Introduction

When NVIDIA introduced the GT200 GPU in the form of the GeForce GTX 280 and the GeForce GTX 260, was the world for the Californians still in order. The competition had placed still no comparable product on the market. When ATi finally presented the Radeon HD 4800 Series, it was much more difficult for NVIDIA. He hastily made 4850 HD life difficult 9800 GTX + on the market of their GeForce from the Radeon HD, while the Radeon HD 4870 showed the price performance show of the far more expensive GeForce GTX 260. Only the GeForce GTX 280 remained long periods of time without competition, ATi Radeon HD 4870 X 2 on the market brought the multi-GPU card.

An untenable state that for the time being just is to be put in terms of GeForce GTX 260 against Radeon HD 4870. NVIDIA’s trick: on the “GeForce GTX 260 new revision” now 216 Shadereinheiten instead of only 192 beer are enabled, which brings a higher performance and therefore better performance with it. It will nevertheless still be the “old” GeForce GTX 260.

ZOTAC could provide kindly provided a sample of the GeForce GTX 260² for a test us, which we’ll look at in detail. Manages NVIDIA, to achieve a large enough lead to the traditional GeForce GTX 260? And how far can you get away from the Radeon HD 4870? On the following pages, we clarify all this and more.

Technical data

< div >

On the new GeForce GTX 260 is as heavy on all other GeForce GTX-200-cards of 1.4 billion transistors chip of gt200 used, which is produced in the 65-nm process at TSMC. NVIDIA GeForce GTX 260 some additional units enabled new revision on which can be compared to the old GeForce GTX 260, which relies on a partly disabled GT200. So are instead of eight now nine Shadercluster active, which is why no longer 192 scalar beer but whose 216 beer available are.

Since the Shadereinheiten in the GT200-architecture on the texture clusters are docked, there is a change in this: on the GeForce GTX 260 from immediately seven texture clusters and not more six the computation take, making 72 pixels per clock (per texture clusters there eight full-fledged TMUs) can be addressed and bilinear filtered. This was possible only with 64 pixels on the GeForce GTX 260.

As other units such as the ROPs and the memory interface to the shader clusters are linked, there is otherwise no change: the GeForce GTX 260 sets continue to 28 ROPs and a 448-bit wide memory interface. The memory capacity is again 896 MB of GDDR3 type. At the clock rates, there are no differences. The TMU domain is referred to as 576 MHz, the Shaderdomäne, however, with brisk 1,242 MHz. The VRAM added 999 MHz.

Impressions

ZOTAC GeForce GTX 260²

With the revised GeForce GTX 260, NVIDIA wants to reshuffling the cards in the lower high end and the Radeon HD 4870, which is wacker beats against the “old” GeForce GTX 260 (Yes sometimes even outdated), keep at a distance. The graphics card manufacturer ZOTAC is again even with the new card of the party and has baptized the in-house copy to the name GeForce GTX 260² so that you can distinguish the different types. Listed the video card is isolated from 260 euro, but the most shops speak rather of EUR 270 to 280 euros. The Accelerator from 290 euro – but not yet by ZOTAC but by EVGA is available.

Visually resembles the ZOTAC GeForce GTX 260² down to the smallest detail of a well-known GeForce GTX 260. At ZOTAC, one can distinguish the run only by the exponents of “square” at the new cards, otherwise the hardware including sticker is absolutely identical.

Even with the GeForce GTX 260² you must infect two six pin plug-in to function correctly. You can get the actual run not to face, because wraps a black Metalummantellung the entire PCB. This will ensure a better heat transfer. Away from the sheath, the cooling system consists of a large heat sink that is equipped with various cooling fins in low-cost aluminium. A sunken cooling block, consists of teurerem, but for more effective copper sits on the GPU itself. In addition, the two cooling units with a heatpipe are connected.

As a fan, a 70-mm radial copy is built on the ZOTAC GeForce GTX 260², which slightly sloping towards GPU is mounted. This, the air pressure should more be bundled and the heatsink to better can cool. The fan takes in the cool air from the housing and it blows through the cooling fins, and thus on the GPU. The heated air is then transported through the bracket out of the case out. The fan works this surprisingly quiet – more on this in the section of volume.

In 2D mode is added down the ZOTAC GeForce GTX 260² on 300 MHz (TMU), 600 MHz (beer) or 100 MHz (memory). At our launch of the GeForce GTX 260 review, there were still 300 / 100 / 100 MHz, which however is not the hardware, but expected to be the latest GeForce drivers. We try to clarify this phenomenon with NVIDIA.

On the bracket, the buyer will encounter two HDCP compatible dual-link DVI outputs that can apply copy protection in high resolutions such as 2560 x 1600. In addition, there is still the mandatory HDTV output. The facilities succeeded ZOTAC in the GeForce GTX 260². In addition to two power cables, there is still a DVI to D-SUB-as well as a DVI to HDMI adapter and still a S-video to S-video/YUV breakout cable. An SPDIF cable is not missing. Praise you must ZOTAC for the software equipment. This is although not versatile, but for this quality. Because in addition to a driver CD still the good racing game “Race Driver Grid” is included in the video card.

Test system

Test system:

• Processor
  • Intel Core 2 extreme QX9770 (overclocked by multiplier to 4 GHz, quad-core)
• CPU cooler
  • Noctua NH-U12P
• Motherboard
  • Asus P5E3 Deluxe WiFi-AP (Intel X 38, BIOS version: 1104) main test Board and for CrossFire systems
  • XFX nForce 790i ultra (NVIDIA nForce 790i, BIOS version: 811N1P01_Beta) for SLI Systems
• Memory
  • 2 x 1024 MB G.skill DDR3-1600 (7-7-7-18)
  • 2 x 1024 MB Patriot DDR3-1600 (7-7-7-18)
• Graphics cards
  • ATi Radeon HD 4870 X 2 (750 / 1,800), 2 x 1024 MB
  • ATi Radeon HD 4870 (750 / 1,800), 512 MB
  • ATi Radeon HD 4850 (625 / 993), 512 MB
  • ATi Radeon HD 4670 (750 / 1,000), 512 MB
  • ATi Radeon HD 3870 X 2 (825 / 900), 2 x 512 MB
  • ATi Radeon HD 3870 (775 / 1,125), 512 MB
  • ATi Radeon HD 3850 (668 / 828), 512 MB
  • ATi Radeon HD 3850 (668 / 828), 256 MB
  • ATi Radeon HD 3650 (725 / 800), 512 MB
  • NVIDIA GeForce GTX 280 (602/1.296/1.107), 1,024 MB
  • ZOTAC GeForce GTX 260² (576/1.242/999), 896 MB
  • NVIDIA GeForce GTX 260 896 MB (576/1.242/999)
  • NVIDIA GeForce 9800 GX2 (600/1.512/1,000), 2 x 512 MB
  • NVIDIA GeForce 9800 GTX + (738/1.836/1.100), 512 MB
  • NVIDIA GeForce 9800 GTX (675/1.675/1.100), 512 MB
  • NVIDIA GeForce 9800 GT 512 MB (600/1.512/900)
  • NVIDIA GeForce 9600 GT 512 MB (650/1.625/900)
  • NVIDIA GeForce 9600 GSO 384 MB (555/1,350/800)
  • NVIDIA GeForce 9500 GT 512 MB (550/1400/800)
  • NVIDIA GeForce 8800 Ultra (612/1,512/1080), 768 MB
  • NVIDIA GeForce 8800 GTX 768 MB (575/1,350/900)
  • NVIDIA GeForce 8800 GTS 512 (650/1.625/970), 512 MB
  • NVIDIA GeForce 8800 GT 512 MB (600/1.512/900)
  • NVIDIA GeForce 8600 GTS (675/1,450/1,000), 256 MB
  • NVIDIA GeForce 8600 GT (540/1.190/700), 256 MB
• Power supply
  • M850 Coolermaster real power Pro modular (850 Watts)
• Peripherals
  • Toshiba SD-H802A HD-DVD drive
  • Pioneer BDC-202BK SATA Blu-ray drive
  • Samsung SpinPoint F1 SATA2 HDD with 750 GB, and 32 MB cache
• Housing
  • Coolermaster of stacker 832
• Driver versions
  • NVIDIA ForceWare 174.16
  • NVIDIA ForceWare 174.53 (9800 GX2, 9800 GTX)
  • NVIDIA GeForce 177.34 (GTX 280)
  • NVIDIA GeForce 177.39 (9800 GTX +, GTX 260)
  • NVIDIA ForceWare 177.72 (9500 GT, 9800 GT)
  • NVIDIA GeForce 177.92 (9600 GSO, GTX 260 NR)
  • ATi catalyst 8.3
  • ATi catalyst 8.6 release 5 (HD 4850, HD 4870)
  • ATi catalyst sample 8-52-2 (HD 4870 X 2)
  • ATi catalyst sample 8 53_RC1 (HD 4670)
• Software
  • Microsoft Windows Vista x 64 SP1
  • Microsoft DirectX 9.0 c
  • Microsoft Direct3D 10

Benchmarks

The following benchmarks were used during our test:

• Synthetic benchmarks:
  • 3DMark06 version 1.0.2
  • 3DMark Vantage 1.0
• Game benchmarks:
  • Assassin’s creed, D3D10(.1), full version, version 1.0
  • Bioshock, D3D10, full version, version 1.1
  • Call of duty 4, full version, version 1.5
  • Call of Juarez, D3D10, full version, version 1.1.0.0
  • Clive Barker’s Jericho, demo
  • Company of heroes, D3D10, full version, version 1.71
  • Crysis, full version, version 1.21
  • F.e.a.r., full version, version 1.08
  • Gothic 3, full version, version 1.12
  • Lost planet, D3D10, full version
  • Rainbow Six Vegas, full version, version 1.06
  • Stalker, full version, version 1.0005
  • Unreal Tournament 3, full version, patch 1.2
  • World in conflict, D3D10, full version, patch 1007

All benchmarks run with maximum details, so that the graphics card will be charged as high. As settings we chose this 1280 x 1024 and 1600 x 1200 (and 2560 x 1600 for graphics cards with 512 MB or more and a corresponding power). So we pay tribute to the modern high end Accelerator, make the resolutions by their computing power lower than 1280 x 1024 CPU limited. In addition to the pure resolutions we can iterate through the benchmark course with two (and if possible eight x) anti-aliasing and 16-fachen anisotropic filter. We use TSSAA (NVIDIA) or AAA (ATi) to smooth alpha-test textures because of compatibility problems no longer within our benchmark show.

After careful consideration and multiple analysis of even recorded game sequences, we have come to the conclusion that the quality of texture filtering on current ATi-and NVIDIA graphics cards in the standard setting (with slight advantages for GeForce products) are comparable. At NVIDIA, we change any settings and we leave in the ATi driver the a.i.-function on “Standard”.

Driver settings: NVIDIA graphics cards (G8x, G9x, GT200)

• Texture filtering: Quality
• Vertical sync: off
• Force MipMaps: no
• Trilinear optimization: A
• Anisotropic sample optimization: from
• Negative LOD bias: clamp
• Gamma adjusted AA: A
• AA mode: 1xAA, 4xAA, 8xQAA
• Transparency AA: from

Driver settings: ATi video cards (R(V)6×0, RV770)

• Catalyst a.i.: Standard
• Mipmap detail level: high quality
• Wait for vertical refresh: always off
• AA mode: 1xAA, 4xAA, 8xAA
• Adaptive anti-aliasing: off

Theoretical benchmarks

FillRate Tester

• This useful little program is used to measure the fill rates of a graphics card. In contrast to the computer or fill rates tests integrated in the 3DMark that primarily measure the bandwidth in the case of single-texturing, this program can be quite differentiated information on different types of fill rate, including the PixelShader fill rates, which we want to look at.

  Refreshrate has been tested in 1024 x 768 in 32-bit with 24 bit Z and 8 bit buffer and 60 Hz.

• Download: Fillrate Tester

Fablemark

• The Fablemark was developed as the subsequent Templemark of PowerVR and serves despite a very high share in Overdraw of the display of the strengths of the KYRO chip regarding the stencil buffer.
  Of course the performance of the stencil calls also on all other cards strongly so that this test can be an indication of upcoming games, insert a Z / stencil-only pass prior to actual rendering, to prevent advance any Overdraw.
  Has been tested using the following command line: [InstallDir] D3DFablemark.exe benchmark = 1 – width = xxxx-height = xxxx-bpp = 32 ”

• More information: PowerVR.com
• Download: PowerVR.com

ShaderMark

• The ShaderMark is currently available in the current version 2.1 and was developed by Tommti systems. Thanks to numerous updates, is the benchmark is still on the amount of time, and measures the performance of the shader units of modern graphics cards. It supports also the shader model 3.0, which is why it is suitable for a comparison of current architectures. Tested up to 25 different shader instructions in the resolution are 1920 x 1200, which are all written in HLSL (high level shader language) high-level language.

• Download: ShaderMark.de

&Lt; div > ‘ D3DRightmark 4 beta “benchmarks < div >

Synthetic benchmarks

3DMark06

The well-known series of benchmark from Futuremark is now published in version 2006 and consequently listens to the name “3DMark06″. Of the six test scenes, measure the performance of the graphics card four sequences and show a graphic splendour that is unparalleled. To achieve that Finns set which is why not only massively the shader model 3.0 is used on most modern 3D technology, extremely elaborate textures, spectacular particle effects, shadow complex computations and as another highlight “High Dynamic Range Rendering” – short HDRR – are used. While Futuremark opts for FP16 HDR, which currently provides the best possible image quality, but is also expensive to calculate. There are more details about this program in our detailed article.