ASRock 775Dual-880Pro

Another one of VIA's innovations with the VIA PT880 Pro, in addition to the VIA Universal Graphics Interface, is the VIA DualGFX Express. The DualGFX technology allows the AGP and PCI Express slots to be occupied on the motherboard and running simultaneously, thus allowing for a greater number of possible displays. As the GeForce 5900XT we used for the AGP testing contains an Arctic Cooling NV Silencer 3, which takes up an entire expansion slot, we pulled out a few more AGP and PCI Express graphics cards to try this technology. On the AGP side of things we attempted to run an ATI RADEON 9200 and 9250 and even an old SiS Xabre. With the PCI Express slot we tried the Leadtek 6600GT and a Sapphire ATI X300SE. Even with the number of card combinations, we weren't able to find a successful combination where the graphics cards would work simultaneously. No matter the combination, both graphics cards would fail to initialize on boot. In an attempt to fix the situation, we altered the BIOS, changed around graphics cards, and even installed a PCI (NVIDIA GeForce FX5200) graphics card in hopes of initializing the display while the AGP and PCI Express slots were being occupied. At this time, it looks as if ASRock engineers simply didn't support VIA's DualGFX Express, or that the technology is extremely restrictive as to what graphics cards it can utilize.

The first benchmark we decided to utilize for testing the ASRock 775Dual-880Pro review is the Unreal Tournament 2004 Demo (3334). In conjunction with UMark for Linux Beta 3, we ran DM-Rankin with 12 bots at 800 x 600 - High Performance, 1280 x 1024 - High Performance, and 1280 x 1024 - High Image Quality. The next benchmark from our usual slew of tests comes in way of Return to Castle Wolfeinstein: Enemy Territory (2.60). The in game display settings used during the time demo benchmarking was 32-bit color depth, high geometric detail; high texture detail, 32-bit texture quality, trilinear texture filtering, 24-bit depth buffer, and single pass dynamic lighting. The Enemy Territory time demo used was captured from the Railgun map and is available at 3dcenter.de. With Enemy Territory, we simply ran the game with a resolution of 1280 x 1024. The next and final gaming benchmark is Doom 3 (1.3.1302). With Doom 3 we ran it at 1280 x 1024 - High Quality with no antialiasing or Anisotropic Filtering and then again with 2x Bilinear AA and 2x AF. For a measure of the I/O performance with the integrated IDE controller, we also performed timed disk reads (hdparm -t) to the 7200RPM 8MB cache IDE drive. Also taking into effect the disk performance while also relying more on the CPU and memory, we measured the time required to Gunzip armyops230.tar, which was an archive of the 745MB America's Army Linux binary. For some compiling action, we measured the time required to compile the LAME MP3 encoder (3.96.1). Also using LAME (3.96.1) the time to encode an 81.3MB WAV file as an MP3 was measured. For the last set of benchmarks that are CPU centric, we ran Opstone Sparse-Vector Scalar-Product and Opstone Singular Value Decomposition benchmarks from BlueSailSoftware. For this testing, we used the Pentium 4 optimized build. As always, unless otherwise specified, each test was run three times with the average of the three being recorded.