DFI 748-AL: First Look at SiS 748
by Wesley Fink on August 30, 2003 12:00 AM EST- Posted in
- Motherboards
DFI 748-AL: Stress Testing
We performed stress tests on the DFI 748-AL in these areas and configurations:1. Chipset and motherboard stress testing, which was conducted by running the FSB at 212MHz; and
2. Memory stress testing, which was conducted by running RAM at 400MHz with 2 DIMM slots filled and at 400MHz with 3 DIMM slots filled at the lowest memory timings possible.
Front Side Bus Stress Test Results:
As standard practice, we ran a full range of stress tests and benchmarks to ensure the DFI 748-AL was absolutely stable at each overclocked FSB speed. These stress tests included Prime95 torture tests, which were run in the background for a total of 24 hours.In addition, we ran several other tasks — data compression, various DX8 and DX9 games, and apps, like Word and Excel — while Prime95 was running in the background. Finally, we ran our benchmark suite, which includes ZD Winstone suite, Unreal Tournament 2003, SPECViewperf 7.0, and Gun Metal Benchmark 2. While we were able to boot and run some tests at speeds as high as 217MHz FSB at default voltage on the 748-AL, 212MHz was the highest FSB overclock that we were able to achieve at default voltage with the Asus without encountering any reliability issues.
Memory Stress Test Results:
The memory stress test is very simple, as it tests the ability of the 748-AL to operate at its officially supported memory frequency (400MHz DDR) at the lowest supported memory timings that our Corsair TwinX LL modules support. The SPD timings for the Corsair v1.2 modules is 2-2-3-6, but without any real means to set individual memory timings on the 748-AL, we had to rely on other utilities to report real memory timings. CPUZ (www.cpuid.com) does not recognize the SiS 748 chipset memory controller, and we could not use it to determine memory timings. We had to depend on results reported in SiSoft Sandra Max3 Memory Test for memory timings. Sandra recognized the chipset correctly, but reported memory timings may not be 100% correct.| Stable Dual DDR400 Timings (2/3 banks populated) |
|
| Clock Speed: | 200MHz |
| Timing Mode: | Performance |
| CAS Latency: | 2.5 |
| Bank Interleave: | 4 |
| RAS to CAS Delay: | 4T |
| RAS Precharge: | 5T |
| Precharge Delay: | 4T |
| Command Rate: | N/A |
We did some experimenting with the “Normal” and “Performance” selections in BIOS. Normal memory timings are reported to be 2.5-4-4-8 in Sandra. When Performance is selected, CAS settings can be chosen, but CAS 2 would not work with the Corsair 3200LL modules or any other high-speed module that was tested. The lowest setting that works is “Performance”, “CAS 2.5” and Sandra timings are reported as 2.5-4-4-5.
Filling all 3 available memory banks is much more strenuous on the memory subsystem than testing 2 banks. With all 3 banks filled, the DFI 748-AL had no problems with the same timings that worked with 2 dimms.
| Stable DDR400 Timings (3/3 banks populated) |
|
| Clock Speed: | 200MHz |
| Timing Mode: | Performance |
| CAS Latency: | 2.5 |
| Bank Interleave: | 4 |
| RAS to CAS Delay: | 4T |
| RAS Precharge: | 5T |
| Precharge Delay: | 4T |
| Command Rate: | N/A |
We were pleased to see the DFI run with stability with 3 DDR400 dimms. With the limited Memory Timing options available, we really can't tell you much about memory timings with 2 and 3 dimms. Normally, 3 dimms requires timings a bit slower than 2, but until DFI adds memory timings in the release BIOS, we really can't tell you the fastest timings that will work on the 748-AL. It’s worth stating again that the real world performance difference between aggressive memory timings and more relaxed memory timings, such as SPD, are often very small. That is more than just words in this 748-AL review, as you will see in performance tests using these very slow memory timings.
We tested all these memory timings using several stress tests and general applications to guarantee stability. Prime95 torture tests were successfully run at the timings listed in the above charts. We also ran 3DMark and Super Pi. None of these stress tests created stability problems for the DFI 748-AL at these memory timings.
Facebook
Twitter
RSS
22 Comments
View All Comments
JAGedlion - Saturday, August 30, 2003 - link
... The DFI 748-AL has 5 IDE ... I see you are referring to PCI slots, can they be called IDE slots too?Anonymous User - Saturday, August 30, 2003 - link
wonder how far would the FSB go if the vdd was pushed into its max. i guess it would definetly go beyond 212MHz, maybe you should have decreased the multiplier.In fact, today I ran a few benchmarks with my epox 8k5a2 KT333 and reached as far as 215. pushing it further the AGP bus became bottleneck and i was using ATI's oldskool's rage128. it amazes me how come brand new hardware can't handle even 72MHz agp clock, whereas my 4 years old card handles 86MHz. in FSB200 (agp/pci 80/40) i'm not experiencing any staiblity issues burning it with prime or cpuburnK7 even 3days. odd
Anonymous User - Saturday, August 30, 2003 - link
One can easily see from his comment that Jeff7181 doesn't know much about these things, but still thinks he's got an argument. Anandtech crew are admirably wasting their efforts in trying to reason with such people.About the board. Two phase and no 12V connector drives me away from it just by looking at it.
About SiS. They have nice chipsets, but the motherboard makers skimp on the design because they use the SiS chipsets for another market than they do for instance with their nForce2 boards. That's unfortunate since I would rather go with SiS drivers than nVidias. Performance is often overemphasised when reliability and ease of use becomes a second thought for many buyers.
Anonymous User - Saturday, August 30, 2003 - link
May i make a helpful suggestion/request for the next (or even current) review? Could you please post the ranges of the bois setting instead of just the maximums? for example, instead of saying "cpu voltage in the full range, up to 1.85v" could you specify exactly what that range is? like 1.1-1.85v? I know the nForce boards can do this, but I'm unclear about 748 boards.I'm personally looking at this board for a home theater PC that I'm going to try and underclock the CPU and see if i can get it to run without a fan and still be fast enough to play DVD's... because of this, I'd like to know the minimum voltage you can run the CPU at, and I don't find that in a lot of reviews. Just that little change would satisfy the speed freaks as well as us quiet freaks.
just a suggestion...
Evan Lieb - Saturday, August 30, 2003 - link
Jeff,As Wesley mentioned and as I said before, Epox and ASUS' nForce2 Ultra 400 motherboards perform nearly exactly the same as the Gigabyte and DFI nForce2 Ultra 400 motherboards (and others) we use in AMD motherboard reviews. I have both boards in my labs and have decided not to send them to Wesley for retesting because I wouldn't want to waste his time with something like that.
Again, we have tested ASUS and Epox boards before and found them to perform within about 1% of other nForce Ultra 400 motherboards. We changed our testbed and made sure a few good nForce2 Ultra 400 motherboards we added for comparison to non-NVIDIA chipsets.
We may add those very popular Epox and ASUS boards to an nForce2 roundup of some type. But for a SiS 748 review like this one, the boards we used were perfectly adequate.
Dennis Travis - Saturday, August 30, 2003 - link
Very good review Wes, keep up the great work.Wesley Fink - Saturday, August 30, 2003 - link
Jeff7181 -I'm asking, Jeff, that you Please give me a break. I took over AMD about 6 weeks ago, and all Athlon reviews prior to that were done by Evan. Evan has answered here, on every post you have made on every AMD review, that the performance difference for Ultra 400 versions was very small, which is why he chose NOT to review the Ultra 400 updates of the boards you mention. The results of our tests of the regular nForce2 versions of the boards you mention are in our database. We also have the two top Ultra 400 performers - the DFI NFII Ultra LANParty and Gigabyte 7NNXP with UPDATED benchmarks in our tests for your comparison. If you recall, we also changed our video cards and benchmark suite recently, and we are finally accumulating a useful database of new benchmark data.
My point, Jeff, is I do NOT have access to the motherboards you mention and cannot add them to our database. I am doing my best to provide useful information to AMD fans, but I cannot provide what you ask. My next AMD review is a Ahtlon64 Preview, and with the launch only 3 weeks away it just doesn't make good use of review time to request an Asus A7N8X Deluxe and retest it.
If you can please show a little patience we will do our best to provide the kind of information you are looking for in the future.
Thank you.
Wesley Fink - Saturday, August 30, 2003 - link
Charts are in Flash because it uses less bandwidth than any of the alternatives. With over 3300 articles and reviews on-line plus News and Forums, bandwidth is a very important subject to a site like AnandTech.Anonymous User - Saturday, August 30, 2003 - link
Why are the charts in Flash? A simple jpg or gif would do.Jeff7181 - Saturday, August 30, 2003 - link
Again I ask, where the hell is the comparisons to the motherboards we all know? The A7N8X Deluxe, NF-7, 8RDA+. And what's with the cheesy title? Can SIS outperform nVidia's nForce2? I could have answered that without even testing them! Geeze... what's going on at AnandTech? These reviews suck!