• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Strive to get OVER 3ghz, but dont understadnd 3:4 setting

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

oc jason

Senior Member
Joined
Feb 12, 2001
Location
Fayetteville, AR
Currently my 2.26b can boot at 17x176, that is at 2-5-2-2, vcore +15% and 2.7 mem v. Now on other AMD systems the mem has been able to boot up all the way up to 220mhz fsb, everyday setting was 185mhz fsb. In my Intel system it does not want to go higher then 176. I assume this is due to the locked multiplier, but I know the chip has more as the majority of 2.26b that I have dealt with and have seem overclocked usually top out about 3200 so im pretty sure there is more in store in the chip.

My 2 options for getting more out of it are 1: Vid Pin Mod for more vcore, but I dont think that is the issue personally. 2: Find a way to overclock the FSB more with the 3:4 setting.

On my BD7II I have the custom BIOS done by someone that I cant remember his name but he is very known for making great bios. 17x176 2-5-2-2, +15%vcore , 2.7mem, CPU is By SPD, and set to 1:1.

I am under the impression now that the 1:1 ratio means the CPU and the MEM run at 176 correct. So if 176 is the limit of the chip than I will not be able to go any higher, but if i set to 3:4 than the chip only runs at 75% or 3:4 of the mem speed? is that right?

At at 3:4 if I know my mem will do 200 and my chip will do 176 then I cna set mem at 200 3:4 and my chip will be running at 150 ?
 
At at 3:4 if I know my mem will do 200 and my chip will do 176 then I cna set mem at 200 3:4 and my chip will be running at 150 ?

I started to loose you untill this....

btw, yes you could run your chip at 150fsb and that would make your mem run at 200.

mica
 
The 3:4 memory setting is the memory bus divider. The first number in the ratio is the FSB and the second is the memory. Think of it this way 3/3 to 4/3, which means the memory bus runs 4/3 as fast as the FSB (or in other words, the memory bus runs 1.33 times faster than the FSB). For example: 150 FSB X 1.33 = 200 MHz memory bus (like Mica said).

Not sure where you got the idea that most 2.26 CPUs will do 3.2 gig, I'd say probably only half of them will even reach 3 gig. You might be reaching the limit of your CPU.
 
Agreed, but for a few minor terminology points.

3:4 is a the cpu:memory ratio. It is exactly true that for every 3MHz of cpu clock, we produce 4MHz of memory base frequency, which is then effectively doubled by the DDR's properties. Another way of saying the same thing is that we have a 1.33 mulitiplier in effect.

Two ways of doing the math here. One in terms of ratio, and one in terms of multiplier.

(176MHz) x 4 / 3 = 234.66MHz

(176MHz) x 1.33 = 234.66MHz

This shows the equivalence of these terms, as 4/3=1.33 and multiplication and division are communitative, allowing us to do the 4/3 division first without changing the result.

As such your DDR would have to run at almost 470MHz to run the 3:4 ratio at 176fsb. While this is possible, you are not going to do it at 2.7V on tight timings.

I agree with Batboy entirely on the potential of a 2.26b. Very few exceed 176fsb, and those that do require a lot more than the ~1.64V acual Vcore the BD7-II delivers when set to +15%. While you can vid pin mod the cpu for more voltage, the memory speed is more important. Lowering the fsb until 3:4 works will yield the best performance. Relaxing the memory timing will allow this to happen at a higher fsb. Try 2-7-3-3 on the memory and back that fsb down until 3:4 works. While it may break your heart to admit to running less than 3GHz it will produce better performance. The fact that running 1:1 doesn't allow more than 176fsb points to the fact that that is all your chip will do on the voltage it is running on. Relaxing the timings to make sure the ram is not limiting the fsb is the only other test needed to demonstrate your CPU is at its limits.
 
Thanks Professor Larva, you are correct, 3:4 is indeed the FSB to memory ratio (I got the terminology slightly mixed up in the first sentence of my last post). Dr. Larva is also correct in recommending max memory bandwidth rather than max CPU overclock, assuming you want the best performance from your system.
 
oc jason said:
but I know the chip has more as the majority of 2.26b that I have dealt with and have seem overclocked usually top out about 3200 so im pretty sure there is more in store in the chip.



According to the database the 2.26 Average overclocked speed is 2858 Mhz with the highest at 3289 :rolleyes:


oc jason said:


On my BD7II I have the custom BIOS done by someone that I cant remember his name but he is very known for making great bios.

You must be refering to Mr.Natural ;)
 
batboy said:
Thanks Professor Larva, you are correct, 3:4 is indeed the FSB to memory ratio

Heh, I make the same mistake at times too. It is my early new year's resolution to never again utter the words "3:4 mulitiplier" :D

I am also trying to swear off the bad habit I have of typing "where" when I mean "were", but this has proven more challenging than quitting smoking...

While these are minor points confusion regarding the basics of PC's is compounded as we try to analyze the more complex aspects present. Now if I could just get people to realize the DDR IS SDRAM, rather than referring to the two as seperate entities when they mean the difference between SDR and DDR SDRAM :)
 
Last edited:
(176MHz) x 4 / 3 = 234.66MHz

(176MHz) x 1.33 = 234.66MHz

not that im a big mathamatician but doesnt BODMAS show that the second one you put on there is irrelevant...

you should have put (176 x 4) / 3 to make the first method show true...
 
ninthebin said:


not that im a big mathamatician but doesnt BODMAS show that the second one you put on there is irrelevant...

you should have put (176 x 4) / 3 to make the first method show true...

Since there is no order of operations priority between multiplication and division it does not matter if you multiply the 176 by 4 first or do the 4 divided by 3 division first. Try it both ways if you don't see the point.

The whole point of the second expression is to show the equivalence of the two expressions. You can calculate the result in either of the two ways you prefer.
 
no its not that - I could see what you were getting at with it - it was just that its division first so the second version of 1.33 isnt really the second version its just the logical working out from the first

but then if the DM goes in no particular order I guess not - as I said im no mathematician
 
oc jason's memory is at 200 Mhz not to bad.

PC3000 = (183 MHz Operating Speed) x (2x Rising & Falling) x (64-bit Bus) / (8 bits per byte) = 2928 MB/s available bandwidth.
 
Back