Q9450 OC / Temps / Settings

[stoked]

L803A756, OEM X3350 from lagoom.

Just got my retail X3350 from the same batch L803A756 from Moogr.com. Gonna rebuild my system with a GA-P35-DS3R. Might lend my chip to a friend to review on his site first though.

[RunawayPrisoner]

I think it's more an FSB limitation of the CPU. On my old E6600 I was limited to 430 when it was on air but when I put it under the Vapo LS I could do 478 stable.

Try reducing the multiplier and try to reach as high of a FSB as you can. If it doesn't go any higher than 430 after vCore has been maxed out then yes, it's the chip limit. But if vCore is low enough, and all of your motherboard's voltages have been maxed out, then it's the motherboard limit.

[Fluke_Dallas]

Try reducing the multiplier and try to reach as high of a FSB as you can. If it doesn't go any higher than 430 after vCore has been maxed out then yes, it's the chip limit. But if vCore is low enough, and all of your motherboard's voltages have been maxed out, then it's the motherboard limit.

Thanks for the response. That's exactly what I did, running an OEM Q9450 on a DFI LanParty LT X48-T2R. But first I confirmed the CPU would do 3.2ghz by running it at 8x and 400 FSB. No problem, stock voltages, 4x prime95 stable overnight. Also tested the RAM at 500. No problem, but it's rated DDR2 1000. Then I dropped the multi to 6x and set the RAM to 1to1 to see how high I could get the FSB. Much to my surprise I could only get to 425 FSB. With the CPU at 2.55ghz I know that wasn't the limiting factor. Also, cooling is not an issue. Water on CPU and northbridge.

So I go ok this sucks, and start some troubleshooting by putting in an E6600. Go through the entire process above and guess what. In the final step the board goes straight to 500 FSB easy as pie. 2x prime95 stable overnight.

It's almost as if Intel has put an FSB limiter in the Q9450. The CPU is clearly capable of higher speeds, as is demonstrated by the test run at 8x. But problems at FSBs over 425 stop you from ever doing serious overclocks with it.

Could it be an electrical design problem with the chip? The Board? A bios problem? Dunno, but this build plan is not working out. Maybe I'll try a Q6600 I have laying around next.

[Hamilton]

Try reducing the multiplier and try to reach as high of a FSB as you can. If it doesn't go any higher than 430 after vCore has been maxed out then yes, it's the chip limit. But if vCore is low enough, and all of your motherboard's voltages have been maxed out, then it's the motherboard limit.

That's the first thing I do when I start out with a new CPU.
With a 6x multi my old E6600 would reach 430 under air and 478 under Vapo!

This new Q9450 does 455 on air so it will be very interesting to put on the Vapo in a couple of days...

[LockBreaker]

I am looking forward to my own build.

Q9450 and maybe 2 GPUs.

And, of course, water cooling.

[mu||et]

So you're saying a G35 is better than a X38 at clocking this chip?

And by the way, my chip ended with 4A758 just FYI. I don't think low numbers have anything to do with the clockability of the chip... but we can all hope.

Im just saying that I have seen quit a few 790i mb's that are running the x3350 and q9450 at 450fsb easy and your mb but my mb bios have been a mess. Every time I go to a new bios I have to redo the dividers for my memory so its stable. I am just waiting for ASUS to get off there ass and give is a good bios, I just think thats what is holding me back from 450 fsb.

[RunawayPrisoner]

Actually, my P5E-VM HDMI board had only 5 BIOS updates, and the only one that supposedly "supports" this CPU is a beta one with a whole bunch of other issues... which made me go back down to the one that did not support the CPU to begin with.
It seems to me... up until this point, like it's the board's limit rather than the chip's. Most X38/X48 boards are reporting 425-440 FSB walls, most P35 boards are reporting 450-460 FSB walls, and... most 780i boards are reporting 470-475 FSB walls. Coincidence?

[mu||et]

Actually, my P5E-VM HDMI board had only 5 BIOS updates, and the only one that supposedly "supports" this CPU is a beta one with a whole bunch of other issues... which made me go back down to the one that did not support the CPU to begin with.
It seems to me... up until this point, like it's the board's limit rather than the chip's. Most X38/X48 boards are reporting 425-440 FSB walls, most P35 boards are reporting 450-460 FSB walls, and... most 780i boards are reporting 470-475 FSB walls. Coincidence?

Yeah thats what I am talking about, hopefully we will get the bios's that will remedy this problems over time. ASUS has so many damn motherboards out no wonder bios updates are slow.

[RunawayPrisoner]

But there is hope. World record belongs to a P35 board. Maybe someone with the same board should ask the guy how he did it.

[Kensek]

My Maximus Formula (X38) could do 440x8 with a Q9450, but couldn't do 450x8.

Sold it on eBay.

Awaiting my X3350 and X48 Asus Rampage. Really want to run it at 475x8 for Benchmarking.

[road-runner]

My X38 DS4 will do 484....

[motopen1s]

Lapped my scythe zipang today (was anything but flat), now though I feel like lapping q9450 badly - since it also just has bent in the middle. I know doing so without even testing the chip first is not very smart thing to do... It's just the thinking that you will have to deal with the installation of the heatsink with the ty push-pin again brings crazy ideas to my head

Will proceed with installation tomorrow... hopefully first numbers will also come soon.

p.s. I officially hate push-pin design

[kickerman65]

Yea i agree push pins suck and it seems im always questioning if they have seated properly and if they are actually holding the heatsink down properly.

Ive got a dumb question that i know probaly anyone on here can answer but im a newbie when it comes to overclocking so here goes: What are the key voltages when it comes to overclocking? Cpu Core, Fsb, And northbridge?

and is 1.5 about the most you wanna go before your really pushin it?
and also when it comes to these voltages, the lower you can get all these and still be stable the better?

[RunawayPrisoner]

Lapped my scythe zipang today (was anything but flat), now though I feel like lapping q9450 badly - since it also just has bent in the middle. I know doing so without even testing the chip first is not very smart thing to do... It's just the thinking that you will have to deal with the installation of the heatsink with the ty push-pin again brings crazy ideas to my head

Will proceed with installation tomorrow... hopefully first numbers will also come soon.

p.s. I officially hate push-pin design

You know... Big Typhoon with really tightened screws, and a lapped X3350 never went over 55C no matter what. :p

[motopen1s]

You know... Big Typhoon with really tightened screws, and a lapped X3350 never went over 55C no matter what. :p

it seems that after cpu will be lapped - zipang will shine (seems like the contact between the two is pretty bad; core temps vary quite a lot)

anyway, everything is working from the first boot, here is the first screen (all voltages in bios are set to auto):




I will try out what the cpu can do, but it already seems/feels it will be a keeper

[motopen1s]

it seems that after cpu will be lapped - zipang will shine (seems like the contact between the two is pretty bad; core temps vary quite a lot)

anyway, everything is working from the first boot, here is the first screen (all voltages in bios are set to auto):




I will try out what the cpu can do, but it already seems/feels it will be a keeper

UPDATE: seems like for now 3.2ghz is the sweet spot for 24/7. 8gb of RAM really add some stress to the NB. @425 FSB running RAM 1:1 4-4-4-12, and setting perfomance level to 7 -> no boot at NB@1.65v. Perf level 8 will error in Prime95 pretty much straight away (did not try to put more on NB).

That is, now running cpu @3.2, DDR800 4-4-4-12, perf level 7 - stable (30min+ in prime95) + low temps (in bios: vcore 1.27, in windows reads 1.21; NB 1.6v, FSB VTT 1.34v). Max temp running prime95 in Real Temp v2.24 is 48°C (idle temp calibration @0). Room temp 24°C.


Any hints for overclocking with 8gb of RAM? (I know it likes 1:1)

[dertimaushh]

My Maximus Formula (X38) could do 440x8 with a Q9450, but couldn't do 450x8.

Same Problem here. 8*450 is a no go with my Maximus Formula and Q9450.

440Mhz works fine.

Will stay with my Q6600 @3800Mhz.

[RunawayPrisoner]

UPDATE: seems like for now 3.2ghz is the sweet spot for 24/7. 8gb of RAM really add some stress to the NB. @425 FSB running RAM 1:1 4-4-4-12, and setting perfomance level to 7 -> no boot at NB@1.65v. Perf level 8 will error in Prime95 pretty much straight away (did not try to put more on NB).

That is, now running cpu @3.2, DDR800 4-4-4-12, perf level 7 - stable (30min+ in prime95) + low temps (in bios: vcore 1.27, in windows reads 1.21; NB 1.6v, FSB VTT 1.34v). Max temp running prime95 in Real Temp v2.24 is 48°C (idle temp calibration @0). Room temp 24°C.


Any hints for overclocking with 8gb of RAM? (I know it likes 1:1)

Umm... loosen the timings to 5-5-5-15. But sincerely, I got 6GB at 475MHz 4-4-4-12 before. It might not be because you have too much RAM. Try taking two sticks out?

[RAID]

RP, did you sell your Q9450? Looks like you have some new parts coming in

[RunawayPrisoner]

Yeah, I sold my X3350. I wanted to buy another one... but then the deal with the Q6700 popped up, and I went with the Q6700 instead.

[RAID]

well, here's to hoping you get a good chip! and enjoy your new card

[RunawayPrisoner]

Thanks. At some point, I may just try my luck with another Q9450... or actually, I want to experiment with another one, to see if it was the board... or the chip that was being held back.

[Richard Dower]

My Q9450 is due in the next day or so....god willing i don't get a dud.

[RunawayPrisoner]

Check the VID of the chip. Seems like chips under 1.200v clock really nice!

[Clint]

Check the VID of the chip. Seems like chips under 1.200v clock really nice!

Well, except from my ubercrappolachip that has 1.125 VID and didn't even manage to be stable at 2.9GHz on stock volts...

I did try it on 2 other boards and with a brand new Corsair 620W PSU, still the same crappy clock. Maxed out at 3.2GHz on 1.45v, and that was under a lapped TRUE with two 120mm 73CFM fans in push/pull on a open bench.

Load temp was only 48C on 1.4volts, but still not over 3.2GHz.

If i dropped vcore one notch it failed, and I tried various configs, it scaled terribly bad. Without any doubt the worst chip I've ever tried in 20+ years of hardware experience....taken the expectations in consideration.

It's sold btw, luckily enough I didn't loose much.