p4b fsb:mem ratios

[Zeus]

Ive got a bit of a problem with my board and how its clocking the ram.

If I add pc100 at a fsb of 120 and a fsb:mem ratio of 1:1, I get a mem speed in cpuid of 160mhz.
If I add pc133 I get the same.

basically its giving a ratio of fsb:mem of 3:4.

The bios specifically says for the ratio auto,1:1 or 3:4 but I always get 3:4.

As most of my ram (640mbs) is pc100 I use that and not the 128mbs of pc133.

Ive updated my bios today, this didnt help.

Has anyone got any ideas about why I dont get the 1:1 ratio I select?

Also, I set my fsb to 126 (cpu=2520mhz from 2ghz) with pc100 (@168) and it works fine. Its completed the tests I give it anyway.

But it wont reach a fsb of 127. Well it does but it aint stable when benchmarking. Now Im persuming its my ram but when I add the pc133 it still isnt stable. The ram timing settings are set to the slowest and the voltages are cpu=1.625 and mem=3.5v.

Is there a way ( apart from setting my ratio to 1:1 )to make this more stable? For some reason my cpu dont like 1.65v though the temp is only 50ish at full load.

Im really confused about this memory thing. 168mhz stable as a rock with pc100 seems a little long in the tooth, but thats what cpuid says, not me!!

Cheers for any help guys

 

[Grande Juan]

That's about where mine craps out at also, using pc150. So I'd say that is some good memory. Maybe turn up the juice going to your memory. It didn't help me though. Also I have a hard time getting much over 128fsb even with relaxed memory timings. Same thing if I try using jumper mode to.

 

[Chicane]

Re:

Trying to OC P4 1.5GHz in P4B. At anything higher than 110FSB the scores in 3DMark2001SE are lower than at 110. However, they pass fine without errors. Any idea what causes the lower scores? CPU temps hover around 45*C. Heatsink is SLK-900U...

 

[Grande Juan]

This mobo needs all the memory bandwidth it can get, thats what it needs period. So I think (this worked for me) run your memory as fast as you can with everything else at default, then up the fsb as far as you dare, corrupted files? Oh yea WELCOME CHICANE your temps seem reasonable, must be some sort of divider kick'n in. Check your bios memory settings "auto" setting could be the culprit.

 

[Zeus]

If grande juan can only about 128, and I get a max of 128 but 126 stable could it be the board? It says it only supports 2.4ghz max, I run at 2.5ghz but it doesnt like much more.

What cpu you got GJ?

O and Chicane, I used to max out at 110, it never used to start at 111.
Put your fsb:mem ratio to 1:1 and up the core voltage. Im not sure what to cos my cpu is very different from yours. Its a P4a, introduced at 2ghz. You should be able to get past 110 though.

Its definatly the agp and pci ports which stop you from going further than about 128. You can lower the pci:fsb ration to 1:4 with the jumpers but you cant increase the core voltage when you use the jumpers.
I dont know whats better......

by the way, sandra benchmark tool gives the memory reading of the fsb when the ratio is 1:1, so it might be cpuid messing up not the board

fsb 128
mem 128/170.6
pci 42.6
agp 85.3

or
fsb 120
mem 120/160
pci 40
agp 80

cos they are the approx maxes with this board. Regardless of pc100, pc133 or pc150, mem voltage or cpu.
Not unless chicane with his 1.6ghz can reach the next increment on the jumpers.... 125mhz with 1.5v core voltage. My 2ghz certainly cant.

 

[Zeus]

Chicane, Ive just found out you can get a 1.6 and 1.8 P4a. If you've got one of these you will beable to overclock quite a bit! More than 110 anyway.

Use your bios overclocking things first but if its looking really good try the 133mhz fsb with the 1:4 pci:fsb ratio.

It may well be your ram which stops you and not your core voltage.

 

[Chicane]

Re:

Thanks, guys. I will try playing around with it some more and see what happens...

 

[Chicane]

Re:

Well, I got some good news. I've decided to try higher FSB speeds than 110MHz and the score jumped up! It would be lower up to 115 but then I decided to try 120, and it would jump back up higher and higher up until 127MHz FSB at default voltage! So, this 1.5GHz is running at 1905MHz now. Higher FSBs than 127 don't seem stable yet even with increased voltage but maybe after a burn-in it might. Still, knowing what a bad overclocker 1.5GHz is it's pretty amazing it's gone that far

 

[Chicane]

Re:

Just a small update -- after several more tests the CPU now seems to be stable at 128MHz FSB and 1.825volts. I guess it "burned" itself in.

 

[(HR)Titan]

not necessery

 

[Chicane]

Re:

Yet another update -- it now runs with 3:4 CPU-to-Memory ratio, which improved the 3DMark2001SE score by 700 marks!

Titan, what isn't necessary, to burn it in? It seemed to make a difference...

 

[Zeus]

Thats a good overclock on your ram!
What kind of temps are you getting with 1.825v?
Its quite high, but I dont know what they start at.
Your agp and pci ports holding out too. 128fsb aint bad at all!
28% overclock! If you can make 133fsb you can use the jumpers and change the pci divider to 4. Then set your ram to 1:1
That way you can have a quick fsb, with normal ram and port speeds.

If it gets tough to go any further it will be your ports, but running at 3:4 cpu:mem ratio you might start getting held back on the ram. Of course you can up the mem voltage too

Thats a good overclock! I cant go past 124 if I want to play games, it wont benchmark at 126, 127 is not very stable and 128 struggles to boot. 126,127and 128 is tempremental on passing post too. Im stuck at 120 for rock solid and 125 for my best stable overclock.
My main problem is not getting enough power to the cpu when using the jumpers. I need to use the jumpers to get past 125mhz fsb.

no matter what I buy I wont get past 125mhz, not unless I get a new motherboard and ram Im sick of the rams bandwidth!

 

[Grande Juan]

I started with 1.5 willy now using a 1.6 northie both chips overclocked to the same fsb 128 stable. The 1.6 did run at 134 fsb with relaxed memory settings, however the performence is better with cranked memory, and a lower fsb, on this mobo anyway. Even using the jumpers (thus locking agp and pci frequencys) I reached the same results.