Difference between revisions of "Sega NAOMI 2"
From Sega Retro
Line 96: | Line 96: | ||
** PowerVR2: 2.1 GFLOPS rendering | ** PowerVR2: 2.1 GFLOPS rendering | ||
* T&L geometry: 8.7 GFLOPS{{ref|Elan: 7.5 GFLOPS (75 floating-point operations per cycle) <br> SH‑4 SIMD: 180 MHz available (10% load, 20 MHz used), 1.26 GFLOPS (90% of 1.4 GFLOPS)|group=n}} | * T&L geometry: 8.7 GFLOPS{{ref|Elan: 7.5 GFLOPS (75 floating-point operations per cycle) <br> SH‑4 SIMD: 180 MHz available (10% load, 20 MHz used), 1.26 GFLOPS (90% of 1.4 GFLOPS)|group=n}} | ||
− | ** Matrix transformations: 240 million vertices/sec{{ref|Elan: 200 million vertices/sec (28 | + | ** Matrix transformations: 240 million vertices/sec{{ref|Elan: 200 million vertices/sec (28 floating-point operations per transform,{{ref|1=[https://books.google.co.uk/books?id=iAvHt5RCHbMC&pg=PA95 ''Design of Digital Systems and Devices'' (page 95)]}} 2 transforms per cycle) <br> SH-4: 45 million vertices/sec (4 cycles per transform){{fileref|SH-4 Next-Generation DSP Architecture.pdf|page=12}}|group=n}} |
− | ** Perspective transformations: 210 million vertices/sec{{ref|Elan: 200 million vertices/sec, 31 | + | ** Perspective transformations: 210 million vertices/sec{{ref|Elan: 200 million vertices/sec, 31 floating-point operations per vertex (28 operations for matrix transform,{{ref|1=[https://books.google.co.uk/books?id=iAvHt5RCHbMC&pg=PA95 ''Design of Digital Systems and Devices'' (page 95)]}} 3 operations for perspective division),{{ref|[http://gamedev.allusion.net/docs/kos-current/matrix_8h.html Dreamcast: Basic matrix operations (KallistiOS)]}} 2 transforms per cycle <br> SH-4: 15 million vertices/sec, 12 cycles per vertex (4 cycles matrix transform,{{fileref|SH-4 Next-Generation DSP Architecture.pdf|page=12}} 5 cycles perspective division),{{ref|[http://gamedev.allusion.net/docs/kos-current/matrix_8h.html Dreamcast: Basic matrix operations (KallistiOS)]}} 12 cycles division latency{{fileref|SH-4 Software Manual.pdf|page=211}}|group=n}} |
− | ** 1 light source: 110 million vertices/sec{{ref|Elan: 100 million vertices/sec (63 | + | ** 1 light source: 110 million vertices/sec{{ref|Elan: 100 million vertices/sec (63 floating-point operations per vertex,{{ref|1=[https://books.google.co.uk/books?id=iAvHt5RCHbMC&pg=PA96 ''Design of Digital Systems and Devices'' (page 96)]}} 1 vertex per cycle) <br> SH-4: 12 million vertices/sec, 14 cycles per vertex (4 cycles matrix transformation, 5 cycles perspective division, 1 cycle surface normal, 4 cycles lighting matrix){{fileref|SH-4 Software Manual.pdf|page=151}}{{fileref|SH-4 Next-Generation DSP Architecture.pdf|page=31}}|group=n}} |
** 4 light sources: 26 million vertices/sec{{ref|Elan: 20 million vertices/sec, 5 cycles per vertex (1 cycle transform, 1 cycle per light source) <br> SH-4: 6 million vertices/sec, 29 cycles per vertex (4 cycles matrix transformation, 5 cycles perspective division, 1 cycle per surface normal, 4 cycles per lighting matrix)|group=n}} | ** 4 light sources: 26 million vertices/sec{{ref|Elan: 20 million vertices/sec, 5 cycles per vertex (1 cycle transform, 1 cycle per light source) <br> SH-4: 6 million vertices/sec, 29 cycles per vertex (4 cycles matrix transformation, 5 cycles perspective division, 1 cycle per surface normal, 4 cycles per lighting matrix)|group=n}} | ||
** 6 light sources: 18 million vertices/sec{{ref|Elan: 14 million vertices/sec, 7 cycles per vertex (1 cycle transform, 1 cycle per light source) <br> SH-4: 4 million vertices/sec, 39 cycles per vertex (4 cycles matrix transformation, 5 cycles perspective division, 1 cycle per surface normal, 4 cycles per lighting matrix)|group=n}} | ** 6 light sources: 18 million vertices/sec{{ref|Elan: 14 million vertices/sec, 7 cycles per vertex (1 cycle transform, 1 cycle per light source) <br> SH-4: 4 million vertices/sec, 39 cycles per vertex (4 cycles matrix transformation, 5 cycles perspective division, 1 cycle per surface normal, 4 cycles per lighting matrix)|group=n}} |
Revision as of 16:21, 21 December 2016
This teeny-tiny article needs some work. You can help us by expanding it.
Sega NAOMI 2 | |||||
---|---|---|---|---|---|
Manufacturer: Sega | |||||
Variants: Sega NAOMI 2 GD-ROM, Sega NAOMI 2 Satellite Terminal | |||||
Add-ons: GD-ROM | |||||
|
The Sega NAOMI 2 is an arcade board developed by Sega and is a successor to Sega NAOMI hardware. It was originally released in 2000. Since it uses similar NAOMI architecture (but significantly beefed up), it is also fully backwards compatible with its predecessor.
The NAOMI 2 is significantly more powerful than the NAOMI, including a dual CPU setup, new T&L GPU, dual rasterizer GPU, increased memory, and faster bandwidth. This leads to games with much more polygons than a NAOMI game, rendered at much faster speeds, while the new T&L GPU adds advanced lighting and particle effects. It was also more affordable than the very expensive (and difficult to program) Sega Hikaru arcade system that preceded it.[1]
As with the NAOMI, the NAOMI 2 was also available in GD-ROM and Satellite Terminal variants. It was Sega's last proprietary arcade system board; subsequent Sega arcade boards have been based on console and PC hardware.
Contents
Development
VideoLogic's Elan, the T&L geometry GPU coprocessor used in the NAOMI 2, had been in development since 1998, when the original NAOMI arcade system and Dreamcast console launched.[2] Yu Suzuki was involved in its development, insisting that it must have enough power to sustain in-game performance of at least 10 million polygons per second will all effects enabled.[3]
Technical Specifications
NAOMI 2 Specifications
- Main CPU: Hitachi SH‑4[4] @ 200 MHz[5][6]
- Units: 128‑bit SIMD vector units with graphic functions, 2× 64‑bit floating‑point units, 2× 32‑bit fixed‑point units
- Bus width: 128‑bit internal, 64‑bit external
- Bandwidth: 3.2 GB/s internal, 1.6 GB/s external
- Fixed‑point performance: 360 MIPS
- SH‑4 floating‑point performance: 1.4 GFLOPS
- Note: With Elan used as geometry coprocessor, the SH‑4's 128‑bit SIMD matrix unit can be dedicated to game physics, artificial intelligence, collision detection, overall game code, or further enhancing graphics. CPU load is reduced by 90% with Elan.[7]
- Sound engine: Yamaha AICA Super Intelligent Sound Processor @ 67 MHz
- PLD: 4 PLD[n 1]
- Operating systems:
- Sega native operating system
- Custom Windows CE, with DirectX 6.0, Direct3D and OpenGL support
- Storage media: ROM cartridge
- Extensions: communication, 4‑channel surround sound, PCI, MIDI, RS‑232C
- Connection: JAMMA Video compliant
Graphics
- GPU: 6 core processors (Elan, SH‑4 SIMD, 2× PowerVR2, 2 DAC)
- Core units: 14 units (Elan, SH‑4 SIMD, 10 PowerVR2 cores, 2 DAC)
- Clock rate: 200 MHz
- GPU T&L geometry coprocessor: VideoLogic Elan @ 100 MHz
- Bus width: 512‑bit (4× 128‑bit)[11]
- Lighting: Up to 16 light sources per polygon, ambient lighting, parallel lighting, point lighting, spotlight lighting
- Vertex support: Combined dynamic and static model processing
- Features: Reduces CPU load to 1/10th (90% reduction), multiple light type support (ambient, parallel, point, spot), hardware Z clipping, offscreen & backface culling[7]
- Elan floating‑point performance: 7.5 GFLOPS[12][n 6]
- GPU rasterizers: 2× NEC‑VideoLogic PowerVR2 @ 100 MHz
- Revision: Dual PowerVR2 doubles rendering performance over NAOMI, which in turn had twice the rendering performance of the Dreamcast, as NAOMI revision has dual ISP cores in each PowerVR2[n 7]
- Bus width: 128‑bit (external)
- Cores: 2x TA (Tile Accelerators), 4x ISP (Image Synthesis Processors), 2x TSP (Texture & Shading Processor), 6x Triangle Setup FPU, 2x RAMDAC
- Units: 176 rendering units (148 ISP units, 20 TSP units, 6 FPU units, 2 RAMDAC)
- ISP units: 4x ISP Precalc Units, 4x ISP PE Arrays (128 PE processor elements), 4x Depth Accumulation Buffers, 4x Span RLC, 4x Span Sorters, 4x ISP Parameter Cache
- TSP units: 2x TSP Precalc, 2x Parameter Cache, 2x Texture Cache, 2x Iterator Arrays, 2x Pixel Processing Engines, 2x Tile Accumulation Buffers, 2x Secondary Accumulation Buffers, 2x Combine & Bump Map Units, 2x Fog Units, 2x Alpha Blending Units[17]
- Triangle Setup FPU: 6 FPU rendering units, 2.1 GFLOPS
- RAMDAC: 230 MHz
- Effects: Bump mapping, multi‑texturing, fog, alpha blending, mipmapping, bilinear filtering, trilinear filtering, anti‑aliasing, environment mapping, specular effects,[20][21] normal mapping
- Features: Tiled rendering, deferred rendering, back‑face culling, hidden surface removal
- Defails: See NAOMI Specifications and Dreamcast Specifications for more details on PowerVR2 graphics system.
- Video DAC: 2× Rohm BU1426KS @ 35.4695 MHz[22]
- Bus width: 48‑bit (2× 24‑bit)
- Color depth: 32‑bit ARGB, 16,777,216 colors (24‑bit color) with 8‑bit (256 levels) alpha blending, YUV and RGB color spaces, color key overlay[23]
- Display resolution: 31 kHz horizontal sync, 60 Hz refresh rate, JAMMA/VGA,[24] progressive scan
- Rendering fillrate:
- Texture fillrate:[n 12]
- Textures per pass: 10 texture layers[12]
- Floating-point performance: 11 GFLOPS
- Elan: 7.5 GFLOPS geometry
- SH-4 SIMD: 1.4 GFLOPS geometry
- PowerVR2: 2.1 GFLOPS rendering
- T&L geometry: 8.7 GFLOPS[n 13]
- Polygon rendering:
- 100 million polygons/sec:[12] 1 light, flat shading, opaque polygons
- 26 million polygons/sec: 4 lights, texture mapping, Gouraud shading, opaque/translucent polygons
- 10 million polygons/sec: 6 lights, texture mapping, Gouraud shading, opaque/translucent polygons
Memory
Bandwidth
- Internal processor cache bandwidth:
- RAM/ROM memory bandwidth: 16.1 GB/s (15.1 GB/s system, 1 GB/s cartridge)
- Video memory: 14.01 GB/s (13.01 GB/s VRAM, 900 MB/s ROM)
- System RAM bandwidth: 10 GB/s[4]
- System ROM bandwidth: 88 MB/s[4]
- Cartridge ROM bandwidth: 900 MB/s[n 42]
- Note: High‑speed access allows ROM to effectively be used as RAM, and textures streamed directly from ROM.[47]
- Cartridge RAM bandwidth: 100 MB/s[n 43]
NAOMI 2 GD-ROM Specifications
The NAOMI GD‑ROM, released in 2001, is identical to the standard NAOMI, but uses GD‑ROM discs for storage instead of ROM cartridges. It comes with a DIMM Board, which is very similar to a ROM cartridge, but with RAM instead of ROM. When a game is installed, the GD‑ROM content is loaded onto the DIMM Board RAM, so that the game data runs from the DIMM Board rather than the GD‑ROM disc. The NAOMI 2 GD‑ROM specification includes the following differences:
- Board composition: Motherboard + Daughter Board + DIMM Board
- Storage media: GD‑ROM drive
- GD‑ROM transfer rate: 1.8 MB/s (1800 KB/sec)
Memory
Bandwidth
- RAM bandwidth: 11–12 GB/s
- Main RAM: 1.6 GB/s
- VRAM: 8.4 GB/s
- Sound RAM: 132 MB/s
- SRAM: 44 MB/s
- DIMM RAM: 1.1–2.13 GB/s[n 45]
List of Games
NAOMI 2 Games
- Jet Squadron (prototype) (2000)
- Virtua Fighter 4 (2001)
- Virtua Fighter 4: Evolution (2002)
- Virtua Fighter 4 Final Tuned (2004)
- Virtua Striker 3 (2001)
- Wild Riders (2001)
- Club Kart: European Session (2002)
- King of Route 66 (2002)
- Sega Driving Simulator (2002)
- Soul Surfer (2002)
- Club Kart Prize (2003)
NAOMI 2 GD-ROM Games
- Beach Spikers (2001)
- Virtua Fighter 4 (2001)
- Virtua Fighter 4: Evolution (2002)
- Virtua Fighter 4 Evolution Ver. B (2003)
- Virtua Fighter 4 Final Tuned (2004)
- Virtua Fighter 4 Ver. B (2001)
- Virtua Fighter 4 Ver. B (2001)
- Virtua Fighter 4 Ver. C (2002)
- Virtua Striker 3 (2001)
- Initial D: Arcade Stage (2002)
- Initial D: Arcade Stage 2 (2003)
- Initial D: Version 3 (2004)
NAOMI 2 Satellite Terminal Games
Notes
- ↑ [49 units, 656‑bit internal, 224‑bit external, 125 MHz, 9.25 GB/s[4] 49 units, 656‑bit internal, 224‑bit external, 125 MHz, 9.25 GB/s[4]]
- ↑ [42 units, 336‑bit (42× 8‑bit) internal, 120‑bit external,[8] 5.3 GB/s 42 units, 336‑bit (42× 8‑bit) internal, 120‑bit external,[8] 5.3 GB/s]
- ↑ [8‑bit,[9] 6 MB/s 8‑bit,[9] 6 MB/s]
- ↑ [2 units, 104‑bit (2× 52‑bit) internal, 32‑bit (2× 16‑bit) external,[10] 1.3403 GB/s 2 units, 104‑bit (2× 52‑bit) internal, 32‑bit (2× 16‑bit) external,[10] 1.3403 GB/s]
- ↑ [4 units, 208‑bit (4× 52‑bit) internal, 64‑bit (4× 16‑bit) external,[10] 2.6 GB/s 4 units, 208‑bit (4× 52‑bit) internal, 64‑bit (4× 16‑bit) external,[10] 2.6 GB/s]
- ↑ [75 floating-point operations per cycle 75 floating-point operations per cycle]
- ↑ [Scaled for high-end arcade technology,[13] with parallel ISP cores and increased PE processing elements within processor.[14] NAOMI 2 has average fillrate of 2 gigapixels/sec, twice that of the NAOMI's average 1 gigapixel/sec fillrate,[15] which in turn is twice that of the Dreamcast's average 500 megapixels/sec fillrate.[16] Scaled for high-end arcade technology,[13] with parallel ISP cores and increased PE processing elements within processor.[14] NAOMI 2 has average fillrate of 2 gigapixels/sec, twice that of the NAOMI's average 1 gigapixel/sec fillrate,[15] which in turn is twice that of the Dreamcast's average 500 megapixels/sec fillrate.[16]]
- ↑ [14 cycles/polygon per ISP FPU, 51 floating-point operations per polygon, 204 floating-point operations per 14 cycles[18][19] 14 cycles/polygon per ISP FPU, 51 floating-point operations per polygon, 204 floating-point operations per 14 cycles[18][19]]
- ↑ [32 pixels/cycle per ISP,1 pixel per PE (processor element), 128 PE (32 PE per ISP, 64 PE per PowerVR2), 6 gigapixels/sec per PowerVR2 (3.2 gigapixels/sec per ISP) 32 pixels/cycle per ISP,1 pixel per PE (processor element), 128 PE (32 PE per ISP, 64 PE per PowerVR2), 6 gigapixels/sec per PowerVR2 (3.2 gigapixels/sec per ISP)]
- ↑ [20 pixels per cycle, 6 PEs (processor elements) per pixel, 1 gigapixel per PowerVR2 (500 megapixels/sec per ISP) 20 pixels per cycle, 6 PEs (processor elements) per pixel, 1 gigapixel per PowerVR2 (500 megapixels/sec per ISP)]
- ↑ [60 layers depth, 4 pixels per cycle (2 pixels per PowerVR2), 32 PEs per pixel, 200 megapixels/sec per PowerVR2 (100 megapixels/sec per ISP) 60 layers depth, 4 pixels per cycle (2 pixels per PowerVR2), 32 PEs per pixel, 200 megapixels/sec per PowerVR2 (100 megapixels/sec per ISP)]
- ↑ [Same as pixel rendering fillrate Same as pixel rendering fillrate]
- ↑ [Elan: 7.5 GFLOPS (75 floating-point operations per cycle)
SH‑4 SIMD: 180 MHz available (10% load, 20 MHz used), 1.26 GFLOPS (90% of 1.4 GFLOPS) Elan: 7.5 GFLOPS (75 floating-point operations per cycle)
SH‑4 SIMD: 180 MHz available (10% load, 20 MHz used), 1.26 GFLOPS (90% of 1.4 GFLOPS)] - ↑ [Elan: 200 million vertices/sec (28 floating-point operations per transform,[26] 2 transforms per cycle)
SH-4: 45 million vertices/sec (4 cycles per transform)[27] Elan: 200 million vertices/sec (28 floating-point operations per transform,[26] 2 transforms per cycle)
SH-4: 45 million vertices/sec (4 cycles per transform)[27]] - ↑ [Elan: 200 million vertices/sec, 31 floating-point operations per vertex (28 operations for matrix transform,[26] 3 operations for perspective division),[28] 2 transforms per cycle
SH-4: 15 million vertices/sec, 12 cycles per vertex (4 cycles matrix transform,[27] 5 cycles perspective division),[28] 12 cycles division latency[29] Elan: 200 million vertices/sec, 31 floating-point operations per vertex (28 operations for matrix transform,[26] 3 operations for perspective division),[28] 2 transforms per cycle
SH-4: 15 million vertices/sec, 12 cycles per vertex (4 cycles matrix transform,[27] 5 cycles perspective division),[28] 12 cycles division latency[29]] - ↑ [Elan: 100 million vertices/sec (63 floating-point operations per vertex,[30] 1 vertex per cycle)
SH-4: 12 million vertices/sec, 14 cycles per vertex (4 cycles matrix transformation, 5 cycles perspective division, 1 cycle surface normal, 4 cycles lighting matrix)[31][32] Elan: 100 million vertices/sec (63 floating-point operations per vertex,[30] 1 vertex per cycle)
SH-4: 12 million vertices/sec, 14 cycles per vertex (4 cycles matrix transformation, 5 cycles perspective division, 1 cycle surface normal, 4 cycles lighting matrix)[31][32]] - ↑ [Elan: 20 million vertices/sec, 5 cycles per vertex (1 cycle transform, 1 cycle per light source)
SH-4: 6 million vertices/sec, 29 cycles per vertex (4 cycles matrix transformation, 5 cycles perspective division, 1 cycle per surface normal, 4 cycles per lighting matrix) Elan: 20 million vertices/sec, 5 cycles per vertex (1 cycle transform, 1 cycle per light source)
SH-4: 6 million vertices/sec, 29 cycles per vertex (4 cycles matrix transformation, 5 cycles perspective division, 1 cycle per surface normal, 4 cycles per lighting matrix)] - ↑ [Elan: 14 million vertices/sec, 7 cycles per vertex (1 cycle transform, 1 cycle per light source)
SH-4: 4 million vertices/sec, 39 cycles per vertex (4 cycles matrix transformation, 5 cycles perspective division, 1 cycle per surface normal, 4 cycles per lighting matrix) Elan: 14 million vertices/sec, 7 cycles per vertex (1 cycle transform, 1 cycle per light source)
SH-4: 4 million vertices/sec, 39 cycles per vertex (4 cycles matrix transformation, 5 cycles perspective division, 1 cycle per surface normal, 4 cycles per lighting matrix)] - ↑ [432,206 bytes 432,206 bytes]
- ↑ [288,322 bytes: 8 KB instruction cache, 16 KB data cache, 64 bytes store queue cache, 1538 bytes registers, 256 KB L2 cache[12] 288,322 bytes: 8 KB instruction cache, 16 KB data cache, 64 bytes store queue cache, 1538 bytes registers, 256 KB L2 cache[12]]
- ↑ [94,208 bytes: 16.5 KB register memory, 49 KB ISP cache, 26 KB TSP cache, 512 bytes FIFO buffer 94,208 bytes: 16.5 KB register memory, 49 KB ISP cache, 26 KB TSP cache, 512 bytes FIFO buffer]
- ↑ [32,780 bytes: 32 KB sound registers, 8 bytes RTC registers, 4 bytes FIFO buffer 32,780 bytes: 32 KB sound registers, 8 bytes RTC registers, 4 bytes FIFO buffer]
- ↑ [16,896 bytes: 512 bytes RAM, 16 KB ROM[34] 16,896 bytes: 512 bytes RAM, 16 KB ROM[34]]
- ↑ [2× 32 MB 2× 32 MB]
- ↑ [2 MB BIOS EPROM, 256 bytes EEPROM[4] 2 MB BIOS EPROM, 256 bytes EEPROM[4]]
- ↑ [24 MB EPROM,[36] 144–256 MB FlashROM/MROM 24 MB EPROM,[36] 144–256 MB FlashROM/MROM]
- ↑ [128–448 MB FlashROM, 0–40 MB EPROM, 128 KB Flash PROM[37] 128–448 MB FlashROM, 0–40 MB EPROM, 128 KB Flash PROM[37]]
- ↑ [128‑bit, 200 MHz 128‑bit, 200 MHz]
- ↑ [512‑bit, 100 MHz 512‑bit, 100 MHz]
- ↑ [2x 2304‑bit, 100 MHz: 2x 32-bit TA tile buffer,[38] 4x 32-bit ISP registers, 2x 32-bit TSP registers,[39] 4x 1024-bit ISP PE Arrays,[14] 2x 64-bit TSP Texture Cache,[40] 2x 32-bit TSP Tile Accumulation Buffer, 2x 32-bit Secondary Accumulation Buffer 2x 2304‑bit, 100 MHz: 2x 32-bit TA tile buffer,[38] 4x 32-bit ISP registers, 2x 32-bit TSP registers,[39] 4x 1024-bit ISP PE Arrays,[14] 2x 64-bit TSP Texture Cache,[40] 2x 32-bit TSP Tile Accumulation Buffer, 2x 32-bit Secondary Accumulation Buffer]
- ↑ [48‑bit, 35.4695 MHz 48‑bit, 35.4695 MHz]
- ↑ [32‑bit, 67 MHz 32‑bit, 67 MHz]
- ↑ [656‑bit, 125 MHz 656‑bit, 125 MHz]
- ↑ [128‑bit, 100 MHz[41] 128‑bit, 100 MHz[41]]
- ↑ [512‑bit, 100 MHz[11] 512‑bit, 100 MHz[11]]
- ↑ [128‑bit, 125 MHz[42] 128‑bit, 125 MHz[42]]
- ↑ [16‑bit, 66 MHz 16‑bit, 66 MHz]
- ↑ [16‑bit, 22 MHz[35] 16‑bit, 22 MHz[35]]
- ↑ [8‑bit, 6 MHz[9] 8‑bit, 6 MHz[9]]
- ↑ [16‑bit, 40 MHz[43][44] 16‑bit, 40 MHz[43][44]]
- ↑ [2× 16‑bit, 2 MHz[45] 2× 16‑bit, 2 MHz[45]]
- ↑ [50 MHz[46] 50 MHz[46]]
- ↑ [16‑bit, 50 MHz 16‑bit, 50 MHz]
- ↑ [24 MB BIOS EPROM, 256 bytes EEPROM 24 MB BIOS EPROM, 256 bytes EEPROM]
- ↑ [1/2× 64‑bit, 133 MHz[49][50] 1/2× 64‑bit, 133 MHz[49][50]]
References
- ↑ File:NextGeneration US 76.pdf, page 37
- ↑ NEC Introduces PowerVR 3-D Engine (09/23/98) (Wayback Machine: 1998-12-06 11:10)
- ↑ File:NextGeneration US 77.pdf, page 61
- ↑ 4.0 4.1 4.2 4.3 4.4 4.5 4.6 Sega NAOMI / NAOMI 2 (MAME)
- ↑ 5.0 5.1 File:DCUK 16.pdf, page 41
- ↑ File:SH-4 Software Manual.pdf
- ↑ 7.0 7.1 Press release: 2000-09-21: Sega Announces NAOMI2 Next Generation Arcade Systems Using Imagination Technologies’ PowerVR Graphics Architecture
- ↑ File:EPF8452A datasheet.pdf
- ↑ 9.0 9.1 9.2 File:EPC1064 datasheet.pdf
- ↑ 10.0 10.1 File:EPM7032AE datasheet.pdf
- ↑ 11.0 11.1 File:UPD4564323 datasheet.pdf
- ↑ 12.0 12.1 12.2 12.3 NAOMI 2 Specifications (May 31, 2001)
- ↑ File:PowerVR.pdf, page 2
- ↑ 14.0 14.1 14.2 File:PowerVR.pdf, page 3
- ↑ Press release: 1998-09-17: SEGA SELECTS POWERVR SERIES2 AS 3D GRAPHICS TECHNOLOGY FOR NEW ARCADE SYSTEM
- ↑ File:Edge UK 067.pdf, page 11
- ↑ File:DreamcastDevBoxSystemArchitecture.pdf, page 110
- ↑ 18.0 18.1 File:DreamcastDevBoxSystemArchitecture.pdf, page 95
- ↑ File:DreamcastDevBoxSystemArchitecture.pdf, page 203
- ↑ JAMMA 2000: NAOMI 2 Revealed (September 20, 2000)
- ↑ File:NAOMI 1998 Press Release JP.pdf
- ↑ File:BU142 datasheet.pdf
- ↑ Neon 250 Specs & Features (Wayback Machine: 2007-08-11 10:20)
- ↑ Sega Naomi Universal
- ↑ Dreamcast Video (KallistiOS)
- ↑ 26.0 26.1 Design of Digital Systems and Devices (page 95)
- ↑ 27.0 27.1 File:SH-4 Next-Generation DSP Architecture.pdf, page 12
- ↑ 28.0 28.1 Dreamcast: Basic matrix operations (KallistiOS)
- ↑ File:SH-4 Software Manual.pdf, page 211
- ↑ Design of Digital Systems and Devices (page 96)
- ↑ File:SH-4 Software Manual.pdf, page 151
- ↑ File:SH-4 Next-Generation DSP Architecture.pdf, page 31
- ↑ File:DreamcastDevBoxSystemArchitecture.pdf
- ↑ File:TMP90PH44 datasheet.pdf
- ↑ 35.0 35.1 File:HM62256B datasheet.pdf
- ↑ Club Kart: European Session (MAME)
- ↑ File:XCF01S datasheet.pdf
- ↑ File:DreamcastDevBoxSystemArchitecture.pdf, page 165
- ↑ PowerVR (Dreamcast Hardware)
- ↑ File:DreamcastDevBoxSystemArchitecture.pdf, page 96
- ↑ File:HM5264 datasheet.pdf
- ↑ File:HY57V161610D datasheet.pdf
- ↑ File:CY2292 datasheet.pdf
- ↑ File:M27C160 datasheet.pdf
- ↑ File:AT93C46 datasheet.pdf
- ↑ File:S29GL-N datasheet.pdf
- ↑ Hideki Sato Sega Inteview (Edge)
- ↑ Sega NAOMI DIMM board and GD-ROM
- ↑ Sega Naomi DIMM board and GD-ROM
- ↑ File:M366S3323CT0 datasheet.pdf
Sega arcade boards |
---|
Originating in arcades |
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
|
Console-based hardware |
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
00
01
02
03
04
05
06
07
08
09
10
11
12
13
14
|
PC-based hardware |
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
22
23
|