X3350 / Q9450 Overclocking Thread

[bryanW1995]

have you tried it on different computers? it's probably due to your internet security settings.

 

[tenax]

nice, but 1.45 volts on the cpu..wasn't prepared to do this for long. it also only stayed stable for about 45 minutes. 460 was completely stable:

http://valid.x86-secret.com/show_oc.php?id=336500


this is 7.5 x 470, completely stable and at about 1.4 real volts on the cpu. i've done 7.5 x 483 before but can't remember it's stability or voltage. i never planned to go with above about 3400 to 3500 long term anyways:

http://valid.x86-secret.com/show_oc.php?id=336486


one thing i needs to be pointed out in all this is if the exercise really is to get the best stable oc with these quads..there's some other factors to consider..i.e. what might be useful to you, may not be to me. for example, look at the fan speeds the guy at xs is running in his system. it must sound like an air tunnel. to me, that is about as useful an overclock as a system that will not run stable as validated by prime, orthos, etc.

my qualifications for myself are:

it must be stable 24/7, it must be quiet, it must not jeopardize the hardware short or long term.

my cpu fan while capable of running 2300 rpm, must be able to run under 900rpm, better yet, 600rpm is where i want it to be absolutely silent from 8 ft away sitting position.

 

[tenax]

Originally posted by: bryanW1995
have you tried it on different computers? it's probably due to your internet security settings.

never thought to, bryan..i hadn't changed anything and just chalked it up to the other log in issues with AT that crop up at times. i will give it a go on another machine. i thought it weird i could read and reply to pms..but no post button

 

[tenax]

3600 or 8 x 450 is a solid go at voltages referred to earlier of 1.3625 on cpu and 1.55 on northbridge, but i would like to see that northbridge temp go down. i fear it's going to take 1.4+ plus voltage on the cpu no matter other settings to get up to 3800 stable, if possible..and that just ain't my cup o tea. and i still don't think 8 x 500 is possible without some extreme measures based on evidence to date.

 

[bka4u2c]

Hello tenax,

What motherboard are you using and type of cooling? I'm not exactly sure about pumping my NB higher than 1.45 I had it at last night to get to 450 FSB. Although, systems temps were still pretty low. What are you using to monitor your NB temps? I also noticed your Q9450 is a B1, where did you get that one from? Mines is a C1 as shown in my earlier post.

 

[tenax]

it's the abit ip35-pro, revision 1, using the 2nd latest beta bios available for it (last official was not penryn quad compatible) so i use revision 16 b09. the basic specs on my system otherwise are:

penryn quad 9450 rev B1 (Engineering sample revision)
vid of 1.018
4 gigs g.skill 8500 memory (4 x 1 gig)


i am using an enzotech extreme cooler with variable speed scythe 120mm fan and mx2 paste

 

[dasracht]

Mines running at 3.2 (400x8) at 1.200 vcore in the bios
56c or 46c depending on whether you believe coretemp or realtemp.

screenshot

 

[bryanW1995]

Originally posted by: tenax
it's the abit ip35-pro, revision 1, using the 2nd latest beta bios available for it (last official was not penryn quad compatible) so i use revision 16 b09. the basic specs on my system otherwise are:

penryn quad 9450 rev B1 (Engineering sample revision)
vid of 1.018
4 gigs g.skill 8500 memory (4 x 1 gig)


i am using an enzotech extreme cooler with variable speed scythe 120mm fan and mx2 paste

hey, where did you get your beta bios? I'm going home tonight and want to update to the beta bios in preparation for the (supposed) arrival of an x3350 from moogr.com in the next week.

 

[dasracht]

Is there any agreement on what the TjMax for the q9450 is? Coretemp has it at 105c but realtemp has it at 95c.

Any generally agreed upon value?

 

[ArizonaSteve]

My temperatures suck.

I have a pretty decent Lian-Li (1200, I think) case with two 120mm fans. I have the fan at the front blowing into the case, and the fan at the back blowing air out of the case, so there should be good airflow inside the case.

I have a Zalman CNPS9500 heatsink on the Xeon 3350 CPU that is orientated so that the air is blown towards the exhaust fan at the back of the case.

The motherboard, as mentioned before, is a Gigabyte GA-EP35-DS3R. The Zalman is connected to the CPU fan connector. Case fans are connected to System fans 1 and 2 on the motherboard.

The other main heat producer in this system is the 3870X2 video card.

Cpu voltage is 1.1V. I am running it at stock speed (2.66GHz).

So far so good. So why are my CPU temperatures (using Vista X64 and CoreTemp) at 63/63/63/63 at idle and into the high 70's when running Prime 95 across all four cores?

I believe my fans might be running too slow (this system is a helluva lot quieter than the Athlon 64 I had in there before, which idled at 28c). Any suggestions?

 

[tenax]

Originally posted by: bryanW1995

Originally posted by: tenax
it's the abit ip35-pro, revision 1, using the 2nd latest beta bios available for it (last official was not penryn quad compatible) so i use revision 16 b09. the basic specs on my system otherwise are:

penryn quad 9450 rev B1 (Engineering sample revision)
vid of 1.018
4 gigs g.skill 8500 memory (4 x 1 gig)


i am using an enzotech extreme cooler with variable speed scythe 120mm fan and mx2 paste

hey, where did you get your beta bios? I'm going home tonight and want to update to the beta bios in preparation for the (supposed) arrival of an x3350 from moogr.com in the next week.

it's available here, bryan:

http://forum.abit-usa.com/showthread.php?t=124502

go down the page and you'll find 2 of the 16 versions..otherwise, you'll have to dig through that thread a bit to or try a search of b09 or b10 and you'll find the newer ones. i'm b09 is better than b10. also, you the bios doesn't come with the flash file so you need to take the one from the 14 bios and edit the ini or whatever it is in the file page and rename the bios so it's a .bin extension, and you call it 14, rather than 16b09 so the flash program sees it. my explanation may not be the best..sorry..there are lots of tips on how to do this at that thread if you need it and lots of good help to be found there.

 

[tenax]

good screenie to be had here from another guy with one:

http://www.hardforum.com/showt...2289565#post1032289565

 

[lopri]

Originally posted by: ArizonaSteve
My temperatures suck.

I have a pretty decent Lian-Li (1200, I think) case with two 120mm fans. I have the fan at the front blowing into the case, and the fan at the back blowing air out of the case, so there should be good airflow inside the case.
...

I believe my fans might be running too slow (this system is a helluva lot quieter than the Athlon 64 I had in there before, which idled at 28c). Any suggestions?

You know that the back-side fan on that case is NOT an exhaust fan, right? (unless you flipped it) It blows air into the case, so your Zalman should be blowing towards the front.

 

[lopri]

Does anyone experience the early-reported problem with 45nm quads, that one of the cores is noticeably slower than the rest?

http://www.anandtech.com/cpuch...howdoc.aspx?i=3184&p=4

 

[ArizonaSteve]

No I didn't know that. And I did change the original fans for quieter ones some time back. So the fan at the back sucks air into the case and the fan at the front is the exhaust. Right?

Time to open up the case again methinks...

EDIT: Or maybe not. I did some further checking and according to this review - http://www.bjorn3d.com/read.php?cID=819 - and this one - http://www.xyzcomputing.com/in...2&limit=1&limitstart=3 - I have the fans set up correctly.

 

[themattman]

I'm glad that these Q9450's can easily push 3Ghz and many users are approaching (or passing) 3.5 Ghz. I've been planning to get a completely new system, but the infinite delay of E8400's and Q9450's has kept me waiting. (My AMD Athlon XP-M 2500+ @ 2.3 is starting to get a bit long in the tooth)

For the average overclocker with a Zalman or a TRUE, what would be an attainable overclock without having to resort to extreme measures like changing more than just the cpu/ram voltages? I know YMMV, but is around 3.4 out of the question?

Btw, I would love to see some review sites put the Oc'ed E8400 against the Oc'ed Q9450. I want the quad for folding, but I want to make sure that the performance is up there with the E8400.

 

[lopri]

http://www.lian-li.com/v2/en/p...dex=25&ss_index=62&g=t

Check the arrows indicating air flow. I know it's kinda weird but that's how they did it. I have one (V1000) and I flipped the intake fan from the back to exhaust.

Oh, and in the original design the side fan is a MUST since that's the only active exhaust.

 

[ArizonaSteve]

Ah. I see the problem. Looks like my 1200 is an older design (the current one has the designation PC-V1200 Plus II). My case does not have the side exhaust fan. I believe the two reviews I linked reference the design that I have.

 

[arthas76]

just installed my oem Q9450 from tankguys - asus p5b-e bios says its an unknown processor (w/ the latest 1704 bios) but it boots fine. idle temps are 40 40 35 35 w/ coretemp. why would they be so diff between the cores? i applied the AS5 like the guide said (line acress the middle) - gonna try ocing later after a bit of burn in

 

[dasracht]

Originally posted by: lopri
Does anyone experience the early-reported problem with 45nm quads, that one of the cores is noticeably slower than the rest?

http://www.anandtech.com/cpuch...howdoc.aspx?i=3184&p=4

My core 1 is a little slower than 2 and 3. After an hour or so of prime its a test and a half behind. Core 4 is in between.

That's not bad at all, but it's a little different than the q6600

 

[ArizonaSteve]

So which one is correct - CoreTemp or Real Temp?

I have CoreTemp reporting 67C and Real Temp reporting 57C, while running Prime95 across all four cores. Definite progress, since a few hours ago CoreTemp was reporting 79C with Prime95 running.

Since I last posted, I did some investigating of the system and CPU fans, and did indeed find that the exhaust fan was actually blowing air into the case. That was duly fixed, and I also redid the Arctic Silver on the processor as per the standard instructions (line across the middle of the heatspreader).

Idle temperatures, according to Real Temp at least, are in the mid 40s.

Going to let things bed down a bit, then try some mild overclocks...

 

[dasracht]

Originally posted by: ArizonaSteve
So which one is correct - CoreTemp or Real Temp?

That's the question I want to know. Coretemp sees the tjmax as 105, realtemp sees it as 95, but no one seems to agree on what it really is. Intel doesn't release those numbers for their non-mobile processors as I understand it. So until theres a consensus I guess we're stuck with the 10c difference in the temp programs.

Anyone have more insight on this than me?

 

[tenax]

Originally posted by: themattman
I'm glad that these Q9450's can easily push 3Ghz and many users are approaching (or passing) 3.5 Ghz. I've been planning to get a completely new system, but the infinite delay of E8400's and Q9450's has kept me waiting. (My AMD Athlon XP-M 2500+ @ 2.3 is starting to get a bit long in the tooth)

For the average overclocker with a Zalman or a TRUE, what would be an attainable overclock without having to resort to extreme measures like changing more than just the cpu/ram voltages? I know YMMV, but is around 3.4 out of the question?

Btw, I would love to see some review sites put the Oc'ed E8400 against the Oc'ed Q9450. I want the quad for folding, but I want to make sure that the performance is up there with the E8400.

3400 with only a cpu voltage change is doable and stable for a 9450. it's above 3400 that starts getting a bit dicey. 3500-3600 is easily obtainable, but you start trading off higher voltages. by 3800, you're pressing the voltage up in the 1.4 range on the cpu which is getting a bit deadly and same deal with mch..up in the 1.5 range or higher. i think once you start pushing a penryn past 1.35, you should ask yourself what your goal was in buying..i know mine was cool and quiet with a decent overclock so with that in mind..i want less than 1.3 volts, 3200-3400 stable and a very quiet computer..so i'm happy.

how are you juding performance of an oc'ed quad with an oc'8400? the 8400 will kick the 9450's butt around the block, between having one full multiplier more, able to easily get over 4000 compared to the easily over 3200 of the quad

 

[bryanW1995]

Originally posted by: themattman
I'm glad that these Q9450's can easily push 3Ghz and many users are approaching (or passing) 3.5 Ghz. I've been planning to get a completely new system, but the infinite delay of E8400's and Q9450's has kept me waiting. (My AMD Athlon XP-M 2500+ @ 2.3 is starting to get a bit long in the tooth)

For the average overclocker with a Zalman or a TRUE, what would be an attainable overclock without having to resort to extreme measures like changing more than just the cpu/ram voltages? I know YMMV, but is around 3.4 out of the question?

Btw, I would love to see some review sites put the Oc'ed E8400 against the Oc'ed Q9450. I want the quad for folding, but I want to make sure that the performance is up there with the E8400.

there's nothing average about a true. the zalman is a turd for overclocking quads, however. according to aigomorla you need a tuniq or true to get a good, solid, long-term stable oc on a 65nm quad. rumor has it that the 45nm chips run cooler, but I'm not so sure that mark agrees with that...

 

[lopri]

If your boards support VTT/GTL value selection, do take advantage of it. Quad-core overclocking is all about those little tiny signal voltages. I also learned the oft-observed huge difference of temps between cores are partly dependent on GTL value. Once you find the right GTL values for FSB and NB, things get whole lot easier.