SiS Announces 748 Chipset

[UtwigMU]

Speaking of chipsets, did you check the KT400A reviews/previews. it comes close to nForce2.

[Nuno]

VIA sucks. That's common knowledge. However they do deliver regularly the fastest chipsets for AMD

Sure. I remember how fun it was to have the fastest chipset at the time (KT133A) and the headaches it gave to me. And that was on 2 boards, Asus and Abit.

[Chrono_Wanderer]

Originally posted by Taz
My Iwill XP333-R, with it's ALi Magik 1 chipset, is working just fine, probably the most stable board I've used and performance is pretty good too

Gotta to agree with Nuno. My Asus A7A266 (ALi MiGiK1) AGP slot was screwed up... can't hardware accelerate any graphics card, not even 2D! Sometimes it doens't detech HDD on the onboard IDE channel. My friend's A7A266 (got it at a completely different store) also has the IDE problem.

Wait... Asus and AMD don't match

But I think it's ALi fault.

Edit: Hey look. I just noticed I have 777 numbers of post beta go to the casino right now

[Taz]

Originally posted by Chrono_Wanderer
But I think it's ALi fault.

None of the three Iwill's I've used have exhibited those problems, it could be because the XP333 uses the later C stepping of the chipset but I thought the only difference between it and the A and B steppings was a performance increase.

[GNEP]

My ECS K7AMA (ALi Magik 1) was OK but never great. Until it died on me. Stable-ish, not very fast, and pretty cheap.

Much prefer both the ECS K7S5A and the MSI 745 Ultra running SiS.

[Nuno]

Ok just to say I´ve bought the XP2400+ now because I did got a nice deal on my XP1600+.
Yes, it isn´t as fast as if it were on a nforce2 board on dual DDR33. But it was the most cost effective hardware path at this time, and pretty much end of the road for the ECS K7S5A as the XP2600+ is the last 133 Mhz fsb XP.

The installation was trouble free. Install CPU, heatsink, boot up, cpu is recognised correctly, everything seems to be fine. Set the fsb to 143Mhz, (cpu 2180Mhz). I ran 3dmark01 and 03, a few benchs and let UT2003 demo on a botmatch benchmark for 4 hours (editing -seconds on the .bat file). Flawless. Seems preety stable to me. This Tbread-B core is really nice, it runs 5º cooler (same heatsink volcano7+) at 680Mhz more.

As expected it was just one more case of nice plug-and-play with this board. It really allowed me to extend a tad more my system useful life, that´s going on for a year and a quarter.

[Indiana]

Planet 3DNow! have a quite large database collecting data for user-satisfaction with specific mainboards.
The results there don't really speak in favor of SIS.

http://www.planet3dnow.de/artikel/diverses/zuverlq103/index.shtml

(in german, but the graphic should speak for itself)

The bad result for SIS based boards is of course more the fault of the manufacturers, but there we have the problem Kurt has earlier adressed: SIS is mostly used by cheap manufacturers like ECS/PC Chips - and even when a "quality brand" uses a SIS chipset, they place it in their budget / low-price area (as it seems unfortunately low quality-wise as well). Best example here is Asus with it's less the great SIS AMD mainboards.

Most people blaming VIA for making bad chipsets have theri experiences from the old buggy KT133 chipsets and never have seen a KT333, which is a fine chipset and never gave me any trouble in the EPoX 8K3A boards I used for several systems.

The NForce2 is fine as well, even speedier than the KT333, but I'd tend to say a slight bit more problematic as well.

[Taz]

VIA's reputation is based on more than just the problems with one particular chipset but those of several chipset. They have had a tendancy to cut corners or not implement things properly, so it's more from that attitude rather than a specific issue. Reputations can change though, SiS used to have a very poor one up until a couple of years ago, they really seem to have turned things around now and just need a few more of the big manufacturers behind them to become one of the best.

[Wombat]

Originally posted by Indiana
P
Most people blaming VIA for making bad chipsets have theri experiences from the old buggy KT133 chipsets and never have seen a KT333, which is a fine chipset and never gave me any trouble in the EPoX 8K3A boards I used for several systems.

I blame Via for the MVP chipsets, the KT133 chipsets, the 686 piece of crap, and having the worst drivers in the industry, even denying a problem existed as George Breese put out a patch to work around their arrogance.

Then there was the KT333, which was followed by the KT400, which brought NOTHING to the table, running DDR 200MHz RAM at lower performance levels than 166MHz.

Indiana, you might have mentioned one good product, but really, they aren't "turning around," they're continuing to produce shoddy products with the occasional exception.

[Indiana]

The 686B bug was a major fault and the even older VIA chipsets all were really crappy, yes. The VIA 4in1 drivers didn't really create problems here - I only left the IDE driver out as the one that comes with windows is faster.
But compared to e.g. NVidias 26MB (!!) NForce2 bloatware drivers, I'm not so sure, whose drivers are to be considered worse...


Still the KT333 was a good chipset and user satisfaction with KT266A and KT333 based boards seems to be very high according to the Planet3DNow database.

The KT400 is not that bad, either (apart from the AGP8x problems in the beginning) and seems to hit stable FSBs of 200MHz much more often than the NForce2. Of course you have to run the memory in synch with the FSB, but VIA is not alone here: just look at the Sis745s abysmal performance in async 133/166 mode or at the less than spectacular performance on NForce2 boards in asynchronous operation.

P.S.: Although that post might sound a bit negative for the NForce2, I have a NForce2 board myself at the time and am happy with it.
But apart from the slightly better speed and the additional onboard stuff there are no advantages to the previous KT333 which did work absolutely problem-free.

[The PIT]

Cougth.

Got a good board no problems bad board same old shit.

Via recyle == recycled bin.

George Beese sorted a load of Via problems out and to be honest probably saved Via lot of market share. At least they now reconise his efforts and he's no longer slagged of by Via as they have done in the past although Via followied is work very closely.

[Marshmallowman]

you need a good manfactuere to make a via board good...I liked my epox 8kta3+, but due to a via chipset bug it could not run XP processors.

sis 735 is so good that a cheap ass manufactuere like ecs can make a good board...quite happy with my temporary mobo k7s5a.It did have some problems which a couple of mods fixed.

I am pissed off with sis for release the 748 chipset, was going to buy a nforce 2 board...now I am going to have to stick with my old board until some decent sis 748 boards come out...,sis is good because they always seem to work in linux(without patches/drivers)

my next board has to have 200mhz support for barton and linux compatibilty

[Nowhere]

Yeah, I'm quite happy with my msi k7t turbo2 (non problematic for me basically)...but I've recommended ecs k7s5a...
But for me it's not a big difference - the same cpu upgrade possibilities, and even if I'd have ecs it would be coupled with sdram...

[Kurt]

Originally posted by Marshmallowman
you need a good manfactuere to make a via board good...I liked my epox 8kta3+, but due to a via chipset bug it could not run XP processors.

sis 735 is so good that a cheap ass manufactuere like ecs can make a good board...quite happy with my temporary mobo k7s5a.It did have some problems which a couple of mods fixed.

I am pissed off with sis for release the 748 chipset, was going to buy a nforce 2 board...now I am going to have to stick with my old board until some decent sis 748 boards come out...,sis is good because they always seem to work in linux(without patches/drivers)

my next board has to have 200mhz support for barton and linux compatibilty

I wouldn't wait for 200MHz FSB if I were you, AMD hasn't decided whether to produce a 200MHz part or not and if they do it will probably be their fastest part so you can expect a 600-700$ price point.

Why don't you go with MSI's 746FX board instead or ECS 746 (which doesn't officially support 166MHz FSB BTW)? Or Epox with nforce2 as they have the more likely chance to hit 200MHz FSB -in case you _might_ need it.

If you plan to wait for SIS 748, you might want to consider the A64 boards...