Overclock.net banner
1 - 13 of 13 Posts

Smash91

· Registered
Joined
·
19 Posts
Discussion starter · #1 ·
Hello Overclock.net,

CPU:
AMD Athlon 64 4200+, @2.2Ghz (939 socket), Core Manchester. Today i`ve tried to break 2.5 Ghz but when I`ve entered OCCT ... bang !! BSOD. I just rised the FSB to 226 and set my multiplier to 11x. Now I`m back to stock, until I get a good OC configuration.

http://valid.canardpc.com/show_oc.php?id=1265363

Recently, I`ve replaced my stock cooler with a Cooler Master TX-3 and my full-load temp is under 50*C. I have Cool`n Quiet feature activated, and in idle, my cpu temp is somewhere around 29*C, and without CnQ, idle temp is around 34.

I want to break the 2.5 Ghz, but I need some help from you guyz.

Later Edit:
My Mobo: AsRock 939n68pv-glan
I must mention that i have 4 sticks of 512MB DDR1(2 Kingmax, 2 Nanya).
 
As a reference, I used to run 2.7GHz (245MHz x11), 1.45v, memory 1:1 using stock speed G-skill PC3200. But that was on low-end H2O with a DFI Lanparty Ultra-D (arguably the best overclocking 939 mobo ever made)

Try raising your vcore to 1.4v and see if you can jump up to 235MHz x11. just watch your temps.
 
Lower ht link to x4(800Mhz)

Fsb to 250

Multi to 10x

Lower ram divider one notch or two to get close to 200Mhz(400Mhz dual)

and try 1.4v like Phille suggested.

Also note that 4 sticks of ram will make it harder to OC than 2 sticks.
 
Discussion starter · #4 ·
I think this video will help me:

film_go.png
YouTube- Overclock X2 4200 3Ghz 10.000+ 3dmark06

But with the FSB at 250. Should i try step by step ... Fsb 225, 10x multi, 1.376 Vcore ... Fsb 250, 10x multi, 1.376Vcore ... Fsb 250, 10x multi, 1.45 Vcore, or jump right to Fsb 250, 10x multi, 1.45 Vcore.
PS: I really don`t know how to work out with the RAM Divider, like MrTOOSHORT told me. I`ll post some pictures with my Bios config soon.
 
HT is unhappy over 1000 mhz(200*5). When you raise fsb over 200, HT goes over 1000 with 5x so lower to 4x. This is most important.
Increase ram and cpu voltage a few notches.
Lower ram divider if it sill don't get stable. 2.5ghz should be easy if stock is 2.2ghz.
 
Quote:

Originally Posted by lurkingdevil View Post
Would you mind reducing the image size and reuploading. Each one is like 2.5mb in size.
Look fine to me.

To the OP, don't be scared. Bump that vcore to 1.40 or even 1.45, raise your FSB to 225 or 230 and see where you are. You will need to change your voltage to get any sort of decent overclock from that chip, so you might as well get used to doing it now. Check your temps and make sure your fans are moving air fast enough to cope with the extra heat from the voltage bump. The HT Link is important only to more picky motherboards, so don't be so concerned with that right away. I used to run mine at 1150 from the stock 1080.

Did I mention don't be scared?

Anf PROBN4LYFE, you don't need to be a showoff.
Image
I could never get that speed out of any one of the 4- socket 939 chips that I owned.
Image
(2x 4400s and 2x 4200s)
 
Here's what you need to do.

250 HTT
10x multiplier
whatever vcore is needed for stability (mine uses 1.425v for 2.5GHz)
133 memory divider (yes, THAT low if you're using 4 sticks)
4x HT Link multiplier

That SHOULD make you stable. Also make sure any spread spectrums are disabled. Make sure your PCIe clock is at 100MHz (not auto), and PCI clock is at 33MHz (if its an option). Also disable Cool N' Quiet. I had nothing but issues with my Manchester with it.

Also, don't expect to pass 2.6GHz. Most Manchester cores won't. They are NOTORIOUS for hitting a wall between 2.5-2.7GHz (depends on how lucky your chip is). Mine takes 1.65v to hit 2.65GHz.

EDIT:
And the reasoning behind you HAVING to run your memory that low:
Socket 939 has a little quirk. The memory controller CANNOT handle DDR speeds past DDR333 without some EXTREME tweaking when using 4 sticks of memory. You are also limited to a 2T Command Rate...PERIOD. You could POSSIBLY run your memory higher, but try for DDR333 to start out with. And loosen your timings up to 3-3-3-8-2T for testing.
 
I run a athlon 3200+ at 2.45 ghz.(2ghz stock) at 1.58V on stock cooling, works just fine. So don't be afraid to bump the voltages. I've also done a +0.1V on NB, for added stability.
Disable cool 'n' quiet in bios.

Quote:

Originally Posted by pioneerisloud View Post
Here's what you need to do.

250 HTT
10x multiplier
whatever vcore is needed for stability (mine uses 1.425v for 2.5GHz)
133 memory divider (yes, THAT low if you're using 4 sticks)
4x HT Link multiplier

That SHOULD make you stable. Also make sure any spread spectrums are disabled. Make sure your PCIe clock is at 100MHz (not auto), and PCI clock is at 33MHz (if its an option). Also disable Cool N' Quiet. I had nothing but issues with my Manchester with it.

Also, don't expect to pass 2.6GHz. Most Manchester cores won't. They are NOTORIOUS for hitting a wall between 2.5-2.7GHz (depends on how lucky your chip is). Mine takes 1.65v to hit 2.65GHz.

EDIT:
And the reasoning behind you HAVING to run your memory that low:
Socket 939 has a little quirk. The memory controller CANNOT handle DDR speeds past DDR333 without some EXTREME tweaking when using 4 sticks of memory. You are also limited to a 2T Command Rate...PERIOD. You could POSSIBLY run your memory higher, but try for DDR333 to start out with. And loosen your timings up to 3-3-3-8-2T for testing.
Why do you need to lower the multiplier? How does it help? Not arguing, just asking.

I have 4 sticks of ram.
In my experience, I couldn't get above ~227 HTT with a 200 memory divider. When lowered to 166, I was able to get 245 HTT. That puts the memory speed to around 202(200 stock). Do you think I could push the processor further with divider set to (133)?

Since I could push the memory to 227(with 200 divider), I figured that was the maximum. With 166 divider and 245 HTT, memory is at 202 well below the maximum(that I found to be 227) and further increase in HTT result in instability. Do you think I've reached my wall? Vcore is at 1.6, I think.
 
Quote:

Originally Posted by pioneerisloud View Post
Here's what you need to do.

250 HTT
10x multiplier
whatever vcore is needed for stability (mine uses 1.425v for 2.5GHz)
133 memory divider (yes, THAT low if you're using 4 sticks)
4x HT Link multiplier

That SHOULD make you stable. Also make sure any spread spectrums are disabled. Make sure your PCIe clock is at 100MHz (not auto), and PCI clock is at 33MHz (if its an option). Also disable Cool N' Quiet. I had nothing but issues with my Manchester with it.

Also, don't expect to pass 2.6GHz. Most Manchester cores won't. They are NOTORIOUS for hitting a wall between 2.5-2.7GHz (depends on how lucky your chip is). Mine takes 1.65v to hit 2.65GHz.

EDIT:
And the reasoning behind you HAVING to run your memory that low:
Socket 939 has a little quirk. The memory controller CANNOT handle DDR speeds past DDR333 without some EXTREME tweaking when using 4 sticks of memory. You are also limited to a 2T Command Rate...PERIOD. You could POSSIBLY run your memory higher, but try for DDR333 to start out with. And loosen your timings up to 3-3-3-8-2T for testing.
I would run it with 1GB so you can utilize the 1T command rate.
I wouldn't run the 10x multiplier either just so you can stretch it a bit more...9x or 9.5 if you can with x3 HT link. running 4x might limit your board to 1000 (and a lame 2.5ghz overclock)
Image


And @philly...I wasn't being a showoff, the link had valuable info such as timings and multiplier I used. Seeing as it was deemed a "golden stepping" afterwards (lol), I sold it shortly after
Image
!
Any other info I have on it can be found via my submissions to hwbot dude
Image
.
 
Quote:

Originally Posted by lurkingdevil View Post
I run a athlon 3200+ at 2.45 ghz.(2ghz stock) at 1.58V on stock cooling, works just fine. So don't be afraid to bump the voltages. I've also done a +0.1V on NB, for added stability.
Disable cool 'n' quiet in bios.

Why do you need to lower the multiplier? How does it help? Not arguing, just asking.

I have 4 sticks of ram.
In my experience, I couldn't get above ~227 HTT with a 200 memory divider. When lowered to 166, I was able to get 245 HTT. That puts the memory speed to around 202(200 stock). Do you think I could push the processor further with divider set to (133)?

Since I could push the memory to 227(with 200 divider), I figured that was the maximum. With 166 divider and 245 HTT, memory is at 202 well below the maximum(that I found to be 227) and further increase in HTT result in instability. Do you think I've reached my wall? Vcore is at 1.6, I think.
You're also comparing a single core to a dual core Manchester
Image
. I've got a single core Venice here, and at 1.70v it loads at a measly 50*C. With that Manny under the hood though, my cooling can BARELY handle 1.45v right now.

And you must have a GOLDEN memory controller on that chip to be able to pull off 4 sticks on the 200 divider. That could be why you needed such high voltage. You probably could pull more CPU speed if you dropped your memory down.

And what's wrong with 250x10? That's already pushing his poor little ASRock board pretty hard. I have no experience with ASRock boards, so I honestly don't know how far the HTT will go. But I wouldn't think it could possibly go much higher than 250.

I was only trying to point out what SHOULD be a fairly simple and stable clock. The OP could eventually work himself up from there once he figures out what all the settings do.

EDIT:
@PROBN4LYFE:
You want to see an insane 939 clock speed? Check my sig
Image
. The very bottom of it in orange.

EDIT2:
I'm not "The Head B00B!" of the Socket 939 club for nothing you know
Image
.
 
1 - 13 of 13 Posts