SiS 756: PCI Express for AMD Socket 939
by Wesley Fink on September 13, 2005 12:05 AM EST- Posted in
- CPUs
Overclocking: SiS 756 Reference Board
*The SiS 756 Reference Board artificially resets the clock to 220 with any setting above 221. This happened with both the stock multiplier (12X) and lower multipliers down to the lowest board multiplier of 8x.
It wasn't enough that the settings for testing overclocking were very limited on the SiS 756. The board artificially set higher clock settings back to 220. We thought that we were testing at 233 CPU clock until we ran CPU-Z, which told us that the board was running at 12 times a clock frequency between 219.5 and 220.3. SiS would have been better to let the board fail overclocking naturally instead of manipulating clock frequencies.
This same phenomenon was seen when we tried to lower multipliers and crank up to the board max frequency of 255. The SiS 756 Reference either dropped to 220 or lowered the multiplier to 133 or both. As a result, we could find no way to clock higher than 220 (+1-%) at any ratio. All-in-all, this was one of the most disappointing performances that we have seen in overclocking a Socket 939 board. Perhaps SiS is actually shipping 756 chipsets with better OC potential. We certainly hope that they are.
| SiS 756 Reference Board | |
| Overclocking Testbed | |
| Processor: | Athlon 64 4000+ (2.4GHz, 1MB Cache) |
| CPU Voltage: | 1.55V (default 1.50V) |
| Cooling: | Thermaltake Silent Boost K8 Heat sink/Fan |
| Power Supply: | OCZ Power Stream 520W |
| Memory: | OCZ PC3200 Platinum Rev. 2 (Samsung TCCD Memory Chips) |
| Hard Drive: | Seagate 120GB 7200RPM SATA 8MB Cache |
| Maximum OC: (Standard Ratio) |
220*x12 (4x HT, 2.5-3-3-8) 2640MHz (+10%) |
| Maximum FSB: (Lower Ratio) |
220* x 9 (4x HT, 1T) (2640MHz, 2 DIMMs in DC mode) (+10% Bus Overclock) |
It wasn't enough that the settings for testing overclocking were very limited on the SiS 756. The board artificially set higher clock settings back to 220. We thought that we were testing at 233 CPU clock until we ran CPU-Z, which told us that the board was running at 12 times a clock frequency between 219.5 and 220.3. SiS would have been better to let the board fail overclocking naturally instead of manipulating clock frequencies.
This same phenomenon was seen when we tried to lower multipliers and crank up to the board max frequency of 255. The SiS 756 Reference either dropped to 220 or lowered the multiplier to 133 or both. As a result, we could find no way to clock higher than 220 (+1-%) at any ratio. All-in-all, this was one of the most disappointing performances that we have seen in overclocking a Socket 939 board. Perhaps SiS is actually shipping 756 chipsets with better OC potential. We certainly hope that they are.
Facebook
Twitter
RSS
31 Comments
View All Comments
DarkFoon - Tuesday, September 20, 2005 - link
Maybe I missed it in the article, but am I the only person who noticed the large "tab" connector (for want of the correct term) on the bottom of the board? It looks like it plugs into a larger board; Is this for blade computing?What is it?
Peter - Friday, September 16, 2005 - link
I can but shake my head in disbelief that there's an entire page dedicated to how the chipset handles RAM ... I thought everyone should have figured by now that the AMD64 platform has the RAM controller inside the CPU, and the chipset has no business on the RAM at all. This should at least have filtered through to reviewers, no?regards,
Peter
Wesley Fink - Saturday, September 17, 2005 - link
Peter -You should have read the RAM page instead of shaking your head. Had you done so you would have discovered that DESPITE the memory controller on Athlon64 being on the chipset, optimal tRas timings are different with each Athlon64 chipset. We proved this long ago.
How the chipset communicates with the on-chip memory controller DOES have an impact on memory performance, though that influence is much less than the chipset-based memory controlers like those for Intel. If you want proof run an industry-standard mem86 test CD ROM on the nForce4, ATI, VIA, ULI, and SiS A64 chipsets. The difference in memory bandwidths for the SAME tested memory using the SAME A64 Processor will surprise you.
As a review site we question and measure conventional wisdom and PR spin. Things are not always as you are told they are by manufacturers or self-appointed experts.
JeSpre - Wednesday, September 14, 2005 - link
I've been after this board since I read the ocworkshop review about a month ago. I've requested it at several online retailers also. Most 939 mATX boards are pretty lousy by comparison. A linux user really can't even consider one with an ATi chipset, while the NForce4 boards (really just one with or without gigabit ethernet) are pretty weak (especially with regard to overclocking), and have their own issues. Meanwhile, SiS already has linux drivers on their site for download. Couple that with the overclocking potential, and this is really the only board I'm after unless some good NVidia C51 mATX boards come along.justly - Tuesday, September 13, 2005 - link
I saw this last night and reread it today to make sure I didn’t miss anything. What I found was disappointing. My disappointment wasn’t in the chipset but your choice to review a reference board that already has production boards made from it (although not available in the US). You even stated (3 time I believe) that reference boards where shipped almost 6 months ago (this isn’t my idea of a timely “review”) especially when OCWorkbench did a review of a production board in the first week of JULY (a full two months ahead of Anandtechs reference board “review”).I was glad to see Anandtech review the ULi M1695 reference and production board so quickly (although I would say that this doesn’t make up for the lack of reviews for any earlier ULi chipset boards, or production ATi chipset boards, or the lack of most VIA board reviews, or this horribly late SiS reference board review… get the point?). The sad thing is I don’t believe Anandteck would have been on top of the ULi chipset reviews if they had not heard about its overclocking potential at a trade show.
Referencing the review at OCWorkbench, this new SiS chipset is able to attain a HTT of 285MHz, so it’s not a dismal overclocker like your “review” (of the reference board) seems to indicate. Then you want to complain about its lack of features, fair enough as long as you compare it to similarly priced boards. After all most of those features do come with a price, do they not?
It was also nice to finally see someone say that (Ethernet) CPU utilization is a worst case scenario that will almost always be much lower than what was measured. I agree SiS needs to work on this, but I don’t think it’s as important as some make it out to be.
Wesley Fink - Wednesday, September 14, 2005 - link
After being severely burned in past SiS chipset reviews - where the chipset was great but no mainstream boards ever appeared - our intention was to wait until retail boards appeared with the SiS 756 before reviewing this SiS chipset. We've waited a long time, and the closest we have to a retail 756 for review is the A9 module for the ECS PF88 modular motherboard. This would not really be a fair review for SiS.OC Workbench is in Singapore, so it is natural they always have a handle on the value chipsets and boards so popular in Asia. Many of thiose boards, as you pointed out, never make it to the US or other parts of the world. We did not receive a similar 756 board for review, though we did try to get one for review.
SiS has complained about our ULi coverage and the absence of SiS coverage, and in the interest of completeness we decided to review the 756 we had bypassed. We asked SIS first if they wanted to provide an updated BIOS for the Reference Board or an updated board, but we heard nothing back from them. Frankly you are right, we should have reviewed a retail board, but if a retail board is not available I will make no apologies for reviewing the only SiS 756 we could get - the Reference Board. We were honest about availability and issues.
SiS did provide us with updated chipset information, so we also provided the latest information available on the entire SiS chipset line and the new Southbridges that will soon be available in this review. This information was not available when the 756 Reference first shipped.
justly - Wednesday, September 14, 2005 - link
I appreciate your candor regarding SiS and the unprofessional business dealings that you experienced or heard about.On the other hand, feeling that you got severely burned because no manufacturer made a board using SiS that was up to your standards is absurd. Just because you gave a chipset a favorable rating doesn’t mean people will buy it, and if people don’t buy it why would a manufacturer want to make it, it should have no effect on how you feel.
SiS is by no means blameless, but you have to realize that even though you gave a SiS chipset an “Editor's Choice” award that is where you stopped. I cant remember the last time Anandtech recommend any chipset other than Nvidia in any guide, in fact every time I have seen an Anandtech staff member respond to a post about SiS it was ALLWAYS negative (complaining about no unified drivers, no beta drivers for an unavailable beta OS, lack of features/overclocking potential).
Even in your writings there is a slight negative undertone about SIS (I don’t think you do this on purpose). Maybe I am wrong but when you refer to SiS boards as “bottom-dwelling” or that “no mainstream boards ever appeared” I have to wonder if the boards are really that bad or are your expectations a little too high for “mainstream”.
I hope you have better luck with SiS in the future, and thanks again for the insight to SiS business practices.
Wesley Fink - Thursday, September 15, 2005 - link
Very few major sites review SiS, ULi, Jetway and Asrock. We do at AnandTech, because we believe a good product deserves a revew, no matter who makes it. So the issue is definitely not bias on the part of AnandTech.SiS fuled one of the best Intel boards I have ever owned. Anand also loved this board made by Gigabyte. However, since the SINXP we have all been patiently waiting for an encore from SiS. Sometimes the story is not as simple as the "big manufacturers won't use my chipset".
nserra - Wednesday, September 14, 2005 - link
Well then you must consider in moving into my country ;)Because here you have them all from the top quality to the lower ones.
All these SIS and Uli boards are seeling here as these nvidia SLI,...
And i agree that you point out the missing "great" features, but don't forget that one board costs 150€ and the other 70€ or less.
Rza79 - Tuesday, September 13, 2005 - link
In the motherboard 'Basic Features' table you state that it's using the 965L while the picture shows it's using the 965.You are testing a reference board!!! Since when does a reference board overclock well or even overclock at all. You shouldn't even test it because a reference board isn't optimised for that.
While Uli's reference board did overclock well, it doesn't mean all of them should.
The ASrock 939S56-M can do 285Mhz FSB.
Can you tell me one chipset that has Firewire by default?
Your conclusion is like making a conclusion for a retail board.
Firewire, overclocking, ... on a reference board??? That's why it's called a reference board. It's a reference for manufacturers ... not overclockers.
Even by today standards the 965 southbridge is ok.
Exept for not having HDA because i consider sata300 a checklist feature.