System Setup and Notes
Armari Magnetar QX2
Hardware
- 2 x Intel Xeon X5355 (2.66GHz, quad-core, 8MiB L2 total, LGA771)
- 4 x 2GB Hynix PC2-5300F-555-11 FB-DIMM, 667MHz, 5-5-5-11
- ASUS DSGC-DW, Intel i5000X + Intel 6321ESB, FB-DIMM, 2 x LGA771
- Enermax Galaxy EGX1000EWL PSU, 1000W peak, modular
- PNY Quadro FX 5600 (G80 1.5GiB), PCIe x16
- Western Digital Raptor 150GB (boot)
- Western Digital Caviar RE2 500GB (data)
Software
Armari configured it as requested with Windows XP Professional x6, the best fit for the software suite we're testing with and the system memory configuration we asked for. With the Quadro FX 5600 having 1.5GiB of board memory, a 64-bit OS is essential to get the most out of it and the host system. NVIDIA supply 64-bit drivers for Windows XP, Windows Vista, Linux and Solaris.- Microsoft Windows XP Professional x64
- NVIDIA ForceWare 97.78 (May 1st 2007, WHQL)
- ATI FireGL Driver 8.353.1.1
- NVIDIA MAXtreme D3D9 64-bit (D3D driver for 3dsmax 9)
- Autodesk 3ds Max 9 64-bit
- Autodesk AutoCAD 2008 64-bit
- Microstation V8 XM Benchmark
- B3D Texture Test
Notes
We tested the Quadro FX 5600 alongside an ATI FireGL V7350 (ATI R580, 1GiB, PCI x16) with the professional apps, and added GeForce 8800 GTX to the mix for the theoretical rate testing and power-level work. All tests were run 3 times and the median reported. Note that the MAXtreme driver is for 3ds Max 9 using the D3D renderer only, and we test using the D3D renderer exclusively (because of problems with the FireGL being unable to complete the render, amusingly).We couldn't make the 32x antialiasing setting work un the 97.78 driver, with image quality no different to the 16x setting and an examination of sample points using OpenGL not detecting any extra multi or coverage samples at 32x. Hooray.
