Page 1:The More The Merrier? SiS655 With Dual DDR333 On Test
Page 2:Wider, Not Faster: Dual Channel DDR Instead Of Speed Increase
Page 3:Waiting For Serial ATA
Page 4:16 P4 Chipsets Compared
Page 5:16 P4 Chipsets Compared, Continued
Page 6:AOpen AX45-4D Max
Page 7:AOpen AX45-4D Max, Continued
Page 8:Asus P4SDX Deluxe
Page 9:Asus P4SDX Deluxe, Continued
Page 10:Gigabyte SINXP-1394
Page 11:Gigabyte SINXP-1394, Continued
Page 12:MSI 655 Max (MS-6730)
Page 13:MSI 655 Max (MS-6730), Continued
Page 14:Test Setup
Page 15:Benchmark Results
Page 16:Comanche 4 Demo
Page 17:Unreal Tournament 2003
Page 19:Office-/Internet-Performance: Sysmark 2002
Page 20:SPECviewperf 7
Page 21:3D Studio Max 5.1
Page 22:Conclusion - Don't Jump The Gun
Conclusion - Don't Jump The Gun
It's a while since we carried out a group motherboard review as confusing as this one. Essentially, the SiS655 shows it has the potential to hit Intel's E7205, alias Granite Bay, where it hurts. The good showing of the Asus P4SDX underlines this, since the slight increase in clock speed is not the only thing that makes it a top performer.
However, it seems that the manufacturers are not yet able to fully realize its potential. In view of the great variations in performance, we have to conclude that no one has quite got to grips with the SiS655 - or that the chipset design is simply not yet fully mature.
Not only that, but we came across number of serious compatibility problems with the memory timings we had chosen (DDR333 CL-2 at optimum settings). The AK45-4D Max from AOpen refused to allow our DDR333 modules to run at 333 MHz and CL-2. Only when we changed to DDR400 memory were we able to run at 333 MHz (!). MSI's 655 Max refused to cooperate at CAS Latency 2 and we could only persuade it to do so at CL-2.5. With the Gigabyte board, on the other hand, it was impossible to run DDR400 modules at 333 MHz. Instead, we were forced to overclock the system.
The one thing all the boards had in common was that they would not work with the maximum timings of the memory modules we used. Even the P4SDX would only complete the SYSmark properly when the memory settings were left on auto. The MSI board was particularly sensitive; anything that was not "normal" caused instability or a refusal to boot.
We fail to understand the obvious overclocking of the Asus P4SDX FSB. The motherboard market leader has no need to resort to such measures.
Therefore, the SiS655 is not a great choice right now. We can't say it clearer than that. We weren't testing second tier motherboards here; these are products from the top manufacturers. If these boards had problems, how do you think boards from the low-cost makers will fare?
We would point the finger at SiS. The dual channel memory interface means that the SiS655 is clearly aimed at the lower price segment of the performance market. However, the 655 cannot satisfy the requirements of a high performance chipset for P4 processors above 3 GHz. The inconsistencies in the memory setup are too great to compensate for the marginal benefits it offers over the Intel E7205. It is food for thought that the 850E, the current version of the two-year old 850 chipset, is yet to be convincingly beaten.
- The More The Merrier? SiS655 With Dual DDR333 On Test
- Wider, Not Faster: Dual Channel DDR Instead Of Speed Increase
- Waiting For Serial ATA
- 16 P4 Chipsets Compared
- 16 P4 Chipsets Compared, Continued
- AOpen AX45-4D Max
- AOpen AX45-4D Max, Continued
- Asus P4SDX Deluxe
- Asus P4SDX Deluxe, Continued
- Gigabyte SINXP-1394
- Gigabyte SINXP-1394, Continued
- MSI 655 Max (MS-6730)
- MSI 655 Max (MS-6730), Continued
- Test Setup
- Benchmark Results
- Comanche 4 Demo
- Unreal Tournament 2003
- Office-/Internet-Performance: Sysmark 2002
- SPECviewperf 7
- 3D Studio Max 5.1
- Conclusion - Don't Jump The Gun