Dothan overclocking thread

[Salkcin]

I'm still a Pentium 4 user because I enjoy the goods of Hyper-Threading. I know you people need the fastest rigs to pump out the highest numbers, but how is the Dothan for normal or workstation use? I tryed a Athlon64 some time ago and I was sitting playing Unreal Tournement 2004 and then I paused the game and went to Windows (alt+tab) to check something on the internet. The internet browser was serveral seconds to open! A Pentium 4 with Hyper-Threading would have opened that browser at once... only thing that kills Hyper-Threading is some heavy loading on the PCI bus - or should I say it kills the use of the computer.

So my question is if the Dothan is only good for benchmarking or also for ordinary use? I imagine it like a Athlon64....

Anyways. Anyone who can confirm the multiplier is unlocked? a multiplier of 6x (or just 10x) would enable a lot greater FSB than default. Max FSB on the DFI is 250MHz - 10x250=2500MHz with DDR500 CAS2-2-2-5?. But it will for sure require some mods like vAGP, vDimm, vCore mod etc.....

[RaptorRaider]

I like Dothans.

900 Mhz with stock cooling and voltage is great. :banana4:

The results seem to be decent: DFI and this review.
All I need to do now is find a nice mATX case.

mATX is backward compatible with ATX, right?

[WesM63]

mATX is backward compatible with ATX, right?

Yes, any mATX board will work in any ATX case.

[Mathias-K]

Depends on where you look...and I have a hard time believeing that a 2.4ghz Winchester vs 2ghz Dothan....Dothan just doesn't seem like the winner
jjcom

None of us are talking about 2GHz vs 2.4GHz. Man, read the thread

Macci told us, that in a specific 2001SE-test, a 2400MHz Dorthan slapped a 2770MHz FX and a 2800MHz winnie.

[macci]

Did the Vcore mod and upped it up to 1.48V so far. Still running with the tiny default cooler

Dothan 1.5G @ 2571MHz, Superpi1M => 30sec
Maxed out the mobo FSB (so far) w/ this one. Chip would have gone higher.

Dothan 1.7G @ 2608MHz, SuperPi 1M => 30sec
PiFast33 => 20.63
Hexus PiFast => 47.19
3DMark2001SE @2515MHz, X800XT PE stock => 28261

The 1.5G did 5Mhz better than the 1.7G w/ default volts/cooler. OVP kicks in at ~1.5V Vcore. Higher Vcore would still give a nice boost for those clocks. But I think that a chip w/ higher multiplier is needed if I decide to up the Vcore and add some better cooling.
At the moment the 1.7G is running at 2.2G 1.25V w/ passive stock heatsink (case fans take care of airflow) - max temp so far +56C (after looping the new crytek demo).

[jjcom]

yeah there, but over all I have a hard time believeing that. 400mhz is alot and both of these chips have short very efficient pipelines. I don't know, maybe this thing is just confusing me...
jjcom

[macci]

Anyways. Anyone who can confirm the multiplier is unlocked? a multiplier of 6x (or just 10x) would enable a lot greater FSB than default. Max FSB on the DFI is 250MHz - 10x250=2500MHz with DDR500 CAS2-2-2-5?. But it will for sure require some mods like vAGP, vDimm, vCore mod etc.....

Multiplier is unlocked from 6x to the max multi of given CPU (15x for 1.5GHz for instance). Multiplier can be adjusted from Windows, same goes to Vcore.
Lower multi would indeed allow higher FSB - if only these boards would do it
MAx I can run past SuperPi 1M test is only 172FSB so far.

[Salkcin]

French review with voltmods: http://www.x86-secret.com/articles/c.../dfi855-10.htm
Translate it to your prefered language: http://babelfish.altavista.com/babelfish/tr

It's just... they use a 2100MHz and looking at their 2,8GHz clock with mods it's running 16x175! - getting past 180Mhz looks like quite a challenge. How much voltage is responsible to apply to the chipset?

[charlie]

reminds me of the good 'ole 845PE days.... by the end of that chipset, we were hitting 200+

C

[death metal]

Maxed out the mobo FSB (so far) w/ this one. Chip would have gone higher.

macci, what was the max voltage handled by the NB chipset? can it be cooled even more to allow for higher FSB? it'd be nice to really push that baby to its full potential. could it be mobos electronic circuitry (i.e. add more components for stability) or pure chipset limitation (i.e. lack of voltage, or NB tops out already)?

[macci]

at this point its very difficult to say whats holding the FSB down. I've only tried adding a fan on top of the NB heatsink, putting some AS5 between the sink and chipset and upping the voltage from 1.55V to 1.85V so far (and I'm not sure if all that helped anything at all since I didn't test the max FSB in 100% stock form - it did a minium of 163FSB 100% stock thou).
TAM from Japan reached FSB of 184 w/ his LN2 cooled 2.1G chip. Might mean that the chipset just needs some better cooling. There are quite a few empty capacitor slots near the NB too. So adding some extra capacitance might help too.

Another thing is that after 133FSB I need to increase AGP/PCI clocks otherwise it will slow down or totally crash - just like i875/865 does after ~303FSB w/ fixed 66/33 mode.

[jjcom]

yeah, maybe the voltage isn't stable enough for the NB to like it....mmm..could just be luck of the draw again. Sometimes you just cant get any further no matter what...lets not hope that to be the case since this is a 250 dollar + board
good luck!

jjcom

[charlie]

So here's the big question.....
Do ANY of the BIG mobo makers realize that a 855 full size ATX mobo designed as a POWER system would be a big HIT?

I think they do.... but Intel came and said "hey, don't steal the thunder from our P4 line, we'll let you put 855 on a micro-atx board for silent desktops, but don't make our 855 a P4 killer!

C

[charlie]

at this point its very difficult to say whats holding the FSB down. I've only tried adding a fan on top of the NB heatsink, putting some AS5 between the sink and chipset and upping the voltage from 1.55V to 1.85V so far (and I'm not sure if all that helped anything at all since I didn't test the max FSB in 100% stock form - it did a minium of 163FSB 100% stock thou).
TAM from Japan reached FSB of 184 w/ his LN2 cooled 2.1G chip. Might mean that the chipset just needs some better cooling. There are quite a few empty capacitor slots near the NB too. So adding some extra capacitance might help too.

Another thing is that after 133FSB I need to increase AGP/PCI clocks otherwise it will slow down or totally crash - just like i875/865 does after ~303FSB w/ fixed 66/33 mode.

Macci,
Sounds like OC'in with the old 3.06HT....but would it be fair to say that at equal CPU speed, the Dothan is faster than P4. A64 and A64 FX?
Also, it seems that the reviews suggest that a 1.6 or 1.7 Dothan will run 2.3gHz with no effort and very little cooling, at 2.3gHz how would this near-silent desktop compare to the big 3.6-3.8 P4's and the A64's???

Oh, and what divider raios are available?


C

[Naughtyboy]

OK guys...my cpu is on its way, a 725 it is..I´ll probably have it on Monday or Tuesday next week....however I´m gonna have to wait for the DFI another week tops
Loooking for some new ram to go along with it...with some luck I´ll get some BH-5 or EB sticks.

Really excited about it.

Macci: Did you tired it with DDR Booster yet..??

[death metal]

at this point its very difficult to say whats holding the FSB down. I've only tried adding a fan on top of the NB heatsink, putting some AS5 between the sink and chipset and upping the voltage from 1.55V to 1.85V so far (and I'm not sure if all that helped anything at all since I didn't test the max FSB in 100% stock form - it did a minium of 163FSB 100% stock thou).
TAM from Japan reached FSB of 184 w/ his LN2 cooled 2.1G chip. Might mean that the chipset just needs some better cooling. There are quite a few empty capacitor slots near the NB too. So adding some extra capacitance might help too.

Another thing is that after 133FSB I need to increase AGP/PCI clocks otherwise it will slow down or totally crash - just like i875/865 does after ~303FSB w/ fixed 66/33 mode.

wow, sounds like some more jobs are in the works =)...please do keep us posted with the project (if you have and once you have the time ), i am in the verge of getting one too, if only to hit 160FSB on stock and 24/7 HTPC rig.....but of course, if there are more mod guides to make this baby run more, i'm very very much eager to see those....

[n00b1337]

Dothans look pimp but how well would it play counter-strike:source considering the DFI only has a 4x AGP?

[Naughtyboy]

Dothans look pimp but how well would it play counter-strike:source considering the DFI only has a 4x AGP?

Mabye you´ll fin some answers HERE

[charlie]

great review link....
Well seems as though a Dothan at 1.7 beats 2.2gHz Newcastle in 3D gaming....
In fact, 2.0gHz Dothan beats 2.2gHz 3500+/939/DC handily in DOOM 3.
And finally, overclocked 2.3gHz Dothan almost beats 2.6gHz FX55 in DOOM 3.

[jjcom]

Dothans look pimp but how well would it play counter-strike:source considering the DFI only has a 4x AGP?

Don't worry about it only having AGP 4x. You'll notice little to no difference when comparing 8x to 4x. 2x causes problems(preformance not as good)..but 4x is fine

jjcom

[WesM63]

Mabye you´ll fin some answers HERE

Thats a great review, but right now my main issue is cost, given the board+cpu cost for stock vs. stock, the A64 wins hands down (~$300 vs $500+). Unless i can get my hands on a Donthan cpu for cheap.

[charlie]

Thats a great review, but right now my main issue is cost, given the board+cpu cost for stock vs. stock, the A64 wins hands down (~$300 vs $500+). Unless i can get my hands on a Donthan cpu for cheap.

True dat!
My 24/7 rig is a 2800+ @ $125 and a Gigabyte K8NS @ $75.....Dothan can't touch that!!!

[macci]

1.6G dothan is a ~200€ product even here in Finland, the Aopen mobo is in the 230€ range. Sure its more than a 2800+ and k8ns but then again I think that it does offer a bit more too
btw, just ordered the AOpen board since it has a lot more space in the CPU socket area PLUS it has connectors for front panel USB

charlie, a chipset limitation (amount of PCI slots) might be a reason why we see only mATX mobos. Not sure about this thou.

"Oh, and what divider raios are available?"

DFI bios offers 1:1, 4:3 and 5:3 for RAM divider, AGP/PCI 1/3, 1/4, 1/5 and 1/6 are listed in the BIOS but I'm not sure if they all work (boot at 150FSB w/ 1/5 for instance wont load OS). AGP/PCI fixed option is available too.

"Macci: Did you tired it with DDR Booster yet..??"

Nope, I just did the easy FB->GND resistor mod, also the mobo has only two ram slots and I'm going to use 2x512MB so there is no space for booster. Also I think I read that the booster wont work w/ the DFI mobo. Vdimm is regulated from 5V rail btw..
So far I've hit 253MHz 2-2-2-5 w/ Mushkin PC3500 256MB stick. Does that already at around 3.35V and increaseing Vdimm didn't seem to improve things at all - so I'm guessing a chipset limitation again..(these sticks pass 3DMark on S939 board in DC mode at 270+).

Might go and pick up a 2G dothan from the stilt today
20x multi should be enough (20x170 ).

[death metal]

macci, thanks a lot for sharing....am holding my purchase until i find out which of the boards OC better and more stable too....

20x multi should be enough (20x170 ).

am drooling....

[sky]

can anyone with the aopen board state the data macci just posted for the dfi, dividers, agp/pci dividers or lock (if available and working).
hm the aopen 855gme is 223€ over here, the 725 (1.6ghz) is at 186€, totalling ~400€. sounds like a nice toy, but i guess i want to see some more benchies.. this would be the uebersweet "in a shoebox" rig to kill any other hardware at a lan .

has any other big mobo-company announced a board for these tiny powerhouses? would be great to see. high fsb would definately see me going intel (dothan) once again.