Need advice on Q9450 @ 3.6Ghz

First post on these forums so hello to you all!

Also, if this post ends up looking like a giant wall of text I apologize. But if I'm going to get help, I might as well provide as much info as possible on my situation.

I built my new system about 4 months ago now. As I have an EVGA 790i Ultra motherboard I have been able to get a lot of help from their forums about overclocking and the specifics of my MB. I am quite comfortable with assembly and tweaking but I'll be honest, I haven't built a PC since my old AMD 2400+

Currently, I have achieved a rock-solid stable OC at 3.6Ghz on a Q9450 CPU. fyi multipliers are locked at 8x on Q9450 so my FSB is running at 1800 Mhz. No heating issues, ~35C at idle and I don't even break 60C under Prime95 load using air cooling. The only issue I am having so far is the amount of voltage I am having to bump up on the VID from a stable 3.4Ghz OC.

Here are the current voltage settings I need to get my stable OC @ 3.6Ghz:
VID =
1.462v (bios setting)
1.43v (real voltage at idle)
1.38v (voltage under 100% load)

VTT / FSB = 1.2v (before you ask I've tested this up to 1.35v with no improvements in stability)

SPP / Northbridge = 1.45v
MCP / Southbridge = 1.55v
DRAM Voltage = 1.9v

So my question to you guys is: how safe is it to be running a Q9450 at my VID voltage levels for extended periods of time? I have gotten very mixed feedback on the EVGA forums so I would like some extra opinions before I decide to leave my OC at 3.6Ghz for 24/7 use.

I am aware Intel suggests 1.36v as a Max for 45nm CPUs. I'm also aware that this doesn't apply to overclocked CPU's. I've read some people say anything below 1.5v on air should be ok as long as temps aren't an issue. On the other hand, I've also been told that leaving my CPU idle above 1.4v is bad. Others say any setting above 1.45v is just bad. I'm just not really sure who to believe anymore.

Also, I was under the impression that 45nm CPU's required less voltage than the previous generation. What bothers me the most about my setup so far is that I seem to need as much voltage as Q6600 to achieve the same overclocks. Second, I have only seen about 3 people on the EVGA forums who need as much voltage as I do to achieve my speeds. Most others are hitting 3.8 Ghz. or if they are running 3.6 they use much lower VID like 1.35v.

Now I know every chip differs, but I just can't help but feel that my voltage requirements for my OC are much higher than average. I should point out that I get a 100% stable OC @ 3.4Ghz with 1.36v VID. This usually translates to 1.32v at idle and 1.29v under load

I would appreciate any help or info on this matter.

Here are my Specs:

Q9450 @ 3.6Ghz
EVGA 790i Ultra MB
2x 2Gb Corsair tw3x4g1600c9dhx @ 1800Mhz link + sync
EVGA GTX 280 Vanilla
500GB Western Digital HD
OCZ 850w GameXstream PSU
Pioneer DVD Burner

I could post the rest of my bios settings if needed but I don't think it will be necessary. at least not for now

Increased voltage can lead to a shorter life-span. However, you'll likely buy a new chip
long before that happens. This is also an indicator that you might be running at the very
edge of what your CPU can handle, so there may be stability risks such as silent data
corruption.

What voltage do you need for 3.5 GHz? That's still a respectable overclock in my book.
You will hardly notice the loss of 100 MHz, and if you can run at 1.36 or lower, you'll
sleep better at night.

Klag

I'd be really surprised if you need that much VID to keep that chip stable at 3.6Ghz... The VTT and the others all make sense, but I'd try lowering that VID quite a bit and testing. My Q9450 would do 3.6Ghz on purely stock voltage; most others can get it there under 1.3v last I did any real checking...

Do Nvidia boards have GTLreference settings? If they do, what are yours?

Required Vcore voltage varies highly from board to board. Your board might just need 1.36 to be stable.

My Q9450 was rock solid at 3.6 on 1.30V with a P5E-VM board. Now I need 1.36 on my X48-DQ6 to get the same stability. The P5E is a $130 board and the DQ6 is a $300 board...go figure!

Every chip is indeed different. You seem to be doing quite well to me, I don't see any issues. Your voltage is a little on the high side, but it's definitely nothing too far out of the ordinary. Temps look good too, I wouldn't worry about it.

Thanks so much for the feedback guys.

To answer a few questions:

3.5Ghz needs almost as much as 3.6Ghz. I still need to set my VID above 1.42v. I haven't really tested much since the voltage levels seems just as bad as 3.6Ghz.
Like I stated in my OP, at 1.36v where I feel very comfortable leaving for 24/7 use, the best I can do is 3.4Ghz.
Which is great and all, but I just can't leave it alone hehhehe. at that speed my stress tests only make my core temps reach 56C at most so I feel like I have a lot of room to play.

The 790i Ultra does have GTLREF settings. This is probably the only thing I haven't had the courage to get into. I leave my settings on auto (-50,-50,0,0).
fyi from what I've read on the EVGA forums it seems these settings aren't working as normal on the 790i Ultra. There have been a couple guides to try and figure this out but since the last bios update nobody really knows anymore lol.

Would proper GTLREF settings allow for lower VID?

Can I also ask what kind of life-span you think my CPU will have at 1.46v? I generally leave the PC on all day and turn it off maybe 4 nights a week. I game a decent amount and work with 3D applications like Maya and 3D studio so I tend to render things overnight. I know nobody can know for sure, I'm just asking for your best guess. what are we talking about here? 6 months? 1 year? 5 years?

Would leaving it on 1.36v really make that much of a difference on lifespan considering my temps?

batut said:

Can I also ask what kind of life-span you think my CPU will have at 1.46v? I generally leave the PC on all day and turn it off maybe 4 nights a week. I game a decent amount and work with 3D applications like Maya and 3D studio so I tend to render things overnight. I know nobody can know for sure, I'm just asking for your best guess. what are we talking about here? 6 months? 1 year? 5 years?

Would leaving it on 1.36v really make that much of a difference on lifespan considering my temps?

Click to expand...

I don't think anyone really knows for sure, I see peeps on here claiming that 1.4 blew up their rig, and others still that say they are 1.45+ for a year with no problems at all...

I think it comes down to how comfortable you are with having to replace teh chip. Personally, I wont run a 45nm CPU >1.36 or >60c unless benching. I just don't wanna shell out the cash for another CPU in that case.

$320 seems like a big loss for a measly 200mhz you wont even notice.

Unfortunately that is my gut feeling also. I guess I was hoping for something to prove me wrong...

im trying to get my q9450 to 3.6 too, im at 3.52, vcore 1.336. I can't seem to get any fruther though, I always crash on prime if I try 3.56, I tried increasing vcore/nb/vtt but still no help, I think I should try and increase a bit more vcore but Im not sure if that will help.

Would proper GTLREF settings allow for lower VID?

Click to expand...

In the case of my q6600 it did. The first thing I noticed after tweaking the GTLREF 1 and 2 values was that the system gained stability the next time I attempted the overclock. The next thing I noticed was that it took .055v less to do it.

In the case of my q9450, getting above 3.6GHz proved very difficult. No amount of brute force would jump that damn chip to 3.8GHz. I had to tweak the CPU and MCH skew values to get it to scale. 1.4v 3.6GHz instantly turned into a 1.5v 3.8GHz with benching and folding stability. CPU skew required more of an adjustment than MCH as is usual.
GTLREF value adjustment, to my surprise, provided no increase in stability and were left at stock.

With my 780i I only need 1.36Vcore in BIOS and 1.3Vcore full load, after playing with some GTL ref voltages. So you probably could lower the Vcore if you try different settings, however it is different for each processor.

Salmon91 said:

im trying to get my q9450 to 3.6 too, im at 3.52, vcore 1.336. I can't seem to get any fruther though, I always crash on prime if I try 3.56, I tried increasing vcore/nb/vtt but still no help, I think I should try and increase a bit more vcore but Im not sure if that will help.

Click to expand...

I just spent some time asking around a little more about safe Vcore on 45nms. And although nobody can really give you a 100% positive answer it seems the max safe zone for air cooling is generally in the 1.5v - 1.6v range. Although if you are pushing 1.6v you are most likely on water cooling already.


You might have to push the vcore more like I did. like I said, it took 1.462v setting in bios to get stable stress runs @ 3.6Ghz. I get a big Vdroop so you might not need that much.


I will see if I can fool around with the GTLREF settings. although... the guides for this stuff are way over my head lol. It may come down to simple trial and error. Assuming I get the courage to mess with it.

I only need 1.27v to get my Q9450 completely stable at 3.52GHz. It simply doesn't want to do even 1MHz more FSB no matter how much I bump up all the voltages.
VTT/FSBT is only 1.22v.

You will eventually destroy the chip with 1.35v and above. Regardless of cooling, the electron flow will destroy the silicon paths.

Seems like I have the opposite problem, I can get to 3.6 on 1.25V. My problem is my chip runs hot as hell under load. It will idle in the low - mid 40s and within 10sec of linpack or P95 it jumps to upper 60s.

Right now I've backed off the OC to 3.5ghz to keep her under 60c. I'm hoping some fine tuning will let me back off Vcore more to get to 3.6+.

You've got a very good mobo, it handles high fsbs better

I got so frusterated with trying to get my 9450 past the 3.2Ghz barrier that I gave up for a while lol. I am currently running 10hrs+ stable on prime @ 3.01Ghz, but I just cant seem to get past that. I might try upping my vcore tonight to around 1.3, because it seems like alot of people are stable at those voltages for 3.2-3.5Ghz.

TheTorch said:

I only need 1.27v to get my Q9450 completely stable at 3.52GHz. It simply doesn't want to do even 1MHz more FSB no matter how much I bump up all the voltages.
VTT/FSBT is only 1.22v.

You will eventually destroy the chip with 1.35v and above. Regardless of cooling, the electron flow will destroy the silicon paths.

Click to expand...

not 1mhz? what board are you running really sounds like a fsb wall issue....

well if you check intels specs for 45nm cpus. the max voltage that you may them release a cpu at is going to be 1.36725v running 1.35 will not cause that issue your refering too. it would happen for sure with 1.4+ but that also does depend on how well you can keep temps in control.

kkpudge7 said:

I got so frusterated with trying to get my 9450 past the 3.2Ghz barrier that I gave up for a while lol. I am currently running 10hrs+ stable on prime @ 3.01Ghz, but I just cant seem to get past that. I might try upping my vcore tonight to around 1.3, because it seems like alot of people are stable at those voltages for 3.2-3.5Ghz.

Click to expand...

more then likly you hit a fsb wall that i talked about. its known to happen at lower fsb's with 45nm cpus on P35/X38/X48. if you lowered the multi and still tried to get past X fsb and it does the same thing, then you hit the fsb wall. if you were running P4x based board you wouldnt have this issue. what i do find off if you are having this issue on the 8x multi 400mhz fsb is really low to hit for the fsb wall. if you wanted to try it, pad mod the cpu for 400mhz fsb, that should solve the fsb wall issue on the board. in alot of cases it has solved the issue but in the few cases where people are running gigabyte boards it doesnt, has to do with bios used on the boards.

Evilsizer:

I think it is probably my novice OC skills more than anything. My current setup for 3.01Ghz is:

CPU vCore @ 1.20
DRAM v @ 1.9v
FSBv @ 1.20
NB @ 1.25
SBv @ 1.1

I really dont know how to tweak the Gtl ref voltages, I messed around with them by trial and error, but that wasn't working for me lol.

I just tried the CPU core @ 1.31v and upped the FSB to 1.3 and upped the NB to 1.31, and SB to 1.2, it wasn't stable at all, I was getting BSOD on startup.

tell me what options you have for "cpu gtl reference voltage", what is you PLL voltage set to? also you have transaction booster disabled correct?

my mistake after taking a closer look your P5E prem is a X48 based board, just a better binned X38.

Transaction booster has two settings on my MB, auto and Manual. When its set to manual there's common performance level which was at 5 automatically when I switched to manual. then it has 4 differnt "pull in" settings and they were all disabled.

my PLL is at 1.5v

CPU GTL ref 0/2=.630
1/3=.670