q9650 @ 4g

Hi guys

This might not be the right place to ask this, but in my defense i'm not going to be talking about voltages and FSB multipliers.. ie. i don't plan to discuss how to get to 4g on q9650.

The thing is I'm looking for a new cpu-cooler to get my stock cooled q9650 to 4GHz.

Now I'm really fond of the Arctic Cooling Freezer XTREME Rev.2. I like it so much because my rig is all AC cooled and I like the brand, plus it's a rly quiet cooler and extremely easy to install (not much of a grease-monkey so I'm prone to virtually PnP solutions).

So my question is, can this cooler get q9650 to 4G without load temps to rise above 70 (I'm still ok with 65-67, 70 would be too much for my comfort tho).

The reason I'm asking this in here and not some overclockers forum is that I'm not rly an overclocker, I just want to get it running on 4g because it's getting hard to keep up with the latest software and multi-tasking and I'm not planning on spending hundreds of $ on a new cpu..

I'm currently on a P5K, probably going to get a P5K-WS 64 or P5E3 or something on those lines (p35 or x48 ) soon (and going from 800DDRII to 1333DDRIII) and the 4g OC is going to be done on that.

So thanks in advance for all your help, suggestions on alternative air-cooled non-passive cpu-coolers are expected as well =)

A180

The Titan Fenrir would be a great choice.

The Titan Fenrir would be a great choice.

Daytona_178

+1

that cpu should be munching anything up but getting it to 4ghz would be a nice overclock

I may be wrong, but I think it'd run over 70c, (full load @ 4.0), with that cooler. According to this review, http://tech-reviews.co.uk/reviews/arctic-cooling-freezer-xtreme-rev-2/5/, it's not an especially good performer. I have a 9650 in my backup rig and have run it up to 4.2. With the S1283 I could only go to 3.8 or so and keep it under 70C (Core temps or Tjunction, not Tcase temps).

I'd consider the Xigmatek S1283V REV.W, Zalman CNPS10X Performa, or Prolimatech Megahalems Rev.B

Thanks for the fast replies.

I have the ZALMAN CNPS10X EXTREME (not performa) available in stock at my local resellers at around 63€ .. It is almost twice the money of the XTREME.. But I really am planning on getting 4g stable enough to keep it running 24/7 + high end gaming sessions in between.. So this baby should keep me below 70? Low 60s at load?

I would think so. Here's a good review of that cooler, http://www.frostytech.com/articleview.cfm?articleid=2475&page=5. Though it still shows the S1283 a bit better. I'm basing my recommendation for the Performa on this, http://www.frostytech.com/articleview.cfm?articleid=2498&page=5. I own the other two I recommended.

I just noticed the Fenrir in that second review. I does look like a good pick.

did some research. Turns out the performa is a budget minded newer version of the extreme, same design different materials and fan tho.

So where the performa handles, etreme should even surpass http://www.thinkcomputers.org/zalman-cnps10x-performa-cpu-cooler-review/4/

Take a look at the graphs.. Extreme has a 3 degree advantage over the performa, handling an overclocked i7 @ load on 58c. So a Q9650 shouldn't be a problem.

Thanks again for pointing me towards this thing, will get that one instead of the Freezer XTREME

8)

Ok, so I've installed the new rig (rampage extreme, q9650, cnps10x extreme) and I'm having some slight issues with the cpu temps, which is odd since this is supposed to be one of the most powerful cpu-coolers on earth, capable of keeping an overclocked high-end i7 @ below 60.

First when I started up the new system I noticed that core#1 temp was around 60 (in bios, i.e. in total idle) which is absurdly hot for this cooler. So I assumed the thermal grease had not been applied evenly. I then reinstalled the cpu-cooler and got the idle temps down to 40ish.. In load though, core#1 still reports ~55 degrees while the other cores and cpu-general remain at 40ish.. This is really odd in my view.. I do have to note that installing this cooler in rampage extreme is a total pita, since the screws on one side of the cooler are directly underneath the massive 12cm fan..

I know this is minimal information on the situation but does anyone have any suggestions? Why is one core 10c hotter than the other ones? Cooler defect? Uneven thermal paste? Badly tightened screws on one side? i remind you this is all on an unclocked cpu (running @ 3,0Ghz).. I'm kind of unnerved about this since the whole point of the upgrade was to be able to get to 4,0Ghz with good temps.. and if it's reporting 55 on load @ 3,0Ghz, what's it gonna do @ 4,0???

Thanks in advance, feel free to post this problem to overclockers forums, I'm not very much an OC specialist so don't have a lot of connections in this line of stuff =)

Ok, so I've installed the new rig (rampage extreme, q9650, cnps10x extreme) and I'm having some slight issues with the cpu temps, which is odd since this is supposed to be one of the most powerful cpu-coolers on earth, capable of keeping an overclocked high-end i7 @ below 60.

First when I started up the new system I noticed that core#1 temp was around 60 (in bios, i.e. in total idle) which is absurdly hot for this cooler. So I assumed the thermal grease had not been applied evenly. I then reinstalled the cpu-cooler and got the idle temps down to 40ish.. In load though, core#1 still reports ~55 degrees while the other cores and cpu-general remain at 40ish.. This is really odd in my view.. I do have to note that installing this cooler in rampage extreme is a total pita, since the screws on one side of the cooler are directly underneath the massive 12cm fan..

I know this is minimal information on the situation but does anyone have any suggestions? Why is one core 10c hotter than the other ones? Cooler defect? Uneven thermal paste? Badly tightened screws on one side? i remind you this is all on an unclocked cpu (running @ 3,0Ghz).. I'm kind of unnerved about this since the whole point of the upgrade was to be able to get to 4,0Ghz with good temps.. and if it's reporting 55 on load @ 3,0Ghz, what's it gonna do @ 4,0???

Thanks in advance, feel free to post this problem to overclockers forums, I'm not very much an OC specialist so don't have a lot of connections in this line of stuff =)

Angle180

Well the reasons you mentioned need to be addressed.

I dont know the cooler myself, but make sure its firmly in place, thermal compound applied correctly(on quads your best bet is a thin line of tc in the middle of the chip, no spreading no blob in the middle, just a thin approx 1-1,5mm line)

Keep in mind that every chip is different even same models, so temps wont be the same. Different idle core temps is not a problem I have a lapped qx9650 and core0 and 1 idles around 21-27C while core2 and 3 around 33-35C, however under load the gaps should tighten up to 4-5C max.

I think its an issue with the hs and/or tc as you said the screws are pain to tighten.

Keep us updated and good luck

I've always mounted the cooler without the fan to get the screws tightened evenly - Like 1/2 turn each in a circular pattern until all are fully tightened - Then put the fan on.

Sometimes the weight of the big coolers can shift the contact patch toward the bottom half of the chip. Load the processor with Prime95 with the side panel off and watch core temps with RealTemp or CoreTemp while you lift up slightly in the cooler to see if they change significantly. On one of my coolers I had to loosen the two lower screws slightly to get more even temps between the four cores. On my 9650, core 1 runs about 8C hotter at idle and 4c at load - always has.

Now 55c fully loaded (core temp) isn't bad at all. Keep in mind that your BIOS and programs like PC Probe II report Tcase, (casing of the processor), while RealTemp and CoreTemp report Tjunction, (core temps). Tcase is typically quite a bit cooler than Tjunction. Intel only specifies maximum temps for Tcase - I think 68C for the 9650. The chip will start to activate thermal throttling at somewhere around 100c Tjunction for self-protection.

I would think you could game and/or fully load the processor for short periods (such as running benchmarks), at 4.0, but I doubt you'd stay within temp specs loading the processor fully for extended periods at 4.0. (maybe 3.6 - 3.8 ). A lot depends on how low you can get your voltages at those OC's.

I've now gotten the cooler mounted well enough that @ 4,0 GHz (unstable so far) the coretemps don't exceed 68 in real life applications, go to max 72 on prime95 largeFFT..

Since i know quite little about OCng I might as well post my OCC thread here too, maybe someone has any ideas.

I did get it booting to win on just a FSB increase to 444 (3994 MHz).. That didn't prove to be very stable though
I then did a little fiddling and wound up with vcore - 1,3650, DRAM Freq 1335MHz (Auto settings pushed it to 860ish) and forced DRAM latencies to default (9-9-9-24), the rest is MB Auto
To my surprise prime95 managed to do a largeFFT test for over an hour stable @ maxtemp 72c for core#0, only when I ran winamp and did minor browsing (whilst the prime95 was running) did I get a BSOD
What's interesting to note also, is that OCCT has always failed with all OC settings I've tried (including the current values, discribed above), prime95 is stable with both small and large FFTs though, that is, until i bsod

So, any ideas?c

Cheers, thx for all your replies

I've now gotten the cooler mounted well enough that @ 4,0 GHz (unstable so far) the coretemps don't exceed 68 in real life applications, go to max 72 on prime95 largeFFT..

Since i know quite little about OCng I might as well post my OCC thread here too, maybe someone has any ideas.

---

I did get it booting to win on just a FSB increase to 444 (3994 MHz).. That didn't prove to be very stable though
I then did a little fiddling and wound up with vcore - 1,3650, DRAM Freq 1335MHz (Auto settings pushed it to 860ish) and forced DRAM latencies to default (9-9-9-24), the rest is MB Auto
To my surprise prime95 managed to do a largeFFT test for over an hour stable @ maxtemp 72c for core#0, only when I ran winamp and did minor browsing (whilst the prime95 was running) did I get a BSOD
What's interesting to note also, is that OCCT has always failed with all OC settings I've tried (including the current values, discribed above), prime95 is stable with both small and large FFTs though, that is, until i bsod

So, any ideas?c

Cheers, thx for all your replies

Angle180

I recommend to set back your RAM to auto until you manage to get your chip stable mate.

Your BSOD is caused by RAM not your cpu anyway, so lets just do one OC at a time. ;)

So basically you are running 444fsb by 9 multi at 1.3650V and one or more cores fail under small FFTs right? What other options you set?

Load line calibration enabled? Fsb voltage raised?

I recommend to set back your RAM to auto until you manage to get your chip stable mate.

Your BSOD is caused by RAM not your cpu anyway, so lets just do one OC at a time. ;)

So basically you are running 444fsb by 9 multi at 1.3650V and one or more cores fail under small FFTs right? What other options you set?

Load line calibration enabled? Fsb voltage raised? Greyfoo-t

I have now reverted back to all Auto, FSB 444

Here are the voltages (Everest):

vCore- 1.38
NB- 1.40
SB- 1.06
CPU VTT- 1.31
CPU PLL- 1.69
DIMM- 1.59-1.61
DIMM VTT- 0.80
At Auto settings DRAM:FSB is 1:1, so DRAM Freq. is 889MHz so timings should be 6-6-6-16 at this freq.

Right now I'm stable enough to use the pc, occt still fails almost instantly after the idle min ends though.. Pretty sure if I ran prime and did a little browsing etc. I would BSOD.

Yesterday, when I was running at 1.365, loadline was disabled, FSB voltage was auto

[QUOTE="Greyfoo-t"] I recommend to set back your RAM to auto until you manage to get your chip stable mate.

Your BSOD is caused by RAM not your cpu anyway, so lets just do one OC at a time. ;)

So basically you are running 444fsb by 9 multi at 1.3650V and one or more cores fail under small FFTs right? What other options you set?

Load line calibration enabled? Fsb voltage raised? Angle180

I have now reverted back to all Auto, FSB 444

Here are the voltages (Everest):

vCore- 1.38
NB- 1.40
SB- 1.06
CPU VTT- 1.31
CPU PLL- 1.69
DIMM- 1.59-1.61
DIMM VTT- 0.80
At Auto settings DRAM:FSB is 1:1, so DRAM Freq. is 889MHz so timings should be 6-6-6-16 at this freq.

Right now I'm stable enough to use the pc, occt still fails almost instantly after the idle min ends though.. Pretty sure if I ran prime and did a little browsing etc. I would BSOD.

Yesterday, when I was running at 1.365, loadline was disabled, FSB voltage was auto

CPU PLL seems high to me, NB could be raised a wee bit, other than that I`d say you just hit your fsb wall, what chipset you using again?

I`d say crank up fsb to nb strap to 333mhz and see if it helps. As you already over 1.3625vcore you might want to raise that a wee bit but I think its not the chip itself that fails anyway :(

Yeah P35, easy on them voltages...

Rampage Extreme is x48 8)

Right now I've achieved OCCT 1H STABLE! on these settings:

CPU - 1.4v

NB - 1.56397v

DRAM V - 1.64356v

NB GTL Rev - +50mW

DRAM Freq - 1335MHz

All else is Auto

Temps don't go above 71 on OCCT load

Gonna pull an all-nighter prime largeFFT and see if it's stable, then try 4.2 maybe 8)

Any suggestions? The guys at OCC were quite satisfied with current settings..

Thx for all your help anyways, will keep this thread posted

Rampage Extreme is x48 8)

Right now I've achieved OCCT 1H STABLE! on these settings:

CPU - 1.4v

NB - 1.56397v

DRAM V - 1.64356v

NB GTL Rev - +50mW

DRAM Freq - 1335MHz

All else is Auto

Temps don't go above 71 on OCCT load

Gonna pull an all-nighter prime largeFFT and see if it's stable, then try 4.2 maybe 8)

Any suggestions? The guys at OCC were quite satisfied with current settings..

Thx for all your help anyways, will keep this thread posted

Angle180

I thought you still on that p5k board :)

Anyway looking good if its stable, then happy days mate, however after 4.0 you will have a lot of heat from further raising vcore and you`ll barely notice any improvement.(I know I know, if you CAN why not ;) )

Ok, prime95 5h largeFFT got me to a concerning 80c on core#0, while the rest were at up-to 75max..

This got me thinking.. according to intel, the tcore for q9650 is 75c.. Am I exceeding that at current settings or not? Wth is tcore anyway :D

Maybe I should try to lower the voltage and see if it remains stable, less voltage means less heat right?

Probably going to stay at 4.0, what you said about 4+ seems reasonable and I'm not on a high-end case (chieftec aegis).

Cheers

Ok, prime95 5h largeFFT got me to a concerning 80c on core#0, while the rest were at up-to 75max..

This got me thinking.. according to intel, the tcore for q9650 is 75c.. Am I exceeding that at current settings or not? Wth is tcore anyway :D

Maybe I should try to lower the voltage and see if it remains stable, less voltage means less heat right?

Probably going to stay at 4.0, what you said about 4+ seems reasonable and I'm not on a high-end case (chieftec aegis).

Cheers

Angle180

Well that`s high enough, tcore usually means your chip would throttle down after a prolonged period on above 75C, my experience that quad chips shut down around 90C.

Paired with a high vcore usually means that could damage the chip but whats worse imo that it reduces the chip`s performance over time. What I do is always increase OC in weeks and give a chip a strain and sort of break in. Clock it to 3.4 stable, run it for say 2weeks, increase the speed in 0.2ghz segments every forthnight. This will allow the cpu to sort of break in, gotta tell you its only my opinion, but I did seen temp drops after using this method for a while.

If you`d buy a case with better airflow you might be able to reduce temps however if not you might want to consider lapping your heatsink/cpu if you feel confident enough.

That could give you massive temp drops, I managed to get down from 73C to 56C at 4ghz on a x38 board 8)

I'm definetly not confident enough lapping my cpu but heatsink.. :idea: Seems reasonable, and quite an easy task

Thx for the tip

This got me thinking.. according to intel, the tcore for q9650 is 75c.. Angle180

I've searched Intel specs numerous times for temp specs and never seen anything regarding a tcore limit. I assume you're referring to Tjunction. Can you link to your source for that?

[QUOTE="Angle180"]This got me thinking.. according to intel, the tcore for q9650 is 75c.. jevery57

I've searched Intel specs numerous times for temp specs and never seen anything regarding a tcore limit. I assume you're referring to Tjunction. Can you link to your source for that?

Same thing, some people call it tcore, refers to max factory core temp...

Right, but I've never seen an Intel document that lists Tjunction max, only Tcase max which Intel limits to 67.9c for the 9650. I would love to know what Intel recommends for Tjunction for the 9650 and the Bloomfield i7s.

According to coretemp my tjunction is set to 100

source for intel Tcase is here - http://ark.intel.com/Product.aspx?id=35428&processor=Q9650&spec-codes=SLB8W

but what is tcase and how do i know i'm not exceeding it..

Anyway I managed to lap my cooler heatsink with moderate success.. Bios idle temps went from 42,5c - 38,5c

Currently running OCCT linpack test (max memory util.) and 5 mins in the temps haven't exceede 75c.. Although they probably will after an hour.

@ 1.35vcore now, all stable

.

I was wrong on the 9650 specs, Intel specs a maximum Tcase of 71.4 and voltage at 1.362

Tcase is the temperature at the center of the processor casing. It's my understanding that what is reported in the BIOS and by programs such as ASUS PC Probe II, (I believe you have an ASUS board). On my i7 these temps, (Tcase), are about 10-12c less that the highest core temp.

.

.

As Intel only specifies a limit for Tcase, I monitor PC Probe II and keep it under 70c, (which correlates to about 81c on Core 0.)

Tjunction max of 100c in RealTemp is about where the chip begins to initiate thermal management self-protective throttling, which would be way above Intel's limit of 71.4 for Tcase.

I'm conservative so I limit my 24/7 overclock to about 20% over stock. That's 3.6 Ghz for both of my chips and is achievable on stock voltages. A moderate OC also gives me a lot of headroom to maintain temp recommendations under load, but to each his own. Some people will run a constant 45-50% OC and never think twice about it. (e.g. an i7-920 @ 4.0)

Ok, thanks for clearing that up.

After lapping my cpu-cooler my highest coretemp has yet to pass 75c so I should be well under the max tcase of 71.6 according to your theory =)

I'm under 33% OC, planning on 24/7 longterm and @ 1.33-1.35v (setting 1.35) vcore..

Should be a settings for a moderately longlife setup I'm hoping.. I mean I know q-series is already outdated but I just spent a 150$ on the rampage extreme so gonna milk this pupppy for as long as possible =)

btw, niiiiice rig dude.. i should get some pics of my modest pc too ^^

Damn, I wish my QX6700 pulled 4g, hell I wish my car could too when braking.

The Titan Fenrir would be a great choice.

Daytona_178