stupid question?
- Thread starter nicnac1
- Start date
Associate
- Joined
- 15 Jul 2009
- Posts
- 23
Yeah, I haven't tried anything like that, but I did read a nice how-to guide that explained lapping. CPU tops aren't usually perfectly flat, so sanding them down maximises their thermal conductivity both my improving overall surface conductivity and evening that conductivity across the whole chip rather than being slightly focused towards the center.
Good stuff! . . . now hopefully thats stable!
I find it a small advantage to run the memory as fast as possible for the final clock but syncronous RAM [1:1] is generally considered a good setting for the initial overclocking session because it allows you to concentrate on the FSB and CPU clocks without having the memory crashing due to running out of spec. With a 400MHz-FSB (1600MHz System Bus) and memory running 1:1 sync it's only hitting 400MHz (DDR2-800) which means it's been effectively removed from the overclocking equation!
For the moment 1:1 sync memory will allow you to concentrate on the FSB/CPU/NBCC clocks i.e if you have a crash its most likely to *not* be a memory issue! one less ball to juggle!
Speaking of NBCC clocks when you get time can you try
2.4GHz CPU (6x400), Memory [1:1] sync (400MHz/DDR2-800) and 600MHz NBCC
9/6 = 1.5
1.5x400= 600
So waiting for you to confirm:
3.2GHz (8x400), 400MHz-FSB (1600MHz System Bus) Memory [1:1] sync (400MHz/DDR2-800) is stable
2.4GHz (6x400), 400MHz-FSB (1600MHz System Bus) Memory [1:1] sync (400MHz/DDR2-800) is stable
You have multiple ways foward now and if the above settings are stable you can concentrate on the CPU and FSB clocks if you like. We were having problems getting the CPU at 3.4GHz before due to alarms and sirens going off caused by the large increase of vCore that was used so I'm not totally sure if getting the chip to 3.6GHz is gonna be achievable without the (CPU/PWM)temps going off the charts or the fact your specific chip isn't high enough grade.
If you check the vCore-Actual that setter was having to use you can see that his Q6600 was able to attain 3.6GHz with 1.416v flowing through it at load which is what I was hoping yours would reach. The thing that you will discover is that no two chips are identical, some need little voltage while others need a huge amount we can only use setters information as a loose guideline and see how you get on.
All you would need to do is set your chips multi back to [x9] which combined with a 400MHz FSB would give you 3.6GHz on the chip however its not gonna be stable until you give it enough juice while at the same time keeping both the CPU/PWM temps in line, its a bit tricky there but I would give it your best shot!
3.6GHz (9x400), 400MHz-FSB (1600MHz System Bus) Memory [1:1] sync (400MHz/DDR2-800)
If you can't get 3.6GHz (9x400) stable even with a massive amount of vCore and extra cooling from your case side panel off with deskfan blowing away then we will have to try something else . . .
3.4GHz (8x425), 425MHz-FSB (1700MHz System Bus) Memory [1:1] sync (425MHz/DDR2-900)
An handy thing about 425MHz-FSB (1700MHz System Bus) is that you can hopefully use one of the upward memory multipliers to raise your RAM to approx 533MHz (DDR2-1066) which could be great when combined with a fast FSB.
Lots of testing to do but take it one step at a time, if you run into a problem or something strange happens then just post back with questions, in essence we need to find out the answers to the following questions which you will have to work out seperately and one at a time . . .
- How high can the FSB-MHz be raised?
- How high can the CPU-MHz be raised?
- How high can the Memory-MHz be raised?
- How high can the NBCC-MHz be raised?
Last edited:
i think before i do any more testing i think i need to sort out my cooling. last stress test i did started pushing 80C on the cpu and pwm and if i'm going to be upping the vcore even more i think i should look at replacing the case to maximise airflow.
do you think the antec 300 will be enough?
do you think the antec 300 will be enough?
It should do the job very well - it's a great value case for an overclocker - especially if you include the optional 120mm fans.
Sounds like a good idea however you can carry on testing the other areas of your system while leaving the vCore lower for now . . .
- How high can the FSB-MHz be raised?
- How high can the CPU-MHz be raised?
- How high can the Memory-MHz be raised?
- How high can the NBCC-MHz be raised?
Pump the vCore! 
If that doesn't work then it may be worth raising the vTT for some testing . . .
Aside from the CPU tests there are a bunch of other tests waiting to be confirmed!
There's lots of backwards and forwards stuff here, don't worry it should all come together soon!
If that doesn't work then it may be worth raising the vTT for some testing . . .
Aside from the CPU tests there are a bunch of other tests waiting to be confirmed!
There's lots of backwards and forwards stuff here, don't worry it should all come together soon!
Will be back home this afternoon so will hopefully be able to confirm some of the above
not looked into other cooling alternatives at the moment... need to do that soon!
You mean 425x8
Now lets be having some stability test screen shots and see where we are!
An error is an error, doesn't mater what it says!
It could be either a lack of vCore or something to do with the raised FSB/Nortbridge
Drop the CPU multi from [x8] to [x7] and try again, if that fails restore the CPU multi to [x8] and try boosting the vNB!
Last edited:
By this stage in your boots I would have performed quite thorough testing on the processor to see how it scaled and what volts it needed. I would have removed as much stress from all the other hardware as possible so that when I encountered an error it would have 99% been something to do with just the overclocked CPU . . .
What we need is a figure (vCore) that your processor needs to stably run 3.4GHz. At the moment you are winging in with a few things being overclocked at once which is possible but makes it a pain when something goes wrong!
If you want to concentrate on just your CPU clocks and see exactly how many volts the CPU needs then can you try . . .
3.4GHz (9x377), 377MHz-FSB (1508MHz System Bus) Memory [1:1] sync (377MHz/DDR2-754)
If it runs stables at XYZvCore at 9x377 then we can use the same vCore for 8x425, if that isn;t stable we can pretty much assume the chipset needs a little more juice!
What we need is a figure (vCore) that your processor needs to stably run 3.4GHz. At the moment you are winging in with a few things being overclocked at once which is possible but makes it a pain when something goes wrong!
If you want to concentrate on just your CPU clocks and see exactly how many volts the CPU needs then can you try . . .
3.4GHz (9x377), 377MHz-FSB (1508MHz System Bus) Memory [1:1] sync (377MHz/DDR2-754)
If it runs stables at XYZvCore at 9x377 then we can use the same vCore for 8x425, if that isn;t stable we can pretty much assume the chipset needs a little more juice!