Now THIS is a motherboard worth waiting!

[Aenslead]

http://www.myhard.com/image20010518/61654.jpg

Finnaly something worthwile! Even though it is powered by a VIA chipset, it seems quite promissing and the board itself looks SWEET. Clean layout, typical of ASUS, and wacky collors, WHICH is something new to ASUS.

No nForce 3 pics on the wild yet.

 

[merlion]

Linkified

 

[Peter]

I like Tyan's latest better ... it uses all the building blocks from AMD's chipset, thus offering an 8x AGP Pro 110 slot, four PCI-X slots coming from two PCI-X busses, the Gbit Ethernet attached to one of those busses, and one legacy PCI bus and slot.

This ASUS board here once again has just plain normal AGP, and a single wimpy 32-bit 33 MHz bus that is already overly loaded by the FireWire, Gbit Ethernet, and Promise SATA chips.

The former is a serious workstation board, the latter is a gamer's toy at best. Lack of I/O bandwidth is what makes it so.

 

[Peter]

Oh, the link ... here

 

[Aenslead]

WELL! That Tyan board is obviously gorgeous, but... its a high-end workstation motherboard, wich is not the case of the board I have shown above...

It is VERY nice looking, being a Socket 754 and Athlon 64-single channel board.

 

[Peter]

Right ... yet still, the lack of fast PCI(-X) busses is a serious lack today. Where shall the 300 MB/s from the SATA RAID controller, the 50 MB/s from the FireWire and the 250 MB/s from the Gbit Ethernet get their bandwidth from - when they're all on one single 133 MB/s legacy PCI bus?

 

[Corporate Thug]

VIA chipset and a passive northbridge cooler, prolly not the choice of the overclocker

 

[Peter]

Who cares? Bus overclocking is an entirely moot point with an Athlon-64, since the relevant bits are all in the CPU itself. There is no front side bus at all, and the RAM controller is in the processor. Chipsets in this architecture are all about I/O, no more no less.

 

[Mem]

VIA chipset and a passive northbridge cooler, prolly not the choice of the overclocker

Some of the best overclocking boards are based on VIA chipsets,as to the passive heatsink,well nothing is stopping the overclocker putting a fan on .

Now THIS is a motherboard worth waiting!

Hmmm no,it doesn`t have PCI Express(a must in my books for the future) and Asus is overated in my books.

This ASUS board here once again has just plain normal AGP, and a single wimpy 32-bit 33 MHz bus that is already overly loaded by the FireWire, Gbit Ethernet, and Promise SATA chips.

The former is a serious workstation board, the latter is a gamer's toy at best. Lack of I/O bandwidth is what makes it so.

Hehe very true .

 

[Tab]

Where are my PCI-X Slots?

 

[Peter]

On the TYAN board.

 

[tcsenter]

Originally posted by: Peter
Right ... yet still, the lack of fast PCI(-X) busses is a serious lack today. Where shall the 300 MB/s from the SATA RAID controller, the 50 MB/s from the FireWire and the 250 MB/s from the Gbit Ethernet get their bandwidth from - when they're all on one single 133 MB/s legacy PCI bus?

Provided that you will simultaneously utilize at 100% all those I/O devices for any substantial length of time. Although I can fantasize about such a situation, I'll never see it.

For many people, the Asus board will suffice.

 

[Peter]

Well, even if you just use two fast drives in RAID0 on the Promise controller, you'll hit the ceiling already. Then try to use a multichannel sound card at the same time ...

32-bit 33 MHz PCI is what ISA was ten years ago. A bottleneck, but a loved one because so many popular devices haven't moved to more modern busses even though they could make good use of doing so.

 

[Tab]

Originally posted by: tcsenter

Originally posted by: Peter
Right ... yet still, the lack of fast PCI(-X) busses is a serious lack today. Where shall the 300 MB/s from the SATA RAID controller, the 50 MB/s from the FireWire and the 250 MB/s from the Gbit Ethernet get their bandwidth from - when they're all on one single 133 MB/s legacy PCI bus?

Provided that you will simultaneously utilize at 100% all those I/O devices for any substantial length of time. Although I can fantasize about such a situation, I'll never see it.

For many people, the Asus board will suffice.

Reguardless, they shouldn't be on 133mb/s bus.

 

[Peter D]

im personally waiting to see if Abit is going to make an Athlon64/Opteron board, and what kinda features they will throw on that are typical of Abit and their synonymity with overclocking...

 

[Aenslead]

Well, ladies, just to get your panties a little bit more wet, here we have Epox's offer.-

http://www.theinquirer.net/?article=11226

Not the best layout I have seen... but its Epox... and STILL using the VIA chipset. And... no, no PCI-X yet. We should NOT expect that untill the 939-pin A64 appears. Also, DO NOTICE that the Tyan board requires a PCI-X Tunnel, and appart from AMD, I have not seen nor VIA nor nVidia say anything about a similar tunnel.

VIA where nice overclockers. nVidia has better overclocking chipsets simply thanks to the PIC/AGP lock. If VIA did the same, they'd get somewhere.

edit: Even if it IS a VIA chipset, it has a nice southbridge, with Serial ATA, Vinyl Audio, and 8 USB 2.0 + 6 PCI. Not bad. Feature wise, it is quite atractive. Lets not forget LAN as well.

 

[Peter]

Well, it's all HyperTransport. It's an open standard - board designers are perfectly free to use AMD's or TI's HT-to-PCIX tunnel chip along with anyone else's AGP tunnel/southbridge combo.

It's only that PCIX is not an easy technology. 133 MHz bus speed with slots is not something you get done on a low cost 4- or 6-layer design. And the tunnel chips aren't exactly cheap either.

On the bottom line, it's about the bottom line. Want real features? Spend real money.

And once again: Get that bus speed overclocking thing out of your head. Athlon architecture does EVERYTHING relevant inside the CPU itself. There is no front side bus, the RAM controller is in the CPU.
Chipsets and their usable frequency combinations are an absolute non-issue here. There are no choices. It's HyperTransport at fixed frequencies in and out, and expansion busses that have their own standardized frequencies anyhow.

 

[Aenslead]

Hmm... that's strange, Peter.

Then how come there are some reports out there about overclocking the A64?

Heck, even Opteron has been overclocked by AMDzone. com via de FSB, shown in the ASUS SK8N BIOS?

Care to explain?

 

[Peter]

You can overclock it - but that's all between the clock synthesizer chip and the CPU. Of course there is a feed of manier frequencies into the CPU, and they can be tuned out of their intended range if the clock synthesizer supports that. But whoever says they did an FSB overclock don't really know what they're talking about. There is no front side bus here.

THE CHIPSET IS NOT INVOLVED IN ANY OF THIS.

Learn something new every day

 

[Aenslead]

The fact that memory controller is on-chip does not mean that the motherboard does not have access to override frequencies, and whether you understand it or not, it still has a BUS it runs on. A BUS is any electrical path made to connect two or more devices.

Anyway... today's the day you could learn:
http://www.amdzone.com/articleview.cfm?articleid=1329&page=7

 

[Peter]

I happen to understand these things ... I'm a BIOS engineer you know. And I tell you, one last time, that in AMD64 architecture, the chipset is not involved in any of the overclocking activities. As I said already further above, it's all between the clock synthesizer and the CPU. Rise the CPU clocks, fine, but the chipset just hangs off a HyperTransport I/O link and is nothing to do with these clocks.