How to OC the M17x-R2 using SetFSB

I would imagine the bios cuts out certain functions when rigged for ati gpus

Setfsb could work it's just a matter of coding .... Could soda a permanent over clock it's been done

 

I am trying this on my Asus n61jq with a 720qm and under my 10 my offset is 2F. Any insight on what this may mean (I do not know any coding, but may this represent another number, and if so how should I manipulate it)

 

the offsets we change are unique to the M17x and that specific PLL. This in no way translates to other computers, unless they use that exact PLL. The only other computer that does is the Asus G73JH

 

I did get mine working with different changes to the PLL. How do you apply these changes everytime the computer is restarted?

 

the same way you changed them In the first place. there's no way to save your changes once you reboot

 

You can write anbatch file and have it execute every time your computer starts

 

that's what I do, but it's easier said than done.

 

Do you guys know any software to adjust ram latencies in windows?

 

they exist (forgot the name - sorry) but you cannot change the cas latency which is the main one

 

can you please try to find a name?

 

These are my stable settings very disappointed
Maybe with some extra cooling I can go a bit further...
http://img813.imageshack.us/f/stable.png

 

Guys, I've been doing some testing and found the newest SetFSB using the Asus G73 PLL isn't 100% stable on my system no matter what clock I set. If I use my original method outlined in the OP by editing the specific registers I talked about, it's 100% stable.

 

Ouch, 143 fsb is a pretty weak OC. Did you up the voltage in the bios?

 

All of them all the way...
Maybe it's the ambient temperature that it is quite high... I don't know...

 

Yeah your temps are pretty high. You should consider some better thermal paste application. I can do 145 fsb stable on mine with stock voltage, pic coming in a min.

 

For what it is worth, your GFlops seem kind of low.....right now my laptop is configured as the cpu being a desktop Q9550 (8.5x 1333 FSB) with 4GB of DDR3-1333 Cas 7 ram and I got higher numbers

 

Mine finished slower than his with a higher bus. My multiplier drops down low though during full loads. If we're going to compare results, we all need to use the same program with the same settings. I'm doing 8 threads with max ram right now as well as running furmark + a very slight OC on the GPU's at the same time. I want to see if this system can handle all that load at once.

 

I applied OCZ freeze, maybe I should use the Shin Etsu that I have from the QX9300...

Could you run the intelburn test so I can have a better picture? (what's your ambient temp btw?)

 

Well we are doing the same LinPack, just a different GUI showing it off

Running FurMark may eat up some of that bandwidth though

Where that is going to hit your system is the power supply haha

update: will get the IBT so it is all comparable

Here is the IBT result - still getting a lot more bandwidth at standard. Let me run it with Max memory used now

 

Ok here's my pic:

 

For sure my i7 throttles like crazy...

 

How much does this test rely on memory speed? Mine's 1066 mhz cas 9.

 

Cas 9 at 1066 is SLOW....sure it aint Cas 7 1066? That is kind of the norm

Well I ran it at Max Memory and the GFlops dropped to the 33.5-33.9 range, still higher than what you are getting though. That being said, it is also taking me 165 seconds at 4 threads so not sure how to decipher this all. I do know that more GFlops is better though...

 

I'm running 8 threads on mine, try that with max memory usage. I thought all standard dell 1066 mhz ram is cas 9. I could be wrong though.

 

If I use the max memory usage on IBT a message pops up on the second run and says I am unstable. Damn (btw 89 seconds per run... )

 

Restart the program, set the number of threads to 8 first and then select max ram and it'll work.

 

It does not really make sense for me to run 8 threads...I have a Core 2 quad with 4 physical cores, you have a Core i7 with 4 physical cores + 4 logical cores making yours 8 total. We are both maxing out our hardware

udpate: Ran 8 threads to see if what I was saying is valid, turns out the end result is not much different (the CPU-z is showing speedstep kicking in, is regularly 8.5x for 2.83 GHz)

And to offer some sort of comparison, my desktop with an i7 965 (3.2 GHz and 1333 BCLK) and DDR3-1600 Cas 8 ram gets 43 GFlops

 

can you share your batch file?

 

I am reapplying some MX-3 right now and see what happens...

 

What are your results with stress level at max + 8 threads? I think this specific test relies on memory speed and that's why my numbers are even lower than statmatisx.

 

Remembered the name

Tweakers.fr

 

netted a 232 sec/22 gflops running 2 cores / 4 threads.
netted a 121 sec/48 gflops running 4 cores / 4 threads.

 

I assume that is your 980x John? lol

Joker, here is 8 threads Max Memory from my laptop

 

yes, but ran in dual core and quad core mode. cpu's are backward compatable.

it looks like he is seriously trotting out or not all cores are running at max speeds....

 

That is what I was thinking as well. Turbo Boost most likely has something to do with this. My numbers are just too high compared to the Lynnfield stuff....the integrated memory controller should be thrashing a 1333 FSB no?

 

They throttle down to 16x-17x at max load like they should because of the TDP limit. I think a lot of this just has to do with memory type and not the actual CPU power. Scook's latest run with 8 threads was slower than mine (I also had furmark running in the background, monitoring programs + firefox) in total time taken despite the faster memory and higher GFlops

 

I think the reason for the faster run despite the lower GFlops is simply the more advanced architecture running more efficiently. Your cpu can do more with less compared to my dinosaur

 

try running it at stock voltage and stock clocks and it will more than likely do the same thing.
my i3-330m and i5-430m we're jumping all over the place, when they should be locked at max multiplier....wonder it if this goes for the mx17.

what happen when you did it at stock?

ran the same test with furmark/prime/linx & intel burn in running. it lost 1 gflop and gained 1 second per time or 5 seconds over all..

and im inclinded to agree with joker..i think it's more of a ram bench than a cpu bench.....but that's just an opinion.

 

Well I definitely have an advantage in Ram speed over him at the moment but again, through a slower memory controller

 

yep.
i was running my ram at 530/ 7-7-7-20
you can try memset and change yours:
7-7-7-20
to
7-6-5-10 59

and see what happens

 

I have been playing with nvidia system tools to try to boost my fsb past 1333 but anything done while the ATI cards are being used results in a system freeze (not bsod)

Not sure how to work around that

 

msn it.....

and i must thank joker for this trophy!
1. 2134 mhz johnksss Intel Core i3 330M @2134MHz Stock 2.0
2. 2133.75 mhz oDDiTy Intel Core i3 330M @2133.8MHz Stock 1.5 OCClub
3. 2130.6 mhz racer Intel Core i3 330M @2130.6MHz 1.0 Hellas Overclocking Team
4. 2129.9 mhz oksboht Intel Core i3 330M @2129.9MHz Stock 0.8 PC Games Hardware
5. 2129 mhz logan Intel Core i3 330M @2129MHz Stock 0.6 PureOverclock
6. 2129 mhz TK-OC Intel Core i3 330M @2129MHz Stock 0.5 XFsatest
7. 2128.87 mhz Hampti Intel Core i3 330M @2128.9MHz Stock 0.4 Overclocking

 

If I use the maximum ram, it will throw a window saying that I am out of memory and I have either to close the program or if I continue it halts

with 2% OC on the BIOS and stock voltage
Imageshack - 2perc.png

It seems that with lower temperatures the throttling is less and I can maintain higher clocks, that's why I get better results.

Next step is to disable turbo and raise the BCLK with the setfsb as high as possible and see if I can get better results

 

I wouldn't worry too much about the IBT results. You just need to focus on getting your overclocks stable.

 

When I overclocked my 9800m GTX i had an option to load the modified profile at start up, you may be able to boot with the 9400m set your cpu clocks to load at start up then reboot with the Ati cards

 

I am able to use it with the ATI's I just have to make small adjustments. It will take a specific boot sequence and overclocking method to make it stable and work is what I have found so far

 

I get better Vantage results when I *don't* overvolt the 920XM. Done half a dozen tests with the voltage being the only difference and it's completely consistent. Heat seems to be the limiter with our beasts.

 

The overvolting will lead to more heat, but if you can manage that heat it will also enable a higher overclock. So as long as you can keep it cold enough, the overvolting will still be to your benefit

 

Any advantage of overclocking with this method on a 920xm over eleet?

 

Finer tuning than an entire multiplier.