company is thirty days from going out of business".<ref name="Witt" /> 

==Architecture==
[[Image:RIVA 128 GPU.jpg|thumb|RIVA 128 GPU]]
[[Image:RIVA 128ZX GPU.jpg|thumb|RIVA 128ZX GPU]]
[[File:NVIDIA@350nm@Fixed-pipeline@NV3 ZX@Riva 128ZX@A22AL9828 MALTA J822XFJ-A DSCx8 polysilicon@makrolens.jpg|thumb|Die shot of the RIVA 128ZX]]
The RIVA 128 was built to render within the [[Direct3D]] 5 and OpenGL [[Application programming interface|API]] specifications. It was designed to accelerate Direct3D to the utmost extent possible, as a departure from Nvidia's NV1 chip. The NV1 chip had been designed for a completely different type of rendering technology called quadratic texture mapping, which is not supported by Direct3D.<ref name="Witt"/><ref name="wsj cohen 2024"/><ref name="Peddie">{{cite news |last1=Peddie |first1=Jon |title=Nvidia's RIVA 128 |url=https://www.electronicdesign.com/technologies/embedded/article/21131077/jon-peddie-research-nvidias-riva-128 |access-date=August 27, 2024 |work=Electronic Design |date=September 23, 2019 |archive-date=January 19, 2024 |archive-url=https://web.archive.org/web/20240119155034/https://www.electronicdesign.com/technologies/embedded/article/21131077/jon-peddie-research-nvidias-riva-128 |url-status=live}}</ref>

The graphics accelerator consists of 3.5 million [[transistor]]s built on [[STMicroelectronics|SGS-Thomson]]'s 5LM [[350 nm]] [[Semiconductor device fabrication|fabrication process]] and is clocked at 100 [[Hertz|MHz]].<ref name=NVRIVA128brochure /><ref name=NVRIVA128ZXpressrelease>[https://web.archive.org/web/19980613043223fw_/http://www.nvidia.com/news/pr/pr022398cFR.html Nvidia sets unprecedented 3D performance level with new RIVA 128ZX processor press release], Nvidia, accessed December 3, 2023.</ref><ref name=NVRIVA128D3Dpressrelease>[https://web.archive.org/web/19980613043338fw_/http://www.nvidia.com/news/pr/pr091097FR.html RIVA 128 gains support as preferred Direct3D developer platform press release], Nvidia, accessed December 3, 2023.</ref> RIVA 128 has a single [[pixel pipeline]] capable of 1 pixel per clock when sampling one texture. It is specified to output pixels at a rate of 100 million per second and 25-pixel triangles at 1.5 million per second.<ref name=NVRIVA128brochure /> There are 12 [[kibibyte|KiB]] of on-chip memory used for pixel and vertex caches.<ref name=NVRIVA128brochure /> The chip was limited to a 16-bit ([[Highcolour|Highcolor]]) pixel format when performing 3D acceleration and a 16-bit [[Z-buffering|Z-buffer]].

The 2D accelerator engine within the RIVA 128 is 128 bits wide and also operates at 100&nbsp;MHz. In this "fast and wide" configuration, as Nvidia referred to it, the RIVA 128 performed admirably for GUI acceleration compared to competitors.<ref name=ragesvelocity128>[http://members.tripod.com/~FireEYE/riva.html STB VELOCITY 128 REVIEW (PCI)], Rage's Hardware, February 7, 1998. [https://web.archive.org/web/19981201220515/http://members.tripod.com/~FireEYE/riva.html Archived] on December 1, 1998.</ref> A 32-bit hardware [[VESA BIOS Extensions|VESA]]-compliant [[Super Video Graphics Array|SVGA]]/[[Video Graphics Array|VGA]] core was implemented as well. Video acceleration aboard the chip is optimized for [[MPEG-2]] but lacks full acceleration of that standard. Final picture output is routed through an integrated 206&nbsp;MHz [[RAMDAC]].<ref name=NVRIVA128brochure /> RIVA 128 had the advantage of being a combination 2D/3D graphics chip, unlike Voodoo Graphics. This meant that the computer did not require a separate 2D card for output outside of 3D applications. It also allowed 3D rendering within a window. The ability to build a system with just one graphics card, and still have it be feature-complete for the time, made the RIVA 128 a lower-cost high-performance solution.

Nvidia equipped the RIVA 128 with 4 [[Mebibyte|MiB]] of [[Synchronous dynamic random-access memory#SGRAM|SGRAM]], a new memory technology for the time, clocked at 100&nbsp;MHz and connected to the graphics processor via a 128-bit [[bus (computing)|memory bus]].<ref name=NVRIVA128brochure /> This provides memory bandwidth of 1.60 [[gigabyte]]s per second. The memory was used in a [[Shared memory architecture|unified memory architecture]] that shared the whole RAM pool with both [[framebuffer]] and texture storage. The main benefit of this, over a split design such as that on Voodoo Graphics and [[Voodoo2|Voodoo²]], was support for 3D resolutions of 800×600 and 960×720, higher than Voodoo's 640×480.<ref name=tomshardwarer128>[http://www.thg.ru/graphic/19971027/print.html Review AGP Graphic Cards], Tom's Hardware, October 27, 1997.</ref>

In early 1998, Nvidia released a refreshed version called the '''RIVA 128 ZX'''. This refreshed design of NV3 increased memory support to 8 MiB and increased RAMDAC frequency to 250&nbsp;MHz. These additions allowed the RIVA 128 ZX to support higher resolutions and [[refresh rate]]s.<ref>Covey, Alf. [http://groups.google.com/group/comp.sys.ibm.pc.hardware.video/msg/b2003d8585319cb2?dmode=source STB Velocity 128 vs STB Velocity 128zx What is the difference?], STB Technical Support, June 3, 1998.</ref> The ZX version was fabricated by [[STMicroelectronics|SGS-Thomson]] and [[TSMC]], and uses an 8{{nbsp}}[[Mebibyte|MB]] [[SGRAM]] memory chip, clocked at 125{{nbsp}}MHz, from [[Samsung Electronics]].<ref name=NVRIVA128ZXpressrelease/><ref>{{cite news |last1=Hong |first1=Samuel Setoh Kok |title=ELSA Victory Erazor LT Riva 128ZX Review |url=https://assets.hardwarezone.com/2009/reviews/video/erazorlt/erazorlt.htm |access-date=31 December 2009 |work=[[HardwareZone]] |date=11 October 1998}}</ref>

The RIVA 128 had an [[Accelerated Graphics Port|AGP]] 1X [[Input/output|bus interface]], whereas the ZX version of it was one of the early AGP 2X parts, giving it some more marketing headroom by being on the forefront of interface technology. The graphics processor was built around [[Intel Corporation|Intel]]'s AGP specification targeting the Intel [[List of Intel chipsets|440LX]] chipset for the [[Pentium II]]. Nvidia designed the RIVA 128 with a maximum memory capacity of 4 MiB because, at the time, this was the cost-optimal approach for a consumer 3D accelerator.<ref name=NVRIVA128FAQ>[http://developer.nvidia.com/object/RIVA_128_FAQ.html RIVA 128/ZX/TNT FAQ] {{webarchive|url=https://web.archive.org/web/20061006100032/http://developer.nvidia.com/object/RIVA_128_FAQ.html |date=2006-10-06 }}, Nvidia, accessed October 9, 2007.</ref> This was the case partly because of the chip's capability to store textures in off-screen system RAM in both [[Peripheral Component Interconnect|PCI]] or AGP configurations.<ref name=NVRIVA128FAQ />

The next major chip from Nvidia would be the [[RIVA TNT]].<ref name=salon>{{cite web|url=https://www.salon.com/2002/05/16/nvidia2/|title=The prince of polygons|last=Turner|first=Daniel Drew|work=Salon|date=May 16, 2002|access-date=August 28, 2024|archiveurl=https://web.archive.org/web/20210224235252/https://www.salon.com/2002/05/16/nvidia2/|archivedate=February 24, 2021}}</ref><ref name=spectra2500>{{cite web|url=https://www.anandtech.com/show/206|archive-url=https://web.archive.org/web/20120505165341/http://www.anandtech.com/show/206|url-status=dead|archive-date=May 5, 2012|website=[[AnandTech]]|title=Canopus SPECTRA 2500 Riva TNT|author=Anand Lal Shimpi|date=September 6, 1998|access-date=August 28, 2024}}</ref>

===Image quality===
[[File:Quake2onRIVA128.jpg|thumb|[[Quake II]] on RIVA 128 (final drivers)]]
At the time of the RIVA 128's release, [[3dfx|3Dfx Voodoo Graphics]] had firmly established itself as the 3D hardware benchmark against which all newcomers were compared. The Voodoo was the first 3D game accelerator to offer exceptional performance and quality. The RIVA 128 was scorned for its lower quality rendering (compared to the Voodoo) and rendering errors.<ref name="Peddie"/>

With initial drivers, the RIVA 128 used per-polygon [[mipmap]]ping instead of the much higher-quality, but more demanding, per-pixel variety.<ref name="Peddie"/> This caused the different texture detail levels to "pop" into place as the player moved through a game and approached each polygon, instead of allowing a seamless, gradual per-pixel transition. Nvidia eventually released drivers that allowed a per-pixel mode. Another issue with the card's texturing was its use of automated mipmap generation. While this improves visual quality and performance in games without mipmaps, it also caused unforeseen problems because it forced games to render in a way that they were not programmed for.

NV3's bilinear filtering was actually "sharper" than that of 3Dfx Voodoo Graphics. But, while it did not blur textures as much as Voodoo, it did instead add some light noise to textures, because of a lower-fidelity filtering algorithm. There were also problems with noticeable seams between polygons.

While initial drivers did present these image-quality problems, later drivers offered image quality arguably matching that of Voodoo Graphics. In addition, because the RIVA 128 can render at resolutions higher than 640×480, the card can offer quality superior to that of Voodoo Graphics, as shown in the above [[Quake II]] screenshot. The final drivers released for the RIVA 128 support per-pixel mipmapping, full-scene anti-aliasing ([[supersampling]]), and a number of options to fine-tune features in order to optimize quality and performance.

== Drivers and APIs ==

Drivers were, for a significant portion of the card's life, rather rough. Not only were the aforementioned Direct3D issues apparent, but the card lacked good OpenGL support.<ref name=ragesvelocity128 /> With RIVA 128, Nvidia began their quest for top-quality OpenGL support, eventually resulting in the board being a capable OpenGL performer. One major disadvantage for Nvidia was that many games during RIVA 128's lifetime used 3Dfx's proprietary [[Glide API]]. Legally, only 3Dfx cards could use 3Dfx's Glide API.

Like the competing [[ATI Rage|ATI Rage Pro]], RIVA 128 was never able to accelerate the popular [[Unreal Engine]] in Direct3D mode due to missing hardware features. It was, however, possible to use the engine's OpenGL renderer, but unfortunately OpenGL support was quite slow and buggy in the original Unreal Engine. Performance in [[Quake III Arena]], a game using an engine more advanced than Unreal Engine 1, was better due to the engine having been designed for OpenGL.{{citation needed|date=August 2017}}

Nvidia's final RIVA 128 drivers for [[Windows 9x]] include a full OpenGL driver. However, for this driver to function, Windows must be set with a desktop [[color depth]] of 16-bit.

A driver for RIVA 128 is also included in Windows 2000 and XP, but lacks 3D support. A beta driver with OpenGL support was once leaked by Nvidia but was canceled later, and there is no Windows 2000 driver for RIVA 128 on Nvidia's driver site today. Neither the beta driver nor the ones come with Windows 2000/XP could support Direct3D.

== Performance ==
At the time, RIVA 128 was one of the first combination 2D/3D cards that could rival Voodoo Graphics. RIVA 128's 2D capability was seen as impressive for its time and was competitive with even high-end 2D-only graphics cards in both quality and performance.<ref name=ragesvelocity128 /><ref name=tomshardwarer128 />
== Chipset table ==
{{Row hover highlight}}
{| class="mw-datatable wikitable sortable" style="font-size:85%; text-align:center;"
|-
! rowspan="2" style="vertical-align: bottom" {{Vert header|Model}}
! rowspan="2" style="vertical-align: bottom" {{Vert header|Launch}}
! rowspan="2" {{Vert header|[[Code name]]}}
! rowspan="2" {{Vert header|[[Semiconductor device fabrication|Fab]] ([[nanometer|nm]])<ref name="vintage3d">{{cite web |title=3D accelerator database |url=http://vintage3d.org/dbn.php |website=Vintage 3D |access-date=30 August 2024 |archive-url=https://web.archive.org/web/20181023222614/http://www.vintage3d.org/dbn.php |archive-date=23 October 2018 |url-status=live }}</ref>}}
! rowspan="2" {{Vert header|Transistors (million)}}
! rowspan="2" {{Vert header|Die size (mm<sup>2</sup>)}}
! rowspan="2" {{Vert header|[[Computer bus|Bus]] [[I/O interface|interface]]}}
! rowspan="2" {{Vert header|Core clock ([[Hertz|MHz]])}}
! rowspan="2" {{Vert header|Memory clock ([[Hertz|MHz]])}}
! rowspan="2" {{Vert header|Core config{{efn|name=geforce 256 1|[[Pixel pipeline]]s: [[texture mapping unit]]s: [[render output unit]]s}}}}
! colspan="4" |[[Fillrate]]
! colspan="4" |Memory
! rowspan="2" {{Vert header|TDP (Watts)}}
! colspan="2" |Latest [[Application programming interface|API]] support
|-
!{{Vert header|MOperations/s}}
!{{Vert header|MPixels/s}}
!{{Vert header|MTexels/s}}
!{{Vert header|MVertices/s}}
!{{Vert header|Size ([[Megabyte|MB]])}}
!{{Vert header|Bandwidth ([[Gigabyte|GB]]/s)}}
!{{Vert header|Bus type}}
!{{Vert header|Bus width ([[bit]])}}
!{{Vert header|[[Direct3D]]}}
!{{Vert header|[[OpenGL]]}}
|-
! style="text-align:left" |[[Riva 128]]
|August 25, 1997
| rowspan="2" |NV3
|SGS [[350&nbsp;nm]]
| rowspan="2" |4<ref>{{Cite web|url=https://www.techpowerup.com/gpu-specs/nvidia-nv3.g227|title=NVIDIA NV3 GPU Specs &#124; TechPowerUp GPU Database|date=30 August 2024 }}</ref>
| rowspan="2" |90
|AGP 1x,<ref>{{cite web |url=http://www.nvidia.com/object/RIVA_128_FAQ.html |title=RIVA 128/ZX/TNT FAQ |access-date=August 30, 2024 |archive-url=https://web.archive.org/web/20180616103247/http://www.nvidia.com/object/RIVA_128_FAQ.html |archive-date=June 16, 2018 |url-status=live }}</ref> PCI
| rowspan="2" |100
| rowspan="2" |100
|rowspan="2" |1:1:1
| rowspan="2" |100
| rowspan="2" |100
| rowspan="2" |100
|rowspan="2" |0
|4
| rowspan="2" |1.6
| rowspan="2" |SDR
| rowspan="2" |128
|?
| rowspan="2" |5.0
| rowspan="2" |1.0
|-
! style="text-align:left" |[[Riva 128]]ZX
|February 23, 1998
|SGS/[[TSMC]] 350&nbsp;nm
| rowspan="1" |AGP 2x, PCI
|8
|?
|}
{{notelist}}

== Competing chipsets ==
* [[Matrox Mystique|Matrox Mystique 220]]
* [[3dfx|3DFX Voodoo Graphics]], [[Voodoo 2|Voodoo²]] (3D-only boards)
* [[ATI Rage|ATI Rage series]] (Pro was the most recent at the time)
* [[S3 ViRGE]], [[S3 Savage|Savage 3D]]
* [[Rendition (company)|Rendition Vérité V1000 & V2x00]]
* [[PowerVR#Series1 (NEC)|PowerVR PCX2]]

== See also ==
* [[Comparison of Nvidia graphics processing units]]

== References ==
{{Reflist}}

== External links ==
{{Commons|Nvidia RIVA 128|Riva 128 series}}
* [https://web.archive.org/web/20061006100032/http://developer.nvidia.com/object/RIVA_128_FAQ.html Nvidia's RIVA 128 FAQ]

{{Nvidia}}

[[Category:Computer-related introductions in 1997]]
[[Category:Nvidia graphics processors]]
[[Category:Graphics cards]]