Gigabyte GA-7DX+

[JEWilson]

Re: Gigabyte GA-7DX+

Ok

I purchased a mobile Athlon XP with a Barton Core
Model # - AXMA2600FKT4C

Installed in the mobo and posts as a Mobile Athlon 800MHz

The mobo auto detects the part at six times multiplier.
The mobo has DIP switches for setting the FSB and the multiplier.
see attached pdf

The mobo is set a FSB 133MHz and the multiplier set at auto.

Problem - When set as above, bios id's mobile athlon 800MHz
If I manually set the jumper to x12 - the bios id's an Athlon XP 2100+
Despite repeated attempts, the multiplier cannot synthesis a
15x mult, that is, the actual multiplier for the Barton XP-M.

As position 5 is used for auto, the DIP switch has a further position
6 which appears to be unused. Position 1 is used for a 0.5 mult
increase.

So... positions 2, 3 and 4 determine the multiplier and being 8
possible settings are available, the max. mulitplier on this mobo
is 12 (plus 0.5 of course for position 1 on the DIP switch).
This accounts for a mult of 5 to 12 inclusive.

It appears the auto setting cannot ID the barton XP-M part.
So... it appears, in order to get the XP-M to be correctly identified,
I will require to add a microcode update to the BIOS to enable
the BIOS to do this correctly and automatically

Any comments to add?

Thanks for your consideration

[Scenic]

Re: Gigabyte GA-7DX+

is there any option in the bios to set the FSB? or only by jumper/switch ?

had those problems on another board.
but instead of modifying the bios i just set the FSB higher and chose a multiplier that matches the original cpu speed with the FSB..

example:
if the Athlon XP-M has 2GHz as real speed (not the + rating):
166MHz FSB and multiplier 12 -> 2GHz
the BIOS will still detect the CPU wrong, but it will work at the standard clock speed..

problem is that you'll need matching RAM for that (166MHz FSB -> DDR333 or DDR266 in async. mode to the FSB.. but this will reduce overall performance)

BTW: i have lots of socket A boards here... so if u need another board

edit: i had a board here that used a multiplier solution for multipliers higher than 12x

for example.. if you set the multiplier to.. um.. 8x it would use a multiplier of 14x if a CPU with this multiplier is used...

dunno how to explain this in english..

maybe there's something similar on your board...!?

edit2: lowering the multiplier will work on most boards allthough the multiplier is locked by the CPU as this lock is only for setting the multiplier higher... not lower..

[JEWilson]

Re: Gigabyte GA-7DX+

Hi

Can only set the FSB by DIP switch
It appears only allows max. of 12.5 per the diagram

No options in the BIOS to set the multiplier
Can set the FSB but this mobo is max. 133MHz so cannot
overclock. Don't want to do that anyway!

I acknowledge your advice in respect of weird multiplier settings.
I am beginning to wonder if switch position 6 (not shown
in the diagram), present in the mobo has something to do
with a mult over 12.5.

This is a new mobo but has remained unused for some time.
I recently recapped it - is working perfectly ok.

Thanks

[kikkoman]

Re: Gigabyte GA-7DX+

have you tried changing the multi via software?
SoftFSB was a good tool. they stopped updating it a while ago, but you may be lucky. depends on the mobo.

CPUMSR (http://www.cpuheat.wz.cz/html/CPUMSR_main.htm) works fine, too (I think it's a bit more universal).

[JEWilson]

Re: Gigabyte GA-7DX+

I should add...

This being a mobile Barton, it starts a 6x multiplier at POST.
I believe alternatively, the issue may be Windows 2000 lacks the
AMD PowerNow! driver and dashboard.

I will try setting the SW multiplier back to auto, install
the driver and determine if I can run the PC as always on
with no battery saving measures.

I will let you know how I progress.

[JEWilson]

Re: Gigabyte GA-7DX+

Update

The PowerNow! driver require to access to a PowerNow! compliant
BIOS as with a laptop. As this mobo does not have this feature,
the driver will not install. So a solution via this route is not available.

Also, the CPUMSR softFSB util came with a driver for WinXP, installed
ok (apparently) but, I am using Win2K SP4.
When I tried to raise the multiplier and apply it, the system locked
up....
Reset button and 2 hours of boot time chkdsk's and a boot time registry
chk and I'm back up again.

Any other suggestions?

[zandrax]

Re: Gigabyte GA-7DX+

No luck: another owner tried to use a mobile Athlon and it booted always at lowest multiplier available (6x).
My suggestion are updating bios to revision F7 and change the multiplier into OS: beware that most desktop boards aren't as flexible as mobile one and, since AMD specs didn't require a compatibility test, most mobos crash when changing multiplier. Moreover CPUs are unstable at some multiplier, so you have to test all possible combinations then exclude the dangerous ones.
My favourite tool is RMClock, but CpuMSR is fine too (being Athlon oriented, it may be even better); SoftFSB is unsuitable because it changes the FSB, not the multiplier.

Zandrax

[JEWilson]

Re: Gigabyte GA-7DX+

Thanks

I already have the latest BIOS at F8a
I also possess the last BIOS at F7 for this mobo

Q - If I manually set the DIP SW to a mult of 12 and
then I try to raise the mult to 15 - this all that is req'd
and AMD CPUInfo does state the max speed is 2.0GHz,
will any soft changes to the clock gen interfere with
the manual setting of 12?

Normally, the mobo is set to auto detect and with
the barton XP-M this defaults to a mult of 6 and a
speed of 800MHz, would it be better to leave at auto
and then attempt to raise the multiplier?

Only a guess all the same.

Even so, correctly the Barton XP-M is supposed to
default to a x6 mult at POST
see further http://fab51.com/cpu/guide/cpu_guide1-e.html

I'm going to contact Gigabyte on this one as well to establish
if a x15 mult or better can be established with the DIP SW.
Would appear odd it it were not to as Tbred on auto setting
can be detected up to a multiplier of 16x @ XP 2600+/133 or 2.1GHz

Thanks

[zandrax]

Re: Gigabyte GA-7DX+

Q - If I manually set the DIP SW to a mult of 12 and
then I try to raise the mult to 15 - this all that is req'd
and AMD CPUInfo does state the max speed is 2.0GHz,
will any soft changes to the clock gen interfere with
the manual setting of 12?

Normally, the mobo is set to auto detect and with
the barton XP-M this defaults to a mult of 6 and a
speed of 800MHz, would it be better to leave at auto
and then attempt to raise the multiplier?

If I remember correctly, the multiplier shouldn't be a big issue since the cpu multiplies fsb internally (fixed multiplier for desktop Athlons, range for mobile ones).
Honestly I prefer to stictly follow specs on boot, so I'd set the autodetection and let the Athlon boot at 800 MHz (6x133): it shouldn't be too slow (my father owns a notebook with a Mobile Barton 2800+ and it boots at 800 MHz too).

I'm going to contact Gigabyte on this one as well to establish
if a x15 mult or better can be established with the DIP SW.
Would appear odd it it were not to as Tbred on auto setting
can be detected up to a multiplier of 16x @ XP 2600+/133 or 2.1GHz

Because it recognizes Thoroughbred' cpuid and it implies that switch settings are useful only for boot (internal multiplier).
Since you can't use Powernow driver (bios lack the voltage-multiplier table), then use cpumsr: this should work by directly programming both chipset and cpu.

Zandrax

[JEWilson]

Re: Gigabyte GA-7DX+

I've set the DIP SW back to auto. Starts at x6 mult as expected.

I've tried CrystalCPUID and it will raise the multiplier to x15.
There is a feature to run it at boot and implement this change.

You suggest CPUMSR, is this a better option to follow for Athlons?

Please advise.

Thanks.

[zandrax]

Re: Gigabyte GA-7DX+

I've tried CrystalCPUID and it will raise the multiplier to x15.
There is a feature to run it at boot and implement this change.

You suggest CPUMSR, is this a better option to follow for Athlons?

Years ago CPUMSR was the first (and only) solution available to manage Athlon's multipliers,but years passed and its developement stopped: if CrystalCPUID works for you, then keep it because it's more recent and (perhaps) has more features.

Zandrax

[JEWilson]

Re: Gigabyte GA-7DX+

Y - agreed

Tried CPUMSR and killed the PC again.
Two hours later... got the PC back on line.

Thanks

[JEWilson]

Re: Gigabyte GA-7DX+

Using CrystalCPUID, after applying a mult of x15 by software,
I have a very slight overclock condition as with;

1. FSB 134.02 MHz as opposed to 133
2. DDR FSB 268.05 MHz as opposed to 266.67
3. Core speed 2010.4 MHz as opposed to 2000

CrystalCPUID reports an overclock of 0.52%

Should this be a concern?

[kikkoman]

Re: Gigabyte GA-7DX+

not really. usually the XP-Ms can take a lot of abuse.
if the RAM timings are within specs, it shouldn't be a problem.

have you tried how high it goes before it becomes unstable?
good to hear that it's finally working. i take it that you didn't update the microcode?

[JEWilson]

Re: Gigabyte GA-7DX+

The BIOS F8a correctly identifies the new XP-M Barton as
a Mobile Athlon at POST

F7 does not, it details Unnkown CPU at POST

They both support the CPU even so.

[JEWilson]

Re: Gigabyte GA-7DX+

As to the matter of the o/c at 0.52%

This system runs a matched pair of Infineon PC2100 ECC Reg
Part # H7S72D64500GR-7-B
The datasheet can be found at


The datasheet details SPD programming at CL2, that is, as with
these DIMMs an fck max. of 133MHz.

Thus, I have had to manually set the CAS Latency to 2.5 in the
BIOS. The datasheet details CL2.5 programming to have an
fck max. of 143MHz.

[JEWilson]

Re: Gigabyte GA-7DX+

I've changed the O/P side of the VRM back to;
10 of 1200uF 6.3V Panasonic FM

The problems have disappeared.

If I want to put polymers in this mobo, will have to consider
installing higher capacitance values.

[gdement]

Re: Gigabyte GA-7DX+

Using CrystalCPUID, after applying a mult of x15 by software,
I have a very slight overclock condition as with;

1. FSB 134.02 MHz as opposed to 133
2. DDR FSB 268.05 MHz as opposed to 266.67
3. Core speed 2010.4 MHz as opposed to 2000

CrystalCPUID reports an overclock of 0.52%

Should this be a concern?

No. Every motherboard is a little off the mark, and there is a reasonable tolerance built in to the specs. If you want to confirm that everything is stable, I'd set the FSB to 138-140MHz (3-5% over) and run stress tests at that speed. If it holds up, then you'll have nothing to worry about at stock speeds over the long term.

The datasheet details SPD programming at CL2, that is, as with
these DIMMs an fck max. of 133MHz.

Thus, I have had to manually set the CAS Latency to 2.5 in the
BIOS. The datasheet details CL2.5 programming to have an
fck max. of 143MHz.

I doubt that's really necessary - it's only running very slightly above the 133MHz (I think actually 133.33) rating. See if it passes memtest at ~ 138-140MHz CL2. If it does, you have nothing to worry about.