How to Overclock ( noob edition )

ES or OEM?

 

ES but again 3.2ghz in my M17 all day long ?.

 

I didn't realize that the M17 used the MCP79 chipset as well.

 

They are not ... Pm45 on the m17

doesn't exain why quad gets 2.9 n man gets 3.3

 

Still not exactly the same thing.

 

Dont feel like getting into a discussion about the differences of OEM/Retail/ES/QS and the steppings etc . So ill just leave it at that .

 

Oh yes you do. Starting with post #29. http://forum.notebookreview.com/showthread.php?t=402927&page=3

 

From what I've read, the MCP79 isn't a very good chipset for overclocking. And from looking at the architecture of the M17 in general, the M17x is a completely different beast and they should not even be compared.

I do not expect newer BIOS releases to increase the overclocking potential of the M17x.

 

It has proven to be a great overclocker actually and better to date then any system i have seen including the Clevos,W90,Whitebook/M17.

I have never seen anyone hit 3.35 ghz stable or not on the M17 nor the W90 and not to sure about the 5797.

 

You're talking about unlocked multipliers. I'm talking about simply FSB overclocking.

 

No im talking about FSB overclocking as well because how do you think mandrake got his CPU so high ? multipliers ? nope.

 

I'm not going to get into it. The FSB taps out before 350, which limits the overclock. I don't see a BIOS update fixing that.

 

Lol Soviet i know all about this stuff but thanks and FYI mine is final stepping etc otherwise i would not have touched it with a ten foot pole.

 

Since purchasing my new memory I've been OC'ing via multipliers only. Before that I was changing multi/fsb/mem. With my x9100 it was just fsb/mem. I just keep it on 12 to get 3.2Ghz and I still keep a 1:2 memory divider ratio.

There so much more to it than just manipulating the fsb.

 

My point was for comparison purposes not the quality of the ES proc.

 

Yeah I know there is more to it but the FSB imposes a huge limit on how much you can OC.

I haven't tried playing with memory settings or anything yet. I just left it at 1066.

 

Which limit are you referring to? There is definitely a bandwidth limitation compared to the i7 architecture if that's what you are referring to. If you're comparing the same architecture then I'm not sure what you are referring to.

 

What do you mean? Are you saying that oem's overclock higher then es? Please explain

 

Send me a ES QX9300 and I'll let you know.

What I was referring to was Scook and I were getting similar results and we have the exact same setup. Maybe the ES procs have some kind of adverse affect in the M17x. Without having multiple samples side by side I wouldn't know for sure.

 

The FSB on the MCP79 has not hit 350 (1400) from what I'm aware. That means that if you have a locked multiplier, say 8.5 for example, the highest processor overclock you will get is going to be less than 2.9GHz. The highest that MY MCP79 seems capable of is 325 (1300) which on my Q9000 leaves me with a max clock of 2437.5

 

Got it, so you were referring to a specific proc. I was thinking more along the line of the Centrino 2 platform.

 

No I am just referring to the maximum front side bus limitation of the chipset being a deciding factor in how high of an overclock can be achieved.

I believe that Moo said the highest he could get was 347 and as we all know he has a super chilled machine so I don't see overclocking getting much better.

EDIT: Actually, Moo got to 362.

 

He had two cpu's. And that is why I asked him, I almost beg him to put the es in his unit.


You can rule out the oem overclcoking higher then the es because kciessem(don't know how you spell that) also had trouble goinng that high.

Jubal also got a hp oem qx9300 and it didn't budge past 2.93.

Quad gets 3.2ghz on the m17 and can pump it to 3.33ghz but can't past 2.93 in his m17x(this proves that his cpu has potentail but it's not working on the m17x for some reason.) quad has an oem

Axman also hit 3.2ghz on his oem on his whitebook but couldn't past 2.93 on his m17x..

Zfactor and alitunay should be able to give us some data on why it's on going higher then it should.



On a side note. Clauss came over my house and he had two m17x and with his t9600(I believe.) on his first unit can hit 3.33ghz and on his second was 3.1ghz ( or 3.2ghz) I wanted to swap them out for better testing but it wasn't mines. but this means that his cpu's were causing this.....

 

Moo is well beyond the specs of the proc. I don't think it's chipset specific in that case.

A 400 FSB would get us a better OC because we would be able to achieve a 1:1 FSB:Mem divider ratio but yes the MCP would probably fry at those speeds.

 

Quad just said he tested the M17x with an ES. As for the others I don't know what they tried. Unless I tested it myself I wouldn't be able to say for sure.

 

Buy another one.
But I posted what they had.
I looked over everything and i'm trying to narrow everything down but it's not there.

 

I've honestly never worried about the FSB:MEM ratio and just overclocked the FSB. I don't feel like digging for the CMOS battery.

 

stable I hit 352 on the fsb all day

I can hit 362 but it's unstable

 

The testing would have to be in a controlled environment. In the end it wouldn't really matter what they tried. Our machines are not 100% the same.

 

hold the right shift key while powering up your system to reset the overclock settings.

fsb:mem ratio has everything to do with how high you can OC.

 

I think I got into the 370s myself when I was trying for 1:1.

 

Well there is so much that can be controlled. Get me?

You have ram, cpu's, and that's really it. Maybe the same gpu's but that shouldn't matter....

 

The person doing it is different.

 

Thank you for the tip. Moo should add that to his first post because that is a REAL time saver.

Okay so what ratio should the FSB be to the memory? I doubt the stock DDR3 in the thing is going to go much past 1066... though they ARE Hynix chips...

 

Once again I correct you

I got my stock ram to ....... 1420 before crashing

 

You don't count with your super hyper cooling magical M17x! LOL

EDIT: Wait, you got the stock 1066 mem to 1420?! You can't change the memory timings in the BIOS that I can see so how on Earth did you achieve that on C7?

 

Okay this machine impresses me more and more... it changed the CAS setting automatically lol. Wish my Bad Axe 2 was this easy... I'm not going to miss all the hunting for the reset jumper... that thing was next to impossible to pull with the board in an Antec 900.

 

Best way to answer this is test. For me 1:2 has been the best so far. The ram will go past 1066 easily but you start getting worse timings. It all depends on the spd of the particular module.

His would have been CL11 during his test.

If you get better memory you can achieve higher speeds without losing too much in timings. I'll post my spd in a sec.

 

This is what my timings look like at different speeds.

 

Lol thanks for that low blow . JK i know your not referring to any one person in that comment but i have literally tried every combo imaginable with no success . Im throwing the Qx9300 back into the M17x now with the A02 bios to see if i make any more headway .

I told don this awhile ago about it could be a machine to machine thing vs just the chip and some machines just might not let you go that high and it could very well be just that .

 

I don't understand the ratio... I'm running 1285:1285 right now... Obviously kicking my CL up.

If I want my FSB at 1285 (the most stable), what should my RAM be? I'm confused. I never really messed with the RAM before since I had a Bad Axe 2.

 

Yeah mine hits CL8 at 609.

 

lol, no I wasn't trying to talk down about anyone. Just as soon as you change the person the testing will be different. If someone had 3 ES procs and 3 OEM procs they could do a much better test.

You are right about each machine being different. If the theory is that each CPU will act differently then what about each MCP or memory module? That's why the testing has to be controlled.

 

That's really the part to test. The M17x comes either 1:2 or 4:10. If I had my fsb at 1285 then I would put my memory at 1285.

 

Well I definitely don't know if CL11 is worth the overclock... Seems like they'd cancel each other out.

Doing a Vantage run at 680/950/1700 to test...

So now that my FSB and RAM are 1:2, you're saying I should theoretically be able to get a higher CPU overclock?

 

CL11 seems excessive to me also. That's why I went out and bought new ram. Bump your GPU memory up to at least 1025 for your test. I test a lot with my mem at 1065.

1:2 gave me the most stable results but I don't have a q9000. That's why you need to see what works best.

 

Thanks for the help Mandrake. I came pretty close to your Vantage score... Lets see what 06 has to say.

I guess I'll try pushing my clocks back to 2.5 again and trying to keep the memory as close as possible and see if I can get a stable overclock. I really need a QX9300 but its not in the budget right now.

 

Really? My score is with PhysX off. If your score is also with PhysX off then you need to shoot for Scooks score which is 1228x.

 

Quad I don't think it's the mobo. Your p8600 hit 3.0ghz/1333fsb/333 bus speed. And with the qx9300 it can't pass 2.93ghz. So your mobo is fine.,

The only incident we had was someone got there mobo to 1200fsb and the mobo fried.

Case rear. It only happened once. On nbr at least.
If the mobo was the limt, there would be alot more fried mobo. This is certainly true since everyone wants to run occt on there notebook.

Therel- alot more but my hands are cramped on this phone.

 

It's not that cut and dry. Different proc will act differently. Even the heat output will be different. 1 person, 1 system, multiple procs. The only way to test this.