From: Philippe B. <phi...@un...> - 2005-08-17 15:21:24
|
Hi Folks, Here are my conclusions about the nVidia graphic cards. The benchmarks found on Tom's hardware web site show that the =20 performances of the AGP8x FX 1100 are about the same as the PCIe FX =20 1300. Taking that assumption I just took the average spec out of the 3 =20= given on nVidia's web page (see LINKS below). *** This gives the following table: *** Graphic card FX 4400 FX 3400 FX 1400 FX 1300 FX 1100 Bus type PCIe PCIe PCIe = PCIe AGP 8x Perf(rel to 1100) 1.97 1.73 1.65 = 1 1 Price(rel to 1100) 2.95 1.8 0.8 = 0.5 1 Ratio Perf/Price 0.66 0.96 2.06 = 2 1 (FX 1100 costs about 780 Euros) These numbers are probably not extremely accurate, but the tendency =20 shown in this table looks in pretty good agreement with the numerous =20 specs reported on hardware.fr (see LINKS below; unfortunately this =20 paper written in French, but charts and graphics are easy to catch). They also look in agreement with the answers I got from our =20 crystallographic bulletin boards (see QUOTES below). *** CONCLUSION: we'll go for an FX 1400 ! *** *** ABOUT RivaTuner (suggested by Wang Chern Hoe): *** I'll keep that trick for my personal record and use, but we don't want =20= to use any dual boot solution in the lab (at least not yet). For a =20 whole lab solution I will agree with Cameron (cm...@uc...) who =20 suggests to opt for a real hardware stereo card. But for personal purposes I find interesting to turn a Geforce 6800 =20 into a Quadro and be able to run O, NOC and PyMol from the XP side, but =20= I've read: - this will turn it into a Quadro 540 but not 4400 (COULD YOU CONFIRM =20= THIS, WANG ?) - this could be done only on some graphic boards from UNIKA (without =20 any reseller in France I think) - as asked Sabuj (sab...@va...) what is the magic =20 taht modify the pin connector for the emitter. Wang could you tell us how physically the connection with the emitter =20= is made ? *** LINKS *** BENCHMARK RESULTS: FX 1100, ATIfireGLx2(t): http://graphics.tomshardware.com/graphic/20040323/opengl-fx1100-x2t=20 -08.html FX 1300, FX 3400, ATIfireGLv5100: http://graphics.tomshardware.com/graphic/20040816/opengl-pcie-14.html FX 1400, FX 3400, FX 4400, ATIfireGLv5100: http://www.hardware.fr/articles/560-1/comparatif-7-cartes-3d-pro-pci-=20 express.html AND: nVIDIA graphic cards specs AGP8x/PCIe http://www.nvidia.fr/object/IO_11761_fr.html *** QUOTES *** [J Mesters; me...@bi...] We have two FX1100 (128 MB ram) up and running just fine under linux. =20= The speed for openGL applications is more than satisfactory. .../... =20 Only Pymol will benefit clearly from faster cards with more memory when =20= rendering etc.. The Iiyama HM204DT is the best (reference class) monitor around with =20 142 kHZ and 200 Hz maximum! Great stereo quality. If that is too expensive, buy the Belinea 108035 with 125 kHz and 160 =20= Hz maximum for half the price. --- [Doug Bailley; ba...@nc...] Our "standard" linux workstation is as follows. Dell Precision 370/380, Pent.4-3.6, 1GB RAM 120GB SATA, DVD+-RW, Quadro =20= FX1400 NEC-Mitsubishi FP2141SB-BK 22" CRT NuVision 60GX Wireless stereo emitter and glasses RedHat Enterprise Linux WS. For our usage, we did not see a significant difference in performance =20= between the FX1400 and some of the 3xxx series Quadro products we =20 tested to justify the higher cost. The NEC-Mitsubishi FP2141SB-BK 22" CRT is an excellent monitor. Cannot =20= comment on any others as the NEC is all that we use. --- [Kay Diederichs; kay...@un...] .../... a) the nVidia FX3450 is a bit expensive (=801300); you won't notice a =20= difference to a card of less than half the price (FX1300) in O. .../... --- Thanks to all and to any other contributor that I would have forgotten, Philippe ------------------------------------------------------------------------ Philippe BENAS, Ph.D. X-ray diffraction facility manager Laboratoire de Cristallographie et RMN Biologiques, UMR 8015 CNRS Faculte de Pharmacie - Universite Paris 5 4, Avenue de l'Observatoire - Case 48 F-75270 Paris cedex 06 --- Office: +33.(0)1-5373-9831 (use ext. 9977 instead to leave a voice mail) Fax: +33.(0)1-5373-9925 X-ray Room (rotating anode): +33.(0)1-5373-9977 X-ray Room (small molecules): +33.(0)1-5373-9679 --- Email: phi...@un... Url: http://lcrbw.pharmacie.univ-paris5.fr/ ------------------------------------------------------------------------ |