Apollo Dual BX 450

When Premio sent me its latest workstation, the Apollo Dual BX 450, I wasn't sure what to expect. I discovered that this midsized manufacturer's high-end workstation is a lot like the high-end systems I have seen from almost every other major manufacturer. The Apollo Dual BX arrived in the Windows NT Magazine Lab with two 9GB 10,000rpm SCSI hard disks, one 256MB nonvolatile RAM (NVRAM) DIMM, an Adaptec Array PCI RAID controller (PCI based—not RAIDport based), a 3Com Fast Etherlink XL 10/100Mbps PCI card, a Creative Labs 2X PC-DVD drive with accompanying Dxr2 decoder PCI card, a Creative Labs 64-bit sound card with a pair of monstrous 36-watt speakers, an ELSA GLoria-XXL Accelerated Graphics Port (AGP) video card, and a 21" MAG monitor with 1800 * 1440 maximum resolution. In a word—loaded.

The Apollo Dual BX has four PCI slots, one shared PCI/ISA slot, one ISA slot, and one AGP slot on the motherboard, providing reasonable expandability under normal circumstances. However, the Lab's test system's DVD encoder card, Ethernet card, RAIDcontroller card, sound card, and graphics card took all the slots but one. And while I'm on the subject of expansion, my test system's midtower case had only two hard disks.

In addition to incorporating the first PC-DVD drive I've seen on a production system running NT Workstation, the Apollo Dual BX is the first Pentium II system I've seen with fans integrated into the heat sinks for each of the two processors. A quick perusal of the AMI BIOS reveals the fans' rotational speed and the processors' temperature.

To test the system performance of the Apollo Dual BX, I ran AIM Technology's WNT workstation benchmark. The two most telling metrics from the AIM tests are the AIM WNT Peak Performance and AIM WNT Sustained Performance values. The peak performance value is a good general measure of how well a system will perform for an extreme power user; the sustained performance value is a good general measure of a system's performance handling continuous loads over an extended period. The Apollo Dual BX's peak performance score was 770.6 applications per second; the sustained performance score was 214.3 applications per second. These metrics are lower than I expected but might be higher with a different hard-disk configuration. (For more information about the AIM benchmark tests, go to the AIM Technology Web site at http://www.aim.com.)

To test the OpenGL capabilities of the Apollo Dual BX's ELSA GLoria-XXL graphics card, I ran three viewsets from the Viewperf benchmark: CDRS, DX, and Lightscape. Each metric is measured in frames rendered per second. The Apollo Dual BX's CDRS score was 94.899, its DX score was 17.797, and its Lightscape score was 1.975. (For more information about OpenGL and the Viewperf benchmark, and to compare the Lab's Apollo Dual BX test system's scores with other systems' Viewperf scores, go to http://www.specbench.org.)

Although the ELSA GLoria-XXL graphics card's performance is impressive for an OpenGL card, the Apollo Dual BX's software drivers could stand improvement. When I tried to replace the ELSA card with a Matrox Millennium II graphics card to run the AIM WNT tests (the Lab's standard configuration for workstation tests), I found the ELSA software all but intractable. I stopped every ELSA-related device in each device's Control Panel applet, removed the drivers and utilities through Control Panel's Add/Remove Programs applet, and even—at the recommendation of an ELSA tech-support engineer—searched for and deleted every ELSA entry in the Registry I could find. However, my efforts were to no avail. Nothing short of reinstalling NT Workstation let me run any graphics card other than the ELSA GLoria-XXL on the Apollo Dual BX.

The Apollo Dual BX proved to be stable and fast. I was particularly impressed by the monitoring features of the Pentium II processors—nice touches on a very nice system.

End of Article