Slashdot Mirror
Socket-A Chipset Roundup
EconolineCrush writes: "The Tech Report has a review up of VIA's new KT333 Socket A chipset. Though it's really a review of the KT333, a total of seven different chipsets from VIA, SiS, AMD, and NVIDIA are compared to determine the uniprocessor Socket A performance king. This is definitely worth checking out if you're in the market for an AMD platform, or are curious to see how your current chipset stacks up against the latest and greatest."
Tom's Hardware has a good article up.
"A total of 26 various benchmark tests clearly shows that the VIA KT333 chipset the best and most capable chipset for AMD CPUs. With only a few exceptions, not even the Nvidia nForce with its expensive dual-channel technology (DDR266) can put up a real fight against the newcomer KT333. With the launch of the KT333, the KT266A will become a thing of the past - you simply won't want to miss out on all the new features such as ATA/133, USB 2.0 or DDR333 support."
So does Anandtech: here.
"When the KT266A was launched it completely blew us away; the performance of the chipset was spectacular and it was clear that it would quickly become a top pick for all Athlon owners. The KT333 doesn't have nearly as great of an impact but the reasons behind that are understandable; both new features supported by the chipset, DDR333 and Ultra ATA 133 aren't features that will result in tangible improvements in performance today. Instead the KT333 is more of a technology enabling platform for VIA. The chipset will not cost any more to manufacture than the KT266A and thus motherboards won't increase in price. While DDR333 SDRAM isn't officially available today (the specification isn't complete), when it is first made available it will carry a price premium over DDR266 SDRAM."
There's a bunch of other good reviews of the set in all its forms and splendor.
Digit-Life
HardOCP
AnandTech
AMDDb
Via Hardware
</karmawhoring>
This tagline is umop apisdn.
They're to control timing so that instructions leaving certain areas of the CPU do not hit the bus before others. My friend who use to map PCB boards for a living calls them "speed bumps".
"...and generally behaved in a manner one can only describe as despicable." - February 27 2001, Michael Sims
I would definitely recommend going with the K7S5A. I've had one running on a Debian file server/net gateway machine on my home network for the last 4 months. I haven't had a single lockup, crash or hang in all that time. I would definitely recommend it, esp. for a machine with a budget ($95 CDN - probably ~$65 US with onboard LAN and audio), or even for a performance system.
My other sig is funny!
I've yet to see a crash on my AMD760 ASUS board. It's middle of the pack in performance (read: more than you'll ever need) and rock solid. VIA has a bad rep, no doubt, but this box just works. Maybe you're running a microsoft OS? Or using lots of 3rd-party binary-only drivers including one that's known to trigger AGP-related crashes? My system is Microsoft, nvidia, and binary-only-driver free; it seems to help a lot.
Clock frequency isn't the issue. Rise and fall times are what drives a board into high speed design rules.
Clock frequency obviously sets an UPPER limit on the rise and fall times - the total rise and fall must be equal to or less than the one cycle. But most logic signals have edges that are much faster than that.
Mr. Fourier tells us that the bandwidth of a perfect square wave is infinite. A 333 MHz clock with 333 MHz bandwidth is a sine wave. Extending it out to the next harmonic found in a square wave requires 1 GHz bandwidth, and the signal starts to look vaguely squarish. 3 GHz gets you something that anyone would recognize as a square wave, but with lots of ripple.
The time/distance relationship to pay attention to is not one cycle, but the rise or fall time. If the travel time is more than the rise or fall, then the trace is a transmission line. If the signal gets to the far end before the rise or fall finish, then you can ignore transmission line effects.
Good point about the MPX chipset. I've been waiting patiently for the Asus A7M266-D and now that it's out, the USB is busted! Plus the boards I've seen have built-in audio, which I'd actually rather *not* have. And what's the deal with that little daughter-card looking thing where the power comes in? Looks kludgey. But the PCI USB 2.0 card is supposed to work in the 64-bit slot, no? So maybe 'losing' a PCI slot isn't that bad. I'm assuming the two 64-bit PCI slots are on a separate bus from the 32-bit slots... would this mean less IRQ conflicts?
So the Tyan Tiger MPX was starting to look good, but then I hear the USB card Tyan is shipping is NOT a USB 2.0 card. So now it seems like you're really losing a PCI slot, since there don't seem to be that many 64-bit PCI cards out there now (or you buy a new 64-bit USB card and junk the free one). Built-in LAN sounds great, and the board looks a lot cleaner, especially the standard hard drive power connector by the ATX power connector for more current. Tyan probably learned a lot from their first SMP Athlon boards.
Anyone else have any experience with SMP Athlons? Oh, and did you use Athlon MP's or did you cheat and use the XP's?