Comparison Of Graphic Chips And Test Configurations
HIS and Sapphire sent Radeon HD 4870 X2 cards set to reference clock rates—any additional overclocking was performed in our own lab. All settings were set via the AMD graphics driver. As a baseline for analysis, the HIS card was tested with both normal frequencies and overclocked. All models with increased frequencies are marked “OC.”
When we started benchmarking for this project, we were using the most current drivers for this test, Catalyst 8.10 from AMD and Nvidia’s beta driver 180.42. On the following page, we will talk about the difficulties and problems we noticed with Catalyst 8.10. The Nvidia beta driver is stable with single cards and the Intel chipset (X38), but with SLI on the nForce 780i, we sometimes experienced problems in Assassin’s Creed and Mass Effect causing the whole system to freeze. As far as performance is concerned, the latest Catalyst 8.11 drivers improve frame rates in Far Cry 2 and Stalker: Clear Sky, according to AMD.
| Card Manufacturer and Chip | Code Name | Memory | GPU Clock | Shader | Memory Clock | SPs |
|---|---|---|---|---|---|---|
| GeForce GTX 280 SLI | GT200 | 1024 GDDR3 | 602 MHz | 4.0, 1296 MHz | 2214 MHz | 240 |
| GeForce GTX 280 | GT200 | 1024 GDDR3 | 602 MHz | 4.0, 1296 MHz | 2214 MHz | 240 |
| GeForce GTX 260 SLI | GT200 | 896 GDDR3 | 576 MHz | 4.0, 1242 MHz | 1998 MHz | 192 |
| GeForce GTX 260 | GT200 | 896 GDDR3 | 576 MHz | 4.0, 1242 MHz | 1998 MHz | 192 |
| GeForce 9800 GX2 | 2xG92 | 2x512 MB GDDR3 | 600 MHz | 4.0, 1500 MHz | 2000 MHz | 2x128 |
| GeForce 9800 GTX+ | G92b | 512 MB GDDR3 | 738 MHz | 4.0, 1836 MHz | 2200 MHz | 128 |
| Card Manufacturer and Chip | Memory Bus | Manufacturing Process | Transistors | Interface |
|---|---|---|---|---|
| GeForce GTX 280 SLI | 512 Bit | 65 nm | 1400 million | PCIe 2.0 |
| GeForce GTX 280 | 512 Bit | 65 nm | 1400 million | PCIe 2.0 |
| GeForce GTX 260 SLI | 448 Bit | 65 nm | 1400 million | PCIe 2.0 |
| GeForce GTX 260 | 448 Bit | 65 nm | 1400 million | PCIe 2.0 |
| GeForce 9800 GX2 | 2x256 Bit | 65 nm | 2x754 million | PCIe 2.0 |
| GeForce 9800 GTX+ | 256 Bit | 55 nm | 754 million | PCIe 2.0 |
| Manufacturer and Chip | Codename | Memory | GPU Clock | Shader | Memory Clock | SPs |
|---|---|---|---|---|---|---|
| Radeon HD 4870 X2 4CF | 2xR700 | 4x1024 MB GDDR5 | 750 MHz | 4.1 | 3600 MHz | 4x800 |
| Radeon HD 4870 X2 | R700 | 2x1024 MB GDDR5 | 750 MHz | 4.1 | 3600 MHz | 2x800 |
| HIS Radeon HD 4870 X2 | R700 | 2x1024 MB GDDR5 | 750 MHz | 4.1 | 3600 MHz | 2x800 |
| HIS Radeon HD 4870 X2 OC | R700 | 2x1024 MB GDDR5 | 800 MHz | 4.1 | 3860 MHz | 2x800 |
| Asus Radeon HD 4870 X2 Top OC | R700 | 2x1024 MB GDDR5 | 790 MHz | 4.1 | 3660 MHz | 2x800 |
| MSI Radeon HD 4870 X2 OC | R700 | 2x1024 MB GDDR5 | 780 MHz | 4.1 | 3600 MHz | 2x800 |
| Sapphire Radeon HD 4870 X2 | R700 | 2x1024 MB GDDR5 | 750 MHz | 4.1 | 3600 MHz | 2x800 |
| Sapphire Radeon HD 4870 X2 OC | R700 | 2x1024 MB GDDR5 | 790 MHz | 4.1 | 3760 MHz | 2x800 |
| Radeon HD 4870 CF | RV770 | 512 MB GDDR5 | 750 MHz | 4.1 | 3600 MHz | 800 |
| Radeon HD 4870 | RV770 | 512 MB GDDR5 | 750 MHz | 4.1 | 3600 MHz | 800 |
| Radeon HD 4850 CF | RV770 | 512 MB GDDR3 | 625 MHz | 4.1 | 1986 MHz | 800 |
| Radeon HD 4850 | RV770 | 512 MB GDDR3 | 625 MHz | 4.1 | 1986 MHz | 800 |
| Radeon HD 4670 | RV730 | 512 MB GDDR3 | 750 MHz | 4.1 | 2000 MHz | 320 |
| Manufacturer and Chip | Memory Bus | Manufacturing Process | Transistors | Interface |
|---|---|---|---|---|
| Radeon HD 4870 X2 4CF | 4x256 Bit | 55 nm | 4x956 million | PCIe 2.0 |
| Radeon HD 4870 X2 | 2x256 Bit | 55 nm | 2x956 million | PCIe 2.0 |
| HIS Radeon HD 4870 X2 | 2x256 Bit | 55 nm | 2x956 million | PCIe 2.0 |
| HIS Radeon HD 4870 X2 OC | 2x256 Bit | 55 nm | 2x956 million | PCIe 2.0 |
| Asus Radeon HD 4870 X2 Top OC | 2x256 Bit | 55 nm | 2x956 million | PCIe 2.0 |
| MSI Radeon HD 4870 X2 OC | 2x256 Bit | 55 nm | 2x956 million | PCIe 2.0 |
| Sapphire Radeon HD 4870 X2 | 2x256 Bit | 55 nm | 2x956 million | PCIe 2 .0 |
| Sapphire Radeon HD 4870 X2 OC | 2x256 Bit | 55 nm | 2x956 million | PCIe 2.0 |
| Radeon HD 4870 CF | 256 Bit | 55 nm | 956 million | PCIe 2.0 |
| Radeon HD 4870 | 256 Bit | 55 nm | 956 million | PCIe 2.0 |
| Radeon HD 4850 CF | 256 Bit | 55 nm | 956 million | PCIe 2.0 |
| Radeon HD 4850 | 256 Bit | 55 nm | 956 million | PCIe 2.0 |
| Radeon HD 4670 | 128 Bit | 55 nm | 514 million | PCIe 2.0 |
Memory Clock=DDR clock rate doubled; physical clock rate is half
DDR5-DDR clock rate quadrupled; physical clock rate is one quarter
SPs=Stream Processors
OC=Overclocked (clock rate higher than standard)
SLI= 2 Nvidia cards in parallel
CF=CrossFire, 2 AMD cards in parallel
4CF=CrossFire, 4 AMD graphics chips in parallel
R700=2xRV770
Shader 2.0=DirectX 9.0; 3.0=DirectX 9.0c; 4.0=DirectX 10; Shader 4.1=DirectX 10.1
The standard CPU for our graphic card charts is an X6800 Extreme Edition running at 2.93 GHz, which is comparable to an E6750 or E8200. For this high-performance test we used a quad-core CPU, which runs at 3.67 GHz, overclocked using the multiplier with settings of 11x333 MHz. If you’ve read the article Wake 3D Power Wake-up, you already know all the steps we took: in order to find the highest overall result, the fastest double chip card should get the fastest CPU partner. We had to choose between an E8500 at 3.8 GHz and the over clocked quad core CPU with 3.67 GHz; the latter turned out to be faster by 3.5%.
| CPU 1 | Intel Core 2 Quad Extreme QX6850 @ 3.673 GHz (11x333 MHz), Socket 775, 1.328 V, 65 nm, L2 Cache 2x4096 KB |
| CPU 2 | Intel Core 2 Extreme X6800 @ 2.93 GHz (11x266 MHz), Socket 775, 1.28 V, 65 nm, L2 Cache 4096 KB |
| CPU 3 | Intel Core 2 Extreme X6800 @ 3.47 GHz (13x266 MHz), Socket 775, 1.328 V, 65 nm, L2 Cache 4096 KB |
| CPU 4 | Intel Core 2 E8500 @ 3.804 GHz (9.5x400 MHz), Socket 775, 1.272 V, 45 nm, L2 Cache 6144 KB |
| FSB | 1066 MHz (4x266 MHz) |
| Motherboard | Asus P5E3 Deluxe, PCIe 2.0 2x16, ICH9R |
| Chipset | Intel X38 |
| Memory | 2x1 GB, Ballistix (Crucial Technology) 1.5 V, DDR3 1066 7-7-7-20 (2x533 MHz) |
| Audio | Intel High Definition Audio |
| LAN | Intel 1000 Pro |
| Hard Drives | Western Digital WD5000AAKS 500 GB, SATA, 16 MB Cache, Hitachi 120 GB, SATA, 8 MB Cache |
| DVD | Gigabyte GO-D1600C |
| Power Supply | CoolerMaster RS-850-EMBA 850 W |
| Motherboard | Asus P5N-T Deluxe, PCIe 2.0 2x16 |
| Chipset | Nvidia nforce 780i SLI |
| Memory | 2x1 GB, A-Data Technology 1.8 V, DDR2 800 5-5-5-18 (2x400 MHz) |
| Audio | ADI 1988B SoundMax |
| LAN | Marvell 88E1116 Gigabit |
| Graphic | AMD Catalyst 8.6 and 8.9, HD 4870 X2 ATI Catalyst 8.10, 9800 GX2 Nvidia ForceWare 175.16, GTX 260 und GTX 280 Forceware 177.39, Nvidia GeForce 180.42 |
| Operating System | Windows Vista Enterprise SP1 |
| DirectX | 10 and 10.1 |
| Chip Set Driver | X38 Intel 8.3.1.1009 780i Nvidia nForce 9.64 |

I Liked this Article, didn't read it word for word so im not going to nit-pic for the sake of it, it confirmed my choice of video card (asus top)and gave benchmarks based on the Processor and speeds i run at....
also goes to show that anyone who bought a 9800gx2 is still sitting pretty.
Thanks Guys.
erm, are you sure about those figures for the 9800GX2 for crysis VHQ @ 1920x1200?
less than 1 frame a second???
otherwise graphics cards comparison article, keep 'em comin'!
cheers,
bill
p.s. stuff and nonsense: http://www.eupeople.net/forum
I believe the less than 1fps crysis result was the problem they were referring to in the text, it is a driver bug with the card if I'm not mistaken.
Would have been nice to see the 4650 in crossfire since it'd be comparable with my 2x3850 setup.
Well done Tino.
Great article.
What is the cooler you have on that processor? Would you recommend that.
Got a quad core thats reaching 70 degrees celsius, seems to high. So i am looking to invest in some non-stock cooling
Great article btw, love the ATI's for now
What was the actaul timedemo used for the Crysis VHQ benchmark ?
hi,
i have a 4870 (1024mb) and i am concidering to get a second, with a asus p6t mobo, will i just beable to plug it in then add the crossfire cable and then use it or is there some fiddling involved. also there is a fairly small gap inbetween the 2 card about 3mm will that still be sufficiant for cooling
kit cowan