Start OC my 4000+ A8Npremium help plz

Associate
Joined
23 Apr 2006
Posts
152
Im starting to over my rig here is the spec

AMD64 4000+(NinJa SilentX 120mm) San Diego ADA4000 AA5BN
2x1 GB Corsair XMS 3200 C2PT2.5-3-3-6
A8N SLI premuim 1009
X1900XT+VF900
Samaung 250 gb
P180 (3x 120mmfans)+ Tangan580w

So far, im trying to find out what the maximum limit of my
1. HTT, FSB
2. Max chip capable
3. RAM Speed and timming

i use this link as guide line

http://forums.anandtech.com/messageview.aspx?catid=28&threadid=1497607&enterthread=y

Since time is my problem right now. I would like to ask poeple who might have same components or any ideas of the components limmitation, i would like to know their limmit roughly so i can jump start my FSB, RAM testing.

I also have try to do this
FSB 215 x12 = 2580
RAM 400 2.5-3-3-6
HT x4
PCI 33 Mhz
PCI Express 100 Mhz
Vcore = 1.5
Vram = 2.6
Temp CPU Load/idle 47/34
prime 95 test (torture blend)for 11 mins stable.

I also try
FSB 250 x12 = 3000
RAM 400 2.5-3-3-6
HT x4
rest are same
But this one not even boot up... too bad

i understand this is the beginning of my OC path and i would like to make it as fast as possible (surely safety first), so please help
 
Last edited:
You can look here and here for OCs with your chip (although take some of the data from the first link with a pinch of salt - look for the average clocks). Hope that helps.

Don't rush your clocking though - it's better to do it over time.
 
Your first oc there is a good starting point.

I would raise the FSb in 5MHz increments until it becomes unstable and raise Vcore etc as per This Sticky

if you follow the sticky, you won't go far wrong. If there's anything you don't understand then don't be afraid to ask :)

It's really not advisable to go by other people's experiences with similar equipment as each CPU is different.

Having said that, one of my rigs has very similar spec to yours (except the gfx). I only managed about 2.64 on mine (220x12) whereas other people were getting over 3.0 on their's.

Stan :)
 
few questions
It seems like it might take few weeks for my project overclock.

1. While i am testing my componet to its limit, is it ok to turn them all back on stock when im not testing it. Or i must continue until each component reach its limit. I mean like im testing my RAM at 200 and 205 today. Once i finish i turn everything back to Stock and come back later for further test.
Is it safe?

2. what is the stock voltage of my CPU? ADA4000 DAA5BN it seems to be vaiable in AMD website of sticky or shall leave it on auto?

3. To find out the Max FSB, how long of prime95 i should be running each time i increase the FSB
 
Last edited:
This morning i raise my HTT to 230 on 12 multi (2760) and fail the prime95 in 4 mins. I lower the HTT down to 226x12(2712) and it seems stable for in prim95 10 mins test.
My ram is on low speed.

Is this my cpu limit? with this auto Vcore? or should i drop multi to 11 and raise the cpu abit more to 245 to be equal to (2695)

or should i continue with x12 and raise Vcore which i could not do on my current Bios version 1009 (how unlucky)
 
Now My HTT is at 226 x12 ( i it wont go any higher )
Im trying to do 1:1 with RAM 2.5-3-3-6

So far i run prime95 for 10 mins it pass

does this also mean my ram is running at 452 speed? seems abit too high?
 
If i run 12x226, Im kind of dont like my RAM to run at 226.
So i lower the multiply to x 11 and
raise HTT to 250 Mhz
Vcore 1.525
HT x 4
RAM 333 (in bios) (running at 196.4)

Is this one better? comment please
 
Last edited:
If it is stable, you can't run your memory too fast. I would rather raise the Vdimm, as I talk about below, and have it run faster. It might have been your CPU limit that you were running into, or it might have been the memory, or even the chipset, but that's the least likely.

You need a BIOS you can raise voltages on.

It is easiest to find the limits of things in a particular order. First find the maximum FSB/HTT speed. Then the memory. Then the CPU. If you follow that order it's easiest to isolate the problem components.

To do this, lower the multiplier of the processor and run the RAM on a divider. This will mean setting the RAM to DDR 333, or even DDR 266 just now. The reason for doing this is to eliminate the RAM and the CPU as the sources of instability for just now. Also, drop the HTT multiplier to 4, and then above 250 MHz, to 3. Then raise the HTT frequency a bit at a time.

I reckon you'll be able to pass 300 MHz HTT. By the time you are reaching that stage you'll need to have brought the CPU multiplier down to 8 to prevent you from having overclocked the processor at this stage. Also, this doesn't need to be a very thorough and time consuming stage. Up the HTT freq. in steps of 5 MHz or so and only test with Prime95 for 5-10 minutes or so. When it fails, go into the BIOS and raise the chipset voltage a notch, and test again. When you finally get bored of finding the maximum HTT freq. move on to the RAM. I personally wouldn't like to run my chipset at the maximum voltage I can, because it's going to get a bit toasty, but it's safe to raise it a notch or two anyway and might get you stable at a slightly higher speed.

To test the RAM you'll need Memtest. You can get a version that runs from a floppy, or a bootable CD. It doesn't take you into Windows, it loads a very basic OS simply to test the memory for stability.

What to do is to set the HTT freq. back to stock, but keep the HTT multi at 4 up to a freq. of 250 MHz and then 3 above that. Set the RAM back to a 1:1 ratio, in an Asus board this will probably be called DDR 400. Whilst keeping the multi of the processor down to prevent overclocking it, start raising the HTT freq. Let it go into Memtest and run through a loop. For speed, to begin with I only use test numbers 5 and 8. If you pass them it's almost certain you'll pass the others. I'm not sure what you'll get from the RAM, but you can expect it to get a decent overclock. When the RAM fails memtest, go into the BIOS and raise the Vdimm a notch.

There is a wide variety of actual memory chips that have been used on XMS3200C2PT memory. Without a revision number I can't tell you what that is. You're safe to a Vdimm of about 2.9V for most types of memory, Winbond stuff will be safe to far higher than that. XMS3200C2PT however is all rated at a RAS to CAS of 2. So when you listed your memory timings as being 2.5, 3, 3, 6 then that one at 2.5 can be changed to 2. Looser timings however does sometimes let you overclock the memory further.

Once you have eventually found the maximum speed of the memory, put the HTT freq. back to normal (but still not the multiplier) and put the CPU multiplier back to 12. Then start raising the HTT freq. to overclock the memory and the processor. Whilst you are within the stable limits of the RAM, just keep raising the HTT freq. and boot into Windows and run Prime95. When it fails Prime, go into BIOS and raise the Vcore a notch.

I know this sounds like a lot of work, but it isn't really. It can be time consuming, but you can get a close approximation to the best overclock within a few hours. To tweak every last little thing can take forever though. At the end of the day it will be worth it if you get an extra 25% from your hardware for nothing but a bit of your time.
 
Last edited:
Thanks for the big reply

Firstly, i sort of went through your process in different order.
I was trying to find the cpu limit first by doing this
1. Lower Ram to 266 (in MB) to get the RAM out of equation but keep the timming 2.5-3-3-6
2. Lower HT to x 3
3. Keep the multi of cpu to 12
4. Set Vcore to 1.5
5. Chipset 1.5
6. HT volatge 1.2
7. PCI exp 100
8 DDR 2.6 V
9 PCI clock 33 mHz

Now i start raising HTT from 200 with 5 at a time.
At Vcore 1.5 i got 226 stable
At Vcore 1.525 i got 230 stable (2760 Mhz)
At vore 1.5375-1.550 235 UNstable

Then i go back to 1-9 setting but LOWER cpu multi to 11
At Vcore 1.5 250 stable (2750 Mhz)
At Vcore 1.525 and bove 255 unstable

Then i go back to 1-9 setting but LOWER CPU mlti to 10
At Vcore 1.5 270 stable
At Vcore 1.525 275 stable (2750 Mhz)
At VCore 1.575-1.55 280 UNstable

At this point i run out of time

For RAM
1. Set at 400 (1:1) with 2.5-3-3-6
2. Lower HT to x 3
3. Keep the multi of cpu to 10
4. Set Vcore to 1.5
5. Chipset 1.5
6. HT volatge 1.2
7. PCI exp 100
8 DDR 2.6 V
9 PCI clock 33 mHz

I start raising from 200 to 225
I kind of not sure if i should go further so i stop at 226 (memtest run stable from 2 hrs 3 standard pass)

It took me 4 days to do these thing.
At this point feel that 275 on HTT and 226 on RAM is in pro overclockers area.
I need to consult this and go slow from now.

Please guide me more to maximise these setting, i My temp is still low MAx(49 on any set up above)
thank you very much
 
Last edited:
I'd drop the multi to something like 8 and then start pushing the HTT speed again. With the RAM set on DDR 266, you have a 2:3 ratio. So you could get the HTT to 300 MHz without having the RAM overclocked. You want to find it's maximum. You already know it will do 280, but that was probably maxing out the CPU, not the HTT frequency. You may need to raise the chipset voltage to get maximum performance.

Once you have the maximum HTT speed, you know how much harder you can push the CPU. You may find with more volts it'll go further. But if you've not found the max HTT speed, then you'll never know whether it is the CPU or the board that is the source of instability. It looks like the CPU is going to reach a limit at 2.75 GHz or so, at least without another voltage bump, but we need to isolate what does what.

Find the revision number for the RAM and I can tell you what to do with it, and what you can expect from it.

Remember, more ideal settings would be 9 x 300 HTT, with memory on DDR 333. That'd give a memory speed of 250 MHz. Same sort of CPU speed, huge increase in memory bandwidth.

You can get better yet from your hardware. All it takes is a bit of practice and experience, and we're here to help. We're not even expecting you to take soldering irons or scalpels to your hardware. Although that's always a possibility...
 
Biggles 266 said:
I'd drop the multi to something like 8 and then start pushing the HTT speed again. With the RAM set on DDR 266, you have a 2:3 ratio. So you could get the HTT to 300 MHz without having the RAM overclocked. You want to find it's maximum. You already know it will do 280, but that was probably maxing out the CPU, not the HTT frequency. You may need to raise the chipset voltage to get maximum performance.

Once you have the maximum HTT speed, you know how much harder you can push the CPU. You may find with more volts it'll go further. But if you've not found the max HTT speed, then you'll never know whether it is the CPU or the board that is the source of instability. It looks like the CPU is going to reach a limit at 2.75 GHz or so, at least without another voltage bump, but we need to isolate what does what.

Find the revision number for the RAM and I can tell you what to do with it, and what you can expect from it.

Remember, more ideal settings would be 9 x 300 HTT, with memory on DDR 333. That'd give a memory speed of 250 MHz. Same sort of CPU speed, huge increase in memory bandwidth.

You can get better yet from your hardware. All it takes is a bit of practice and experience, and we're here to help. We're not even expecting you to take soldering irons or scalpels to your hardware. Although that's always a possibility...

thanks for the reply
I will try 9x255 --> 300 today.

Here is the info on my ram
CMX 1024- 3200 C2PT XMS3202 v.1.2
XMS 3200 1024 400Mhz CL2

and after i get more info on my ram i will try and see how far my ram could go

Right now the CPU temp is going to 54 in "Oblivion" Mode, however i may need better fan CMF with low noise.

Cheers
 
I believe that revision 1.2 used Winbond CH-6 chips. That means they'll perform at their best with lots of volts. They'll probably do 2-2-2-5 timings if given enough and I'd expect 230 MHz at least from them. But you'll need a lot of Vdimm for best performance, probably more than your board can supply. You'd maybe need a DDR Booster to give the 3.6V to make them sing.

Either way, my point is, use volts. It won't hurt them and they will go further and tighter.
 
what is DDR booster ? is it a software ?
I have heard that DDR booster is for OCZ ram and they are hardware mod.

please advice.
 
With the above setting

1. Lower Ram to 266 (in MB) to get the RAM out of equation but keep the timming 2.5-3-3-6
2. Lower HT to x 3
3. Keep the multi of cpu to 8
4. Set Vcore to 1.5
5. Chipset 1.5
6. HT volatge 1.2
7. PCI exp 100
8 DDR 2.6 V
9 PCI clock 33 mHz

RAM
1. Set at 400 (1:1) with 2.5-3-3-6
2. Lower HT to x 3
3. Keep the multi of cpu to 10
4. Set Vcore to 1.5
5. Chipset 1.5
6. HT volatge 1.2
7. PCI exp 100
8 DDR 2.6 V
9 PCI clock 33 mHz

I try to keep the timming at 2.5-3-3-6 all time
I now can overclock my RAM at
258 Mhz With test 5 pass 2 times, test 6 pass time and test 8 pass 2 times
I was trying to rasie voltage up to 3.0 V but it make no different to the speed at all most of the test fail at test 5.

This is quite strange. Same thing happen at the cpu speed too

at reach 10x275 and 11x250 with Vcore 1.5 V and even i try to raise Vcore up to 1.55 V it wont go further.

Any more advice? pleasssss.....
 
Last edited:
Back
Top Bottom