X3350 / Q9450 Overclocking Thread

[continuum]

Hmm, I'm at 3.28ghz on my Q6600 G0 right now at 1.3625v, I get the feeling I can push it harder pretty easily.

Q9450 is my next purchase though-- if I can get 3.5ghz or 3.6ghz (3.8ghz would probably be ideal) with some DDR2-1066 I'd be happy.... sounds like you guys are hitting my desired speed range most of the time. Might be time for me to buy soon!

 

[runawayprisoner]

Originally posted by: aigomorla

Originally posted by: runawayprisoner
No prob. Mine defaulted to 1.2250v, too... for some reason. Jeebus, that voltage was enough for 3.6GHz! Too bad 3.80GHz required 1.304v, though.

can i ask how your testing your overclocks? What stress program are you using?

windows bootable and prime stable are two completely different things on chips.

Windows stable: Look at vcore-
http://i125.photobucket.com/al.../p73/aigomorla/OMG.jpg
But no way in hell she'll hold prime at that voltage. She's completely windows stable but not prime stable.

Prime Stable: 10hours. I recomend no less then 7hours tho.
http://i125.photobucket.com/al...orla/Clipboard01-1.jpg


As you can see it takes a considerable amount of more voltage to get the thing to prime stable. <also yes i know the clock is different, but the voltage required for both 4.2 and 4.0 for prime stable was the same. This is why i keep it at 420-425

I use OCCT to stress the machine for about a few hours. It gives quicker errors than Prime95. But just in case you are interested:

http://i235.photobucket.com/al...etpham/PC/PrimeEnd.jpg

That'll be stable for 10 hours (disregard the 54000+ hours thing please...). But that was with vdrooping... (vCore load is at 1.288v) After vdroop modding, the thing would idle at 1.288v and load maximum at 1.304v, which is ideal to me.

Turned out 1.344v load was enough for 3.90GHz stable in the end, so my guess is around 1.38v to 1.400v would be necessary for 4GHz. Seems like I was right... at least in the case of your QX9650.

And by the way, booting into Windows Vista is a whole different story compared to booting into Windows XP. I've had my machine booting into Windows XP stable at 3.80GHz 1.200v vCore... but it wouldn't boot into Vista until 1.280v vCore.

 

[bka4u2c]

Originally posted by: bka4u2c

Originally posted by: runawayprisoner
Yeah. You need to manually undervolt the CPU. Mine ran 3.2GHz at 1.14v without any problem.

Good tip, thanks. I'll lower it until Windows becomes flaky. This might also be a good start to see how far FSB will go on this voltage.

Tested down to 1.16 and passed OCCT before I got tired and went to bed. I'm thinking thats pretty low but if I can go lower. I love the steady vcore on this board. VCORE

 

[ArchAngel777]

Originally posted by: bryanW1995

Originally posted by: Foxery

Originally posted by: bryanW1995
ok, so there's a beta 17 bios for the ip35-e? What does it fix? Did your ip35-e run ok right out of the box with the Q9300? Man, this is getting more and more difficult to decide what to do...

There's a sticky in the Abit Forum: BIOS Section where a member hosts all recent betas; however, they almost never have release notes. No one knows what's in 17b02.

I flashed my BIOS with my old chip before the new one came. New-in-box boards are highly unlikely too boot up without doing this first.

As a point of interest, my board defaulted to 1.1625 volts VCore, and has no settings lower than this while the Q9300 is plugged in.

yeah, that's an annoying abit strategy. My ip 35 pro won't do less than 1.25v on my Q9450 and wouldn't do less than 1.35v on my e6750!!!

All the Abit P35 based boards set their lowest voltage point at the VID of the CPU. You cannot undervolt with the IP35's with any official BIOS or Beta BIOS from Abit.

 

[imported_BigJim]

While this article does not review our favorite heatsinks, and at the same time amazingly pulls together over a dozen models I've never heard of, its demonstrates that most of them work better when the pipes are horizontal (usually when the case is vertical)

"In every case where the CPU cooler used a "U" shaped heat pipe rod in their design, having the unit positioned with the rods running horizontally proved to offer a cooling benefit... This goes to show you that heat-pipes are in fact prone to suffer the effects of gravity in their design. Once the vapor cools and becomes liquid, it seems to have an easier time completing the thermal circuit from side to side than it does from top to bottom."

http://benchmarkreviews.com/in...9&limit=1&limitstart=8

So get those pipes horizontal for best overclocks!

Originally posted by: bryanW1995
aigomorla and markfw900 are the quad kings here, and I've never heard either of them recommend an arctic cooling cpu hsf for a quad, certainly not one that you want really good oc's on. Up to 3.0 or 3.2 even you might be ok, heck my zalman 9700 will probably get me to 3.0 or 3.2 ok, but if you want a seriously high OC then you should spring for a tuniq, a thermalright ultra 120 extreme or one of those new noctua coolers that AT wrote about the other day.

EDIT: what I'm trying to say is that a QUAD generates a LOT MORE heat than a dualie, and a cooler like the AC7 has its weakness exposed quickly by a quad.

Cool. This must be the new Noctua NH-U12P. http://www.anandtech.com/showdoc.aspx?i=3268&p=2 Interestingly in the benchmarkreviews review, that Noctua comes out decidedly in the middle of the pack of 12. And in the Anandtech review it performs just behind the TRUE. Did AT get the heatpipes horizontal?

Back in the benchmarkreviews review, the Noctua and the TRUE also show similar performance, but bested by half a dozen other products. The OCZ Vendetta 2 wins; its also cheaper.

 

[lytalbayre]

Here are my preliminary X3350 OC results.

1.2250vcore, 3.52Ghz at 440mhz bus x 8 mulitplier. Memory running at 880mhz, 5-5-5-15.

Cooling with a thermalright 120 extreme with a noctua fan. Have a rocket fish case with two Silenx 120x38mm fans.

Temps are about 30-40c idle, and 50-60c under load. Ambient temp is pretty low however, at around 65farenheit.

The memory is 1066 memory, would l be better off not running it in a 1 to 1 configuration and maxing out the memory speed??

 

[imported_BigJim]

@lytalbayre my 2 cents is, look at the RAM using cpu-z and see if there is an EPP setting, and if that requires more voltage. And feel if its getting too hot and needs a fan pointed at it. What is your current RAM voltage, current MCH voltage? Probably there are forums dedicated to your G.Skill and you can find out what other people are doing. I bet you can bump the memory multiplier to 2.4 or 2.5; possible this needs voltage bump but be careful. Are you OCCT stable? Folks, what memory benchmark tool can he use to measure any improvement?

 

[GarfieldtheCat]

Originally posted by: lytalbayre
Here are my preliminary X3350 OC results.

1.2250vcore, 3.52Ghz at 440mhz bus x 8 mulitplier. Memory running at 880mhz, 5-5-5-15.

Cooling with a thermalright 120 extreme with a noctua fan. Have a rocket fish case with two Silenx 120x38mm fans.

Temps are about 30-40c idle, and 50-60c under load. Ambient temp is pretty low however, at around 65farenheit.

The memory is 1066 memory, would l be better off not running it in a 1 to 1 configuration and maxing out the memory speed??

IMHO, it's not worth it....you won't gain much in the way of a speed increase in applications, and it will put more stress on the NB and RAM.

Take advantage of the RAM running a bit under spec and see if you can decrease the voltage to them, cooler RAM is happy RAM

 

[bryanW1995]

Originally posted by: Foxery
I keep forgetting... here's a side by side comparison of the Q9300's Retail HSF next to my E6750 Retail HSF. Pardon the low quality camera.

Side view
Looking at the bases

Is the retail 9450's HSF this shitty, too?

yes the 9450 is the same. the hsf is the same as the e8xxx chips except that it has that strip of copper on the bottom iirc. No worries, if you don't oc it's more than adequate, and if you DO oc then you'd be a fool to not spend a few bucks on an upgraded hsf. With an AC unit for $20 or a tuniq for $45 then even a cheap bastard like me can get a hsf within budget!

 

[bryanW1995]

Originally posted by: ArchAngel777

Originally posted by: bryanW1995

Originally posted by: Foxery

Originally posted by: bryanW1995
ok, so there's a beta 17 bios for the ip35-e? What does it fix? Did your ip35-e run ok right out of the box with the Q9300? Man, this is getting more and more difficult to decide what to do...

There's a sticky in the Abit Forum: BIOS Section where a member hosts all recent betas; however, they almost never have release notes. No one knows what's in 17b02.

I flashed my BIOS with my old chip before the new one came. New-in-box boards are highly unlikely too boot up without doing this first.

As a point of interest, my board defaulted to 1.1625 volts VCore, and has no settings lower than this while the Q9300 is plugged in.

yeah, that's an annoying abit strategy. My ip 35 pro won't do less than 1.25v on my Q9450 and wouldn't do less than 1.35v on my e6750!!!

All the Abit P35 based boards set their lowest voltage point at the VID of the CPU. You cannot undervolt with the IP35's with any official BIOS or Beta BIOS from Abit.

yeah, I know, it's just frustrating. my e6750 wouldn't let me go below 1.35 for that reason. Actually it's 1.20v on the X3350 btw.

 

[imported_ST]

Originally posted by: runawayprisoner

I use OCCT to stress the machine for about a few hours. It gives quicker errors than Prime95. But just in case you are interested:

http://i235.photobucket.com/al...etpham/PC/PrimeEnd.jpg

That'll be stable for 10 hours (disregard the 54000+ hours thing please...). But that was with vdrooping... (vCore load is at 1.288v) After vdroop modding, the thing would idle at 1.288v and load maximum at 1.304v, which is ideal to me.

Turned out 1.344v load was enough for 3.90GHz stable in the end, so my guess is around 1.38v to 1.400v would be necessary for 4GHz. Seems like I was right... at least in the case of your QX9650.

And by the way, booting into Windows Vista is a whole different story compared to booting into Windows XP. I've had my machine booting into Windows XP stable at 3.80GHz 1.200v vCore... but it wouldn't boot into Vista until 1.280v vCore.

Runaway - Have you tried to undervolt your Q9450 with your Asus mobo? If so, how low can you go? The reason why i ask is because i will be making my setup similar to yours, but during idle / low cpu% time, i want to undervolt my setup.

 

[runawayprisoner]

Originally posted by: ST

Originally posted by: runawayprisoner

I use OCCT to stress the machine for about a few hours. It gives quicker errors than Prime95. But just in case you are interested:

http://i235.photobucket.com/al...etpham/PC/PrimeEnd.jpg

That'll be stable for 10 hours (disregard the 54000+ hours thing please...). But that was with vdrooping... (vCore load is at 1.288v) After vdroop modding, the thing would idle at 1.288v and load maximum at 1.304v, which is ideal to me.

Turned out 1.344v load was enough for 3.90GHz stable in the end, so my guess is around 1.38v to 1.400v would be necessary for 4GHz. Seems like I was right... at least in the case of your QX9650.

And by the way, booting into Windows Vista is a whole different story compared to booting into Windows XP. I've had my machine booting into Windows XP stable at 3.80GHz 1.200v vCore... but it wouldn't boot into Vista until 1.280v vCore.

Runaway - Have you tried to undervolt your Q9450 with your Asus mobo? If so, how low can you go? The reason why i ask is because i will be making my setup similar to yours, but during idle / low cpu% time, i want to undervolt my setup.

Uhm... 1.08v for stock clocks, 1.12v for 3.2GHz, 1.14v for 3.4GHz, 1.20v for 3.6GHz, 1.288v for 3.8GHz, 1.328v for 3.90GHz. Load volts are:

1.10v stock clocks
1.14v 3.2GHz
1.16v 3.4GHz
1.23v 3.6GHz
1.31v 3.8GHz
1.34v 3.9GHz

 

[Foxery]

So, stupid question of the day... What is the default, stock voltage of the 9300/9450 supposed to be? I cannot for the life of me find anyone on the web who quotes it; I only see the wide range of 0.85-1.3625, as given by Intel's Processor Finder. Since my BIOS doesn't let me go lower than Abit's preset, I don't know if I should trust their figure.

 

[aigomorla]

Originally posted by: Foxery
So, stupid question of the day... What is the default, stock voltage of the 9300/9450 supposed to be?

Typically its the number your VID displays.

So if coretemp or realtemp says 1.135 that is your stock voltage.


Lastly i am going to say this one more time. If you have speedstep, or C1E or your chip is throttling, that is not a true overclock. Windows will automatically scale down incase of throttle, and by doing so reduces your multi and also reduces the power draw and heat output.

This is not a true test in overclock. You need to disable all those and closely monitor your temps using some from of load program. Typical load testing involves several hours to label a system "Rock Solid Stable". And some programs are greater then others.

For quadcores I DONT RECOMEND Orthos. Its crap, and it passes where Prime fails. My favorate is Prime95 25.6 on small FFT's. Some people also like OCCT, it gives nice graphs after test is complete.

Mark lives by F@H and requires many hours of running it without errors, which i also feel is the ultimate stress program along with linpack. The definition of stable will vary greatly from one person to another.

 

[Foxery]

I bought this chip specifically for F@H, and test with Prime95. I'm on top of all of that

I thought you couldn't go by software readings for voltage due to vdrop and vdroop... I want to know the official specification. Usually review sites mention this, and I just don't see it anywhere.

What does OCCT stand for? I just started seeing it referenced recently, and I don't know that one.

 

[aigomorla]

Originally posted by: Foxery
What does OCCT stand for? I just started seeing it referenced recently, and I don't know that one.

This i have no idea. Ruby told me the new version was good, the old version blowed hardcore.

But when you goto there site, it feels communistic. You'll understand what i mean when you type occt in google and look for downloads.

http://www.ocbase.com/perestroika_en/index.php?Download

 

[bryanW1995]

Originally posted by: Foxery
I bought this chip specifically for F@H, and test with Prime95. I'm on top of all of that

I thought you couldn't go by software readings for voltage due to vdrop and vdroop... I want to know the official specification. Usually review sites mention this, and I just don't see it anywhere.

What does OCCT stand for? I just started seeing it referenced recently, and I don't know that one.

I can't comment on occt but my x3350 defaults to 1.2on my mobo while coretemp shows 1.0375. im pretty sure that abit is right but I would like to know for sure

 

[tenax]

there's OCCT

http://www.neowin.net/news/sof...-checking-toolocct-200

 

[runawayprisoner]

Or for those who are too lazy to check, OCCT stands for: OverClock Checking Tool

 

[Foxery]

Ooh, thanks. I'll have to play with this tomorrow. I love the release notes:

You may encounter my weird error message "Save a programmer, kill a user". Starting from 1.1.1a, this message isn't in OCCT anymore.

OCCT's "Communist" theme is for fun. I didn't do that for politics, just for the spirit.

 

[aigomorla]

Originally posted by: tenax
there's OCCT

http://www.neowin.net/news/sof...-checking-toolocct-200

my link is from the dev direct. :T

http://www.ocbase.com/perestroika_en/index.php?Download

 

[imported_BigJim]

Originally posted by: BigJim
"In every case where the CPU cooler used a "U" shaped heat pipe rod in their design, having the unit positioned with the rods running horizontally proved to offer a cooling benefit... This goes to show you that heat-pipes are in fact prone to suffer the effects of gravity in their design. Once the vapor cools and becomes liquid, it seems to have an easier time completing the thermal circuit from side to side than it does from top to bottom."

http://benchmarkreviews.com/in...9&limit=1&limitstart=8

Its been a day and no replies. People, people, can I get my props? Comparing a vertical TRUE to a horizontal Vendetta 2, I just saved you 7.75 degrees, Celsius no less, and $25! Even when both motherboards are vertical, I still saved ya' 6 degrees. That oughta translate to some hella MHz. Hear me now, believe me later! ;-)

*Q6600 G0 Stepping Core 2 Quad 2.4GHz (Overclocked to 3.3 GHz)

 

[bka4u2c]

Originally posted by: BigJim

Originally posted by: BigJim
"In every case where the CPU cooler used a "U" shaped heat pipe rod in their design, having the unit positioned with the rods running horizontally proved to offer a cooling benefit... This goes to show you that heat-pipes are in fact prone to suffer the effects of gravity in their design. Once the vapor cools and becomes liquid, it seems to have an easier time completing the thermal circuit from side to side than it does from top to bottom."

http://benchmarkreviews.com/in...9&limit=1&limitstart=8

Its been a day and no replies. People, people, can I get my props? Comparing a vertical TRUE to a horizontal Vendetta 2, I just saved you 7.75 degrees, Celsius no less, and $25! Even when both motherboards are vertical, I still saved ya' 6 degrees. That oughta translate to some hella MHz. Hear me now, believe me later! ;-)

*Q6600 G0 Stepping Core 2 Quad 2.4GHz (Overclocked to 3.3 GHz)

I gave it consideration, but my hard drives are directly above the CPU so If I turn it horizontal the fan would blow directly onto the hard drives. The hard drives do have an exhaust fan behind them that exhaust directly out of the case (Thermaltake VA8000). If I position it so the fan blows down then it goes directly on my video card and NB. Solution?

 

[runawayprisoner]

Originally posted by: bka4u2c

Originally posted by: BigJim

Originally posted by: BigJim
"In every case where the CPU cooler used a "U" shaped heat pipe rod in their design, having the unit positioned with the rods running horizontally proved to offer a cooling benefit... This goes to show you that heat-pipes are in fact prone to suffer the effects of gravity in their design. Once the vapor cools and becomes liquid, it seems to have an easier time completing the thermal circuit from side to side than it does from top to bottom."

http://benchmarkreviews.com/in...9&limit=1&limitstart=8

Its been a day and no replies. People, people, can I get my props? Comparing a vertical TRUE to a horizontal Vendetta 2, I just saved you 7.75 degrees, Celsius no less, and $25! Even when both motherboards are vertical, I still saved ya' 6 degrees. That oughta translate to some hella MHz. Hear me now, believe me later! ;-)

*Q6600 G0 Stepping Core 2 Quad 2.4GHz (Overclocked to 3.3 GHz)

I gave it consideration, but my hard drives are directly above the CPU so If I turn it horizontal the fan would blow directly onto the hard drives. The hard drives do have an exhaust fan behind them that exhaust directly out of the case (Thermaltake VA8000). If I position it so the fan blows down then it goes directly on my video card and NB. Solution?

Get a new case? Most cases have top exhaust fan now.

 

[bka4u2c]

Well my case does have a top exhaust fan but it's a 92mm fan and its not directly over the hard drives and CPU. Now you've given me reason to spend more money on upgrades. That should make the wife ecstatic. But I can always tell her a "runawayprisoner" told me I need a need new case so I can turn my CPU heatsink horizontal and drop my temps by 4-5 degrees. At which point she'll say, "HUH?" that just might work, thanks. Off to look at new cases.