Dothan overclocking thread

@ ALL : Could be that you all are right, about that bios boot problem - MoBo boots (when Bios checking hardware for first few seconds on AUTO settings (bios boot_block defaults) & only than a bit later applies settings determined in Cmos; many boards arround have this issue as I see ... so previously set multi/Fsb/core_frequency & Vcore are not engaged instantaneusly first moment!

Guys, Abit & their Bios manufacturer made stuff to be safe for use for normal users (NOT OC-ers eXtremes), so do not think that there would be available some "enhanced" bios uppgrade for you from them! You need to find someone else really capable to hex_edit the bios code & prepare you a "hacked" Bios for your purposes to re-flash the MoBo resolving certain problems. IMHO this would be the best way.

The next "safest" way would be to try that "dirty Vid Pin wire trick" on P-M socket/Cpu to trick the MoBo of higher boot Vcore if works. If does not, the next step would be leaving that dirty wire mod trick engaged & again reflash Bios with the same Bios version. Could happen that then you will be having a default Vcore higher than before & could be raised in bios even more (maybe up to 2V-attention!-too much!). This trick worked many times in the past, maybe will work also here. Here you have an mine example of tipical FlashBoot diskette ready to be than custom modified if works for you (read_me inside,file): http://users.volja.net/jerman55/FlashBiosFD.zip

The third "safe" way would be that my mentioned Hardware way Fixing Vcore on the MoBo. But that also involves some reading & knowledge too: Vcore Mosfets on the MoBo are driven by an output of 8-bit input digital PWM chip there, which gets combinations on its state binary pins by another smaller digital chip, which gets serial data on MoBo´s i2c (SMBus) data link from Bios (that last chip on old boards with only jumpers or microSwitches for Vcore wasn´t there needed)! Find searching on Google datasheet for that mosfet driver chip on your MoBo, to know which its pins to solder (or terminate with pull_up/down resistors) to a 8-way microSwitch setup. (not sure if w/o need for cutting traces near that chip on MoBo will work, maybe will!). So instead of controled Vcore by Bios, you have a set of switches, which is IMHO better than have on MoBo totally fixed just single Vcore! So that microSwitchSet controls the input logic states of a Mosfet driver chip to procure "right" Vcore to the Cpu!

The forth "safe" mod would be that (added later in my last post on previous site) "PwrGood" trick; see an example picture on my site under comp/mods, how I did similar stuff for Tuallies; the wire is fixed with transparent sellotape & goes shortly into a hole of an orange isolating small tube used for isolation not to tuch internal contacts base (to prevent any electrical shorts-"Electronics runs on Smoke; if it comes out of it, its a BS !") :D :D :D ... You better practice modding/opening socket before on some dead slotket/socket/MoBo, to make it later "tiny", nice & clear & not "dirty"! Do mods on the table not already mounted inside PC!

The last one less safe is the first socket/2-pins mod I proposed (VssSense+VccSense), but you have to study yourself datasheets further before attempting that. I hope that you know, that all these my modding ideas can be tried at your own risk, nobody guaranties you success & many times things can go "south"; sh1t happens! :) But usually people with guts have at least some success (not always)!

Further ideas for better OC will follow maybe already today later !
--
Regards, Spajky, the "member of Silent computer owners Team" ... :)

Quote:

---

Originally Posted by STEvil

Now that we know what it is maybe the bios code could be extracted and injected into IC7's or other boards? Or vmods might be the easier way to go since it seems they might need one anyways? Nice work Spajky ;)

---

i tried that with ic7 .. injected all dothan microcodes from aopen 915 mobo (for some odd reason, dfi mobos don't have microcodes for c0 step).
then fired up and no POST whatsoever.. no beeps no nothing.
then i measured vcore from mosfets, got ~1v, maybe something below.. as my pm 730 has 0.988v "battery optimized mode voltage" so it must've been it.
then i tried to increase vcore all the way up to 1.5v but no go.
i don't have port80 card lying around so i really don't know what's going on..

btw, does mobo need a "confirmation" about vccp and/or gtlref, too?
with pent M and adapter, they are naturally too low or missing (because adapter generates 'em and mobo gets too low feedback voltage or none)

hmm... mby i'll try to tinker with it this weekend as i still haven't sold my ic7.
i kinda hate asus and ami bios, so i'm interested :)

Quote:

---

Originally Posted by Frozendeath

...
The question this brings up to me is:

When does the multiplier get applied to the cpu? If the set vcore does not kick in immediately, how long does the multi take to be put in?

For example, trying to boot at 9 * 265 is actually like 12 * 265 causing the initial load to be high enough to pull the 1.2v "vcoreboot" to a no post flag?
...

---

Yes, this makes so much sence to me, I bet it's the case here. I really hope there will soon be a cure for this through a new beta or hacked BIOS or something. My 730's max is at about 2600MHz and that would mean my max boot FSB is a bit under 220MHz.. Haven't tested this yet since I've been booting at 200 or 201 but I'll bet.

I had to add something to my previous post about hardware modding Vcore on MoBo: here is also a picture of a so called "DroopMod": http://malvescorner.net/images/p4cdroop.jpg . The guy did it analogically with a precision trimmer Pot across the signal line electronics partially bypassing it to MosFets directly from that big PWM driver chip (as I mentioned it in my previous post) sitting down left on the slide. But this IMHO does not resolve the Vcc,boot problem! It looks like just helps keeping Vcore stable under load, but IMHO that is NOT needed for Dothans, since even if max.OC-ed they do not consume more than 2/3 of current like P4 can & does!

So if droppin´ Vcore under heavy load exists, probably there is some not good Caps or MosFets overheating problem (maybe also PWM controler overheating too; you know what to do then :) )

@ caater : dothan microcodes injected into bios code alone does not help much (only post show propper Cpu) but after recognition of Cpu, other Bios routines coded inside bios do other job (involved with CPU recognition) & that has to be hacked too!

[Q]does mobo need a "confirmation" about vccp and/or gtlref, too?[/Q]

IMHO not necessary, but depends on MoBo design/Bios code; looks like on Asus boards it does. Thats why I proposed modding the adapter as an alternative ...

But guys, are you sure that you have a locked PCI/Agp clock inside bios enabled? If not, this could be also a problem with all kind of peripherrals like integrated audio, Lan, firewire, video card, IDE & especially Sata controller not tolerating too much out of specs clocks! Try to disable temporary ALL except FD controller & insert an old PCI graphic card just for test & start a PC with a Dos/W98 boot diskette! Does it boot with higher Fsb than before? If so, have enabled PCI/Agp lock in Bios but also enable one by one alone those Sata controllers. Can happen that that Bios lock affects only one Sata controller (could be same problem like with some NF4 AMD board as I recall) ! This is to check too!

If PC boots by only FD & IDE controller enabled on higher Fsb than before, take some old Ide spare drive formated Fat16/32 & insert it as an only HD enabled inside temporary, but before go here: http://users.volja.net/image/Files/ResQsys.htm & have some fun ... :D :D :D ...

Hasta La Vista, I´ll be back soon ... :)

I know that's a little off topic but had anyone here try the vdimm mod on P4P800 VM ?
THx

well, my experience with pm730 and adapter was very brief.
got my system working and tuned up, when just suddenly, p4c800e died. with smoke.
i hope that adapter didn't die :(

@ spajky: that's what i was afraid of.. only microcodes won't do..
all we need is one interested bios engineer :rolleyes:

btw, how to explain that some non-asus mobos post with pentium M?

Quote:

---

Originally Posted by caater

@ spajky: that's what i was afraid of.. only microcodes won't do..
all we need is one interested bios engineer :rolleyes:

---

if only somebody could contact tictac or merlin

*hint* *hint* ;)

@ Nazaar :

*hint* *hint* : ... or any other knowing someone "close" to /Bios engineer/ @ Phoenix, Award or Ami ...

@ caater : <how to explain that some non-asus mobos post with pentium M?>

A bit different MoBo/Bios design; no hardware checking routines @ post; barebone features w/o added newer peripherrals, who knows what else. The ECS one is so old that even is not supporting 200Fsb & runs even with regular SDRAM ! Since I did not read before whole thread (I am on DialUp!):

What are other non-Asus MoBo (except that mentioned) that work; do you have a list? Make some research for me please!
--
Regards, Spajky, :) passed months ago -
3rd_Ann.: "Tualatin OC-ed/BX-Slot1/inaudible setup!"

Quote:

---

Originally Posted by Spajky

What are other non-Asus MoBo (except that mentioned) that work; do you have a list? Make some research for me please!

---

here is a sticky thread for unsupported mobos:
http://www.xtremesystems.org/forums/...ad.php?t=65261

First test run of my Pentium M 740, Philliphines pack date 28/4/05
PiFast @ 2700mhz Stock cooled
SuperPi @ 2700mhz Stock cooled
3dmark @ 2700mhz very bad 6800LE :slapass:

Quote:

---

Originally Posted by Spajky

@ ALL : Could be that you all are right, about that bios boot problem - MoBo boots (when Bios checking hardware for first few seconds on AUTO settings (bios boot_block defaults) & only than a bit later applies settings determined in Cmos; many boards arround have this issue as I see ... so previously set multi/Fsb/core_frequency & Vcore are not engaged instantaneusly first moment!

Guys, Abit & their Bios manufacturer made stuff to be safe for use for normal users (NOT OC-ers eXtremes), so do not think that there would be available some "enhanced" bios uppgrade for you from them! You need to find someone else really capable to hex_edit the bios code & prepare you a "hacked" Bios for your purposes to re-flash the MoBo resolving certain problems. IMHO this would be the best way.

The next "safest" way would be to try that "dirty Vid Pin wire trick" on P-M socket/Cpu to trick the MoBo of higher boot Vcore if works. If does not, the next step would be leaving that dirty wire mod trick engaged & again reflash Bios with the same Bios version. Could happen that then you will be having a default Vcore higher than before & could be raised in bios even more (maybe up to 2V-attention!-too much!). This trick worked many times in the past, maybe will work also here. Here you have an mine example of tipical FlashBoot diskette ready to be than custom modified if works for you (read_me inside,file): http://users.volja.net/jerman55/FlashBiosFD.zip

The third "safe" way would be that my mentioned Hardware way Fixing Vcore on the MoBo. But that also involves some reading & knowledge too: Vcore Mosfets on the MoBo are driven by an output of 8-bit input digital PWM chip there, which gets combinations on its state binary pins by another smaller digital chip, which gets serial data on MoBo´s i2c (SMBus) data link from Bios (that last chip on old boards with only jumpers or microSwitches for Vcore wasn´t there needed)! Find searching on Google datasheet for that mosfet driver chip on your MoBo, to know which its pins to solder (or terminate with pull_up/down resistors) to a 8-way microSwitch setup. (not sure if w/o need for cutting traces near that chip on MoBo will work, maybe will!). So instead of controled Vcore by Bios, you have a set of switches, which is IMHO better than have on MoBo totally fixed just single Vcore! So that microSwitchSet controls the input logic states of a Mosfet driver chip to procure "right" Vcore to the Cpu!

The forth "safe" mod would be that (added later in my last post on previous site) "PwrGood" trick; see an example picture on my site under comp/mods, how I did similar stuff for Tuallies; the wire is fixed with transparent sellotape & goes shortly into a hole of an orange isolating small tube used for isolation not to tuch internal contacts base (to prevent any electrical shorts-"Electronics runs on Smoke; if it comes out of it, its a BS !") :D :D :D ... You better practice modding/opening socket before on some dead slotket/socket/MoBo, to make it later "tiny", nice & clear & not "dirty"! Do mods on the table not already mounted inside PC!

The last one less safe is the first socket/2-pins mod I proposed (VssSense+VccSense), but you have to study yourself datasheets further before attempting that. I hope that you know, that all these my modding ideas can be tried at your own risk, nobody guaranties you success & many times things can go "south"; sh1t happens! :) But usually people with guts have at least some success (not always)!

Further ideas for better OC will follow maybe already today later !
--
Regards, Spajky, the "member of Silent computer owners Team" ... :)

---

In your third option, this would involve a set of dipswitches to manipulate vid's(0-5 on and off) and set vcore directly to ic right? This could be done on the ct-479 itself locations u4 or u5, or on the p4c800-e at this location?

i tried to flash aopen 915 mobo bootblock to ic7 today.. wouldn't work :)
i suspected it, but i had to try :)

´ALO ´ALO,
Listen to me carefully, because I am gonna tell You only once! ... :D :D :D


@ ALL : Some additional ideas for better final OC : "MODding the Asus CT-479 ADAPTER!" further ... :) [final part]


[datasheet for Dothan & P4 required too]


If you watch your CT-479 from top view having power connector on your left /according to best pics I have not having the adapter myself/ :

- 3 chips on the right are IMO bidirectional cmos digital/analogic voltage clamp Bus switch ICs [Purpose: translating both way i/o Cpu signals to chipset & viceversa by adapting voltage levels; on P_M native boards not needed! Inividual quality of them determines adapters posibility of highest OCs; as an example: generic modded slotkets for Tualies OC-ed them better than PowerLeap slotket having this kind of ICs on-board]

- on left upper corner: IMO all-in-one switching (PWM) regulator/doubler IC [purpose: generating all other supply voltages needed for circuitry on board not using +5 or +12V, negative ones for example etc.] Certain SMD elements around it /caps & resistors/ determine its output voltages.

- big linear voltage regulator on the top of socket (probably using power feed_in +12V line, not sure) [IMO it procures 1,05 VccP voltage for the CPU i/o stages = 2nd core voltage - 1st time after 8y & P233mmx again true CPU differential/double powering!!!] Max current rating 2,5A !

- smaller linear voltage regulator (transistor?) taking previous source (VccP) to generate 0,63V i/o digital signals recognition state reference voltage (GTLREF). It is situated IMO in the left middle between power connector & that PWM chip.

- maybe there around is some other SMD transistor or lin.regulator (can´t find or see it-maybe in the middle between previous mentioned smaller voltage regulator & left upper socket corner!) to procure VccA voltages 1,5 or 1,8V for CPUs internal analog PLL clock generator (4 pins on CPU, see datasheet), but maybe is not even there & previous PWM chip takes care also for that!

- "missing" LowESR cap close upper to power connnector! Designed but not soldered maybe because of space problem with default cooler mounting? :) IMO serves filtering entering power line.

Linear voltage regulators (=LVR) have usually right pin Input, left pin Pilot-V & middle one (connected to its under metal base) is Output-lower/regulated_voltage. Mounted classic SMD transistor usually has terminals: single upper one Collector, right down one Emitter & Base /control/ pin on left. Both types can be positive (I+O or C-npn_Q terminals) on more positive potential) or negative (viceversa) as regulators mainly or even combined /positive regulator and pnp transistor instead & viceversa/.

Positive type "LVR" output voltage is determined/piloted usually by a middle point of a 2-SMD_resistors voltage divider going from its positive O terminal to Gnd (ground, mass, zero volt, black power conn.wire) & usually its minimal voltage is around 0,65V if its P terminal is directly [w/o resistor] tied to the Gnd (no need than for the other resistor!) if the LVR has fixed out voltage.

So to change its output voltage you just need to change those resistors value ratio, maybe by replacing them with other values or changing/diminishing/lowering their value by adding paralell to one of them much (10x?) bigger value resistor to lower a bit their value. If you add additional resistor to one going to Gnd, you will lower that output voltage; if you do that to the other one going to + , you will rise (overvolting!) output voltage. Easy ! :)

For maybe changing that PWM chip voltages (I can´t decipher markings on it), you are on your own by finding apropriate datasheet for it for further procedures if needed!


Before you try anything (sure like always on your own risk not to blame anyone if sh1t happens): do modding stuff on the "Lab Table-non conducting surface" with only adapter involved /no Cpu, no MoBo!/; when measuring voltages & results of mods you will have to power it up by inserting the power connector powered from some PSU on. Should not happen anything, myself I powered many times MoBo´s On w/o CPU or anything inserted in & had no problems later.

Hope I do not have to mention, you have to have some practice with soldering tiny & delicate stuff & have proper soldering equipment with magnifying glass & light, calm hands & digital multimeter (DVM) too. If not (or not having guts), let someone else capable & equiped do it 4 U instead. But if you guys do mods on MoBo, this should not be a problem ... :) But first carefully find & measure default voltages on CT-479 /VccP, VccA & GTLREF ones/ !!! And for checking & doing mods is nice to have a spare adapter (cheap), MoBo (not so cheap) & some older Celeron_M (much cheaper than P_M!) if something goes wrong. :)


What is to accomplish? Overvolting over defaults certain voltages just like Vcore on MoBo! :D for same purposes!

- VccP = Cpu signal i/o stages powering them higher to 1,15V (max.1,2V!-same as "Vcc,BOOT"-default Vcore boot voltage); same effect like upping Vcore, but it has NEVER to be higher than VCcc,BOOT one to prevent reverse currents from Vcore inside the Cpu; files data corruption may occur! [I experieced this with my Tuallie time ago!]

- GTLREF = signal reference point /threshold between signal binary states/ setting it higher to 0,7V; helps chipset having higher margins between 0&1 digital states - better Fsb OC at higher FSB.

Those two task accomplished, Cpu & chipset will be consuming a tiny bit more current & heat a bit more, but only few % ! Shouldn´t be a problem! Max rating: GTLREF 2/3 VccP; max VccP same as Vcore = 1,6V but only if that low 1,2Vcore Vcc,BOOT problem is resolved before somehow, but even with that fixed, I woudn´t go higher for safe than 1,5 V VccP & 1V for GTLREF maybe with additional cooling of those 2 voltage regulators, not to overheat & kill them or damage the MoBo´s chipset or CPU !!!

- 3rd check: VccA voltage for internal Cpu PLL generator: for P4 & P_M is/should be 1,8V, while for Celeron_M can be also 1,5V (or 1,8V) max 1,57 (or 1,89V). Maybe this is selectable somehow by that CT-479 Asus adapter when C_M is inserted (maybe even Bios way) & that could be one of the reasons (lower V option!), why people using it could not OC for example that CM_350J more than max.170-180Fsb/2,3GHz regardless Vcore & cooling !!! (could be that SMD transistor/if there in the middle between mentioned smaller GTLREF voltage regulator & left upper socket corner is responable for this or even that PWM chip, who knows; a need 4 voltage measuring to know that!). Max current rating to Cpu there = 120mA !

- next possible enhancement to the adapter: adding some min.1000µF/16V "missing" LowESR cap there; since can not be mounted normaly because of HSF, would have longer connecting wires (not best), to fix that is a need to add on its contacts under the slotket an additional 100nF block-decoupling capacitor to further "clean" incoming power to the adapter (power stability)!


The effect of this "enhancements" (not counting some more heat generated) is that currents running thru Cpu & adapter/chipset are a bit higher thus lowering signal impedances & so less succeptible to disturbances (signal "ringing") following to beter stability of setup alowing higher OC. But is never good to exagerate with limit values (long term success & running it for years like my Tuallie) !!!

So ... "dr.Spajky" has finished his homework for You all & left the building, now its your turn to proceed ... :D

"...We will Clock it to the Moon, We will OC-it to the Moon to the end of this year, not because is easy, but because is Hard! ... J.OC.K ! :D :D :D

My goal for Christmass: Celeron-M_350J 100% OC on Air, to replace my current Tuallie ... :)

Space (1,3@2,6GHz), the Final Frontier ... :D :D :D

Please snip a bit quoting! when replying ...

Quote:

---

Originally Posted by k|ngp|n

In your third option, this would involve a set of dipswitches to manipulate vid's(0-5 on and off) and set vcore directly to ic right? This could be done on the ct-479 itself locations u4 or u5, or on the p4c800-e at this location?

---

k|ngp|n , Yes & No; cheaper & older boards can function like this, but looks like newer ones do not. Bios reads Vid definitions (wire trick! on socket for default Vcore) & after first second checking all voltages recieves from Cpu boot Ok command & gives than digitally thru SMBus data to voltage regulator what "default" Vcore or rised in Cmos settings to set. Thats why that Vcc,BOOT 1,2Vcore problem is present; with my Idea modding digital states directly on the MoBo chip for Vcore might circumvent that problem!

That location on the slide, IMHO has no direct effect on the Vcore regulator chip (one on the right, partially shaded with much pins), but I could be also wrong about that ...

I've up'ed the voltage using all three methods- using a trimmer attached to the voltage regulation chip, the vid pin mod, and the bios control. I've found no difference between the vcore p4c800 mod and the bios control. Both are great. I think the vid pin mod is a bad idea because you can't finely tune the voltage! For example, I'm running my 730 at 2.9 using 1.478v

Have any1 used dothan 760? Got ne luck with those?... i could get one for cheap(finnish prizes) ~300euros. Just wondering is it worth buying. In theory it could run faster because of the higher 2.0ghz core speed.

Quote:

---

Originally Posted by datura3

I think the vid pin mod is a bad idea because you can't finely tune the voltage! For example, I'm running my 730 at 2.9 using 1.478v

---

IMHO Vid pin mod is a great idea to determine default Vcore (lower/higher just a bit if needed to expand/shrink Bios Vcore setting range) if needed & done properly. In case of my Tuallie & my MoBo bios, I couldn´t go lower than 1,5 & max was at more than 2V. After that mod I could go lower & max Vcore could be set to 1,75V (max safe long term for Tuallies! with better cooling 4 OC-ing) & all values in_between like used to be before.

The problem with Centrino platform CPUs is that there are 2 (two) default Vcores for them !!! And there is a possible another problem too: the third voltage on socket: supply one for internal PLL that can be also double option !!! [P4 instead has single high one, because it needs it for few GHz (main double speed! ALU units clock, but consumes for PLL just half the current than on Dothans-60mA only!). Someone should really measure that VccA on CT-479, maybe Asus decided to use lower fixed (1,5V) option on their adapter for that ...

I need to ask a question about cooling and these beast. I have my 730-pm now @2880 PI stable. Now cpu runs -11c idle/2c load. It is still with stock CT-479 adapter in my chilled air chaimber. Now do you guys think that it be wort to put it on chilled WC? I would probably be -20c idle/ -7 load.

I just have seen that most people just gained 100-200 or so Mhz from sub temps with cascades. Ln2 is dif story. I have hit a wall right now @2880 benchable. Without chiled air it was 2.65. I could take cpuz screen @2.960 but that dont make me anything since i'm building 24/7 sys to play games. Wc is installed in chillbox but i realy dont want to oppen her up, playing even @2400 just kicks a**

Quote:

---

Originally Posted by railer

I need to ask a question about cooling and these beast.

I just have seen that most people just gained 100-200 or so Mhz from sub temps with cascades. Ln2 is dif story.

---

Yep, looks like 0,09µ process & Dothan core design has a limit @ 3GHz even with Phase cooling!

Quote:

---

Originally Posted by Spajky

Yep, looks like 0,09µ process & Dothan core design has a limit @ 3GHz even with Phase cooling!

---

But "miraculously" on LN2 they can go to 4gHz++ at 1.65v...

My Dothan is benchable at 3106mHz @ -65C

C

Quote:

---

My Dothan is benchable at 3106mHz @ -65C

---

What was it benchable on moderate air cooling? you have 750?

I'm hoping to gain like 3G bench with chilled water. Only 120Mhz to go. Besides on water the temps @load would be realy stable. Same WC holded my Pressy @4.3 @8c load. Thats like atlest 150W of heat. Also with this 730 my temps in the box just stay on same level, much less heat. Onother part is that my PSU rails are also got stable as hell 12v is @ 12.22 @load and it use to drop to 11.67 with pressy.
Anyhow i love this setup. This is a truly overcloker dream cpu. :fact:

Quote:

---

Originally Posted by railer

What was it benchable on moderate air cooling? you have 750?

I'm hoping to gain like 3G bench with chilled water. Only 120Mhz to go. Besides on water the temps @load would be realy stable. Same WC holded my Pressy @4.3 @8c load. Thats like atlest 150W of heat. Also with this 730 my temps in the box just stay on same level, much less heat. Onother part is that my PSU rails are also got stable as hell 12v is @ 12.22 @load and it use to drop to 11.67 with pressy.
Anyhow i love this setup. This is a truly overcloker dream cpu. :fact:

---

On air I'm good @ 2800ish... and yeah, it's a 750.

C

Few screns to support my words.
First one is on regular air stock heatsink 50c @load. Second with chilled air stock heatsink -11idle/2c. Third oh hell 25c patched SSE2 PI just for fun.

The mem timings hurt my PI performance. These are good chips Hynix b5 280mhz stable only @2.8v but wont hold tight timings. Best i could do is 2.4.3.8 @240 mHz. With 3.2v max i could probably go pass 300 but i dobt there will be a need for such with my setup
I will need to sell them now.......

1GB of OCZ EL DDR PC-4000 Gold VX memory with OCZ Booster are on the way ;0]
You think those should do well with my MB? Man i hope they will, I serched OCZ site and it says they are "compatible" With natural cooling they will get in the chillbox i hope i can up the voltage quite a bit :sonic:

Quote:

---

Originally Posted by railer

Few screns to support my words.
First one is on regular air stock heatsink 50c @load. Second with chilled air stock heatsink -11idle/2c. Third oh hell 25c patched SSE2 PI just for fun.

The mem timings hurt my PI performance. These are good chips Hynix b5 280mhz stable only @2.8v but wont hold tight timings. Best i could do is 2.4.3.8 @240 mHz. With 3.2v max i could probably go pass 300 but i dobt there will be a need for such with my setup
I will need to sell them now.......

1GB of OCZ EL DDR PC-4000 Gold VX memory with OCZ Booster are on the way ;0]
You think those should do well with my MB? Man i hope they will, I serched OCZ site and it says they are "compatible" With natural cooling they will get in the chillbox i hope i can up the voltage quite a bit :sonic:

---

Low timings DO matter. I gained about a second in superpi when going from 2,5-3-3-8 to 2-3-2-5 :)

Anyone running over 1.7v ???