Overclocking and Need Help....

[Keith75]

I have had a similar problem with my machine with the last 3 motherboards I have had in it. I notice that when running moderately overclocked that after running for 4 or 5 days this will happen. It is usually triggered by my double clicking on a folder or trying to open my computer or a folder. What happens is everything except the desktop picture will dissappear, explorer will quit, and then reappear and I can then continue as normal. I am not 100% sure this is caused by the overclocking, I am just assuming it is. Has anyone else had anything like this on a machine? If it is caused by overclocking it seems to be the fault of the FSB setting because it will do it with the CPU running at stock speed but the FSB speed cranked up somewhat.

Keith

[chaz]

I have had that problem. Usually after running a heavy application on my gaming machine, but like you said everything return to normal. It happens almost every time I quit Battlefield 1942.

My machine is an overclocked xp2400, but the fsb isnt pushed, its running 133*16.5

It hasnt seemed to cause any real probs so i let it go.

Just my 2 cents

[Beerknurd]

I need to learn how to OC... Anyone know the easiest way to do it??

[Jeff]

You guys should try running Prime95 in stress-test mode to see how stable your overclocked system is.

Beerknerd...what are your system specs?

[Beerknurd]

Pentium® 4 Processor 550 with HT Technology (3.40GHz, 800 FSB)

Microsoft® Windows® XP Home Edition

1GB Dual Channel DDR2 SDRAM at 400MHz (2x512M)

120GB Serial ATA Hard Drive (7200RPM)

3.5 in Floppy Drive

Dual Drives: 16x DVD-ROM Drive + 48x CD-RW Drive


Is that what you needed??? Did I forget anything??

[Anonymous]

Keith75
What is your hardware CPU, motherboard/chipset, Ram, PSU rails.

What do you have FSB, multiplier and Vcore set too and what do you use for cooling?

[TA_Super_10]

Sorry it was me posting that last one, I logged in but when i clicked on forum it logged me back out again.

Oh also Beerknurd
You probably have a option to set AGP\PCI Frequency set them to 66/33

[Keith75]

Well, I have since turned it down but haven't been running long enough to know if it is going to happen again. Last time it happened I was running at 207MHz FSB and the multiplier is set to 11 which is factory. I had the Vcore set to 1.55. It seems like it is my memory causing the problem but I am not positive. I am running the DIMMS at 1.75v.

System:
Athlon 64 3200 Newcastle
MSI K8N Neo Platinum nForce 3 250
1 Gig. PC3200 DDR CL2
Radeon 9800 Pro

[vaughan]

Well I tried overclocking my XP3200 and failed. Its running on a Gigabyte GA-7n400-pro2v2 motherboard and has 1 GB of DDR333 RAM. It has a pair of Seagate Barracuda 80GB SATA drives in mirror configuration. When I tried to modify the BIOS settings to 5% higher for fsb, voltage and AGP it froze. On re-boot - black screen. I reset the BIOS by shorting the CMOS jumpers as per the Gigabyte instructions. Since then it has worked fine at Stock speeds except for 2 things.
1. The Ricoh 9120 CD burner / DVD reader combo drive is detected in BIOS but won't read anything,
2. M$ Office XP won't work. I got a message to the effect that Office had detected significant hardware changes and I had to re-install. I can't as (1) stops me.

H E L P !

Vaughan

[BC]

Well I tried overclocking my XP3200 and failed. Its running on a Gigabyte GA-7n400-pro2v2 motherboard and has 1 GB of DDR333 RAM. It has a pair of Seagate Barracuda 80GB SATA drives in mirror configuration. When I tried to modify the BIOS settings to 5% higher for fsb, voltage and AGP it froze. On re-boot - black screen. I reset the BIOS by shorting the CMOS jumpers as per the Gigabyte instructions. Since then it has worked fine at Stock speeds except for 2 things.
1. The Ricoh 9120 CD burner / DVD reader combo drive is detected in BIOS but won't read anything,
2. M$ Office XP won't work. I got a message to the effect that Office had detected significant hardware changes and I had to re-install. I can't as (1) stops me.

H E L P !

Vaughan

chkdsk /f your boot drive.... xp does not always recover on its own.
get XP to delete the CD rom.....
my comp - prop - hardware - dev mgr... and delete it.
back up to most recent sys restore checkpoint if within 24 hours.
it will reboot as normal.

do a full power off again (unplug PS or switch off the PS too)... take it cold during the sys restore reboot cycle.

after sys restore & reboot... cd will be back and should read.

the registry obviously got hosed in crash. happens to me all the time :(
(all my drives have the 8mb buffer on them)

the buffer (8mb) on the drives is what bit you during this.... xp does not always flush it completely... the registry is the last to be written out during reboot... remembr, it exists as 5 files.

[BC]

Well, I have since turned it down but haven't been running long enough to know if it is going to happen again. Last time it happened I was running at 207MHz FSB and the multiplier is set to 11 which is factory. I had the Vcore set to 1.55. It seems like it is my memory causing the problem but I am not positive. I am running the DIMMS at 1.75v.

System:
Athlon 64 3200 Newcastle
MSI K8N Neo Platinum nForce 3 250
1 Gig. PC3200 DDR CL2
Radeon 9800 Pro

Keith... both the 3200 5AP (clawhammer @ 10x) and 4AX (NewCastle @ 11x) have locked multipliers.... may I ask how you are doing it? I know in the lab, clocking, multiplying, wait-states, voltages, etc, etc, is a snap as we always have to set all the dip switches for each PUT (processor under test).

I see a spot on my MB for a 4 position DIP switch set which will do 'hard set' of the multipliers to override bios... right now, the DIP switch is NOT installed... This switch (if I followed the traces correctly with the meter) will allows manual settings / override of the BIOS CPU multiplier... it will start to CPU standard (which is 10 for Claw) to CPU standard+8 in 0.5 step increments (hence the 16 possible combinations of 4 switches). Do I have to simply put myself 'back in the lab' at home with the ASUS board?
(NOTE: This DIP also ups the clocks for AGP and PCI... there is a table actually printed on the MB)

My Claw is currently running at FSB 228 (perfect multiple of the RAM, which is Corsair 2-3-2-6), Core is 1.55, RAM is 2.7v and AGP is bios (1.5) standard, BUT am holding back the AGP to 4x. I didn't see a need to push to 8x. I can't however get the BIOS to cooperate and push it past 230 FSB to 234, which is the next perfect multiple. I assume the ASUS NB and/or hypertransport are getting in the way.

The NewCastle is running FSB 210 (as a compromise between the math multiples).. the math works out that it will go to 220 FSB on Corsair CMS512 (3-3-3-8) with no mods either (that's 2420 Mhz and also it's internal cap). The NewCastle was perfectly stable at that FSB as well.

You guys clearly have MB's that are/seem more tolerant or more 'free' than the standard ASUS type... and seem more like my lab machines. So, may I ask your indulgance and ask you to tell me how the HECK you get a machine which is locked at 11 to go over that???? Nothing I use software wise at home will unlock & force it up... Do you have something that I don't ... or do I need to think 'this is my lab' at home and just go for it? (I dislike production boards without dip switches & jumpers, FYI)


Thanks in advance
BC

[vaughan]

BC,
Didn't work. I have no previous System Restor Points.
When I uninstalled using Device Manager, then Cold Boot the system detected new h/w & installed the Ricoh MP9120 DVD/CDW OK. It shows up in My Computer but cannot be accessed, ie if I put a CD in it the thing won't recognize it has a CD in it.
I tried having booting from ONLY the CD drive with a Win XP Pro CD in it. Somehow the PC booted normally to Windows. Its as if the BIOS gets ignored and it just reads the HDD. Weird.
I'll take it to the computer store near me when I get a chance.
Thanks for trying.
Vaughan

[BC]

BC,
Didn't work. I have no previous System Restor Points.
When I uninstalled using Device Manager, then Cold Boot the system detected new h/w & installed the Ricoh MP9120 DVD/CDW OK. It shows up in My Computer but cannot be accessed, ie if I put a CD in it the thing won't recognize it has a CD in it.
I tried having booting from ONLY the CD drive with a Win XP Pro CD in it. Somehow the PC booted normally to Windows. Its as if the BIOS gets ignored and it just reads the HDD. Weird.
I'll take it to the computer store near me when I get a chance.
Thanks for trying.
Vaughan

Vaughan,
sounds like you have a legitimate read problem... I had a problem with a sony drive once that could not read the TOC in session 1, but session 2 was fine.... turns out the plastic gear had worn out.... that is why I don't buy sony drives anymore... their lifespan is too short for the price here..
If your drive is under warranty... seee if you can get warranty replacement.... get an RMA and just swap it out.

BC

[Keith75]

BC,

Ok, I am not sure what you mean. I haven't changed the multiplier for the chip. I keep it at x11. I am just increasing the FSB setting up from 200 to 207 at the moment.

Keith

[BC]

BC,

Ok, I am not sure what you mean. I haven't changed the multiplier for the chip. I keep it at x11. I am just increasing the FSB setting up from 200 to 207 at the moment.

Keith

Keith,
from the words of your previous post... you said the multiplier was 'back' at 11.... i figured you had pushed the clock above 11x.

as for only getting 207... try pushing AGP down to 4x.. and pushing
hypertransport down to 400 mhz. see if your FSB comes up.

It will help you find the weak spot.

BC.

[Keith75]

Ok, I will do that. I have the AGP and PCI bus speeds locked at 66/33MHz. The symptoms I had when it was overclocked more were errors saying that a program had a serious error and had to close and in windows it gave me some type of error I can't remember in a little dialog box.

Keith

[BC]

Ok, I will do that. I have the AGP and PCI bus speeds locked at 66/33MHz. The symptoms I had when it was overclocked more were errors saying that a program had a serious error and had to close and in windows it gave me some type of error I can't remember in a little dialog box.

Keith

If that error shows up again, please let me know and I will look it up

I am suspecting a NB/AGP timing conflict. Normally, when the FSB goes up, so should the PCI & AGP in this new architecture because it's driven by the CPU, which is driven by the FSB and it all stays in sync. The kicker w/ NewCastle & Claw is that any changes to FSB have a 4x effect on the Hypertransport in the NB. At 200 FSB, it's 800 Mhz x 16 bits... hence the North Bridge's chipset *AND* cooling quality come into question when above that.

for what it's worth... I did remove the factory compound from my NewCastle's NB and heatsink, and I polished the dickens out of bottom of that heatsink. I did it because the board/chip was hot at 200 and the heatsink was cool... Now the situation is reversed (as it should be). I used just enough AS-5 to get it to fill the center depression area and start to spread onto the main surface. Once I achieved that, I locked the heatsink down (hoping simple spring pressure would do the rest and it has). So far, it has without failure. That includes operating at 2.37 Ghz (2396 mhz). Since Dad has that machine for now, no further tests can be performed.

The last report from my Dad (and you will ALL love this).... his Pentium-III (which I tweaked until it hurt) at 1.2G is outrun by the NewCastle by 4.5:1 ... He renders Mpeg video on it. It used to take 6-6.5 hours to render a 2 hour video. Now it renders in 80 minutes. so the 1:3 ratio became a 3:2. Needless to say, there is a about to be a check in the mail for me to build a replacement... LOL I did go shopping... All the top computer places here are out of chips / MBs.. I suspect the Semprons / FX-53's, etc are hitting the market officially....

Has anyone seen any of the new chips officially for sale?



--- on a side note --- (since my hands work today)
I have also determined that the Clawhammer, although limited internally at 2400 mhz, is not capable of sustaining it, regardless of voltages.. It will eventually 'hang'.

Someone made the suggestion of lowering the multiplier to 9 and then upping the FSB. Given I'm already getting 1.9+ out of single channel memory mode, I'm just going to accept it can't go all the way to 2400, stop at 2304 and let it rest. There are newer chips for us to play with.




---- to finish -- per your machine...

But, as for your machine... we definately need to get it stable at 800x16 Hypertransport, 4x AGP, and at least 210 FSB (2310 mhz) so you can find the sweet spot and take it up to 2.37 or better.

Let me know what math / dependencies you need (if any). Let me know anything I can do to help you get that puppy singing.... Only 207 it a terrible waste.

BC

[Keith75]

It will be the weekend before I have time to mess with it again. I noticed that when I was running at a FSB of 210 my memory bandwidth in Sandra was around 2800. When I lowered it back down to 206 which I am running now at 204 FSB, but with the memory Command Rate set to 1T instead of 2T and now at the slower FSB speed I am getting around 3100! This setting seems to really be important. The Northbridge cooler does look pretty wimpy but lots of people seem to be getting 230 to 300 MHz FSB setting on this board. I just keep thinking it is my memory holding me back. I am going to bump it up and record that error next time so I can give it to you.

Thanks,
Keith

[BC]

It will be the weekend before I have time to mess with it again. I noticed that when I was running at a FSB of 210 my memory bandwidth in Sandra was around 2800. When I lowered it back down to 206 which I am running now at 204 FSB, but with the memory Command Rate set to 1T instead of 2T and now at the slower FSB speed I am getting around 3100! This setting seems to really be important. The Northbridge cooler does look pretty wimpy but lots of people seem to be getting 230 to 300 MHz FSB setting on this board. I just keep thinking it is my memory holding me back. I am going to bump it up and record that error next time so I can give it to you.

Thanks,
Keith


I think you just stumbled on the magic.... you are running only 1T state for memory? I am running 2T. Therefore, the question is: Can the memory keep up with a 1T delay state while running more than 204.?
Your memory performance is obviously incredible.

I will experiment snce I have control over 1T vs 2T ... you may have answered my problem as well... I felt that CL2, single channel mode required a 2T recovery time with the processor cranked up.... and maybe it does... I will dig into the math and see where things are falling down for me.... I am sure that it's falling down for you as well, and the others have managed to get past this hurdle without realizing it or having forgotten to mention it as a tuning parameter limitation above certain speeds. I will hit the timing spec docs for certain.

BC.

[Keith75]

Hey BC, when I was running at the faster FSB and it was giving me the error I had the memory timing at 2T. I only changed it to 1T after turning it back down to 204 FSB. So far after 3 days it hasn't puked on me or given me any errors.

Keith

[BC]

Hey BC, when I was running at the faster FSB and it was giving me the error I had the memory timing at 2T. I only changed it to 1T after turning it back down to 204 FSB. So far after 3 days it hasn't puked on me or given me any errors.

Keith


Ok... so running at 204 FSB, you can get away with a 1-tick command delay.

the previous FSB speed was too much for 2T (it must have been a 3 or 4T need then... something was wacko.... or is the BIOS backwards??? )


May I suggest that once you are certain you are stable ... write down your settings... and then just switch to 2T at 204. See if that runs ok. It better, otherwise you have a hardware problem and need warranty replacement.

Once 204-2T is stable (2T because you have AT LEAST CL2) start
stepping up 1 or 2 Mhz at a time... you better make it to 210-2T.

210 vs 200 = 5%... I'm sure your board is not THAT restrictive. Most
specs are +/- 10%... unless you require tight tolerances (like in power regulation which is +/- 3%).

If you can't get even 204 at 2T to work, I'ld get the warrantly replacement saying the board wont run stable at specs. All you have to say is you have a NewCastle instead of a Clawhammer core and their tech support folks will understand.

let me know please how it works out... I'm bigtime curious now.

BC.

[Keith75]

I am anxious to try OCing more this weekend. I downloaded a new BIOS that finally fixed my temperature reading 70-75 deg Celsius under a load so atleast now I know where I am running. My die temp. is running at 123 deg. Farenheit under a load which is cooler than my XPs do. My Asus board confused me because it didn't read the die temperature so it was always reading about 10-15 deg Farenheit cooler on the same chip.

Keith

[BC]

I am anxious to try OCing more this weekend. I downloaded a new BIOS that finally fixed my temperature reading 70-75 deg Celsius under a load so atleast now I know where I am running. My die temp. is running at 123 deg. Farenheit under a load which is cooler than my XPs do. My Asus board confused me because it didn't read the die temperature so it was always reading about 10-15 deg Farenheit cooler on the same chip.

Keith

Keith...

PLEASE tell me that's 123 F ?????? Otherwise Glad you found out where the readings were coming from and why... ALWAYS good to know.

The chip should run cooler overall.. My NewCastle (at Dad's) is running 42C typical under load... 44-45 MAX under extreme load at 210 w/ 2T.... but it is still a 32 bit bus... so the dual channel-ness of the Newcastle can't be used (it's in a 754 pin package).

If you read my other posts, you'll find Im in a BIOS or burnt hardware
situation... and it's also memory related.

1T vs 2T is available on Newcastles and dual channel machines.... Please watch out and make sure you have a dual channel machine with that 1T setting.

I'm trying to get some details from ASUS about how they implement the 1T... if it has anything to do with single vs dual sided memory boards, then I definately know they have a bug and aren't taking the switchover time into account.

I'm going bios hunting tonight myself.... I'll report back if I find anything.

I think I found a bug. I may drop back a rev and see what happens.

Any suggestions?

Chuck

[Ototero]

I managed to OC my amd 2400+ into a 2600+ by upping the fsb.

But before switching it on, I reversed the air flow on the cooler as Chuck suggested somewhere. When I started, temp under predictor load went up to 60C.

So I reset the fsb but the temp stayed around 57-58C.

Question: could I have upset the "gunk" between cooler and chip?
should I reseat it with as5?

http://boinc.mundayweb.com/predictor/stats.php?userID=43

[BC]

Stu,
Obviously, put the fan back the way it was for now and
get AS-5 ... it has the best thermal conductivity going except for
certain very expensive ceramic glues.

just remember that w/ as-5 a b-b size (or gain of rice) is all you
need.

I dont understand the temp rising when you flipped the fan, but if you
can tell me which one you have, it is possible your model doesn't
draw air cleanly that way...

You will gain more from better compound than flipping the fan.

a rule of thumb i always use... If you break the bond between cpu
and heatsink... you wipe it all off with alcohol and start fresh.


let me know if i can be of any other help
chuck

[Ototero]

Yes, I think that is what happened. The bond broke when unscrewing and rescrewing. I could see the whole thing rocking as I was reseating it.

So, I had a Copper Silent 2L from Arctic Cooling sitting around, added a rice grain of AS that came with it (sorry no AS5) and fitted that.

The cpu is now oc'ed at 2600+ from 2400+ and runng at 55C under load.

I'll leave it for a few days to cook.