Sign in with
Sign up | Sign in
Your question

Hot hot bug bug again with the new AMD/via chipse

Last response: in CPUs
Share
July 12, 2001 3:37:24 AM

I don't know if you like the title.

<font color=blue>Your mouse moved. WINDOWS NT must restart for changes to take affect. Restart Now?[OK]</font color=blue>
Anonymous
a b à CPUs
July 12, 2001 3:47:09 AM

Actually, all you need to read is the title.

It's talking about incompatibilities between the AMD 762 north bridge and the Via VT82C686B south bridge.

The 762 is the north bridge for the dual athlon platform (760MP). There is only one MB available that uses the 762 northbridge, and it doesn't use the Via south bridge (it uses the AMD-766).

So, this issue is only of concern to motherboard manufactures (& Via) as they work on more dual athlon products.


<i>The early bird gets the worm...
but the second mouse gets the cheese!</i>
Related resources
July 12, 2001 3:33:01 PM

Yeah, but thanks for crying wolf, juin.
If you ever have anything that actually matters to say, don't bother.

-----------------
Whoever thinks up a good sig for me gets a prize :wink:
July 12, 2001 9:58:34 PM

Amd incompatible with VIA LOL! this is worth 10. :-)

"AMD/VIA...you <i>still</i> are the weakest link, good bye!"
July 12, 2001 10:09:43 PM

Wow the best of both worlds. A P4 killer.
July 12, 2001 10:11:28 PM

Noko whats your choice:

P4 Northwood / ATi Chipset vs. Athlon 4 / Nforce

Both will support dual channel DDR memory.

Nice Intel and AMD users get a Cookie.... :smile: Yummy :smile:
July 12, 2001 10:21:00 PM

Depends if the Nforce supports the Radeon 2 or not. From what I can tell the Athlon 4 and Nforce will be a better performing setup. I am eagerly awaiting for the Northwood and hope Intel opens up the throttle wide open and let the chip perform. As in 512k of cache and max out what ever the chip can perform in speed (ghz). Intel needs to skip this lame 100mhz speed increments on ghz chips trying to bleed dry everyone possible. If Intel just let loose and blow away AMD processors by cracking up the speed in the Northwood then that maybe my next chipset. Not sure if I will buy a ATI chipset because I don't like integrated video solutions unless it is hydravision capable with the AGP card then that would be cool. As for DDR or RDRam I could care less, I will buy the one that performs the best for the money.
Anonymous
a b à CPUs
July 12, 2001 10:37:21 PM

>Amd incompatible with VIA LOL! this is worth 10. :-)

*shrug*

An incompatibility that no end user (except maybe some beta testers) has ever seen. It's a developmental bug, caught before the release (unlike some other cases I don't need to mention).

Sorry, no points for that one.



<i>The early bird gets the worm...
but the second mouse gets the cheese!</i>
July 12, 2001 10:45:12 PM

Well i heard 8x AGP on the ati chipset

Nice Intel and AMD users get a Cookie.... :smile: Yummy :smile:
July 12, 2001 10:52:53 PM

8x AGP? 4x isn't even maxed out yet. Can we really use 8x?
I mean sure, it's good for the future, but will anything use it when the MB comes out?

-----------------
Whoever thinks up a good sig for me gets a prize :wink:
July 12, 2001 10:55:08 PM

NV25 - Geforce 4
NV30 - Geforce 5 ???
----
R300 - Radeon 3

Nice Intel and AMD users get a Cookie.... :smile: Yummy :smile:
July 12, 2001 10:56:17 PM

Bear in mind the current problems intel is having with the p3 XEON when adding L2 cache.

A little bit of knowledge is a dangerous thing!
July 12, 2001 11:03:45 PM

The P4 was design to have 512kb cache.

Nice Intel and AMD users get a Cookie.... :smile: Yummy :smile:
July 13, 2001 12:58:02 AM

point being?

A little bit of knowledge is a dangerous thing!
July 13, 2001 1:02:45 AM

Well they strip it down to 256kb and it sucks. 512kb should give it the muscles it wants.

Nice Intel and AMD users get a Cookie.... :smile: Yummy :smile:
July 13, 2001 3:29:45 AM

I am not planning on any major system upgrade until late next year. I may end up with a Radeon 2, maybe not. For sure a Radeon 3 or GF4(5)? if they are really good. So then I will seriously consider my options. Intel should really lay down the speed in the Northwood, that is what the P4 is designed to do instead of playing tag and teeter-tooter with the T-Birds. Prices for Intel cpu's for the performance is just plain ridicules at the moment. Maybe Intel is worried about being labelled as a monopoly, I don't know but the last two years is not Intels best.

<P ID="edit"><FONT SIZE=-1><EM>Edited by noko on 07/12/01 11:32 PM.</EM></FONT></P>
July 15, 2001 4:06:58 AM

Maybe Intel is worried about being labelled as a monopoly, I don't know but the last two years is not Intels best.

I already been do.I dont read all on this but i think intel have decide to negocite.That 2 year old i think.
July 15, 2001 4:12:26 AM

K7D Master

The only mobo that will be release with "normal" price.AMD 761 have the same prob.The probleme is cause be the implantation.
Anonymous
a b à CPUs
July 15, 2001 1:36:33 PM

>K7D Master
I didn't find any specs to confirm whether they use the problem combination or not. I didn't look to hard either, no time right now.

But the board is not released yet (at least it's not on pricewatch). If it is released with those bugs intact, then you can bitch about it all you want.


<i>The early bird gets the worm...
but the second mouse gets the cheese!</i>
July 15, 2001 5:42:28 PM

I dug a little deeper. The problem exists in the AMD761 chipset, but AMD outlines a component-level fix for it--the addition of a few resistors on the motherboard takes care of the problem. The fact that I have this combination, have the IO-APIC enabled in the kernel, and still have a rock-solid system puts the nail in the coffin. Zero points for this one...

As for the K7D, I don't even care until the board's on PriceWatch. I probably won't care then either, cos I'd go for the Tyan anyways, just to get away from VIA crap.

There is also an obscure issue with the 761+VIA 686B+registered memory+ACPI STR for which AMD has no fix. I suppose that's worth maybe a half point.

Now here's the ultimate question: would you consider this an AMD incompatibility or a VIA incompatibility?

Kelledin

"/join #hackerz. See the Web. DoS interesting people."
Anonymous
a b à CPUs
July 15, 2001 6:11:41 PM

"Now here's the ultimate question: would you consider this an AMD incompatibility or a VIA incompatibility?"

Well, since the northbridge is designed for AMD's product by AMD, and the south bridge is designed by a third party for the product, I would guess that it's *another* VIA issue.

Aklein

It's July. O the Joy of Summer!
July 15, 2001 8:27:05 PM

just to get away from VIA crap.

WOW i never expect that.So you admit that there a prob and it been like that all the time with VIA.ALI suck too less that VIA but suck.SIS the only left SIS 735 is still to be prove.

<P ID="edit"><FONT SIZE=-1><EM>Edited by juin on 07/15/01 04:37 PM.</EM></FONT></P>
July 15, 2001 9:37:31 PM

What is your point? Intel has it share of bad technology as in 820 MTH, P3 1.13 orginal. So what?
July 16, 2001 3:14:32 AM

I don't believe VIA's poor quality was ever in dispute. VIA sucks badly. But hey, VIA makes motherboards for Intel CPUs as well--AMD isn't the only one saddled with VIA.

It's not like there aren't good chipsets for AMD though. I'd have to hear a lot of bad things about the ALi chipset to write it off as crap. It's about 2-3% behind the AMD760 chipset performance-wise but offers good stability. Plus, there <i>are</i> AMD761-based motherboards without the VIA southbridge.

Kelledin

"/join #hackerz. See the Web. DoS interesting people."
Anonymous
a b à CPUs
July 16, 2001 3:48:22 PM

>The problem exists in the AMD761 chipset, but AMD outlines
>a component-level fix for it--the addition of a few
>resistors on the motherboard takes care of the problem.

Actually, the link that juin gave is about the AMD762 chipset, which is used in the 760MP board(s).

They specifically note that the fix for the 761 bug doesn't work for the 762 (page 6 of the pdf).


<i>The early bird gets the worm...
but the second mouse gets the cheese!</i>
July 17, 2001 3:47:00 AM

Ja, I noticed; that bit was part in reply to juin's claim that it exists on the AMD761 as well. I noticed the lack of a fix for a 762...I'm just committed to not caring :wink: .

Kelledin

"/join #hackerz. See the Web. DoS interesting people."
July 17, 2001 3:55:19 AM

I just hope there will be 1 benchmark for stability like sysmark.

Stability we can only argue with personal impression.In general intel have the reputation to be more stable.The only reference is reputation and what website say.""web site say"" who cant Toms say is rock solid he have test it for 3 day with all new driver and just got format HD just install windows.The condition was perfect for stability.The same thing goes for brookdale how can you know???

<P ID="edit"><FONT SIZE=-1><EM>Edited by juin on 07/17/01 00:05 AM.</EM></FONT></P>
July 17, 2001 4:03:34 AM

The KT133a looks to be a very stable VIA chipset for T-Birds. If implemented right it functions virtually flawless. I never had a problem with the 686b southbridge myself and since then the problem has been fixed. I am sure a 1.4ghz T-Bird on a KT133a would beat out a 1.4ghz P4 Northwood on a Brooksdale chipset both using PC133 ram on virtually every benchmark. I also have a KX133 via chipset which wasn't the best setup if the bios was incorrectly set, in that case VIA wasn't as dependable but then again it was VIA's first attempt with the Athlon, not bad at all. I am looking forward to the new SIS735, ALI revised Magik and Nvidia Nforce chipsets, they can only increase the performance of the T-Bird and Palimino further. KT266???????

<P ID="edit"><FONT SIZE=-1><EM>Edited by noko on 07/17/01 00:04 AM.</EM></FONT></P>
July 17, 2001 4:28:19 AM

mmm hard to get reliable benchmarks of true processor performance when one is using different mobo's AND ram.
how does the brockdale 1.4 compare to the sdram athie 1.4?

"i love the smell of Overclocking in the morning!" Says my Hamster.
July 17, 2001 9:06:24 AM

<b><A HREF="http://www.aceshardware.com/board/general/read.php?mess..." target="_new">http://www.aceshardware.com/board/general/read.php?mess...;/A>

Folks, that 762+686B thing is not a chip bug in itself, it's simply an interoperational issue (on non-PCI sideband signals) of two chips that weren't designed to work together. VIA chipsets use a different protocol between north and south for buffer flush requests than does the AMD 76x series - and that screws the interrupt acknowledge handling up.

While in single-CPU configuration this can be worked around to work perfectly fine (see AMD appnote on 761+686B), it can't be done for dual-CPU configurations.

Again, what 686B does is perfectly OK - when paired with a VIA north bridge. The other way round, AMD 766 south bridge with any dual-processor-capable VIA north, wouldn't work either. Whose fault? Certainly neither AMD's nor VIA's - blame those mainboard designers that tried to combine stuff that doesn't belong together. They got away with it with 751 and 761 north bridges, but with 762, it plain and simply doesn't work. That's what the 762+686B appnote describes.

With AMD 762 north, what we really want is not a special version of VIA's 686B to be used, but a 66-MHz-capable PCI south bridge that allows the 762's PCI bus to run at that speed, and that has another 33 MHz PCI bus hanging out of its butt. AMD 768 I'd say. That'll make the "760MPX" chipset then. I guess VIA might not really be interested in making a revised 686 since the AMD768 shouldn't be too far away - and by the time it arrives, noone will care much about building Bi-Athlon boards with no 66 MHz PCI bus.</b>


:smile: Good or Bad have no meaning at all, depends on what your point of view is.
Anonymous
a b à CPUs
July 17, 2001 1:55:16 PM

>Ja, I noticed; that bit was part in reply to juin's claim
>that it exists on the AMD761 as well.

Oops, sorry. I missed that bit. Read first, then post :smile:

<i>The early bird gets the worm...
but the second mouse gets the cheese!</i>
Anonymous
a b à CPUs
July 17, 2001 2:04:15 PM

>Stability we can only argue with personal impression.

No, stability can be tested & measured. It just isn't ussually done because, by definition, it takes a long time. If you're not up for months straight at full load, you can't talk about stability. Most hardware sites aren't willing to devote those kind of resources to an issue that most people don't care about.

And the vast majority of people don't give a damn about it. This is clearly evidenced by the fact that the vast majority of people use Microsoft operating systems, which makes most hardware stability issues trivial in comparison.


<i>The early bird gets the worm...
but the second mouse gets the cheese!</i>
!