Difference between revisions of "Sega Hikaru"

Revision as of 08:47, 17 October 2018

Sega Hikaru

Manufacturer: Sega

Release

Date

RRP

Code

World

1999-05

?

837-13402

The Sega Hikaru is a successor of the Sega NAOMI and Sega Model 3 arcade systems that was developed in 1998 and debuted in 1999. The Hikaru was used for a handful of deluxe dedicated-cabinet games, beginning with 1999's Brave Fire Fighters, in which the flame and water effects were largely a showpiece for the hardware.

It was significantly more powerful and expensive than the NAOMI. The Hikaru featured a custom Sega GPU with advanced graphical capabilities, additional CPU and sound processors, various custom processors, increased memory, and faster bandwidth. It was the first game platform capable of effective hardware Phong shading, the most intensive form of shading at the time, and was capable of the most complex lighting and particle effects of its time.

It was the most powerful game system of its time (Planet Harriers in particular was regarded as having the best video game graphics at the time), but it was very expensive and difficult to program. Since it was comparatively expensive to produce, Sega soon abandoned the Hikaru in favor of continued NAOMI development. It was succeeded by the NAOMI 2, which was not quite as powerful but more affordable.^[1]

Technical specifications

Board composition: Main Board, ROM Board, AICA Sound Board, I/O Board, Filter Board, Network Board^[2]

Optional: Sound Board 2^[3]^[4]

Operating systems:

Sega native operating system
Custom Windows CE, with DirectX 6.0, Direct3D and OpenGL support

Extensions: communication, 4‑channel surround audio, PCI, MIDI, RS‑232C
Connection: JAMMA Video compliant, VGA

Main

Main CPU: 2× Hitachi SH‑4 @ 200 MHz^[5]

Units: 2× 128‑bit SIMD vector units with graphic functions, 2× 64‑bit floating‑point units, 2× 32‑bit fixed‑point units
Bus width: 256‑bit (2× 128‑bit) internal, 128‑bit (2× 64‑bit) external
Fixed‑point performance: 720 MIPS
Floating‑point performance: 2.8 GFLOPS
Note: With Sega Custom 3D GPU, the SH‑4's 128‑bit SIMD matrix unit can be dedicated to game physics, artificial intelligence, collision detection, overall game code, or additional graphical performance.

MIE bridge MCU: Sega 315‑6146 Maple‑JVS MCU (Zilog Z80)^[6]^[7] @ 14.7456 MHz^[2] (8/16‑bit instructions, 8‑bit bus, 2.14 MIPS)
Memory controllers: 2× Sega 315‑6154 Memory Controller @ 200 MHz (2× 32‑bit, DMA capabilities)^[3]^[2]^[8]
Main Board PLD: 27 units, 928‑bit (25 GB/sec) internal, 640‑bit (21 GB/sec) external^[2]^[9]

2× Sega PAL (Lattice GAL16V8) GAL @ 250 MHz: 16 units (2× 8 units), 128‑bit (2× 64‑bit), DMA control, graphics processing,^[10] 4 GB/sec
Sega 315‑6083A, 315‑6085, 315‑6086 @ 250 MHz: 3 units, 384‑bit (3× 128‑bit), 12 GB/sec
Sega 315‑6202 (Lattice CY37128) CPLD @ 167 MHz: 8 units, 416‑bit (8× 52‑bit) internal (9 GB/sec), 128‑bit (8× 16‑bit) external (3 GB/sec)^[11]

Network Board processors:^[2]

Network CPU: Motorola 68000 @ 40 MHz (16/32‑bit instructions, 16‑bit bus, 7 MIPS)^[2]
Network PLD: FPGA @ 180 MHz (32‑bit),^[12] 3× PAL @ 40 MHz, Sega 315‑5804 CPLD @ 40 MHz
Network processors: 2× Sega 315‑5917 @ 40 MHz

ROM Board processors:^[2]

ROM Board PLD: FPGA @ 180 MHz (32‑bit),^[12] CPLD/PAL @ 182 MHz (32‑bit)^[13]^[14]
Security IC: Sega 315-5881 @ 28 MHz

Sound

Sound engine: 2× Yamaha AICA Super Intelligent Sound Processor (315‑6232) @ 67 MHz

Internal CPU: 2× 32‑bit ARM7 RISC CPU @ 45 MHz
CPU performance: 34 MIPS (2× 17 MIPS)
PCM/ADPCM: 16‑bit depth, 48 kHz sampling rate (DVD quality), 128 channels
Bus width: 32‑bit (2× 16‑bit)
Other features: DSP, sound synthesizer

Graphics

The Sega Hikaru uses custom 3D graphics hardware, which include the following specifications:^[15]^[16]^[3]

Graphics Engine GPU: Sega Custom 3D GPU @ 250 MHz
GPU core processors: 7 processors^[2]^[9]

2× Sega GPU 15 CP Command Processors (315‑6197) @ 250 MHz: 512‑bit (2× 256‑bit), Geometry Processor
Sega GPU 1A Image Generator (315‑6087) @ 250 MHz: 128‑bit, rasterizer/renderer^[17]
2× Sega GPU DMA controllers (315‑6084) @ 250 MHz: 256‑bit (2× 128‑bit)
2× Analog Devices ADV7120 Video DAC @ 80 MHz: 48‑bit (2× 24‑bit)^[18]

GPU Geometry Processors: 2× Sega GPU 15 CP Command Processors

Hardware T&L: Transform, clipping, lighting
Materials: Flat shading, Gouraud shading, Phong shading, diffuse, ambient, specular, unlit
Fog: Color, transparency, density, depth blend, translucency
Rendering: Double‑buffered 3D rendering (odd & even frames), depth cueing, depth buffer, depth bias, face culling, static meshes, dynamic meshes
Shading: Flat shading, Gouraud shading, Phong shading, diffuse, ambient, specular, linear
Modelview matrix: Instanced drawing, multiple instances, shared attributes between models,^[19] modelview stack
Object memory: 8 viewports, 256 modelviews, 16,384 materials (256 LOD levels), 16,384 textures/texheads (256 LOD levels), 1024 lights (256 light sets)

GPU DMA controllers: 2× Sega GPU DMA controllers

GPU IDMA (Indirect DMA) controller: Loads texture data from MaskROM (via external bus) into texture banks (with metadata), allows CPU access to texture banks
DMA controller: Moves textures around in framebuffer, transfers bitmap data to bitmap layers, allows CPU access to framebuffer

Color depth: 32‑bit ARGB, 16,777,216 colors (24‑bit color) with 8‑bit (256 levels) alpha blending, YUV and RGB color space, color key overlay
Display resolution: 31 kHz horizontal sync, 60 Hz refresh rate, 80 MHz Video DAC, JAMMA/VGA output, progressive scan^[2]^[18]

Single monitor display: 496×384 to 800×608 (default 640×480)
Dual monitor display: 992×384 to 1600×608 (default 1280×480)
Video output: 496×384 to 1968×1080 (default 640×480)
Framebuffer: 496×384 to 2048×2048 (default 2048×2048)^[2]

Lighting: 1024 lights per scene, 4 lights per polygon, 256 light sets per scene (4 lights per set), 8 window surfaces

Light types: Diffuse, ambient, specular, horizontal, spot
Emission types: Constant, linear, infinite linear, square, reciprocal, reciprocal squared
Object types: Lights (with individual position, direction and emission properties), lightsets (a set of up to 4 lights that share a mesh)

GPU capabilities: 2 bitmap layers, calendar, 16,384 vertices per mesh,^[20] hidden surface removal, deferred rendering

Framebuffer: 2048×2048 sheet (can be partitioned into framebuffer, tile data, and/or 1‑2 bitmap layers), handled by 2 GPU 1A Image Generator rasterizers/renderers (double‑buffering), accessible by DMA controller
Effects: Stencil, shadows, motion blur, particle effects, fire effects, water effects,^[21] fog, alpha blending, anti‑aliasing, specular effects,^[22]
Features: Tiled rendering, deferred rendering, back‑face culling, hidden surface removal
Phong shading: Per‑pixel lighting/shading computation processed by rasterization pipeline,^[23] deferred rendering can prevent shading of overdrawn pixels to maximize rendering bandwidth,^[24] all compute units could be used as shader units to maximize Phong shading performance

Texture capabilities: 1×1 to 2048×2048 texture sizes, mipmapping, mipmap trees, texture panning, multi‑texturing, bump mapping, normal mapping, texture filtering, bilinear filtering, trilinear filtering, environment mapping^[22]

Texture banks: 2 texture banks (stored as 2× 2048×1024 sheets), stores textures from MaskROM (with 16‑byte metadata per texture in Command RAM)

Floating-point performance: 15–30 GFLOPS^{[n 1]}
Rendering fillrate:

Opaque polygons: 8 GPixels/s^{[n 2]}
Translucent polygons: 1.6 GPixels/s (8bpp), 800 MPixels/s (16bpp), 530 MPixels/s (24bpp)^{[n 3]}

Texture fillrate:

Opaque polygons: 8 GTexels/s
Translucent polygons: 4 GTexels/s (8bpp), 2 GTexels/s (16bpp), 1.3 GTexels/s (24bpp)^{[n 4]}

Phong shading performance:

2 million polygons/sec: 4 lights/polygon, 50-100-pixel polygons^{[n 5]}
3-6 million polygons/sec: 4 lights/polygon, 32-pixel polygons^{[n 6]}
10-20 million polygons/sec: 1 light/polygon, 32-pixel polygons^{[n 7]}

Gouraud shading performance:

30-60 million polygons/sec: 4 lights/polygon,^{[n 8]} 32-pixel polygons
100-200 million polygons/sec: 1 light/polygon,^{[n 9]} 32-pixel polygons

Texture mapping performance:

80 million polygons/sec: 50-100-texel polygons
100-200 million polygons/sec: 32-texel polygons

Memory

Memory: Up to 465 MB^[2]

Main memory: 130 MB^{[n 10]}
Video memory: 286.25 MB^{[n 11]}
Sound memory: 48 MB^{[n 12]}
Other memory: 480 KB^{[n 13]}

RAM: 110.625 MB^[8]^[2]

Main RAM: 64 MB SDRAM^[28]^{[n 14]}
VRAM: 30.25 MB^[29]^[15]

4 MB Command RAM^{[n 15]}
8 MB texture banks^{[n 16]}
8 MB framebuffer SDRAM^{[n 17]}
10 MB other SDRAM^{[n 18]}
256 KB Synchronous SRAM^[30]

Sound RAM: 16 MB SDRAM^{[n 19]}
Other Main Board RAM: 128 KB SRAM^[31]

MIE RAM: 64 KB^[6]
Backup RAM: 64 KB

Network Board RAM: 192 KB SRAM^[31]
ROM Board RAM: 64 KB SRAM^[32]

ROM: Up to 354 MB

Boot ROM: 2 MB EPROM^[2]
Game ROM Board: Up to 352 MB (64 MB main ROM, 256 MB video MaskROM,^[33] 32 MB sound ROM)^[3]
Note: High‑speed access allows ROM to effectively be used as RAM, with polygons and textures streamed directly from game ROM Board.^[34]

Cache: 96 KB^{[n 20]}

Bandwidth

Internal processor cache bandwidth:^[2]

SH‑4 cache: 6.4 GB/s^{[n 21]}
Sega Custom 3D GPU: 32 GB/s
AICA Sound Processor: 536 MB/s^{[n 22]}
Z80 MIE MCU: 15 MB/s^{[n 23]}
315‑6154 Memory Controllers: 1.6 GB/s^{[n 24]}
Main Board PLD: 25 GB/sec^{[n 25]}
ROM Board PLD: 1.45 GB/s^{[n 26]}
Network Board 68000: 80 MB/s^{[n 27]}
Network Board FPGA: 720 MB/s^{[n 28]}

RAM bandwidth: 35.111 GB/s^[2]

Main RAM: 2.4 GB/s^{[n 29]}

SH‑4: 1.6 GB/s^{[n 30]}
Memory Controllers: 800 MB/s^{[n 31]}

VRAM: 32 GB/s^{[n 32]}

Command RAM: 16 GB/s^{[n 33]}
Texture banks: 8 GB/s^{[n 34]}
SDRAM: 7.2 GB/s^{[n 35]}
Synchronous SRAM: 400 MB/s^{[n 36]}

Sound RAM: 268 MB/s^{[n 37]}
HM62256 SRAM: 193 MB/s^{[n 38]}

MIE RAM: 15 MB/s^{[n 23]}
Backup RAM: 44.444444 MB/s^{[n 39]}
Network Board RAM: 133.333333 MB/s^{[n 40]}

ROM Board RAM: 250 MB/s^{[n 41]}

ROM bandwidth: 4 GB/s^[2]

Boot ROM: 800 MB/s^{[n 42]}
ROM Board PLD: 1.45 GB/s^{[n 43]}
ROM Board Connectors: 1.5 GB/s^{[n 44]}

List of games

Brave FireFighters (1999)
NASCAR Arcade (2000)
Planet Harriers (2000)
Star Wars Racer Arcade (2000)
Air Trix (2001)
Cyber Troopers Virtual-On Force (2001)
Cyber Troopers Virtual-On Force Ver.7.7 (2002)

History

According to Sega in 1999:^[21]

“	Brave Firefighters utilizes a slightly modified Naomi Hardware system called Hikaru. Hikaru incorporates a custom Sega graphics chip and possesses larger memory capacity then standard Naomi systems. "These modifications were necessary because in Brave Firefighters, our engineers were faced with the daunting challenge of creating 3d images of flames and sprayed water," stated Sega's Vice President of Sales and Marketing, Barbara Joyiens. "If you stop and think about it, both have an almost infinite number of shapes, sizes, colors, levels of opaqueness, shadings and shadows. And, when you combine the two by simulating the spraying of water on a flame, you create an entirely different set of challenges for our game designers and engineers to overcome; challenges that would be extremely difficult, if not impossible to overcome utilizing existing 3D computers. Hikaru has the horsepower to handle these demanding graphic challenges with clarity, depth and precision."	„

In addition, the Hikaru also uses two Hitachi SH-4 CPU processors, two Yamaha AICA sound engine processors, a Motorola 68000 network CPU, and a dual GPU setup. The Hikaru hardware was largely complete in 1998, before it was released to the public in 1999.^[2] The system was very expensive, and difficult to program.^[1] The word "Hikaru" (ひかる) means "to shine" in Japanese.

Photo gallery

Main PCB board
Sound board
Comm board
ROM board (left side)
ROM board (right side)
Main board (Up Left view)
Main board (Up Right view)
Main board (Down Left view)
Main board (Down Right view)
Custom board A
Custom board B
Outer Case
Inside, assembled

Notes

↑ [Phong shading: 2 million polygons/sec, 4 light sources per polygon, 7,898–15,796 floating-point operations per 100-pixel polygon, 15–30 GFLOPS Phong shading: 2 million polygons/sec, 4 light sources per polygon, 7,898–15,796 floating-point operations per 100-pixel polygon, 15–30 GFLOPS]
↑ [32 pixels per cycle,^[25] 250 MHz 32 pixels per cycle,^[25] 250 MHz]
↑ [3.2 GB/s framebuffer SDRAM bandwidth (double-buffered) 3.2 GB/s framebuffer SDRAM bandwidth (double-buffered)]
↑ [8 GB/s texture bank RAM bandwidth (double-buffered) 8 GB/s texture bank RAM bandwidth (double-buffered)]
↑ [7,898–15,796 floating-point operations per 50-100-pixel polygon for 4 light sources (196 operations setup, 156 operations per pixel),^[26] 15–30 GFLOPS 7,898–15,796 floating-point operations per 50-100-pixel polygon for 4 light sources (196 operations setup, 156 operations per pixel),^[26] 15–30 GFLOPS]
↑ [5188 floating-point operations per 32-pixel polygon for 4 light sources (113 operations setup, 156 operations per pixel )^[26] 5188 floating-point operations per 32-pixel polygon for 4 light sources (113 operations setup, 156 operations per pixel )^[26]]
↑ [1297 floating-point operations per 32-pixel polygon for 1 light source (49 operations setup, 39 operations per pixel)^[26] 1297 floating-point operations per 32-pixel polygon for 1 light source (49 operations setup, 39 operations per pixel)^[26]]
↑ [181 floating-point operations per vertex/polygon: 113 operations for transformation, 68 operations for 4 light sources (17 operations per light source)^[27] 181 floating-point operations per vertex/polygon: 113 operations for transformation, 68 operations for 4 light sources (17 operations per light source)^[27]]
↑ [130 floating-point operations per vertex/polygon: 113 operations for transformation, 17 operations for 1 light source^[27] 130 floating-point operations per vertex/polygon: 113 operations for transformation, 17 operations for 1 light source^[27]]
↑ [64 MB RAM, 66 MB ROM 64 MB RAM, 66 MB ROM]
↑ [30.25 MB RAM, 256 MB ROM 30.25 MB RAM, 256 MB ROM]
↑ [16 MB RAM, 32 MB ROM 16 MB RAM, 32 MB ROM]
↑ [384 KB RAM, 96 KB cache 384 KB RAM, 96 KB cache]
↑ [32 MB per SH‑4 32 MB per SH‑4]
↑ [2× 315‑6197 2× 315‑6197]
↑ [2× 315‑6084 2× 315‑6084]
↑ [IC 44‑47^[2] IC 44‑47^[2]]
↑ [1 MB Geometry Processor, 1 MB Image Generator 1 MB Geometry Processor, 1 MB Image Generator]
↑ [8 MB per AICA 8 MB per AICA]
↑ [48 KB per SH‑4 CPU^[35] 48 KB per SH‑4 CPU^[35]]
↑ [256‑bit, 200 MHz 256‑bit, 200 MHz]
↑ [2× 32‑bit, 67 MHz 2× 32‑bit, 67 MHz]
↑ ^23.0 ^23.1 [8‑bit, 14.7456 MHz 8‑bit, 14.7456 MHz]
↑ [64‑bit, 200 MHz 64‑bit, 200 MHz]
↑ [928‑bit, 250 MHz 928‑bit, 250 MHz]
↑ [2× 32‑bit, 180/182 MHz 2× 32‑bit, 180/182 MHz]
↑ [16‑bit, 40 MHz 16‑bit, 40 MHz]
↑ [32‑bit, 180 MHz^[12] 32‑bit, 180 MHz^[12]]
↑ [192‑bit, 100 MHz, 6 ns^[28] 192‑bit, 100 MHz, 6 ns^[28]]
↑ [128‑bit 128‑bit]
↑ [64‑bit 64‑bit]
↑ [1088‑bit, 4.5 ns 1088‑bit, 4.5 ns]
↑ [315‑6197, 512‑bit, 250 MHz 315‑6197, 512‑bit, 250 MHz]
↑ [315‑6084, 256‑bit, 250 MHz 315‑6084, 256‑bit, 250 MHz]
↑
[9× 32‑bit, 200 MHz, 4.5 ns^[29]
- Framebuffer: 3.2 GB/s (4x 32‑bit,^[2] 200 MHz)
9× 32‑bit, 200 MHz, 4.5 ns^[29]
- Framebuffer: 3.2 GB/s (4x 32‑bit,^[2] 200 MHz)]
↑ [32‑bit, 100 MHz, 5 ns^[30] 32‑bit, 100 MHz, 5 ns^[30]]
↑ [32‑bit, 67 MHz, 6 ns^[36] 32‑bit, 67 MHz, 6 ns^[36]]
↑ [72‑bit, 45 ns^[31] 72‑bit, 45 ns^[31]]
↑ [16‑bit, 22.222222 MHz 16‑bit, 22.222222 MHz]
↑ [48‑bit, 22.222222 MHz 48‑bit, 22.222222 MHz]
↑ [16‑bit, 125 MHz, 8 ns^[32] 16‑bit, 125 MHz, 8 ns^[32]]
↑ [64‑bit, 100 MHz^[37] 64‑bit, 100 MHz^[37]]
↑ [2× 32‑bit, 180/182 MHz^[12]^[13]^[14] 2× 32‑bit, 180/182 MHz^[12]^[13]^[14]]
↑ 2x 32‑bit, 182 MHz (Wayback Machine: 2016-02-28 17:27)

References

↑ ^1.0 ^1.1 File:NextGeneration US 76.pdf, page 37
↑ ^2.00 ^2.01 ^2.02 ^2.03 ^2.04 ^2.05 ^2.06 ^2.07 ^2.08 ^2.09 ^2.10 ^2.11 ^2.12 ^2.13 ^2.14 ^2.15 ^2.16 ^2.17 ^2.18 Sega Hikaru (MAME)
↑ ^3.0 ^3.1 ^3.2 ^3.3 Sega Hikaru Memory Controller (Valkyrie)
↑ Sega Hikaru AICA Sound Boards (Valkyrie)
↑ File:SH-4 Software Manual.pdf
↑ ^6.0 ^6.1 Sega Hikaru MIE (Valkyrie)
↑ Demul 0.56 (1 September 2010)
↑ ^8.0 ^8.1 Sega Hikaru (Valkyrie)
↑ ^9.0 ^9.1 File:Hikaru rombd upright.jpg
↑ File:GAL16V8 datasheet.pdf
↑ File:CY37 datasheet.pdf
↑ ^12.0 ^12.1 ^12.2 ^12.3 File:PLSI2032 datasheet.pdf
↑ ^13.0 ^13.1 File:M4A3 datasheet.pdf
↑ ^14.0 ^14.1 File:MACH111 datasheet.pdf
↑ ^15.0 ^15.1 Sega Hikaru GPU (Valkyrie)
↑ Sega Hikaru GPU CP (Valkyrie)
↑ Sega Hikaru Renderer (Valkyrie)
↑ ^18.0 ^18.1 File:ADV7120 datasheet.pdf
↑ Instanced Rendering
↑ Sega Hikaru GPU Private (Valkyrie)
↑ ^21.0 ^21.1 Sega Confirms Hikaru Does Exist (November 24, 1999)
↑ ^22.0 ^22.1 File:NAOMI 1998 Press Release JP.pdf
↑ Computer Graphics: Principles and Practice (page 871)
↑ Computer Graphics: Principles and Practice (page 900)
↑ Sega NAOMI
↑ ^26.0 ^26.1 ^26.2 Computer Vision and Graphics: International Conference, ICCVG 2004, Warsaw, Poland, September 2004, Proceedings (page 78)
↑ ^27.0 ^27.1 Design of Digital Systems and Devices (pages 95-97)
↑ ^28.0 ^28.1 File:HM5264 datasheet.pdf
↑ ^29.0 ^29.1 File:HY57V161610D datasheet.pdf
↑ ^30.0 ^30.1 File:UPD432232 datasheet.pdf
↑ ^31.0 ^31.1 ^31.2 File:HM62256B datasheet.pdf
↑ ^32.0 ^32.1 File:CY7C199 datasheet.pdf
↑ Planet Harriers (MAME)
↑ Hideki Sato Sega Interview (Edge)
↑ File:SH-4 32-bit CPU Core Architecture.pdf
↑ File:K4S641632 datasheet.pdf
↑ File:CY2292 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 VIC Dual G80 Zaxxon VCO Object System 1 LaserDisc Appoooh Hang-On System 2 System 16 OutRun Gigas X Board Y Board System 24 System 18 System 32 Model 1 Model 2 H1 Board Model 3
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 SG-1000 System E System C Mega-Tech System Mega Play Sega Titan Video NAOMI Hikaru NAOMI 2 Atomiswave System SP Chihiro Triforce
PC-based hardware
05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Lindbergh Europa-R RingEdge RingWide RingEdge 2 Nu Nu 1.1 Nu 2 ALLS

[Phong_shading:_2_million_polygons.2Fsec.2C_4_light_sources_per_polygon.2C_7.2C898.E2.80.9315.2C796_floating-point_operations_per_100-pixel_polygon.2C_15.E2.80.9330_GFLOPS-25] [Phong shading: 2 million polygons/sec, 4 light sources per polygon, 7,898–15,796 floating-point operations per 100-pixel polygon, 15–30 GFLOPS Phong shading: 2 million polygons/sec, 4 light sources per polygon, 7,898–15,796 floating-point operations per 100-pixel polygon, 15–30 GFLOPS]

[32_pixels_per_cycle.2C.7F.27.22.60UNIQ--ref-00000047-QINU.60.22.27.7F_250_MHz-27] [32 pixels per cycle,^[25] 250 MHz 32 pixels per cycle,^[25] 250 MHz]

[3.2_GB.2Fs_framebuffer_SDRAM_bandwidth_.28double-buffered.29-28] [3.2 GB/s framebuffer SDRAM bandwidth (double-buffered) 3.2 GB/s framebuffer SDRAM bandwidth (double-buffered)]

[8_GB.2Fs_texture_bank_RAM_bandwidth_.28double-buffered.29-29] [8 GB/s texture bank RAM bandwidth (double-buffered) 8 GB/s texture bank RAM bandwidth (double-buffered)]

[7.2C898.E2.80.9315.2C796_floating-point_operations_per_50-100-pixel_polygon_for_4_light_sources_.28196_operations_setup.2C_156_operations_per_pixel.29.2C.7F.27.22.60UNIQ--ref-0000004B-QINU.60.22.27.7F_15.E2.80.9330_GFLOPS-31] [7,898–15,796 floating-point operations per 50-100-pixel polygon for 4 light sources (196 operations setup, 156 operations per pixel),^[26] 15–30 GFLOPS 7,898–15,796 floating-point operations per 50-100-pixel polygon for 4 light sources (196 operations setup, 156 operations per pixel),^[26] 15–30 GFLOPS]

[5188_floating-point_operations_per_32-pixel_polygon_for_4_light_sources_.28113_operations_setup.2C_156_operations_per_pixel_.29.7F.27.22.60UNIQ--ref-0000004D-QINU.60.22.27.7F-32] [5188 floating-point operations per 32-pixel polygon for 4 light sources (113 operations setup, 156 operations per pixel )^[26] 5188 floating-point operations per 32-pixel polygon for 4 light sources (113 operations setup, 156 operations per pixel )^[26]]

[1297_floating-point_operations_per_32-pixel_polygon_for_1_light_source_.2849_operations_setup.2C_39_operations_per_pixel.29.7F.27.22.60UNIQ--ref-0000004F-QINU.60.22.27.7F-33] [1297 floating-point operations per 32-pixel polygon for 1 light source (49 operations setup, 39 operations per pixel)^[26] 1297 floating-point operations per 32-pixel polygon for 1 light source (49 operations setup, 39 operations per pixel)^[26]]

[181_floating-point_operations_per_vertex.2Fpolygon:_113_operations_for_transformation.2C_68_operations_for_4_light_sources_.2817_operations_per_light_source.29.7F.27.22.60UNIQ--ref-00000051-QINU.60.22.27.7F-35] [181 floating-point operations per vertex/polygon: 113 operations for transformation, 68 operations for 4 light sources (17 operations per light source)^[27] 181 floating-point operations per vertex/polygon: 113 operations for transformation, 68 operations for 4 light sources (17 operations per light source)^[27]]

[130_floating-point_operations_per_vertex.2Fpolygon:_113_operations_for_transformation.2C_17_operations_for_1_light_source.7F.27.22.60UNIQ--ref-00000053-QINU.60.22.27.7F-36] [130 floating-point operations per vertex/polygon: 113 operations for transformation, 17 operations for 1 light source^[27] 130 floating-point operations per vertex/polygon: 113 operations for transformation, 17 operations for 1 light source^[27]]

[64.26nbsp.3BMB_RAM.2C_66.26nbsp.3BMB_ROM-37] [64 MB RAM, 66 MB ROM 64 MB RAM, 66 MB ROM]

[30.25.26nbsp.3BMB_RAM.2C_256.26nbsp.3BMB_ROM-38] [30.25 MB RAM, 256 MB ROM 30.25 MB RAM, 256 MB ROM]

[16.26nbsp.3BMB_RAM.2C_32.26nbsp.3BMB_ROM-39] [16 MB RAM, 32 MB ROM 16 MB RAM, 32 MB ROM]

[384.26nbsp.3BKB_RAM.2C_96.26nbsp.3BKB_cache-40] [384 KB RAM, 96 KB cache 384 KB RAM, 96 KB cache]

[32.26nbsp.3BMB_per_SH.E2.80.914-42] [32 MB per SH‑4 32 MB per SH‑4]

[2.C3.97_315.E2.80.916197-44] [2× 315‑6197 2× 315‑6197]

[2.C3.97_315.E2.80.916084-45] [2× 315‑6084 2× 315‑6084]

[IC_44.E2.80.9147.7F.27.22.60UNIQ--ref-00000066-QINU.60.22.27.7F-46] [IC 44‑47^[2] IC 44‑47^[2]]

[1.26nbsp.3BMB_Geometry_Processor.2C_1.26nbsp.3BMB_Image_Generator-47] [1 MB Geometry Processor, 1 MB Image Generator 1 MB Geometry Processor, 1 MB Image Generator]

[8.26nbsp.3BMB_per_AICA-49] [8 MB per AICA 8 MB per AICA]

[48.26nbsp.3BKB_per_SH.E2.80.914_CPU.7F.27.22.60UNIQ--ref-0000007D-QINU.60.22.27.7F-55] [48 KB per SH‑4 CPU^[35] 48 KB per SH‑4 CPU^[35]]

[256.E2.80.91bit.2C_200.26nbsp.3BMHz-56] [256‑bit, 200 MHz 256‑bit, 200 MHz]

[2.C3.97.26nbsp.3B32.E2.80.91bit.2C_67.26nbsp.3BMHz-57] [2× 32‑bit, 67 MHz 2× 32‑bit, 67 MHz]

[8.E2.80.91bit.2C_14.7456.26nbsp.3BMHz-58] 23.0 ^23.1 [8‑bit, 14.7456 MHz 8‑bit, 14.7456 MHz]

[64.E2.80.91bit.2C_200.26nbsp.3BMHz-59] [64‑bit, 200 MHz 64‑bit, 200 MHz]

[928.E2.80.91bit.2C_250.26nbsp.3BMHz-60] [928‑bit, 250 MHz 928‑bit, 250 MHz]

[2.C3.97.26nbsp.3B32.E2.80.91bit.2C_180.2F182.26nbsp.3BMHz-61] [2× 32‑bit, 180/182 MHz 2× 32‑bit, 180/182 MHz]

[16.E2.80.91bit.2C_40.26nbsp.3BMHz-62] [16‑bit, 40 MHz 16‑bit, 40 MHz]

[32.E2.80.91bit.2C_180.26nbsp.3BMHz.7F.27.22.60UNIQ--ref-00000089-QINU.60.22.27.7F-63] [32‑bit, 180 MHz^[12] 32‑bit, 180 MHz^[12]]

[192.E2.80.91bit.2C_100.26nbsp.3BMHz.2C_6.26nbsp.3B.5B.5Bwikipedia:Nanosecond.7Cns.5D.5D.7F.27.22.60UNIQ--ref-0000008E-QINU.60.22.27.7F-64] [192‑bit, 100 MHz, 6 ns^[28] 192‑bit, 100 MHz, 6 ns^[28]]

[128.E2.80.91bit-65] [128‑bit 128‑bit]

[64.E2.80.91bit-66] [64‑bit 64‑bit]

[1088.E2.80.91bit.2C_4.5.26nbsp.3Bns-67] [1088‑bit, 4.5 ns 1088‑bit, 4.5 ns]

[315.E2.80.916197.2C_512.E2.80.91bit.2C_250.26nbsp.3BMHz-68] [315‑6197, 512‑bit, 250 MHz 315‑6197, 512‑bit, 250 MHz]

[315.E2.80.916084.2C_256.E2.80.91bit.2C_250.26nbsp.3BMHz-69] [315‑6084, 256‑bit, 250 MHz 315‑6084, 256‑bit, 250 MHz]

[9.C3.97.26nbsp.3B32.E2.80.91bit.2C_200.26nbsp.3BMHz.2C_4.5.26nbsp.3Bns.7F.27.22.60UNIQ--ref-00000097-QINU.60.22.27.7F.0A.2AFramebuffer:_3.2_GB.2Fs_.284x_32.E2.80.91bit.2C.7F.27.22.60UNIQ--ref-00000098-QINU.60.22.27.7F_200_MHz.29-70] [9× 32‑bit, 200 MHz, 4.5 ns^[29]
Framebuffer: 3.2 GB/s (4x 32‑bit,^[2] 200 MHz)
9× 32‑bit, 200 MHz, 4.5 ns^[29]
Framebuffer: 3.2 GB/s (4x 32‑bit,^[2] 200 MHz)]

[36] Framebuffer: 3.2 GB/s (4x 32‑bit,^[2] 200 MHz)

[37] Framebuffer: 3.2 GB/s (4x 32‑bit,^[2] 200 MHz)]

[32.E2.80.91bit.2C_100.26nbsp.3BMHz.2C_5.26nbsp.3Bns.7F.27.22.60UNIQ--ref-0000009C-QINU.60.22.27.7F-71] [32‑bit, 100 MHz, 5 ns^[30] 32‑bit, 100 MHz, 5 ns^[30]]

[32.E2.80.91bit.2C_67.26nbsp.3BMHz.2C_6.26nbsp.3Bns.7F.27.22.60UNIQ--ref-000000A0-QINU.60.22.27.7F-73] [32‑bit, 67 MHz, 6 ns^[36] 32‑bit, 67 MHz, 6 ns^[36]]

[72.E2.80.91bit.2C_45.26nbsp.3Bns.7F.27.22.60UNIQ--ref-000000A4-QINU.60.22.27.7F-74] [72‑bit, 45 ns^[31] 72‑bit, 45 ns^[31]]

[16.E2.80.91bit.2C_22.222222.26nbsp.3BMHz-75] [16‑bit, 22.222222 MHz 16‑bit, 22.222222 MHz]

[48.E2.80.91bit.2C_22.222222.26nbsp.3BMHz-76] [48‑bit, 22.222222 MHz 48‑bit, 22.222222 MHz]

[16.E2.80.91bit.2C_125.26nbsp.3BMHz.2C_8.26nbsp.3Bns.7F.27.22.60UNIQ--ref-000000AB-QINU.60.22.27.7F-77] [16‑bit, 125 MHz, 8 ns^[32] 16‑bit, 125 MHz, 8 ns^[32]]

[64.E2.80.91bit.2C_100.26nbsp.3BMHz.7F.27.22.60UNIQ--ref-000000B0-QINU.60.22.27.7F-79] [64‑bit, 100 MHz^[37] 64‑bit, 100 MHz^[37]]

[2.C3.97.26nbsp.3B32.E2.80.91bit.2C_180.2F182.26nbsp.3BMHz.7F.27.22.60UNIQ--ref-000000B4-QINU.60.22.27.7F.7F.27.22.60UNIQ--ref-000000B7-QINU.60.22.27.7F.7F.27.22.60UNIQ--ref-000000BA-QINU.60.22.27.7F-80] [2× 32‑bit, 180/182 MHz^[12]^[13]^[14] 2× 32‑bit, 180/182 MHz^[12]^[13]^[14]]

[.5Bhttps:.2F.2Fweb.archive.org.2Fweb.2F20160228172702.2Fmembers.iinet.net.au.2F.7Elantra9jp1.2Fgurudumps.2Fwip.2Fhikaru1.jpg_2x.5D_.5Bhttps:.2F.2Fgithub.com.2Fstefanoteso.2Fvalkyrie.2Fblob.2Fmaster.2Fsrc.2Fmach.2Fhikaru.2Fhikaru-gpu.c_32.E2.80.91bit.5D.2C_182.26nbsp.3BMHz-81] 2x 32‑bit, 182 MHz (Wayback Machine: 2016-02-28 17:27)

[:File:NextGeneration_US_76.pdf_p37-1] 1.0 ^1.1 File:NextGeneration US 76.pdf, page 37

[.5Bhttps:.2F.2Fgithub.com.2Fmamedev.2Fmame.2Fblob.2Fmaster.2Fsrc.2Fmame.2Fdrivers.2Fhikaru.cpp_Sega_Hikaru_.28MAME.29.5D-2] 2.00 ^2.01 ^2.02 ^2.03 ^2.04 ^2.05 ^2.06 ^2.07 ^2.08 ^2.09 ^2.10 ^2.11 ^2.12 ^2.13 ^2.14 ^2.15 ^2.16 ^2.17 ^2.18 Sega Hikaru (MAME)

[.5Bhttps:.2F.2Fgithub.com.2Fstefanoteso.2Fvalkyrie.2Fblob.2Fmaster.2Fsrc.2Fmach.2Fhikaru.2Fhikaru-memctl.c_Sega_Hikaru_Memory_Controller_.28Valkyrie.29.5D-3] 3.0 ^3.1 ^3.2 ^3.3 Sega Hikaru Memory Controller (Valkyrie)

[.5Bhttps:.2F.2Fgithub.com.2Fstefanoteso.2Fvalkyrie.2Fblob.2Fmaster.2Fsrc.2Fmach.2Fhikaru.2Fhikaru-aica.c_Sega_Hikaru_AICA_Sound_Boards_.28Valkyrie.29.5D-4] Sega Hikaru AICA Sound Boards (Valkyrie)

[:File:SH-4_Software_Manual.pdf_p-5] File:SH-4 Software Manual.pdf

[.5Bhttps:.2F.2Fgithub.com.2Fstefanoteso.2Fvalkyrie.2Fblob.2Fmaster.2Fsrc.2Fmach.2Fhikaru.2Fhikaru-mie.c_Sega_Hikaru_MIE_.28Valkyrie.29.5D-6] 6.0 ^6.1 Sega Hikaru MIE (Valkyrie)

[.5Bhttp:.2F.2Fdemul.emulation64.com_Demul_0.56_.281_September_2010.29.5D-7] Demul 0.56 (1 September 2010)

[.5Bhttps:.2F.2Fgithub.com.2Fstefanoteso.2Fvalkyrie.2Fblob.2Fmaster.2Fsrc.2Fmach.2Fhikaru.2Fhikaru.c_Sega_Hikaru_.28Valkyrie.29.5D-8] 8.0 ^8.1 Sega Hikaru (Valkyrie)

[:File:Hikaru_rombd_upright.jpg_p-9] 9.0 ^9.1 File:Hikaru rombd upright.jpg

[:File:GAL16V8_datasheet.pdf_p-10] File:GAL16V8 datasheet.pdf

[:File:CY37_datasheet.pdf_p-11] File:CY37 datasheet.pdf

[:File:PLSI2032_datasheet.pdf_p-12] 12.0 ^12.1 ^12.2 ^12.3 File:PLSI2032 datasheet.pdf

[:File:M4A3_datasheet.pdf_p-13] 13.0 ^13.1 File:M4A3 datasheet.pdf

[:File:MACH111_datasheet.pdf_p-14] 14.0 ^14.1 File:MACH111 datasheet.pdf

[.5Bhttps:.2F.2Fgithub.com.2Fstefanoteso.2Fvalkyrie.2Fblob.2Fmaster.2Fsrc.2Fmach.2Fhikaru.2Fhikaru-gpu.c_Sega_Hikaru_GPU_.28Valkyrie.29.5D-15] 15.0 ^15.1 Sega Hikaru GPU (Valkyrie)

[.5Bhttps:.2F.2Fgithub.com.2Fstefanoteso.2Fvalkyrie.2Fblob.2Fmaster.2Fsrc.2Fmach.2Fhikaru.2Fhikaru-gpu-cp.c_Sega_Hikaru_GPU_CP_.28Valkyrie.29.5D-16] Sega Hikaru GPU CP (Valkyrie)

[.5Bhttps:.2F.2Fgithub.com.2Fstefanoteso.2Fvalkyrie.2Fblob.2Fmaster.2Fsrc.2Fmach.2Fhikaru.2Fhikaru-renderer.c_Sega_Hikaru_Renderer_.28Valkyrie.29.5D-17] Sega Hikaru Renderer (Valkyrie)

[:File:ADV7120_datasheet.pdf_p-18] 18.0 ^18.1 File:ADV7120 datasheet.pdf

[.5Bhttp:.2F.2Fogldev.atspace.co.uk.2Fwww.2Ftutorial33.2Ftutorial33.html_Instanced_Rendering.5D-19] Instanced Rendering

[.5Bhttps:.2F.2Fgithub.com.2Fstefanoteso.2Fvalkyrie.2Fblob.2Fmaster.2Fsrc.2Fmach.2Fhikaru.2Fhikaru-gpu-private.h_Sega_Hikaru_GPU_Private_.28Valkyrie.29.5D-20] Sega Hikaru GPU Private (Valkyrie)

[.5Bhttp:.2F.2Farchive.is.2FnyrP3_Sega_Confirms_Hikaru_Does_Exist_.28November_24.2C_1999.29.5D-21] 21.0 ^21.1 Sega Confirms Hikaru Does Exist (November 24, 1999)

[:File:NAOMI_1998_Press_Release_JP.pdf_p-22] 22.0 ^22.1 File:NAOMI 1998 Press Release JP.pdf

[.5Bhttps:.2F.2Fbooks.google.co.uk.2Fbooks.3Fid.3D-4ngT05gmAQC.26pg.3DPA871_.27.27Computer_Graphics:_Principles_and_Practice.27.27_.28page_871.29.5D-23] Computer Graphics: Principles and Practice (page 871)

[.5Bhttps:.2F.2Fbooks.google.co.uk.2Fbooks.3Fid.3D-4ngT05gmAQC.26pg.3DPA900_.27.27Computer_Graphics:_Principles_and_Practice.27.27_.28page_900.29.5D-24] Computer Graphics: Principles and Practice (page 900)

[Sega_NAOMI-26] Sega NAOMI

[.5Bhttps:.2F.2Fbooks.google.co.uk.2Fbooks.3Fid.3DKQnejL0ivfQC.26pg.3DPA78_.27.27Computer_Vision_and_Graphics:_International_Conference.2C_ICCVG_2004.2C_Warsaw.2C_Poland.2C_September_2004.2C_Proceedings.27.27_.28page_78.29.5D-30] 26.0 ^26.1 ^26.2 Computer Vision and Graphics: International Conference, ICCVG 2004, Warsaw, Poland, September 2004, Proceedings (page 78)

[.5Bhttps:.2F.2Fbooks.google.co.uk.2Fbooks.3Fid.3DiAvHt5RCHbMC.26pg.3DPA95_.27.27Design_of_Digital_Systems_and_Devices.27.27_.28pages_95-97.29.5D-34] 27.0 ^27.1 Design of Digital Systems and Devices (pages 95-97)

[:File:HM5264_datasheet.pdf_p-41] 28.0 ^28.1 File:HM5264 datasheet.pdf

[:File:HY57V161610D_datasheet.pdf_p-43] 29.0 ^29.1 File:HY57V161610D datasheet.pdf

[:File:UPD432232_datasheet.pdf_p-48] 30.0 ^30.1 File:UPD432232 datasheet.pdf

[:File:HM62256B_datasheet.pdf_p-50] 31.0 ^31.1 ^31.2 File:HM62256B datasheet.pdf

[:File:CY7C199_datasheet.pdf_p-51] 32.0 ^32.1 File:CY7C199 datasheet.pdf

[.5Bhttp:.2F.2Fwww.mamedb.com.2Fgame.2Fpharrier.html_Planet_Harriers_.28MAME.29.5D-52] Planet Harriers (MAME)

[.5Bhttp:.2F.2Ffarm6.staticflickr.com.2F5471.2F12172411045_18bfc5912f_c.jpg_Hideki_Sato_Sega_Interview_.28Edge.29.5D-53] Hideki Sato Sega Interview (Edge)

[:File:SH-4_32-bit_CPU_Core_Architecture.pdf_p-54] File:SH-4 32-bit CPU Core Architecture.pdf

[:File:K4S641632_datasheet.pdf_p-72] File:K4S641632 datasheet.pdf

[:File:CY2292_datasheet.pdf_p-78] File:CY2292 datasheet.pdf

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[n 1]

[n 2]

[n 3]

[n 4]

[n 5]

[n 6]

[n 7]

[n 8]

[n 9]

[n 10]

[n 11]

[n 12]

[n 13]

[28]

[n 14]

[29]

[n 15]

[n 16]

[n 17]

[n 18]

[30]

[n 19]

[31]

[32]

[33]

[34]

[n 20]

[n 21]

[n 22]

[n 23]

[n 24]

[n 25]

[n 26]

[n 27]

[n 28]

[n 29]

[n 30]

[n 31]

[n 32]

[n 33]

[n 34]

[n 35]

[n 36]

[n 37]

[n 38]

[n 39]

[n 40]

[n 41]

[n 42]

[n 43]

[n 44]

[25]

[26]

[27]

[35]

[36]

[37]

@@ Line 100: / Line 100: @@
 :* Effects: [[wikipedia:Stencil buffer|Stencil]], shadows, motion blur, particle effects, fire effects, [[wikipedia:Fluid simulation|water effects]],{{ref|[http://archive.is/nyrP3 Sega Confirms Hikaru Does Exist (November 24, 1999)]}} fog, alpha blending, anti‑aliasing, specular effects,{{fileref|NAOMI 1998 Press Release JP.pdf}}
 :* Features: [[wikipedia:Tiled rendering|Tiled rendering]], [[wikipedia:Deferred shading|deferred rendering]], [[wikipedia:Back-face culling|back‑face culling]], [[wikipedia:Hidden surface determination|hidden surface removal]]
-:* [[wikipedia:Phong shading|Phong shading]]: Per‑[[pixel]] lighting/shading computation processed by rasterization pipeline,{{ref|1=[https://books.google.co.uk/books?id=-4ngT05gmAQC&pg=PA871 Computer Graphics: Principles and Practice (Page 871)]}} deferred rendering can prevent shading of overdrawn pixels to maximize rendering bandwidth,{{ref|1=[https://books.google.co.uk/books?id=-4ngT05gmAQC&pg=PA900 Computer Graphics: Principles and Practice (Page 900)]}} all compute units could be used as shader units to maximize Phong shading performance
+:* [[wikipedia:Phong shading|Phong shading]]: Per‑[[pixel]] lighting/shading computation processed by rasterization pipeline,{{ref|1=[https://books.google.co.uk/books?id=-4ngT05gmAQC&pg=PA871 ''Computer Graphics: Principles and Practice'' (page 871)]}} deferred rendering can prevent shading of overdrawn pixels to maximize rendering bandwidth,{{ref|1=[https://books.google.co.uk/books?id=-4ngT05gmAQC&pg=PA900 ''Computer Graphics: Principles and Practice'' (page 900)]}} all compute units could be used as shader units to maximize Phong shading performance
 * Texture capabilities: 1×1 to 2048×2048 texture sizes, [[wikipedia:Mipmap|mipmapping]], mipmap trees, texture panning, multi‑texturing, [[wikipedia:Bump mapping|bump mapping]], [[wikipedia:Normal mapping|normal mapping]], texture filtering, bilinear filtering, trilinear filtering, [[wikipedia:Reflection mapping|environment mapping]]{{fileref|NAOMI 1998 Press Release JP.pdf}}
 :* Texture banks: 2 texture banks (stored as 2×&nbsp;2048×1024 sheets), stores textures from MaskROM (with 16‑byte metadata per texture in Command RAM)