• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

ide: Yonah + Aopen i975Xa-YDG

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

mikeguava

Member
Joined
Feb 14, 2005
Location
Miami
Work in progress -will add more later


Yonah on the Aopen i975Xa-YDG

After a painfully slow start Yonah / MODT has finally made it - and actually starting to be fun. This below hopefully should make the transition a bit easier

Helpful things to to get started on the AOpen i975Xa-YDG :


VMods:
http://www.xtremesystems.org/forums/showthread.php?t=95411
Vcore + VDIMM mods are a must for serious overclocking, VCCA mod still needs to be investigated more... Tcore mod might not be necessary with latest bios 1.03C​


Files:

Temp monitoring:


For internal CPU temp diodes:

For Motherboard CPU temp diode:
MBM5 ( use eg. MSI 865 PE Neo2-P motherboard sensors - great find by KIWI ! )
http://users.pandora.be/lsdsmurf/darkvision/mbm/MBM5370.EXE


For Motherboard CPU temp diode:

Speed Step / change multis:


Yonah Datasheet:

The Manual:

BIOS files:

Careful - difficult / not possible to revert back to earlier BIOS using WINFLASH proggy - you might have to go DOS flash. Consensus seems to be that R1.03C is the best bios so far ( not sure about R1.04 yet )

V1.03 - latest offical bios
http://download.aopen.com.tw/Downlo...b/bios/i975xa-ydg/wica13.zip&Section=1&DL=yes

V1.03B - Merom support / Speedstep ON
http://oc-x.de/DOWN/ica13b1.zip

V1.03C - allows to enter bios with subzero temps / Speedstep OFF / NO MEROM SUPPORT!
http://oc-x.de/DOWN/ica13c.rar

V1.04 - no Merom support ( thanks easypanic! )

http://www.ocfantasy.com/ftp/Wica14.zip
Clockgens:


AOpen Coo
http://www.aopen.com

Clockgen from Franck Delattre
http://www.cpuid.com
Getting past FSB wall:

The are several things that influence where your FSB wall is:
Heat
Vcore (+VCCA)
Core

While on air the FSB wall can be caused by 2 things - either a really bad CPU or Heat + Vcore limitations. A good CPU should do around 250 FSB on air ( regardless of multi!), and should go further once Vcore is raised and /or better cooling methods are used. Weird thing is that the FSB can be the same with lower multis - yet once you get colder and raise the Vcore you can get a much higher FSB - until you hit the actual FSB wall of your silicone/CPU. Once that wall is reached even more cooling will not help.

One thing that helped me also greatly to get further on FSB limitation is rasing the FSB slowly. I only use the remote control / Aopen Coo program to raise the FSB. The single mhz steps appears to help not to crash.

Also I noticed that having a certain load helps prevent premature lockups. Once you hit a particular FSB wall - running at that particular max for several minutes can help to get again an extra mhz. It can be a slow process - but this can give you extra 1-6 mhz


The motherboards themselves have no input / very little input on the FSB wall. The same CPUs performed within 1FSB for me on different Aopen i975Xa-YDG boards.


NEW info - Steven has found that during VCORE swings ( droop ) whenever the vocre is on the low end - FSB will lock up much earlier and it is helpful to make sure to rasie FSB whenever the Vcore is at the high end!

With some Yonahs I have found that actually running lower Vcore ( in my case around 1.57V) was helping to get the FSB higher as opposed to running the max 1.62V


Known bugs:

- Can't enter Bios with subzero temps
- Using Molex plug prevents board from booting
- Memset can crash when changing CAS latency
- Clockgen randomly bugs out
- Vcore with current vmod has a max of ~ 1.64V
- Coldbug when CPU is colder than ~ -50C coretemps ( not Evap temps ) This bug prevents you from loading into Windows. For this we might still need the Tcore mod - needs to be checked. This "bug" should only be protection feature of a Windows driver
- Switching between different CPUs on the board without clearing BIOS can cause core temperatures to raise significantly ( +15C air )
- PCI Video cards ( Matrox Millenium maybe others ) and AOpen Coo do NOT get along - prolly becuase of video overlay - causes one core to be fully loaded
- I can not set the bios with my max stable FSB - has to be lower by almost 20MHZ FSB
- ECC memory is always enabled by default in BIOS (might have no impact)

* can't update to older bios from 1.03B ( KIWI )
* board has NO Sata raid, only ata ( KIWI )
* Crossfire only works flawless using ATI driver version 6.4 ( according to Tyrou )
* Crossfire works with ATI driver 6.3 when Write Combining is disabled ( CENTRAL )


Protecting the core:

I hate to admit - but I already managed to kill 2 Yonahs by not following what I constantly preach to my buddies - PROTECT THE CORE!!! One broke ( thank god a free sample ) by simply dropping it on the floor - the other ( of course my golden step 8 CPU I paid mightily for...) by installing a EVAP on one side too tight crushing part of the core.

There are several ways to do so - the easiest using several layers of electrical tape to form a shim (prolly around 4-5 layers depending on the height of tape )
Another easy way is to use the IHS of a socket 478 Intel P3 / P4

The Yonahs appear to be even more fragile to me than Dothan - but that could only be my personal experience..​

Voltages:

VCore: Would like to get input here from more users before I tell everyone to run 1.62V 24/7 lol



VCCA: Stock: 1.5V Kyosen runs it as high as 2.5V :eek: :eek: need to investigate VCCA voltages myself

VDIMM: Fatty D9: 2.3V
On my D9s I run 2.3V max - since we are stuck will a high FSB/MEM ratio for good good ol' fatty D9s there is no need to go wild on the VDIMM - my set could take the clocks easily with what the BIOS allowed it to run... - would love to get some input from user of different types of ram​

PCI-E: I keep AUTO
I have seen no improvement by raising this voltage for FSB / Clocks. Differnt GPUs might benefit from raising this voltage.​

NB voltage: I keep AUTO
I have seen no improvement by raising this voltage for FSB / Clocks.​


OS TWEAKS for benchmarking:

I will have to reserve this space for Tweakmeister Gautam's input - or any other Tweakmeisters. But from my experience there might be very few...but a useMPtimer string and hotfix might be helpful for certain things​

Other information and thoughts:

-Do not challenge Tyrou into a benchmarking competition

-T2700 should come out soon, which should bring the Yonah benchmarking and overclocking to a whole new level

-Intel CAN touch this! :D

-Upcoming Merom CPUs are working flawless with Bios R1.03B on this board

- make sure to set the right CAS latency in BIOS

-BIOS R1.03 should be best for superpi because it has SPeeedstep turned off ( not sure about R1.03C ???)

- we desperately need a BIOS that has all BIOS options unlocked!!!

- the AOpen board only has 2 power phases - there is some minor droop - but nothing too bad!

- WHen setting BIOS clocks above ~ 258FSB the bios will show you incorrect CPU clocks during bootup - the actual clocks will be as set in bios​

Running under cold

- Bios R1.03 , R1.03B will not allow you to enter Bios once you reach below 0*C CPU temps ( read from the motherboard probe ). No biggie if you set everything proper in positive temps. Everything else will run perfectly fine

- Bios R1.03C is supposed to allow you to enter the BIOS with neg. Temps

- If CPU core temps ( read by CPU internal probes ) are below ~ -40*C - Windows will crash at loadup - I suspect this is a driver that is supposed to protect the core. I am not sure if Bios R1.03C will help here - but I doubt it. I have seen this -40C with several different Intel type CPUs - in case of the Yonah at least we can boot...

- I was able to run ~ -160*C core temps - but I had to SLOWLY lower temps. Actually the system booted even @ -180*C but appeared to get VERY buggy and unstable - @ -190C it crashed

- Core Temp program will only read up to ~ +5*C - this is becuase of how Intel calculates core temps and not because of proggy.

- HWiNFO32 can read negatives CPU temps from the motherboard Winbond W83783S probe - but once the negative temps are hit the reading will go downards from 255C ( like with AMD ...255*C = -0.X *C)
This probe becomes very inaccurate ~ - 30C and stops / gets stuck soon thereafter

- a trick I used with my single stage to get into bios was to turn off system while the single stage keeps running for a couple of minutes to let the board ( Winbond CPU temp sensor) to cool below - 45*C - and with the cold temps you can go into bios.

JUMPER settings:


Can't find the manual ATM - input would be helpful

http://www.xtremesystems.org/forums/showpost.php?p=1406514&postcount=1709


Running Merom:

-Make sure to run BIOS R1.03B

-Merom needs less Vcore than Yonah ( ~ 1.3V-1.4V)

-Merom does run hotter

-Coretemp proggy will not show correct temps ( need to subtract 100C minus the shown temp )
 
Last edited:
Nice Mike, sounds like you've been thru a few chips already ;) I'll be trying some CF soon, we'll see what problems that may bring with my setup. Here's a good basic jumper info post from mine.

http://www.xtremesystems.org/forums/showpost.php?p=1380520&postcount=984

I have yet to try the remote. Booting 240 like Team Japan & Clockgen to FSB limit on air was working perfectly. My thoeries on what killed my ES - running high 3.3v line (over 3.7v) in combination with high PCIe frequencies, maybe one or both had something to do with it. Reports on XS of high 3.3v causing instability so maybe I was past that, never had any instability that I could see from it however. I also ran with the molex and had no problems booting on BIOS 1.03.

My first Intel rig and I am pretty impressed with it, lets hope Merom can match Conroe somehow.
 
If Mike gets to double post, so do I.

Absolutely excellent guide! Everything in here is spot on- and extremely useful. Slowly increasing my FSB gave me 90MHz more above going "full force." I found that once you've got a wall at a certain point, a good way to break through it is to do a 32M run at the point just before this wall. It's also crucial to do a 1M run at every 1MHz FSB increment once you get within 10MHz of the top end. Unfortunately, this renders SuperPI times a little worse than they optimally could be. :-/ You can sort of make them more around where they should be by performing the copy-waza tweak at each interval.

BIOS 1.03C does indeed not fully solve the cold booting issues. At about -40C or lower, BIOS reads the temp as "125 C/1 f." At this point, Windows will not boot properly. It'll try to load, but either restart or shutdown just before doing so. You either need the tCore mod to get around this, or raise temps sufficiently until in Windows. Then you can go ahead with the lowest (DI in my case) temperatures.

Registry tweaks that are typically used in AMDs do not apply here. They either do nothing or hurt times. Just use a fresh install of Win2k3 and do as few runs as possible once in Windows. Copy-waza, courtesy of Team Japan, ie copying a large ~1GB file from another drive or partition onto your benching partition helps for sure. However, as far as I can tell, only by about .03 sec or so, but the gain is there. Really the most effective way to shave times down is to tweak the secondary memory timings in I975 tweaker. However, as I frustratingly found, what works for one set of memory often does not work for another, even from the same IC or even manufacturer. tRFC is often overshot, should be reduced in almost all cases. When its too low it can cause instability very easily, so it takes some trial and error to find exactly what your particular modules like. I found that tRAS is best off at 4, no lower, no higher. Might be wrong about this, but that's the way it lined up for me.

For a shim I use Scotch tape, find it to be stickier than electrical tape. Either one's fine. I don't like IHSes as they invariably throw a thermal barrier in the way. Here's what my proc looks like, apologies for the excess Ceramique.
 

Attachments

  • Dscn0331.jpg
    Dscn0331.jpg
    75.1 KB · Views: 829
excellent write up :thup:
(glad it got posted over here too ;))

lookie lookie :eek:




thats a Yonah i borrowed from Sucka (THANKS MAN!!!) i just booted @ 1.5Vcore so my mobo is just fine, it was my Yonah that was gimped... o ya and thats a Merom next to it

making up my tape pads so i dont chip Merom :bang head


-----------


BAHHHHH NO MEROM support in BIOS 1.3C and guess what i have.... C since thats what i needed to run COLD...:bang head

guess ill get over my "mounting" fears faster than i thought... :rolleyes:

thanks for posting NO Merom support, id have been trying for HOURS to get this to work :)

hope i can flash "back" to 3B from 3c :-/
(kick'n back for a bit B4 remount, back to Yonah.... figured i should use a Big Typhoon vs that stock thing.... ya, great idea :rolleyes: )
 
Last edited:
Merom works? Be more specific when you say things that might be exciting Joe :D Not sure on the AOpen board, but BIOS->Advanced->CPU Config for EIST control usually. I haven't seen the AOpen BIOS firsthand, but I know it's pretty light as far as settings go.

Why are you guys using Scotch tape for shims? Wouldn't some double-sided tape with the protective cover left on the top be more like it? Thicker and cushy ;)
 
Joe Camel said:
KICK A$$!! it works!! :attn:

so.... speed step.... any suggestion on how to disable it? (BIOS setting POWER: auto / ? doesn't make a difference :()
Haven't been able to turn it off myself. I think Speedstep in on for good on 1.03B

You didn't get to get it boot in 1.03C - did you?

El<(')>Maxi said:
Nice Mike, sounds like you've been thru a few chips already ;) I'll be trying some CF soon, we'll see what problems that may bring with my setup. Here's a good basic jumper info post from mine.

http://www.xtremesystems.org/forums/showpost.php?p=1380520&postcount=984

I have yet to try the remote. Booting 240 like Team Japan & Clockgen to FSB limit on air was working perfectly. My thoeries on what killed my ES - running high 3.3v line (over 3.7v) in combination with high PCIe frequencies, maybe one or both had something to do with it. Reports on XS of high 3.3v causing instability so maybe I was past that, never had any instability that I could see from it however. I also ran with the molex and had no problems booting on BIOS 1.03.

My first Intel rig and I am pretty impressed with it, lets hope Merom can match Conroe somehow.
Unfortunately yes - went thru a couple of Yonahs - but not as many as Team Japan :santa:

Unfortunately only had 1 Step 8 and this one only lasted very briefly - and that was the one REALLY hurts...lthis cpu should have been easily in the 290s.

I looks like without a good step 8 not much can be done to keep the Japanese SuperPi benchers under control...

Thanks for the pic -will update everything tonight
 
Last edited:
right, i could get into BIOS etc but right @ the point where it would start to boot off of (ANYTHING) id just get a blinking cursor @ bottom left...(with 1.3C)

flashed back to 1.3B with a DOS(5) boot disk and the flash tool that comes with the BIOS no problem there. (after having to swap back to the T2300 chip)


i swear my mobo hates ES chips or i just bought the DOGS of the world's chips.... volt buggeg AGAIN :bang head

when i had Sucka's T2300 i was able to boot fine with 1.5Vcore in BIOS and got up to 240 FSB (but this was only like 5 min of playing around) so i KNOW things CAN work....

anyway, this Merom acts a lot like my buggy T2600 where it cant load windows/stable > XXXX Vcore.... still finding the MAX.... 224 FSB all i could manage with 12x and probably around 1.400Vcore

this thing is FAST, even with this dinky OC, <18.000 Pi 1M *should* be mine.... only .1 to shave off

cooling Big Typhoon (90 CFM fan)


here is a tease of the power of Merom:

merom219pi7wi.th.jpg

totally UN-tweaked...
 
Last edited:
Joe Camel said:
right, i could get into BIOS etc but right @ the point where it would start to boot off of (ANYTHING) id just get a blinking cursor @ bottom left...(with 1.3C)

flashed back to 1.3B with a DOS(5) boot disk and the flash tool that comes with the BIOS no problem there. (after having to swap back to the T2300 chip)


i swear my mobo hates ES chips or i just bought the DOGS of the world's chips.... volt buggeg AGAIN :bang head

when i had Sucka's T2300 i was able to boot fine with 1.5Vcore in BIOS and got up to 240 FSB (but this was only like 5 min of playing around) so i KNOW things CAN work....

anyway, this Merom acts a lot like my buggy T2600 where it cant load windows/stable > XXXX Vcore.... still finding the MAX.... 224 FSB all i could manage with 12x and probably around 1.400Vcore

this thing is FAST, even with this dinky OC, <18.000 Pi 1M *should* be mine.... only .1 to shave off

cooling Big Typhoon (90 CFM fan)


here is a tease of the power of Merom:

merom219pi7wi.th.jpg


Merom not voltbugged just wants LESS vocre - 1.4 is more than enough - 1.45 is about the max vcore before it goews really sour. I would not recommend more than 1.4V for air - even that might be a lot. Under phase 1.45 has not shown any improvement in clocks - maybe under Ln2 it will help
 
Joe Camel said:
but how run Merom COLD without 1.3C .... hard mod :bang head
too bad i wont take a iron to my mobo... NO skillz!!


I recall you mention chilly1 - bypass the froster - turn the SS off during boot and after 5secs of seeing Windows load turn it back on. WIndows will boot at ~ -30*C without a prob - any cooler and it'll crash.
 
Not to devalue the information presented here, but I'm left with but one thought after reading it. These chips might be really something with a motherboard to run them on. This sounds eerily like overclocking on a MSI :eek:
 
larva said:
Not to devalue the information presented here, but I'm left with but one thought after reading it. These chips might be really something with a motherboard to run them on. This sounds eerily like overclocking on a MSI :eek:
lol, something about that just cracks me up bigtime.

:eek: Amazing score Joe!!
 
What clocks are you running? In the screen shots it seems that speedstep has kicked in and your CPU is running a lot slower than I think it should be lol.
 
run into a fare share of restarts/lock-ups now and beyond clearing CMOS often, im finding a run of chkdsk seems to help alot...

ran into a case of "gray screen" EVERY time i tried to boot windows and after swapping out the master card for the 6600GT clear cmos etc, chkdsk /p got me up and booting again :D


EDIT:

trying some 3D action now....

crossfire+merom= cross147682aqua9lk.th.jpg

crosssettimgs147682aqua5np.th.jpg
 
Last edited:
Back