AMD 400 FSB

[HardWareBoss]

Looks like AMD is going to launch 400 FSB soon. XP 3200 around the bend.

 

[Spitfire_x86]

Old News

<b><A HREF="http://geocities.com/spitfire_x86" target="_new"> My Rig</A></b>

 

[HardWareBoss]

Just trying to get a little input from the AMD crowd.

 

[sabbath1]

Yeah, should be released in May. Sadly, they would have to clock it real fast to be able to keep up with Intels upcoming 800 FSB CPU.

 

[Mephistopheles]

not sad at all... just means the newest processors will be fast as hell! And the way I see it, that´s good news. But is it true, the 400Mhz FSB? Last I saw was that they were keeping a 333Mhz FSB on the 3200+...

 

[sabbath1]

Check: http://www.amdzone.com

Looks like Abit, for instance has some confirmation of a 400 Mhz FSB, though I haven't really seen any confirmation myself, but since a mainboard manufacturer says so, I suppose it's true.

 

[hineigger]

My k7nl2 is running at 200fsb.... If it matters.... Only negative side is my athlon 2400 is not made for that, so it needs huge cooling. Willl eventually put fan on northbridge's heatsink.

 

[HardWareBoss]

Ya I read the same piece. Thats why I started this post. No word from AMD.

 

[sabbath1]

My bet is that AMD secretly announced their 400 MHz FSB products under the webcast they had a few days ago. The webcast for those who missed it, can't be viewed for another 3 weeks though.

 

[Mephistopheles]

I haven´t seen any verifiable claims to a 400Mhz FSB capable processor on <A HREF="http://www.amdzone.com" target="_new">http://www.amdzone.com</A>. What I <i>have</i> seen is 400Mhz FSB capable motherboards. That´s actually nothing new. We´ve had motherboards that support x86-64 for quite a while now.

Anyway, isn´t this amdzone site just <i>slightly</i> biased or is it just me?... Not in this case, but overall?

And talking about FSB increases, I am <i>far</i> more interested in the 533Mhz->800Mhz on Canterwood (that´s a <i>+50%</i> increase) than in 333Mhz->400Mhz on Barton CPUs (that´s a +20% increase). The Barton will indeed have to be clocked higher than 2.3Ghz if it´s not to fizzle against the 3.2Ghz, 800Mhz HT Northwood... This is going to be an interesting month!

 

[bikeman]

I dunno a thing about this 200 MHz FSB for Barton fuzz, but hey, Mephisto, aren't you a little overexcited about this i865/i875-thing? Ok, it will be a step forward, and changes are big that at last the i850E will be outperformed in every aspect, but the way you describe it, it looks as if the 50% FSB increase will give a performance boost of about the same proportion ... I'm sorry to tell you, that won't happen. My guess, you ask? At most a 10% improvement over the i850E, on average ...

Greetz,
Bikeman

<i>Then again, that's just my opinion</i>

 

[juin]

I guess this is not a bad guess.

[-peep-] french

 

[Mephistopheles]

Actually, I´d be a complete fool if I was expecting a 50% increase in performance. I´m not, so...

What I <i>am</i> expecting is something in the lines of 3-5% due to PAT (these are official numbers), some 2-5% or so due to the dual-channel DDR400 alone (over i850E), which also sounds completely reasonable from SiS chipsets benchmarks with DDR400, bearing in mind that SiS´ chipsets perform a tad worse than Intel´s and don´t run synchronously, and another 8% or so from the FSB increase, which is also a valid - and actually conservative - figure that comes from <A HREF="http://www.anandtech.com/chipsets/showdoc.html?i=1806&p=11" target="_new">overclocking boards to 800Mhz FSB</A> (also considering that these boards weren´t designed for that, so while the average on that test was like 6-7%, I expect some 8%). So I´m expecting about a 13-18% increase on Canterwood and an overall increase of 10-13% on Springdale. I don´t think that´s unreasonable to expect.

And about the 333Mhz->400Mhz FSB transition, working with DDR400 dual channel... well, I don´t expect more than 10% over current DDR400 dual channel configurations. Not at all. Tell me if my math is wrong here...

 

[sabbath1]

You have to check the link to Abits website, which can be found at the bottom of the news page, or at the latest archives.

Amdzone is *extremely* biased. A very good site with professional people, but they're very biased.

For me, the both FSB increases are equally interesting in a way, I understand why the Intel increase is more interesting for you, and while I do agree on that point, AMDs increase is equally interesting and important, for the sake of competition.

 

[Mephistopheles]

For me, the both FSB increases are equally interesting in a way, I understand why the Intel increase is more interesting for you, and while I do agree on that point, AMDs increase is equally interesting and important, for the sake of competition.

Absolutely, I see what you mean. If it weren´t for competition´s sake, I think Intel wouldn´t go 800Mhz FSB (remember, it´s one of their defense lines against x86-64); neither would AMD go 400Mhz FSB, I guess... That´s good, we customers get faster computers.

 

[AfroGeek1]

I'm not sure about other SIS chipsets. But, I have had pretty good luck with the SINXP (SIS655). So far I have gotten my OCZ memory(which gigabyte says they board will not support) to kick out DDR467@3-6-3-3-3 and at DDR350 I got 2-6-3-3-2. I recall that the all other dual channel intel boards only kick out a DDR266 reading. I saw the review this chipset was given and I'm not sure if it was entirely accurate. It said that the Gigabyte could not run DDR400 dimms at 333. I have not had a problem doing that. In fact I was able to attain a better CL and CAS setting at those speeds.

My question is this. What released P4 motherboard with DDR performs better than the SIS655s?

 

[Mephistopheles]

The SiS655 is very interesting indeed, for its technical features. It´s the only chipset that supports dual-channel DDR333 or DDR400 for the P4 platform, for the moment. What is interesting is that Granite Bay usually performs better than SiS655 when they´re both using DDR266... Not by much, of course, but by a small notch. Intel has supposedly optimised the memory controller for synchronous operation. Maybe I´m being optimistic, but I do expect their memory controller on the DDR400-800Mhz FSB Northbridge to perform better than i850E by as much as a few percent (2~5% or so).

But much of what I´ve been saying is just wild guessing, guys, so don´t take me too seriously. It´s really not worth speculating this much. We´ll see the numbers next week, anyway.

 

[sabbath1]

Bottomline: Every AMD fanboy in the world will fear the 800 mhz FSB P4.

 

[HardWareBoss]

Its about time we get faster computers. Heck my old AMD 500 MHZ CPU I bought way back in Jan 2000. At the end of the year we hit 1000 MHZ Now its April 2003 an we have 3.06 Thought we would have been at 4 -5 Gig by now.

 

[Teq]

Anybody know if this is 100mhz x 4 or is it 200mhz x 2 ?



--->It ain't better if it don't work<---

 

[phial]

200x2 is the next step in the way they are going


dunno tho, just logic

blah.. siggys suck

 

[slvr_phoenix]

I hate to say it, but I think that AMD <i>will</i> release 2x200MHz FSB CPUs. They'll do so <i>not</i> to compete with Intel's performance, but to compete with product numbering. AMD will just use the FSB advancement to raise their product numbers without actually raising the MHz of the CPUs.

That's my take on what'll happen.

<font color=blue><pre>If you don't give me accurate and complete system specs
then I can't give you an accurate and complete answer.</pre><p></font color=blue>

 

[Twitch]

I'm thinking 2.2 ghz on 400 FSB = 3200 + in AMD's collective mind. We'll see, I guess. THG mentioned seeing a 3200+ on display at a trade show, but they said it was on a 333 FSB. Wouldn't a Barton have to be running at 2.33 ghz to qualify as a 3200+ (relative to the 3000+ that is?)

I'll be surprised if Barton can reach 2.33 in sufficient quantities to mass-produce. AMD wouldn't release a 2250 on 166 FSB and call it a 3200+ would they? Would they dare?

2.2 on a 200 FSB sounds logical, though I seriously doubt it will even come close to a 3.2 gig P4 800 FSB. In fact, a 2.2 on a 400 FSB would probably be a more realistic 3000+. Speculating that the 800 FSB will yield a 7-10% (or so) benefit for P4, it seems like a 400 FSB Barton would have to run at 2.4 ghz to realistically be called a 3200+. Just my $.02.



<-----Insert witty sig line here.


<P ID="edit"><FONT SIZE=-1><EM>Edited by Twitch on 04/08/03 10:08 PM.</EM></FONT></P>

 

[HardWareBoss]

That sounds about right with AMD.

 

[sabbath1]

A Barton should be at 2.4 GHz 400 MHz FSB or above to stay even close to that monstrous 800 Mhz FSB P4. In fact, a overclock test (I think it was here on THG) of a P4 2.26 to 800 MHz FSB provided superior performance over the stock P4 at 3.06. No, it was not a typo. Judging from that test, the 800 Mhz FSB P4 looks like to be performing with nearly 1 GHz better performance than the 533 MHz FSB P4 would at the same speed. Therefore, AMD would need 2.4, no 2.5 Ghz or above to remain competitive. With 400 MHz FSB. This raise could be enough, because of AMDs more IPC. It could. But it would be a tough thing to do.

The chance for a real speed bump could be possible though, I know what my 3000+ is capable of, a few reviews around the net shows that it's capable of near 2.5 Ghz. That's with watercooling though, if I remember correctly.
Then if it's capable of touching the 3.2 GHz 800 MHz FSB P4 is another question.