A64 3000+ oc to 3700+

  • Thread starter Thread starter rstlne
  • Start date Start date
R

rstlne

http://www.hardocp.com/article.html?art=NjI5LDE=

Now.. I am sure some of you may have seen it.. BUT
273fsb (how's he get that??)..

I am not up real big on this a64 stuff yet (and I dont see many questions
like this) so is it that the fsb of these new boards will go higher to
support later chips? ..
or am I just missing something..
I seem to have missed the 233, 266, and 300 fsb days being reach'd..
So how high will these boards go fsb wise.. OR since he was just pushing the
high fsb only to the CPU then is that why it could go so high..

Just looking for feedback is all :) ..
Re
RstLne
 
rstlne said:
http://www.hardocp.com/article.html?art=NjI5LDE=

Now.. I am sure some of you may have seen it.. BUT
273fsb (how's he get that??)..

I am not up real big on this a64 stuff yet (and I dont see many questions
like this) so is it that the fsb of these new boards will go higher to
support later chips? ..
or am I just missing something..
I seem to have missed the 233, 266, and 300 fsb days being reach'd..
So how high will these boards go fsb wise.. OR since he was just pushing
the high fsb only to the CPU then is that why it could go so high..

Just looking for feedback is all :) ..
Re
RstLne
Click to expand...

"With that said, see what can be accomplished when scaling the multiplier
down and the FSB up while the PCI/AGP bus is locked to spec speeds."

Ben
 
Ben Pope said:
"With that said, see what can be accomplished when scaling the multiplier
down and the FSB up while the PCI/AGP bus is locked to spec speeds."

Ben
--
Click to expand...

So your saying that there is unlimited room for clockspeed with the pci/agp
bus's lock'd and the ram dividers keeping the speed sane?
could I have say a 600mhz fsb on this board.. I think that's where I was
going with it ;)..
 
So your saying that there is unlimited room for clockspeed with the pci/agp
bus's lock'd and the ram dividers keeping the speed sane?
Click to expand...

I'm not sure what he was trying to say, but that wasn't it.:-)
could I have say a 600mhz fsb on this board.. I think that's where I was
going with it ;)..
Click to expand...

No. I'm not sure what the new chipsets will do, but I'm sure it's not
going to go that high or anywhere close to it. 300 maybe, but I'd have
my doubts about that. And there's other factors to consider too.
 
rstlne said:
http://www.hardocp.com/article.html?art=NjI5LDE=

Now.. I am sure some of you may have seen it.. BUT
273fsb (how's he get that??)..
Click to expand...

I'll try again... My previous post was a bit brief :-p
I am not up real big on this a64 stuff yet (and I dont see many questions
like this) so is it that the fsb of these new boards will go higher to
support later chips? ..
or am I just missing something..
Click to expand...

Well, there are various factors that affect it, but essentially the new
process (with SOI) has more headroom than the old one, so the new processors
can be clocked higher.

Another one is that the "FSB" or more correctly, hypertransport link was
designed from the ground up to operate at frequencies higher than 200MHz. I
think the version 1 spec has clock rates to 1.4GHz, but I'm not sure if
thats practical.
I seem to have missed the 233, 266, and 300 fsb days being reach'd..
Click to expand...

I think this confused me a bit last time I replied. I don't know of any x86
style processors that operate with a FSB above 200Mhz, officially. Many
Intels can be clocked to 250MHz, apparently, and the Bartons to 233 or so.
So how high will these boards go fsb wise.. OR since he was just pushing
the high fsb only to the CPU then is that why it could go so high..
Click to expand...

Well, the link is still there between Northbridge and CPU, but I think it's
designed for higher frequencies (being hypertransport) than the old FSB.
Also, the memory is not clocked that high, it was at around 200MHz, which is
a shame.

Ben
 
He set the host clock to 273MHz from the default of 200. With older
chipsets, this wasn't possible because the PCI/agp buses weren't locked at
33/66MHz, so 273 on the old chipset would have given him pci/agp clock
rates of 45.5/91, at which the components couldn't work. Also earlier
chipsets were only spec'd for 800MHz on the HTT link (600 for nf3-150).
The newer chipsets have the new spec of 1000MHz. He also lowered the
multiplier from 10 to 9 and threw a lot of voltage at it. The board he
used had a lot of seperate control settings that might not be availible on
other boards.

I think that answers the original question, except it's not really called
a FSB any longer, but host clock.. And to be honest, I'm not sure if the
MHz numbers used to define the HTT link or really clock rates. I think
they are data rates with a MHz thrown behind them since they talk of being
2 way. I haven't looked into too much, except that with some/most boards,
changing the host clock also changes the HTT link, and you might notice
that it reports the HTT clock rate at 273MHz. Corrections welcome.:-)
 
Back
Top