Gigabyte NF4 review

http://www.ocworkbench.com/gigabyte
GA-K8nxp-9

On Mon, 15 Nov 2004 23:13:01 +0800, df wrote:

> http://www.ocworkbench.com/gigabyte
> GA-K8nxp-9

Quote from article;

Overclocking wise, this board doesn't seem to provide enough as the FSB is
limited to 250MHz. Although it provides voltage controls for Vcore, Vdimm,
VAGP and Vhtt, it lacks one of the most important feature, multiplier
control. We tried to use software to lower down the multiplier, it works
but after reboot, the system won't POST if we increase the FSB above
220MHz despite running the memory in async mode e.g cpu/mem = 200/166. We
hope the production unit would improve in this aspect. *latest : according
to Gigabyte, a new bios update will include new multiplier option and
improve overclocking.

It's not the memory thats stopping the overclock, it's the HT. The Ht
clock is derived from the system bus setting. For all practical purposes,
this is the FSB, just with another name (HT). So with a 5x (1000) HT at
220MHz system bus, it's running the HT at 1100. If they would have
lowered the Ht multiplier to 4x (800), then they should have been able to
raise the system bus (FSB) all the way to 250MHz (assuming the cpu would
run at that speed) and that would run HT at the rated 1000.

This applies to all K8 boards. One reason i didn't have a problem clocking
my old board without a PCI lock to 233MHz system bus was because I lowered
HT before raising the bus to 233.

Another thing. Don't confuse the 1000MHz you see for Ht as the clock
speed. It isn't.

http://www.hypertransport.org/consortium/cons_faqs.cfm

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm

Will there be an AGP version of this board??

"Wes Newell" <> wrote in message
newsan.2004.11.15.19.28.25.452518@TAKEOUTverizon .net...
> On Mon, 15 Nov 2004 23:13:01 +0800, df wrote:
>
>> http://www.ocworkbench.com/gigabyte
>> GA-K8nxp-9
>
> Quote from article;
>
> Overclocking wise, this board doesn't seem to provide enough as the FSB is
> limited to 250MHz. Although it provides voltage controls for Vcore, Vdimm,
> VAGP and Vhtt, it lacks one of the most important feature, multiplier
> control. We tried to use software to lower down the multiplier, it works
> but after reboot, the system won't POST if we increase the FSB above
> 220MHz despite running the memory in async mode e.g cpu/mem = 200/166. We
> hope the production unit would improve in this aspect. *latest : according
> to Gigabyte, a new bios update will include new multiplier option and
> improve overclocking.
>
> It's not the memory thats stopping the overclock, it's the HT. The Ht
> clock is derived from the system bus setting. For all practical purposes,
> this is the FSB, just with another name (HT). So with a 5x (1000) HT at
> 220MHz system bus, it's running the HT at 1100. If they would have
> lowered the Ht multiplier to 4x (800), then they should have been able to
> raise the system bus (FSB) all the way to 250MHz (assuming the cpu would
> run at that speed) and that would run HT at the rated 1000.
>
> This applies to all K8 boards. One reason i didn't have a problem clocking
> my old board without a PCI lock to 233MHz system bus was because I lowered
> HT before raising the bus to 233.
>
> Another thing. Don't confuse the 1000MHz you see for Ht as the clock
> speed. It isn't.
>
> http://www.hypertransport.org/consortium/cons_faqs.cfm
>
> --
> Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
> http://mysite.verizon.net/res0exft/cpu.htm

i think it won't be
pci-e will come ture