Benefit at running Q6600 multiplier at x9 instead of x8

skiltrip

Distinguished
Jun 3, 2008
20
0
18,510
I recently built a Q6600 machine on an Asus P5K with G.Skill DDR2 800 Ram (4GB - 2GBx2). Using Arctic Cooling Freezer 7 Pro as my heatsink/fan.

I easily got my Q6600 up to 3GHz by upping the fsb to 333, and just adjusting my DRAM back down to 800. Idle it's around 45c in CoreTemp, under max load in Prime95 it sits around 64c. (This is with low speed fan settings)

As that was so easy, I can't help but be curious about getting it up to 3.2.

So I was reading a few other posts on here, and it was suggested to set the fsb at 400, multiplier at x8, and that keeps my RAM running at the stock 800 at 1:1.

Some other posters seemed to "prefer" the x9 mult. What is the benefit to using x9, or to put it another way, was is the down side to using x8?

And if I did go the x8 route, which voltages might need a boost? Currently at 333x9 I don't need to touch any of the voltage settings and it runs in Prime95 for hours and hours with no errors.

Thanks!
 

Lupiron

Distinguished
Feb 9, 2008
1,711
0
19,780
First off, there is no difference. You just use the Multiplier to adjust speeds and other junk as you see fit.

But running a higher FSB is automatically faster. So x 8 @ 400 FSB is way better than 350 x 9 or whatever.

It's simply due to the fact that your FSB is your data pathway to and from all main system memory.

If that is the case, then that means your 333 FSB can support only 667 Mhz of data. (Double data rate!) Yet you have your RAM running at 800 Mhz. It's so nice of all those memory dividers to cloud your vision!

Logically thinking, it would be pretty hard to pack 400 Mhz of data into a 333 Mhz bus. So that means the higher the FSB, the beter your system will perform. You'll note that you can not run your memory Under the FSB's speed... Hmm, suspicious!

BTW, if your Voltages are on Auto, you are prolly getting more voltage than you need to those component locations. The Auto setting tends to overvolt, it's to stupid to adjust for other factors, so the higher the better!

If you wanna Down Load and run the newest Core Temp release, and tell me your VID, I'd be able to tell you the Loaded VCore value you'll need for 3.6 Its quite easy, after all.

--Lupi
 

skiltrip

Distinguished
Jun 3, 2008
20
0
18,510
Could I safely do 3.6 with an Arctic Cooling Freezer 7 Pro air cooler? I used the MX-2 thermal paste that's applied stock to the cooler, and didn't lap or anything fancy.

You made some interesting points... I have a question... So when I'm running at 333x9, and my ratio allows me to keep my ram at 800, am I being "fooled" into thinking my memory is running at full speed or something? Is that why people always recommend the 1:1 ratio?

So if I was going to go for 3.2GHz, running at 1:1.

I could do it at 400x8 with my DRAM set at 800, is it that simple, or am I missing something.

Of course, If I'm understanding correctly, some voltages would need to be tweaked, or could I leave them on auto because they tend to overvolt anyway?
 

cyborg28

Distinguished
Aug 2, 2007
232
0
18,680


Sorry to hijack the thread, but I am in a very similar situation and can't do much with FSB 400. I have a P5k-VM with Q6600, and 4x1GB DDR2 800. My VID is 1.25, and I am using (CPU-Z) 1.24 V Idle, 1.20 V Load with 333 FSB. I Get idle T=~37 C and load T=~50 C (Prime95 & CoreTemp).

I can Run FSB 400 x 7 = 2.8 GHz all day and night, but cant get FSB 400 x 8 even at 1.325V Idle 1.285 Load.

How much voltage can safely be put through the processor? Will I ever get 3.2 or 3.6 GHz? Will I fry the chip if I do.

I have tried a VDroop Pencil Mod to lower my boards Idle volatages, but it did not seem to work. Others report the board I have can have Vdroop almost eliminated but I still get .04V of Droop.

Please help!

Cyrus
 

Lupiron

Distinguished
Feb 9, 2008
1,711
0
19,780
You could prolly leave them on Auto, but really just the North bridge needs to be set around 1.45 to support the higher FSB, and it is that simple. 400 x 8 = 3.2 in sync at 400 Mhz with yer DDR2 ram.

And it really is as simple as it sounds. If your ram is running at 400 Mhz and your bus to and from that ram is running slower, how can you get the faster ram through the slower bus? You simply can't. test it yourself. Its not a HUGE gain, but it's a free decent gain in performance.

Enough that you should Try for 410 FSB after 400 stable, and saving it to your Over clock Profile in the bios. Then Go for 410 FSB in sync and test yer rams read, write and latency. It gets better a lil bit each pass.

Then try 415 FSB for in sync at 430Mhz... any decent brand 800 Mhz ran will get that, just up the voltage by .05 at 415-425, .05 again for a bit more.

You also may want the NB to 1.48 or so for 420+ FSB and up the PLL to 1.60 or so.

Please Down Load and run Core Temp and list the VID of your processor. You'll be glad you got core temp anyways!

--Lupi
 

Lupiron

Distinguished
Feb 9, 2008
1,711
0
19,780
Wanna list the VID of your processor as well, please??

And I am currently doing just as I said to do, BTW, blending @ 428 FSB with my RAM in sync at 4-4-3-15 @ 2.15 volts.

With a few other mem enhancements. oner 8500 Megs, copy.

--Lupi
 

iluvgillgill

Splendid
Jan 1, 2007
3,732
0
22,790
as lupiron said running a lower multiplier will allow you to have higher FSB at the same speed with higher multiplier.so memory subsystem will have a higher performance increase than lower FSB.

but running a higher FSB will subsequently put higher strain on the mobo.but i doubt the mobo will die with 1600FSB.

running the Q6600@3.6Ghz witl AC Freezer 7 Pro is not always impossible.but in my case it wouldnt even do 3.2Ghz.the core temp just simply too high to even look at.
 

Lupiron

Distinguished
Feb 9, 2008
1,711
0
19,780
See, take a peek!

Memorytesting.jpg


I wanna find the edge of FSB stability for the RAM at these timings and levels.

been going up from 810 Mhz to this 856 Mhz, and it smokes!!

--Lupi
 

skiltrip

Distinguished
Jun 3, 2008
20
0
18,510
Thanks for all the detailed responses Lupi. It's greatly appreciated. I do have the lastest CoreTemp, so I'll post my VID tonight.

Never mentioned my ram timings... factory stock it's 5-5-5-15, so I'm not sure if that equates to a much lower ability to be pushed compared to 4-4-4-12 timings.

I'll see if I can't get it running at 3.2 tonight. Wouldn't that be exciting!
 

skiltrip

Distinguished
Jun 3, 2008
20
0
18,510


I'm pretty sure the P5K is ok up to a 1600fsb. There's a sticker on my motherboard box that says 1600FSB! or something like that. I'll double check when I get home.
 

Lupiron

Distinguished
Feb 9, 2008
1,711
0
19,780
Sure, these chips can handle a lot more than the common rumors mention. Hell, I accidentally booted into Vista with my VCore at 1.8125 in the Bios, was only 1.78 in windows, though.

And on a B3 q6600 at that, looked fine, just 80c + cores.

So I dont think my 1.39 will harm it much. After all, Intel warranties them to run in spec at .85 - 1.50 volts. The part that Voids the Warranty is the frequency modification, or as we like to call it, over clocking.

So if its safe to run it at 2.4 @ 1.5000 volts and it will last intels 3 year warranty period, I sure am not scared of less volts and more speed! You should check out my two entries in the over clocking competition that was kinda going on. There is some voltage and speed. And Rather good Voltages for the Speed!

Skiltrip, 400 FSB is common, most boards do it. All 333 boards, pretty much get 400 Mhz and more, so 400 FSB boards are good for 450+ Just takes some minor tweaking, or sometimes, oddly, not much tweaking, like the P5n-d 750i.

--Lupi
 

iluvgillgill

Splendid
Jan 1, 2007
3,732
0
22,790


oh i just saying that in theory it will.its like running a engine at 3000rpm compare to 2000rpm.which 1 will last longer.i think you got the answer.
 

iluvgillgill

Splendid
Jan 1, 2007
3,732
0
22,790


shadow i pretty much agree on your comment.but has anyone you encounter have reported their CPU is dead due to a high overclock.in my case there hasnt been anyone.
 

skiltrip

Distinguished
Jun 3, 2008
20
0
18,510
So I am running Prime95 right now, have been for about 1/2 hour, and so far, no problems...

I'm at 400x8
3.2GHz
RAM is at 800
1:1
Set NB voltage to 1.4v
All my other voltages I left at AUTO

In CoreTemp VID is 1.3250v

With all cores at 100%(prime95)...

My CPU temp is riding right around 50c
Core temps are riding around 58-62, highest core usually around 60-61, lowest around 58.

Anything else I should change, or if I'm able to run Prime95 with no errors am I good??
 

iluvgillgill

Splendid
Jan 1, 2007
3,732
0
22,790
your VID is very high for G0 Q6600.but to comfort you i got the same VID Q6600 as you.

and from my experience in my case you wont be go any further then 3.2ghz.whats your load temp for the cores?i would guess its on 70C.i had the same VID as you and same cooler as you.but at the end i switch to a Zalman 9700 and achieved 425X8.and load temp is 70C.
 

iluvgillgill

Splendid
Jan 1, 2007
3,732
0
22,790
you should run prime for 8 hours before declearing its stable and OC further.once i had prime and fail at the 17th hour.so you see?but 24hours prime is for the final final OC that you want to keep.
 

iluvgillgill

Splendid
Jan 1, 2007
3,732
0
22,790
yeah the value is suppose to move for specific reason.but you dont really need to know about it.that temo you got is fine you should OC it further just dont let the temp get above 70C.
 

skiltrip

Distinguished
Jun 3, 2008
20
0
18,510
After about 40 minutes, I got an error on one of the cores and the test stopped. What should I tweak before I try again.
My setttings again were...
400x8
3.2GHz
RAM is at 800
1:1
Set NB voltage to 1.4v
All my other voltages I left at AUTO

My core temps were peaking at about 64c, but on average stayed closer to 61c

Do I need to tweak one or more of the voltages?? Temperature wise it looked as if I was ok.