The Ultimate GPU Benchmark (2004 - 2007)

Published: 14th June 2020 (last update 14th June 2020 )

It is time to revisit the video card benchmarks. This is the first part of completely new project focused on as thorough GPU benchmarking as possible. Not only average fps, but also frametimes analysis and other attributes measured by MSI Afterburner. All values are shown using interactive charts.

Introduction

I have been testing PC hardware over 10 years and during that time I gained a lot of experience in this field. Also my capabilities of preparing the right SW tools for proper benchmarking are now better than ever. Not to mention the ability to buy the right tools and equipment which wasn't the case before. There are only two thing missing here to make this benchmark project trully the best. One of them is isolated power consumption of the video cards (which is rather difficult to accomplish, especially for multi GPU). And the second one would be true hardware-based frametimes measurement using high framerate capture card. Anyway, Fraps provides frametimes data that are reliable enough to show more in-depth analysis of tested GPUs. There are also attributes like memory usage, GPU Load and total system Power usage monitored by MSI Afterburner and HWiNFO.

The base stone is no doubt the test system. it is rather old for todays standards, but proven by the years of service. Also it matches all my requirements - support of 2-way SLI/CF 16+16 and 3-way SLI/CF 16+8+8. So we are looking at Sandy Bridge-E X79 platform - Asus P9X79 Deluxe and Xeon E5-1650 overclocked to 4.8 GHz. I also considered using something more up-to-date. But there are not many more options in HEDT it seems. Haswell-E wouldn't be that much faster (better IPC, lower clock) and still costs a lot when I want the 40-pcie-lanes variants. Gaming performance of X299 and X399 also isn't all that great and not very cost-effective anyway.

The last but not least - the interactive charts. It took some time to develop, but now they are finally working at their full potential.

Test System - Hardware

Xeon E5-1650 @ 4.8 GHz (6C/6T, fixed clock)
Asus P9X79 Deluxe
4 × 4 GB DDR3 1600 CL9-9-9-24-1T
Kingston HyperX Savage 120 GB
Toshiba 3 TB / 7200 rpm
Corsair RM1000i
Fractal Design Define S
Acer FX270HU (2560×1440)

Test System - OS and Drivers

Windows 7 x64
GeForce 275.33 (for GeForce 6800 GS, 7600 GT, 7800 GTX, Quadro FX 4500)
GeForce 296.10 (for every other GeForce and Quadro)
Catalyst 10.2 Legacy (for Radeon X800/850 Series, X1800/X1900/X1950 Series)
Catalyst 12.1 (for Radeon HD 2000 Series)
Catalyst 15.200.1062.1004 (for Radeon HD 6550D)
Adrenalin 19.12.3 (for Radeon R7 Graphics)
...
For all GPUs forced 16×AF, high quality filtering (if possible), vsync off

Test System - Games

Bioshock [2007, DX9]
Call of Duty: Modern Warfare [2007, DX9]
Call of Duty: World at War [2008, DX9]
Crysis [2007, DX9]
Doom 3 [2004, OGL]
Enemy Territory: Quake Wars [2007, OGL]
Far Cry [2004, DX9]
Far Cry 2 [2008, DX9]
Half-Life 2: Episode Two [2007, DX9]
Mirror's Edge [2009, DX9]
Need for Speed: Most Wanted [2005, DX9]
Serious Sam 2 [2005, DX9]

Tested Video Cards

	Radeon X800 GTO	Radeon X850 XT PE	Radeon X1600 XT	Radeon X1800 XL	Radeon X1800 XT	FireGL V7350	Radeon X1900 XTX	2 × Radeon X1900 XTX
GPU	R430	R480	RV530	R520	R520	R520	R580	2 × R580
Architecture	R3xx/R4xx	R3xx/R4xx	R5xx	R5xx	R5xx	R5xx	R5xx	R5xx
Technology	110 nm	130 nm	90 nm	90 nm	90 nm	90 nm	90 nm	90 nm
Die Size	240 mm²	297 mm²	150 mm²	288 mm²	288 mm²	288 mm²	352 mm²	2 × 352 mm²
Transistor Count	160 M	160 M	157 M	321 M	321 M	321 M	384 M	2 × 384 M
Transistor Density	0.66 M / mm²	0.54 M / mm²	1.05 M / mm²	1.11 M / mm²	1.11 M / mm²	1.11 M / mm²	1.09 M / mm²	1.09 M / mm²
GPU Clock	400 MHz	540 MHz	590 MHz	500 MHz	625 MHz	600 MHz	650 MHz	650 MHz
ROPs	12	16	4	16	16	16	16	2 × 16
TMUs	12	16	4	16	16	16	16	2 × 16
Shaders	12 PS + 6 VS	16 PS + 6 VS	12 PS + 5 VS	16 PS + 8 VS	16 PS + 8 VS	16 PS + 8 VS	48 PS + 8 VS	2 × 48 PS + 8 VS
Memory	256 MB GDDR3	256 MB GDDR3	256 MB GDDR3	256 MB GDDR3	512 MB GDDR3	1024 MB GDDR3	512 MB GDDR3	512 MB GDDR3
Memory Clock	980 MHz	1180 MHz	1400 MHz	1000 MHz	1500 MHz	1300 MHz	1550 MHz	1550 MHz
Bus Width	256 bit	256 bit	128 bit	256 bit	256 bit	256 bit	256 bit	2 × 256 bit
Memory Bandwidth	31.4 GB/s	37.8 GB/s	22.4 GB/s	32 GB/s	48 GB/s	41.6 GB/s	49.6 GB/s	2 × 49.6 GB/s
Fillrate (Pixel)	4800 MP/s	8640 MP/s	2360 MP/s	8000 MP/s	10000 MP/s	9600 MP/s	10400 MP/s	2 × 10400 MP/s
Fillrate (Texel)	4800 MT/s	8640 MT/s	2360 MT/s	8000 MT/s	10000 MT/s	9600 MT/s	10400 MT/s	2 × 10400 MT/s
Bus Type	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0
TDP	49 W	69 W	42 W	70 W	113 W	111 W	135 W	2 × 135 W
DirectX	9b	9b	9c	9c	9c	9c	9c	9c
OpenGL	2.0	2.0	2.0	2.0	2.0	2.0	2.0	2.0
Launch Year	2005	2004	2005	2005	2005	2005	2006	2006

	Radeon X1950 Pro	2 × Radeon X1950 Pro	Radeon X1950 XT	Radeon X1950 XTX	Radeon HD 2600 XT	Radeon HD 2900 GT	Radeon HD 2900 XT	2 × Radeon HD 2900 XT
GPU	RV570	2 × RV570	R580+	R580+	RV630	R600	R600	2 × R600
Architecture	R5xx	R5xx	R5xx	R5xx	Terascale	Terascale	Terascale	Terascale
Technology	80 nm	80 nm	90 nm	90 nm	65 nm	80 nm	80 nm	80 nm
Die Size	230 mm²	2 × 230 mm²	352 mm²	352 mm²	153 mm²	420 mm²	420 mm²	2 × 420 mm²
Transistor Count	330 M	2 × 330 M	384 M	384 M	390 M	720 M	720 M	2 × 720 M
Transistor Density	1.43 M / mm²	1.43 M / mm²	1.09 M / mm²	1.09 M / mm²	2.55 M / mm²	1.71 M / mm²	1.71 M / mm²	1.71 M / mm²
GPU Clock	580 MHz	580 MHz	625 MHz	650 MHz	800 MHz	600 MHz	743 MHz	743 MHz
ROPs	12	2 × 12	16	16	4	12	16	2 × 16
TMUs	12	2 × 12	16	16	8	12	16	2 × 16
Shaders	36 PS + 8 VS	2 × 36 PS + 8 VS	48 PS + 8 VS	48 PS + 8 VS	120 Unified	240 Unified	320 Unified	2 × 320 Unified
Memory	256 MB GDDR3	256 MB GDDR3	256 MB GDDR3	512 MB GDDR4	256 MB GDDR4	256 MB GDDR3	512 MB GDDR3	512 MB GDDR3
Memory Clock	1400 MHz	1400 MHz	1800 MHz	2000 MHz	2200 MHz	1600 MHz	1660 MHz	1660 MHz
Bus Width	256 bit	2 × 256 bit	256 bit	256 bit	128 bit	256 bit	512 bit	2 × 512 bit
Memory Bandwidth	44.8 GB/s	2 × 44.8 GB/s	57.6 GB/s	64 GB/s	35.2 GB/s	51.2 GB/s	106.2 GB/s	2 × 106.2 GB/s
Fillrate (Pixel)	6960 MP/s	2 × 6960 MP/s	10000 MP/s	10400 MP/s	3200 MP/s	7200 MP/s	11840 MP/s	2 × 11840 MP/s
Fillrate (Texel)	6960 MT/s	2 × 6960 MT/s	10000 MT/s	10400 MT/s	6400 MT/s	7200 MT/s	11840 MT/s	2 × 11840 MT/s
Bus Type	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 2.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0
TDP	80 W	2 × 80 W	96 W	125 W	50 W	150 W	215 W	2 × 215 W
DirectX	9c	9c	9c	9c	10	10	10	10
OpenGL	2.0	2.0	2.0	2.0	3.3	3.3	3.3	3.3
Launch Year	2006	2006	2006	2006	2007	2007	2007	2007

	FireGL V8650 OC	Radeon HD 6550D	Radeon R7 Graphics	GeForce 6800 GS	GeForce 7600 GT	GeForce 7800 GTX	Quadro FX 4500	GeForce 7900 GS OC
GPU	R600	Sumo	Spectre	NV42	G73B	G70	G70	G71
Architecture	Terascale	Terascale 2	CGN	NV4x/G7x	NV4x/G7x	NV4x/G7x	NV4x/G7x	NV4x/G7x
Technology	80 nm	32 nm	28 nm	110 nm	80 nm	110 nm	110 nm	90 nm
Die Size	420 mm²	228 mm²	245 mm²	225 mm²	100 mm²	333 mm²	333 mm²	196 mm²
Transistor Count	720 M	1176 M	2410 M	202 M	177 M	302 M	302 M	278 M
Transistor Density	1.71 M / mm²	5.16 M / mm²	9.84 M / mm²	0.9 M / mm²	1.77 M / mm²	0.91 M / mm²	0.91 M / mm²	1.42 M / mm²
GPU Clock	850 MHz	600 MHz	720 MHz	425 MHz	560 MHz	430 MHz	430 MHz	500 MHz
ROPs	16	8	8	12	8	16	16	16
TMUs	16	20	32	12	12	24	24	20
Shaders	320 Unified	400 Unified	512 Unified	12 PS + 5 VS	12 PS + 5 VS	24 PS + 8 VS	24 PS + 8 VS	20 PS + 7 VS
Memory	2048 MB GDDR4	512 MB DDR3	1024 MB DDR3	256 MB GDDR3	256 MB GDDR3	256 MB GDDR3	512 MB GDDR3	256 MB GDDR3
Memory Clock	2000 MHz	1866 MHz	2133 MHz	1000 MHz	1400 MHz	1200 MHz	1050 MHz	1400 MHz
Bus Width	512 bit	128 bit	128 bit	256 bit	128 bit	256 bit	256 bit	256 bit
Memory Bandwidth	128 GB/s	29.9 GB/s	34.1 GB/s	32 GB/s	22.4 GB/s	38.4 GB/s	33.6 GB/s	44.8 GB/s
Fillrate (Pixel)	13600 MP/s	4800 MP/s	5760 MP/s	5100 MP/s	4480 MP/s	6880 MP/s	6880 MP/s	8000 MP/s
Fillrate (Texel)	13600 MT/s	12000 MT/s	23040 MT/s	5100 MT/s	6720 MT/s	10320 MT/s	10320 MT/s	10000 MT/s
Bus Type	PCI-E 1.0	IGP	IGP	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0
TDP	~300 W	~50 W	~32 W	50 W	40 W	86 W	88 W	55 W
DirectX	10	11	12	9c	9c	9c	9c	9c
OpenGL	3.3	4.5	4.6	2.1	2.1	2.1	2.1	2.1
Launch Year	2007	2011	2014	2005	2007	2005	2005	2006

	2 × GeForce 7900 GS OC	GeForce 7950 GT	GeForce 7950 GX2	GeForce 8600 GTS	GeForce 8800 GTS (320 MB)	GeForce 8800 GTS (640 MB)	Quadro FX 4600	GeForce 8800 GTX
GPU	2 × G71	G71	2 × G71	G84	G80	G80	G80	G80
Architecture	NV4x/G7x	NV4x/G7x	NV4x/G7x	Tesla	Tesla	Tesla	Tesla	Tesla
Technology	90 nm	90 nm	90 nm	80 nm	90 nm	90 nm	90 nm	90 nm
Die Size	2 × 196 mm²	196 mm²	2 × 196 mm²	169 mm²	484 mm²	484 mm²	484 mm²	484 mm²
Transistor Count	2 × 278 M	278 M	2 × 278 M	289 M	681 M	681 M	681 M	681 M
Transistor Density	1.42 M / mm²	1.42 M / mm²	1.42 M / mm²	1.71 M / mm²	1.41 M / mm²	1.41 M / mm²	1.41 M / mm²	1.41 M / mm²
GPU Clock	500 MHz	550 MHz	500 MHz	675 MHz	513 MHz	513 MHz	500 MHz	575 MHz
ROPs	2 × 16	16	2 × 16	8	20	20	24	24
TMUs	2 × 20	24	2 × 24	16	24	24	24	32
Shaders	2 × 20 PS + 7 VS	24 PS + 8 VS	2 × 24 PS + 8 VS	32 Unified	96 Unified	96 Unified	69 Unified	128 Unified
Memory	256 MB GDDR3	512 MB GDDR3	512 MB GDDR3	256 MB GDDR3	320 MB GDDR3	640 MB GDDR3	768 MB GDDR3	768 MB GDDR3
Memory Clock	1400 MHz	1400 MHz	1200 MHz	2000 MHz	1600 MHz	1600 MHz	1400 MHz	1800 MHz
Bus Width	2 × 256 bit	256 bit	2 × 256 bit	128 bit	320 bit	320 bit	384 bit	384 bit
Memory Bandwidth	2 × 44.8 GB/s	44.8 GB/s	2 × 38.4 GB/s	32 GB/s	64 GB/s	64 GB/s	67.2 GB/s	86.4 GB/s
Fillrate (Pixel)	2 × 8000 MP/s	8800 MP/s	2 × 8000 MP/s	5400 MP/s	10260 MP/s	10260 MP/s	12000 MP/s	13800 MP/s
Fillrate (Texel)	2 × 10000 MT/s	13200 MT/s	2 × 12000 MT/s	10800 MT/s	12312 MT/s	12312 MT/s	12000 MT/s	18400 MT/s
Bus Type	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0
TDP	2 × 55 W	75 W	110 W	75 W	143 W	143 W	134 W	155 W
DirectX	9c	9c	9c	10	10	10	10	10
OpenGL	2.1	2.1	2.1	3.3	3.3	3.3	3.3	3.3
Launch Year	2006	2006	2006	2007	2007	2006	2007	2006

	2 × GeForce 8800 GTX	GeForce 8800 Ultra OC	Quadro FX 5600 OC
GPU	2 × G80	G80	G80
Architecture	Tesla	Tesla	Tesla
Technology	90 nm	90 nm	90 nm
Die Size	2 × 484 mm²	484 mm²	484 mm²
Transistor Count	2 × 681 M	681 M	681 M
Transistor Density	1.41 M / mm²	1.41 M / mm²	1.41 M / mm²
GPU Clock	575 MHz	650 MHz	600 MHz
ROPs	2 × 24	24	24
TMUs	2 × 32	32	32
Shaders	2 × 128 Unified	128 Unified	128 Unified
Memory	768 MB GDDR3	768 MB GDDR3	1536 MB GDDR3
Memory Clock	1800 MHz	2200 MHz	1900 MHz
Bus Width	2 × 384 bit	384 bit	384 bit
Memory Bandwidth	2 × 86.4 GB/s	105.6 GB/s	91.2 GB/s
Fillrate (Pixel)	2 × 13800 MP/s	15600 MP/s	14400 MP/s
Fillrate (Texel)	2 × 18400 MT/s	20800 MT/s	19200 MT/s
Bus Type	PCI-E 1.0	PCI-E 1.0	PCI-E 1.0
TDP	2 × 155 W	171 W	~171 W
DirectX	10	10	10
OpenGL	3.3	3.3	3.3
Launch Year	2006	2007	2007