Four Chipsets For Xeon Platform
Currently there are four different chipsets for the Socket 604 platform. The oldest is the 7205 (Granite Bay), with an FSB speed of 133 MHz (FSB533) and a dual DDR266 interface. The chipset is also available for the P4 platform with Socket 478. The Granite Bay is the basis for all of the following chipsets with a dual-memory interface: the Intel 875P (Canterwood), the 865 (Springdale), the 7505 (Placer) and 7501 (Plumas 533); additionally, workstation and server chips were developed.
| Chipset | FSB | CPU Plattform | CPU Support | AGP-Graphic | Memory Support | ECC Support |
|---|---|---|---|---|---|---|
| Intel 850 | 100 MHz | Socket 603 | Dual | yes | Dual-RDRAM | yes |
| Intel 860 | 100 MHz | Socket 603 | Dual | yes | Dual-RDRAM | yes |
| Intel 7205 (Granite Bay) | 133 MHz | Socket 603 | Dual | yes | Dual-DDR266 | yes |
| Intel 7500 (Plumas) | 100 MHz | Socket 604 | Quad+ | no | Dual-DDR200 | yes |
| Intel 7501 (Plumas 533) | 100 MHz | Socket 604 | Quad+ | no | Dual-DDR200 | yes |
| Intel 7505 (Placer) | 133 MHz | Socket 604 | Dual | yes | Dual-DDR333 | yes |
| Intel 875P (Canterwood) | 133 MHz | Socket 604 | Dual | yes | Dual-DDR333 | yes |
Performance: Xeon 3.06/ L3 Cache
The performance in dual operation is a significant factor when purchasing a workstation. As the performance of the dual Xeon system with an L3 cache is mainly interesting when compared to a system without the cache, we used the following applications.
Lightwave 7.5b, Cinema 4D XL 8.1 and 3D Studio Max 5.1 are among the most frequently used applications in the professional sector. To supplement this, we use SPECviewperf, which determines an average from various high-end program packages. Oftentimes in video applications, large data quantities must be transcoded by the MPEG process (MPEG-2, MPEG-4, MPEG-7). A common representative of this category is the encoder from MainConcept or Pinnacle Studio 8.5, which tends to serve the low-cost market, although the MPEG-2 encoding works rather well with the studio software. To investigate the effect of expanding the cache with distributed computing applications, we used a SETI text client (Version 3.04) with Intel optimization. For the sake of thoroughness, we also included the results of the SPEC CPU 2000 benchmark, which is the standard for comparisons of cross-platform systems.
In order to get decent graphics performance, we used Nvidia GeForce FX5900 Ultra, which is equipped with a 256 MB texture memory. 512 MB of local memory from Infineon (ECC registered) are also used.
