Need help for my first ever overclock !

A

ahmadka

Limp Gawd
Joined
Jul 11, 2006
Messages
136
Hey guys....Im thinking of finally giving overclocking a try on my year and a half old rig, and see how things turn out.. Im gonna use this thread to walk ask all my questions in... I decided to make a new thread because the last one was only meant for getting some guides, which I found lots of :)...it was getting side tracked....

Anyways, here are my opening questions. I know they are a lot, but Im a first timer at this, so please bear with me, I need all the help I can get. I also like to do my homework before I decide to take the test :D .....

Please answer as many of them as you can, preferably in the same order as they appear:

1) In many OC DB's, I have seen people quoting codes such as CCBBE 0615 EPMW in their entires. I believe they are either called OPN, or PR numbers, dont know which. I found out that they are only written on the CPU chip itself, and on the stock HSF that comes with it. Is there any other way to retrieve this number, because I dont want to open up my system ? :)

2) As far as I know, the acceptable vore range for my processor (Opteron 165 Denmark Dual Core 939 Pin) is 1.30v~1.35v...So wouldnt that limit the max vcore voltage I can use? And hence, also limit the processor's overclocking capabilities ?

3) How previous will be the vcore voltage, if i use the 'Special VID Control', to increase the percentage ? Will the final floating point value be rounded off ?

4) what is the role of the chipset, and the chipset voltage, in the entire overclocking process, if any ?

5) My Memory is rated DDR500, so I should be fine with upto a 250 FSB. Does this means that I dont have to mess with the latency timings, provided I keep the FSB at 250 or less ?

6) Say I want to take the FSB beyond 250, and I still want a 1:1 ratio, then do I need to change the latency timings ? I think thats a yes, but not sure ...If so, how ?

7) what is the difference between the RAM divider and the RAM ratio ? I found a guide claiming that they are different things..

8) Is it required to change the ram's latency timings at all ? When can we get away with use OCing the rest of the system, and leaving the latency timings as they are ? Right now, the latency timings in my system are all set to AUTO.

9) What is the recommended HT Link speed ? Some people say that 1000mhz is ok, but we shouldnt go beyond that...while others suggest to lower this. But isnt lowering the HT Link kind of like 'underclocking' your system ? I know that HT Link doesnt necessarily mean a faster system, but still I thought I should ask this.

10) specifically, which hardware depends on the FSB speed ? I am asking this because if I change the FSB, I would where else to expect changes from in the system...changes like increased temperatures, etc..

11) FSB is something physically deployed on the motherboard itself, as far as I know. So if I increase the FSB, it would put more load on the motherboard itself. Is this true ? If yes, will it increase the motherboard's temperature) ?
 
OK well, I finally decided to give the actual practical stuff a shot...and I enjoyed it quite a lot ;) ...

For the time being, I only tried to find out the max speeds of FSB and CPU.. I didnt take a shot at the RAM right now.....Please have a look at the results and suggest to me what should be my next move:

Finding the MAX FSB speed:

Here's what I did...I set the following in the BIOS:

HT Link Multi = 2x
CPU Multi = 6x
RAM ratio = 133

Then, from windows, I started the 'in-place fft' torture test (aka max heat/power consumption test). I started off from 200mhz FSB. Then every 2 to 5 minutes, I would use ClockGen to increase the FSB by 5 Mhz....Here are the highlights of this procedure:

=> Went upto 255mhz with no problems :)
=> When I tried to go from 255 to 260, the computer froze at the exact moment when I clicked on 'Apply Settings' in ClockGen (like when you see the button as being pressed, and it stays like that). When I rebooted, I went straight to 260mhz, and it worked ! And then after that I restarted the 'in-place fft' torture test in Prime 95....

=> The above again happened when I tried to go from 260 to 265...
=> The above again happened when I tried to go from 265 to 270...
=> When I tried to go from 270 to 275, the computer froze again. But now this transition didnt even work when I tried it *before* restarting prime 95, after rebooting, like it did in the previous cases...

So uptill 255mhz, I had no problems, and the max FSB I could reach through any possible means,was 270..


Finding the MAX CPU speed:

Here's what I did...I set the following in the BIOS:

HT Link Multi = 2x (same as before)
CPU Multi = 9x
RAM ratio = 133 (same as before)

Then, from windows, I started the 'small fft' torture test. I started off from 200mhz FSB. Then every 2 to 4 minutes, I would use ClockGen to increase the FSB by 5 Mhz....Here are the highlights of this procedure:

==> I went upto 225mhz on the FSB (2036mhz on the CPU) with no problems
==> When I tried to go from 225 to 230, ClockGen crashed, although it had managed to increase the FSB before doing so...Everything else was fine...probably just some error :p

==> When I tried to go from 240 to 245, the system rebooted....I when into windows again after reboot, and tried going straight to 245, without Prime 95 running...again, the system rebooted...

So the MAX CPU speed I got was 2166.6 Mhz (when FSB was 240mhz)...

Heat Issues:

In both the above tests, I didnt get any heat issues...There were hardly any temperature changes at all...The CPU was 43~44*C at max load, at 2166 mhz ...so no problem there...

Questions:

1) Regarding the low overclock I got on the FSB (255 with no problem), I read in a guide that one of the causes for this is when you have a SATA plugged into SATA1 or SATA2...This applies to me......Should I try SATA3 or SATA4 then ?

2) I think to get better CPU speeds, I wud need to feed the CPU with more volts, since temperatures werent a problem at all......I know, I know, many Opty 165 owners have successfully added more volts to their Opty 165's, but please have a look at this PDF:

http://products.amd.com/en-us/downloads/AMD_Opteron_First_Generation_Reference_101607.pdf

My CPU is the one right at the bottom on the first page....According to this PDF, the vcore limit is just 1.30v~1.35v......Wouldnt this be a problem ?


Please, at least answer these 2 questions....:rolleyes:
 
Wow... LOTS to tackle here. Let me start from the beginning.

1) In many OC DB's, I have seen people quoting codes such as CCBBE 0615 EPMW in their entires. I believe they are either called OPN, or PR numbers, dont know which. I found out that they are only written on the CPU chip itself, and on the stock HSF that comes with it. Is there any other way to retrieve this number, because I dont want to open up my system ?

These "Codes" are called "Steppings" and are written on the box that the Processor comes in, and engraved into the chip it's self... 99% of the time.

2) As far as I know, the acceptable vore range for my processor (Opteron 165 Denmark Dual Core 939 Pin) is 1.30v~1.35v...So wouldnt that limit the max vcore voltage I can use? And hence, also limit the processor's overclocking capabilities ?

The maximum voltage for an Athlon 64 CPU, is, in my opinion around 1.4v air, and 1.5v on water. The 1.35v guideline can easily be pushed a bit. Just keep it under 55c and you're fine.

3) How previous will be the vcore voltage, if i use the 'Special VID Control', to increase the percentage ? Will the final floating point value be rounded off ?

The "Special VID Control" is different on each motherboard. Every board has different overclocking options... so I'm not sure, but I'd expect the "Default voltage" to be multiplied by this "Special VID Control" as a percentage... So, 1.35 * 1.04 = 1.4v (note, this is increasing the stock VID of 1.35v by 4%). But I'm totally not sure, as I've never dealt with this before.

4) what is the role of the chipset, and the chipset voltage, in the entire overclocking process, if any ?

The chipset is less important in the A64 world than the intel world... Simply because the chipset does not deal with memory functions like it does for Intel. It plays less of a roll, but it's often used to control different busses around the board. As a general rule, this should be increased by 1 point above stock for testing, and reduced back down to stock after the over-volt is deemed stable. If it is still stable after reducing it, you're golden.

5) My Memory is rated DDR500, so I should be fine with upto a 250 FSB. Does this means that I dont have to mess with the latency timings, provided I keep the FSB at 250 or less ?

That's correct. DDR500 RAM is capable of hitting 250FSB with the 1:1 (aka 2:1 or 200%) divider. You can decrease the divider in the BIOS to allow higher FSB speeds without having to touch timings.

6) Say I want to take the FSB beyond 250, and I still want a 1:1 ratio, then do I need to change the latency timings ? I think thats a yes, but not sure ...If so, how ?

If you want to keep the 1:1 ratio, then you should loosen all the timings, and increase the voltage.

7) what is the difference between the RAM divider and the RAM ratio ? I found a guide claiming that they are different things..

Not sure... Should be the same thing as far as I know. However, SOMETIMES Ram speed is calculated as a ratio of CPU speed (So, instead of the 0.87:1 divider (DDR333), you'd have CPUmhz / 12 = RAM speed) but most BIOS's don't use this notation.

8) Is it required to change the ram's latency timings at all ? When can we get away with use OCing the rest of the system, and leaving the latency timings as they are ? Right now, the latency timings in my system are all set to AUTO.

No, you shouldn't have to change them at all, unitl you exceed 250mhz FSB. However, the term "Auto" often sets them to a really loose value, generally looser than what the RAM is rated for. The best bet is to change all the timings so they match what the RAM is rated for.

9) What is the recommended HT Link speed ? Some people say that 1000mhz is ok, but we shouldnt go beyond that...while others suggest to lower this. But isnt lowering the HT Link kind of like 'underclocking' your system ? I know that HT Link doesnt necessarily mean a faster system, but still I thought I should ask this.

It should be kept below 1000mhz. Remember, there's a multiplier with this too. You should never exceed 1000mhz, but going below it doesn't hurt performance at all. Stick to the 4x mult for FSB's between 210-260, and use the 3x mult for FSBs above 261.

10) specifically, which hardware depends on the FSB speed ? I am asking this because if I change the FSB, I would where else to expect changes from in the system...changes like increased temperatures, etc..

Almost everything! RAM runs faster, the CPU runs faster, the NB runs faster. The PCIe and PCI signaling busses *can* run faster if they're not locked (you did lock them, right;)). Also, the MOSFETs and Vreg circuitry on the board runs "faster" to supply more power to the chips that request it.

11) FSB is something physically deployed on the motherboard itself, as far as I know. So if I increase the FSB, it would put more load on the motherboard itself. Is this true ? If yes, will it increase the motherboard's temperature) ?

Yes. Yes and yes.
Click to expand...
 
1) Regarding the low overclock I got on the FSB (255 with no problem), I read in a guide that one of the causes for this is when you have a SATA plugged into SATA1 or SATA2...This applies to me......Should I try SATA3 or SATA4 then ?

2) I think to get better CPU speeds, I wud need to feed the CPU with more volts, since temperatures werent a problem at all......I know, I know, many Opty 165 owners have successfully added more volts to their Opty 165's, but please have a look at this PDF:

http://products.amd.com/en-us/downlo...nce_101607.pdf

My CPU is the one right at the bottom on the first page....According to this PDF, the vcore limit is just 1.30v~1.35v......Wouldnt this be a problem ?
Click to expand...

1) It's POSSIBLE that the SATA being plugged in caused an issue. Try moving the ports around... But, I'd probably attribute the low OC to other factors. You might need to:
  1. Increase your CPU Voltage to 1.4v
  2. Increase your RAM voltage 1 setting above stock
  3. Decrease your Ram Divider to the 0.867:1 setting (the DDR333 setting) or one below the stock setting (DDR400)
  4. Increase the NB voltage (ahem, probably not the problem)

Try these, in order, until one helps. Make sure you do them one at a time so you know what the limiting factor was. If a voltage increase doesn't help, I suggest lowering it back to the previous value, to avoid excess heat generation, then continuing onto the next point.

2)the 1.3-1.35v recommendation is what AMD rates the chips to run at. This is not the "maximum" voltage, but the "maximum non-OC voltage". Every chip requires a different level of power to run, as each chip is slightly unique. Take the modern C2Q's for example: Mine runs stock at 1.21v, where my friends runs stock at 1.24v. All that figure was stating, is the maximum allowable voltage for a stock chip, and the maximum allowable voltage without voiding your warranty. (Note, unless you fry your chip badly, AMD will never know that you exceeded the voltage).

Finding the MAX FSB speed:

Here's what I did...I set the following in the BIOS:

HT Link Multi = 2x
CPU Multi = 6x
RAM ratio = 133

Then, from windows, I started the 'in-place fft' torture test (aka max heat/power consumption test). I started off from 200mhz FSB. Then every 2 to 5 minutes, I would use ClockGen to increase the FSB by 5 Mhz....Here are the highlights of this procedure:

=> Went upto 255mhz with no problems
=> When I tried to go from 255 to 260, the computer froze at the exact moment when I clicked on 'Apply Settings' in ClockGen (like when you see the button as being pressed, and it stays like that). When I rebooted, I went straight to 260mhz, and it worked ! And then after that I restarted the 'in-place fft' torture test in Prime 95....

=> The above again happened when I tried to go from 260 to 265...
=> The above again happened when I tried to go from 265 to 270...
=> When I tried to go from 270 to 275, the computer froze again. But now this transition didnt even work when I tried it *before* restarting prime 95, after rebooting, like it did in the previous cases...

So uptill 255mhz, I had no problems, and the max FSB I could reach through any possible means,was 270..
Click to expand...

I attribute this to low CPU voltage. But, uhm, you should be doing FSB modifications in the BIOS for maximum success and safety :).

*Nothing against clockgen, it has it's place, but it also causes lots of problems that are difficult for new users to identify*
 
OK, like you advised, I am slightly increasing the vcore voltage in the following retake of the above test...Trying to find the MAX FSB for my board again...

Here are some changes I made in how the experiment was to be performed:

==>I changed FSB from the BIOS, instead of using ClockGen
==>I increased the vcore voltage from stock to 1.375v
==>I used Prime95 version 25, instead of the old 24...
==>The 'in-place fft' test was run for longer periods now, between 5 to 10 minutes..

The other properties are as follows:

RAM ratio = 133
HT Multi = 2x
CPU Multi = 6x

I was able to go upto 270 again, like before. But a 275mhz again posed problems: the system wouldnt go into Windows !

I investigated this issue...and found the reason (at least I think) why I cant cross 270mhz on the FSB...

As soon as I changed the RAM ratio from 133 to 120, the system allowed me access to windows again....I ran Prime 95 with no problems...


But when I tried 280mhz from the BIOS, the system again wouldnt go into windows....I tried coming down 1 unit at a time from 280, to find the exact MAX....and suddenly, another problem faced me : My hard disk wouldnt get detected, not even in the BIOS !! Fortunately the problem was fixed, thanx to the CMOS reset concept...

The increase in vcore voltage didnt have any positive effect on the FSB then...

So at 280mhz, and a ram ratio of 120, the RAM is running at just 168 mhz.....

I doubt the RAM is limiting the max FSB....what else could be the cause then ?
 
I suggest you run your HT link at 3x's.... Your board should not have any problems whatsoever hitting 350 FSB.. Unless of course you got a dud DFI NF4. I've seen this happen before and I believe that's what the cause was. So yeah as I mentioned just change your HT multi to 3.

Edit.. It's best to keep your HT link as close to 1000 as possible.
 
Yeah I had that idea too, although I havent tested it so far....

Another reason why I thought this might be was that maybe I need to increase the chipset voltage....although again, i havent tried this yet too...

I want to be sure about these things, before I go trying them out...

OK, another observation I made, while I was going over some OC DB's, that people with original DFI BIOS'es, generally have a lower FSB overclock, than those having 'other' BIOS'es...

'other' BIOS'es, such as OCZ Tony's 704-bta (I think that is what its called)....

People with official DFI BIOSes generally have a max FSB below 300...ranging from 250 to 280..while those with the latter Tony one have max FSB's above 300, ranging from 310 to 350...

Although there were many exceptions to the above observation, I think that the BIOS version as matters when it comes to finding the MAX FSB for your board....What do you think ??
 
Well you could try a different bios. I easily went into 300 FSB with a stock DFI bios. It wouldn't hurt to increase the Northbridge voltage a bit. Just remember, baby steps, a little at a time. If things don't get better with a little bump in voltage, your problem is somewhere else.

Where are your memory timings at too btw?

Edit. There was a DFI overclocking guide over on DFI street. Google it. Just make sure you get the one for NF4. There might be a few good pointers in there. I used to have it, but really never read it.
 
OK guys, I tried the max FSB and max CPU speed experiments AGAIN :D :

Finding MAX FSB speed attempt #3:

BIOS settings:

Constants:
HT Link multi = 3x
RAM ratio = 133:200
CPU Multi = 6x

Variable:
FSB speed

Other Settings:
All changes done through the BIOS, not ClockGen !

Procedure:

I started off from FSB = 270 mhz....

With 270mhz, I could NOT boot into Windows....

==> Changed DRAM voltage from 2.60v, to 2.70v

Now, I COULD boot into Windows, at FSB = 270mhz

Tried 275mhz........wont boot....

==> Changed Chipset voltage from 1.50v to 1.60v

Still could NOT boot into Windows with FSB = 275mhz

==> Change 'PCI eXpress Frequency' from 100mhz to 101mhz !

STILL, I could NOT boot into Windows...

Conclusion:
==> The following factors are NOT limiting the FSB speed:
i) DRAM voltage
ii) Chipset voltage
iii) PCI eXpress Frequency


Question:
If the above factors are not limiting the FSB speed, what is ?????

It wouldn't hurt to increase the Northbridge voltage a bit. Just remember, baby steps, a little at a time. If things don't get better with a little bump in voltage, your problem is somewhere else.
Click to expand...

Already tried that above......No positive effect....:(

Where are your memory timings at too btw?
Click to expand...

Right now, all the RAM timings are set to AUTO.....I have not yet tried inputting the timings at which my RAM is stock rated (3-4-4-8 @ 250mhz)





Finding MAX CPU speed attempt #2:

BIOS settings:

Constants:
HT Link multi = 3x
RAM ratio = 133:200
CPU Multi = 9x
PCI eXpress Frequency = 101mhz

Variable:
FSB speed

Other Settings:
All changes done through the BIOS, not ClockGen !
The in-place fft test from Prime95 v25.5 was run for about 10 minutes, after each change

Procedure:

I started off from FSB = 240 mhz....
Kept increasing 5mhz after each successful attempt...
I went all the way upto 270mhz on the FSB successfully !! Thats 2429 Mhz :D :D !!


Conclusion:

I didnt recieve any errors in the in-place fft test in Prime95 v25.5 ....

So I DONT think the CPU is a limiting factor so far...

I didnt go past 270mhz on the FSB myself, because the FSB has stability issues beyond this point :( ........

Also, the vcore voltage was still at STOCK (1.30v~1.31v)...I didnt need to increase it :D

Even at full load at 2429 mhz, the CPU's temperature remained about 48*C ~ 50*C...the MAX ever reached was 52*C...not bad :)

Question:
What can I do to further test the stability of the CPU at higher FSBs ? I know about running tests for longer hours, etc.... But everyone / every guide gives its own method, duration, 'stress program to be used' details....Can someone recommend the *main* stress tool... ?
 
Ok I suggest you change your memory settings from auto to it's rated timing. I'm sure your board and your 165 have some more in them. Also, try your memory divider at 166. Shoot for a 300 FSB and cross your fingers.

Also, do me a favor. Take some screen shots of your cpu and memory in cpuz. Also, load up clockgen and click on "get clocks" I think it's called. Then take a screen shot of that too.
 
Sta!nless said:
Also, do me a favor. Take some screen shots of your cpu and memory in cpuz. Also, load up clockgen and click on "get clocks" I think it's called. Then take a screen shot of that too.
Click to expand...

Thanks for the reply Sta!nless :) .....Here's one mega picture to give u all the juicy details...

Included are CPU-Z v.1.41 (All 5 pages), ClockGen's 'Clocks', Motherboard Monitor 5.3.7.0's stats, and 'Sensors' from Everest Ultimate Edition v4.20 ...

NOTE: In the CPU-Z display showing RAM info (bottom right), 'Slot #3' is selected...Please note that 'Slot #4' is *exactly* the same as this (Dual Channel)...while Slots 1 and 2 are empty :)...

RAM timings have been set to AUTO in the BIOS (I have never changed this setting)...Also, my RAM is STOCK rated for 3-4-4-8@250 Mhz... (The last one was 7 or 8, can't remember :p )

The PC is in idle state here, by the way ...

Here's the screenshot:

dataez4.png
 
Just tried to boot the system at 166 RAM ratio, and 275mhz FSB....still didnt work :(

I'm running out of options here :confused:
 
Another attempt...which failed...

Settings similar to the ones before, only this time, I tried the using the stock timings of my RAM as well,--> 3-4-4-8-1T.....again, it failed...I couldnt boot windows with 275mhz FSB :( ...tried 3-4-4-8-2T...again failed...

Here are the exact settings I used (for 3-4-4-8-1T):






In addition to that, I also tried shifting the FSB from 133 to 140, 150...still nothing..

In addition to *those*, I then changed vcore to 1.425v (0.125v more) , chipset voltage to 1.6 v (0.1v) more....STILL NOTHING !! :mad:

Dont know what is causing this...

I think that whatever is causing my FSB speed to not reach 270mhz, is something to do with the RAM....why ? Because the ONLY 2 times when I WAS able to boot into windows, at 275mhz (!!!!), was when I did either of these 2 changes:

1) Change RAM ratio from 133 to 120
2) Increase RAM voltage from 2.60v (stock) to 2.70v

But even then, I was again not happy, because then I could not reach 280mhz :p
 
OK, yet *another* attempt ....

Same basic settings as before, but this time, I set the RAM to 4.5-7-7-12-2T......the first 3 numbers are the highest my motherboard cud allow...

Result ? My computer wont even POST now!! thats a first !! :mad:

Had to reset the jumper to get it live and kicking..

Here's the RAM settings I used:

 
I think I found the problem...

It is the RAM that is limiting the max FSB after all :(...

I was having a look at the Memory module details topic at DIY street .... link ... and I figured out the chip used in my RAM:

- OCZ EL DDR PC-4000 Gold Edition [2x1GB] (UCCC) [3-4-4-8]
Part number: OCZ5002048ELGE-K
http://www.ocztechnology.com/produc...pc_4000_1024mb_gold_edition_dual_channel-2gb_
Click to expand...
I also found the following description written there regarding this RAM I am using:

Samsung UCCC
- Maximum voltage needed is 2.8v
- 2.5-3-3-X @200Mhz
- 3-4-4-8 / 3-4-3-8 @250Mhz
- Maximum Overclock ~275Mhz
Click to expand...

So I guess, I really have hit the limit then....my rig's max FSB is 275mhz after all...or 272mhz to be exact :rolleyes:....

The end of the road as far as the FSB goes...what do you guys think ?
 
That's probably a max of 275 on your ram. You can avoid that using a divider. It sounds like you tried all your options. I'm sure your board should at least have more in it. What are your temps at load? If you already posted that I apologize. Also, I wanted those screenshots at your max FSB of 270? Could you do that for me and post another screenshot?
 
YES !!!! IT WORKED !!! 300mhz SUCCESSFULL !! :D

OK,..normal again :rolleyes: .....I was able to boot into windows, with an FSB of 300mhz !! FINALLY I broke the 270mhz~280mhz limit !!

The settings I used were exactly as they were before...The only change was a CPU Multi 9x, instead of 6x...

I was using 1.425v when I booted, which worked.. :)
So I tried switching to a lower vcore..1.375v...still worked :D

Also, just to confirm this, that the 6x CPU Multi was the problem, I tried switching back to 6x CPU Multi..... Result ? The computer wont POST :mad: ...

Had to reset the jumper....Thats twice I have had to bring my rig back from the dead in the last 24 hours :eek: ...


Anyways, what should I do now ?
Should I try to check whether the computer is stable at these settings before going any further ?
Should I try to go for even higher timings ? Beyond 300 ?
Or should I work on the RAM timings, etc ??

Also, should I use a 9x CPU Multi from now on, with higher FSBs ? Note that this wont exactly mean 'removing a variable from the equation' :p

Anyways, here are some screenshots of the EXACT settings I used...







 
I just finished a bowl of cereal, quite a good thread, very informative. I was going to mention memory dividers, usually this works wonders rather than buying higher-rated memory to run sync.
 
ahmadka said:
The settings I used were exactly as they were before...The only change was a CPU Multi 9x, instead of 6x...
Click to expand...

I thought you tried with a mutli of 9 after trying the 6X...

Oh well I'm glad you got it.

Keep your memory timings as they are now.

I would stick with that 9x.

Your chip will more then likely hit a wall before your board will.

Now just keep checking for stability as you increase the FSB.

Then go back and try to tighten your memory timings.

GJ on not giving up.

Also, keep an eye on your temps.
 
OK then, here are the latest headlines :) ...

I used a 166 RAM ratio, a 300mhz FSB, and a 1.35v vCore, along with the previously mentioned RAM settings...

I did 1 hour in Prime 95....no errors...:)...

Same settings, for 310mhz FSB...1 hour in Prime 95...no errors again :D...

Then I ran into a problem...

with a 166 RAM ratio, going any higher on the FSB meant more pressure for the RAM...even at 310mhz FSB (above), the RAM was already operating at 257mhz...7mhz faster then its factory speed...

To deal with this, I tried the following 3 options, all of which resulted in different kinds of errors:

1) FSB = 320mhz, RAM Ratio = 166, RAM latency = 4-5-4-8-1T...(caused a no-POST situation :( )
2) FSB = 333mhz, RAM Ratio = 150, RAM latency = 3-4-4-8-1T (so RAM is at 249mhz)...No Windows...
3) FSB = 320mhz, RAM Ratio = 166, RAM latency = 3-4-4-8-1T ....I COULD boot into windows, but Prime95 immediately threw an error when I started its small fft test :(


I could have tried giving either CPU and/or RAM more voltage in (3) above, but I didnt try that....CPU approaches 53*C when I give it any more than 1.35v...

Need advice...
 
Ok, things you can try.

133 divider.. At a 310 FSB your memory is at 200. At a 333 FSB your memory will be at 214.

Your processor will be at 2.997

Sometimes overclocking is about compromise. Sure everything has a max, but sometimes you have to find a happy medium.

You could try more voltage, but your temps seem a little on the high side. Your still in the safe zone though.

I think what is happening, and I could be wrong, but I think your processor's memory controller is getting too stressed being at 310 with DDR500 performance from your memory.

Most people say that CPU speed is king. So I suggest worrying more about your 165's max stable speed.

GJ so far!!
 
OK, suppose I wanna move onto the RAM now, seeing that I have found the 'limits' for FSB and CPU...I want some advice on how to go about testing the RAM through memtest on boot....

I know that I start from 'loose' timings, and then I start tightening them up, but here's the thing......My RAM is rated at 3-4-4-8-2T@250mhz ... So far, I have only tried 'loosening' the timings twice....both time, the computer went into a no-POST situation...the timings I had tried were 4.5-7-7-8-2T, and 4-5-4-8-2T I think.....So please guide me on this..
 
Well if your satisfied with your current overclock, I would try to tighten the memory timings. I would start with 3338 1t .
 
Well I decided to skip the RAM thing for now, cuz my brain has really stressed out over the past days......phew...time to check the stability..

I ran Orthos for a little over 8 hours @ 2794mhz....Guess what ? No errors, and CPU temperature remained around 50~51 too...

My system is now officially stable at 2794 Mhz :D :D ...... I cant believe I did it :p

I have also made my *very* first OC ever !! Its at DIY Street ...check it out:

http://www.diy-street.com/forum/showpost.php?p=766304&postcount=434

Thanx to everyone who has helped me out here so far :D
 
You must log in or register to reply here.
Back
Top