Now I've finally got my TRUE rev. C installed on my Athlon II X4 630 (Propus), I want to overclock it but I'm totally confused by all the guides on the Net that talk about reducing the NB, HT and RAM multipliers, as my BIOS doesn't seem to have those settings, despite being regarded as a good board generally for overclocking (Biostar TA790GXBE).

I can reduce the HT Link speed to 1600 Mhz (although CPU-Z shows it's running at 1840 Mhz) and increase the FSB but I can't figure out the rest. I've uploaded the pdf of the BIOS manual here if anyone has the time to look at it (p.33 is where the manual overclocking section starts) to see if they can explain what I need to adjust.

This is what I've got/adjusted so far:



With those settings, I've been able to set the FSB to 230 to get 3.22Ghz @1.456v, the RAM runs at 460Mhz and it managed 5 passes of IntelBurnTest OK. CPU temp Min/Max was 24/40c.

Even though it passed IntelBurnTest OK, something didn't quite feel right and there seemed to be strange graphical glitches (I'm running a 6950 now, with the IGP disabled) that I hadn't experienced before, which is why I think I'm missing something important. I really want to get the FSB up to 250 for 3.5Ghz, but I did have a go at 250 and 240 but got BSOD on both. At least it's booting since I reduced the HT Link to 1.6Ghz, 'cos before I did that it wouldn't even POST and I had to keep resetting the CMOS with the jumper!

You can do the old fashion BIOS overclocking but if you do you lose out on Cool n Quiet. You literally have a hotter system, and you'll likely be dealing with more noise for little or no gain vs other methods.

Myself I left the BIOS options at normal/reliable settings (no overclock during boot) then use K10STAT to overclock on the fly if I need it or underclock when I don't.

See viewtopic.php?f=28&t=58198. Working K10STAT profiles have several advantages over BIOS overclocking.


As to the rest of your BIOS settings:

Check with CPU-Z which revision of the X4 630 you have. If it is C3 I'd turn C1E on, if its C2 I'd leave it off.

Spread spectrum I'd leave at the BIOS default (ON or OFF as set by the BIOS), only disable this no matter what if you are doing extreme overclocking which you probably should avoid anyway.

CPU vcore I'd set back to default.

Memory voltage I'd set to the default unless you have problems and then only go 0.1v higher to stabilize if needed. Check the SPD tab of CPU-Z to see what memory settings should be for your RAM. Look for a column labeled JEDEC #x with 400 MHz in the frequency row and use those values.

Chipset and HT voltage should be at defaults, no extra voltage needed.

Be sure to set all the FSB back to default or 200 or whatever after lowering the voltage back to stock.

Yeah, I don't think this will work with the X4 630, as it's multiplier locked so the only way to overclock it is to increase the FSB.

If you are multiplier locked the FID won't change but you can change the DID and voltage. It can be useful for over or undervolting but it won't help you in most overclocking situations.

One option would be to make sure your bios is at least 78DGA902.BST 2009-09-02 then sell your X4 630 and get one of these:

X3 720 BE $75 95W TDP
X4 9950 BE $80 125W TDP (be sure it isn't the 140W version)


another option is to just go for milder BIOS overclocking with the X4 630. Like say stock voltages and the FSB at 210 or 220.

I've taken my X3 720 to 3.2 GHz for short tests but I can't see a good reason to overclock it in daily use so I'm undervolting more than I'm overclocking.

Keep in mind that MHz isn't the only metric that matters for overall performance. The X3 720 has 6MB L3 cache which will make it faster than the X4 630 in some cases. the X4 9950 has 2MB L3 cache and will be faster in all cases than the X4 630 but still might be slower than the X3 720 in some cases.

But any BE CPU will let you use K10STAT for more flexible clocking options.

I'm not interested in changing my processor thanks. I've actually got an X3 720 on my other PC, which I've overclocked and undervolted at 3.2Ghz (whilst it does go a little bit higher, it needs quite a voltage boost to do so and I decided it wasn't worth the substantial temperature increase).

I'm not sure that the L3 Cache would make much difference for gaming, which is what I'll be using this PC for (as well as an HTPC). It seems OK at 230 FSB now, after I bumped up something else in the BIOS (can't remember what just now!), so hopefully I'll be able to get it up to 250 FSB with a bit more tweaking.

I wonder if Phenom MSR Tweaker will be able to downclock the CPU at idle, even if overclocked in the BIOS?

I'm using the Athlon II X4 630 on a different board now, Asrock 880GMH/U3S3, which I'm setting up for my brother.

Just getting round to setting up the overclock. I've got

FSB - 230Mhz
CPU - x14 = 3220Mhz@1.4v
NB - 2300Mhz@1.175v (NB Core - 1.19v)
HT - 1840Mhz (next step is 2070Mhz)
DDR3 - 1532Mhz (next step is 1840Mhz and it's Crucial only rated at 1600Mhz). Bank and Channel Interleave are Enabled, Power Down - Disabled
UCC - Disabled
CPU Active Core Control - Disabled.
Onboard GPU - Default

I need to stability check but as least it's booting now, whereas before it was messed up just changing the RAM from 1333Mhz to 1600Mhz but I've selected Profile 1 now which seems to be needed to set the right timings. It does idle at 920Mhz instead of 800Mhz now though, so it might be worth using PhenomMSRTweaker to reduce that and the idle voltage but I'm not sure it'll make much difference. Strangely, HwInfo still shows the NB as 2000Mhz but CPU-Z shows it correctly as 2300Mhz.

I'm not sure what the NB Core is or if I need to increase any of the voltages. I've kept the HT under 2000Mhz as I understand that's necessary. Any thoughts/advice welcome.

Hmm, IntelBurnTest caught instability on the second loop, so should I increase the CPU voltage?

Yes, it might help. I have a 620X4 (no L3) that runs 3.5Ghz stable (for years now) at 1.3695 using mini_Ninja in a NSK2400.

Hmm, that's lower voltage than what I'm currently using for 3.2Ghz. It's set to 1.4v in the BIOS but I think CPU-Z showed around 1.38v.

fwiw, my mobo is an older Asus M4A75TD. I have memory over volted to 1.695v

Other settings:

292
12x
9x
1.3695
LoadLine cal 19.35%
HT 1400
Memory running ~1100, 2x2Gb

Running Win7 Ult and OpenSUSE 12.3 (dual boot).

This gives me very stable CPU temps (55C) when this machine is at full load (encoding using x264_64) and memory speed doesn't seem to affect performance for this particular load.

This CPU has no L3.

Hmm, looking at my settings I used on the previous board I see I had to set HT to 1600Mhz for it to work. I was assuming setting it to 2000Mhz or just under would be correct, so that's probably why I've been having problems. I even noted that problem in the first post

I'll go and try again!

My board has Load-Line Calibration - Auto and says when manually overclocking it should be set to 1/2 for AM3+ and Disabled for AM3, so I guess I should set that to disabled. There's also a CPU Dynamic Overclocking setting and I don't know what that does but it's set to Disabled by default so I guess I should leave it like that.

Well reducing the HT didn't help.

Asrock pointed me to this table which shows that officially Athlon II X4 630 on a 4-slot board with 2-slots populated only officially supports 1066Mhz RAM http://support.amd.com/us/kbarticles/Pa ... guide.aspx

Testing with the RAM speed set to Auto, which sets it to 960Mhz when overclocking or 1333Mhz otherwise, I managed to Post up to FSB 240Mhz reliably, with the CPU, NB and HT all reduced to keep them at or below their default clocks, so it does seem I need to keep the RAM speed below 1066Mhz. At FSB 250Mhz or even 245 it didn't post reliably though.

I was also able to bump up the CPU to 240*14=3.36Ghz and post OK. Would have liked to get to 3.5Ghz but it's better than nothing. I haven't tried with the HDD connected or tested for stability in Windows yet though. Maybe if I disable the SATA3 chip and just use the SATA2 it will go further or maybe it's the onboard GPU that's holding it back (although the PCI-E speed is pegged to 100Mhz so I don't think the GPU should be overclocked).

I've had some success using OC Mode - CPU Mode, which lets me select a percentage to overclock by. 5% tested stable with IntelBurnTest, so I tried 20% for 3360Mhz. That bumps up the CPU-NB to 1.25v despite underclocking it to 1920Mhz. It set the HT to 2100Mhz so I reduced that to 1920Mhz manually.

It set the CPU to 1.35v and the NB Core to 1.19v (default 1.10v) which seems strange as what I've read says to leave this alone. It set the RAM to 960Mhz. At those settings, IBT failed (well Linpack actually) on the first run, so I bumped up the CPU to 1.40v and that failed on the fourth run.

Temps were CPU 62, Core 62, MB 42. I believe the Athlon II X4 630 is rated for 71c but that's getting a bit close for my liking, so I'm not sure if I should increase the voltage to 1.45v. My CPU fan (on Xigmatek HDT-D1248 top-down cooler) ramps up from 1000 to 1400RPM at 50c and my front fan ramps up from 600 to 750RPM at 60c. My exhaust fan runs at a constant 750RPM.

Here is a shot from today running an encode, it's really cool today, about 70F in my office, so temps are a bit lower than normal. I usually go for the NB Freq setting (in CPUZ) stable as opposed to straight CPU clocks. Notice I'm single channel on the memory, I had a channel go bad a while back on this mobo.

O/C Pic AthlonII 620

Well it's quite cool here today (indoor thermo reads 25c) and the PC is right next to an open window.

Your temps are a lot better than mine but maybe that's partly because IntelBurnTest heats up the CPU far more than anything "normal" like doing an encode and partly because your fans are running much faster than mine.

At idle, I've currently got my exhaust fan running a bit faster, at 930 RPM, my front fan at 590 RPM and my CPU fan at 1054 RPM.

At load, the exhaust (Power) fan stays the same as it's not a variable speed motherboard connector, the CPU fan hits 1490 RPM and the front (Chassis) fan 1000 RPM, so it's mainly your Chassis fan (I assume that's your front intake fan) that's running a lot faster than mine but my Arctic PWM is horribly loud even at 1000RPM but very quiet at 590RPM, so I wouldn't want to run that any faster at idle. For that, I get CPU 62c, Core 62c, MB 37c. I guess I could reduce the target temps and see if that helps (currently CPU 50c, Chassis 60c).

I disabled Spread Spectrum and Cool'n'Quiet in the BIOS but left Load Line Calibration on Auto and it passed five runs of IntelBurnTest Standard at Auto 20% overclock (FSB 240, CPU 3360 @ 1.40v (auto was 1.35v), HT and NB 1920, NB 1.25v, RAM 800Mhz@5-5-5-19-2T (it autoset 960Mhz before so I'm not sure why it choose 800Mhz this time, 800Mhz isnt even presented as an option most of the time, with 960Mhz being the lowest, so if that is the best setting I can't even select it!).

Even though IntelBurnTest passed, I have DVDFab Passkey set to load with Windows and that crashed. This is only the second time this has happened and both times were when overclocking, which makes me think it's not as stable as IBT might suggest.

My fans are mislabeled. The one labeled chassis is actually the CPU fan (mini-Ninja w/80mm fan using SpeedFan/PWN) and the other two fans are 120mm case exit fans (using manual fan controllers/Fanmates) and yes, when encoding, this machine is a bit loud, but in "low power mode", with the fans at min (and the CPU fan off), it's very quiet.

In my O/C testing, I find that doing and encode with x264_64bit using an AVS source with QTGMC to deinterlace an interlaced source provides a serious beating on the CPU and memory, so much so, that I don't even use Prime95 anymore for stability testing.

fwiw, it seems like your voltage is high for a "mere" 20%, but then, all chips are different...what does your NB Freq show in CPU-z?

Ah OK, that might explain why your CPU is much cooler than mine then. What fan do you have on the mini_Ninja though, as I guess a 80/92mm would spin faster anyway to generate the same amount of air? I used to use one of those and found it OK but I'd have thought the Xigmatek HDT-D1248 would do a bit better, or at least as well. Looking at these results, it seems it's not the best heatsink and works better on 65mm Kentsfield CPUs than 45mm Yorkfield CPUs but I'm not sure which the Athlon II X4 is closer to http://www.xbitlabs.com/articles/cooler ... 84_12.html

NB and HT are both at 2000Mhz at default and I set them to 1920Mhz when testing the overclock.

I did have to use 1.456v on the Biostar board with the TRUE to get this CPU to 3.22Ghz, so it does seem like I've just got a poor performer here. Not that this would be a problem if it weren't for the temps. Even if the Xigmatek isn't quite as good as the TRUE, I'm having trouble understanding the much higher temps I'm getting with it:

(I'm being generous and assuming the recorded figure I have with the TRUE is Core and not CPU, I'll see if I've got any fuller records somewhere).

Idle: Xigmatek CPU 37, Core 27 @2.8Ghz(800Mhz)@1.36v(1.10v) vs TRUE Core 24
Load: Xigmatek CPU 62, Core 62@3.36Ghz@1.36v vs TRUE Core 40c@3.22Ghz@1.456v

so either way, that's 22c higher at .1v lower!

I wonder if I didn't apply the TIM properly but I used (the last of my) Arctic MX-2 compound and applied it by filling in the channels on the HDT base and then applying two short lines, as shown here
http://archive.benchmarkreviews.com/ind ... mitstart=5

I was using Arctic Silver with the TRUE but I don't think that should make much difference and the MX-2 should perhaps be better if anything (or safer at least, not containing any metal). It'd be a major pain removing and reseating the heatsink as it would require disconnecting everything and removing the motherboard from the case to get at the nuts, so I obviously don't want to do that if the heatsink is performing as good as can be expected but if it's majorly not working properly then it's best that I fix it, even if only to minimise temperatures at default clocks and prolong the life of the CPU. I'll have to order some more MX-2 to do this but I might as well get some anyway as I'm sure I'll need it for something else.

Hmm, this is interesting http://www.xbitlabs.com/articles/cooler ... 284_5.html

It suggests that the heatpipes need to be pointing upwards/at the bottom of the heatsink or else it doesn't work very well.

Well I took the heatsink off to re-orient it and as you can see from these photos, the TIM coverage was awful, which I don't understand as I filled in the channels and applied two short lines, as shown at the bottom of this page http://archive.benchmarkreviews.com/ind ... mitstart=5

Any advice on how I can get better coverage when I re-apply it this time? I'm using MX-2.

Well this time, instead of putting the two lines of TIM on the dividers, as shown at the bottom of this page http://archive.benchmarkreviews.com/ind ... mitstart=5 I put it on the two centre heatpipes instead and it seem to have spread rather better.

I did use a bit too much TIM on the left-hand middle pipe as it came out a bit quick (new tube probably had something to do with that, nothing to do with my shaky hands and poor co-ordination!). I'm not sure it's ideal and it looks like there might be some uncovered area in the middle of the left-centre pipe. It's hard to tell exactly what part of each pipe is actually making contact with the CPU. I don't think the TIM pattern tells us much and although I place the heatsink on the CPU as square as possible, I have to move it around a fair bit to line up the holes with the crossbar, so any analysis of the TIM spread would probably be skewed by that. With the pipes at the bottom of the heatsink now, I can see that the CPU is slightly larger than the heatsink base, so that the bottom edge of the CPU isn't properly covered and all it has above it are the pipes as they come out of the base and start to bend but I don't suppose that'll matter much, as long as the centre is covered and in good contact.

Looking at my previous heatsink photo, I'm wondering if the dark patch of TIM is where it's heated up, in which case it would seem like the heatsink was only contacting the CPU on one side for some reason, which can't be good.

Anyway, I obviously can't tell exactly how well it's spread without removing it each time, which is obviously stupid, so the only real way to tell if I've done a better job this time is to see if the temperatures are lower, although in large that will probably be due to re-orientating the heatsink, as that review showed that it doesn't work well other than with the pipes at the bottom. I might test with the board laying flat and then with the case upright though to see if there's a major difference.

my voltage on my "crap" amd 1100t system is much higher than the voltages around here. I'm at 1.46 volts to achieve 4.2ghz 21x200 on the amd phenomII x6 1100t. And this is in a crappy dell case with 1 intake fan and 1 exhaust and a corsair h80 in push pull. Prime 95 pushes temps to 58C in an air conditioned house at 67F. But regular usage sees temps in the low 50's maximum while gaming or transcoding. Gelid GC extreme TIM applied to corsair water block

I guess you have to be careful about comparing voltages between different CPUs but anyway, I needed 1.456v to get this Athlon II X4 630 to 3.22Ghz on my old Biostar board. It seems some CPUs (or versions of) just don't overclock as well as others, or need more juice to do so. I don't really expect to be able to hit 3.5Ghz like noee has with his 620, at 1.3695v, when I probably need around 1.45v to hit 3.36Ghz (was unstable at BIOS 1.40v although CPU-Z was only reporting 1.365v so maybe I'll only need 1.4v (actual, not BIOS setting) to run stably at 3.35Ghz).

OK, so either the TIM re-application or the re-orientation or a combination of both has definitely helped.

Stock (FSB 200, CPU 2.8Ghz):
Exhaust fan (Xilence) 920 RPM, Front Intake Fan 610 RPM, CPU Fan 1035 RPM
Idle: CPU 35, Core 24, MB 39
IntelBurnTest (Standard, 5 runs): CPU 50, Core 42, MB 42. CPU Fan 1454 (spun up at around 48c with target temp 45c). Front fan didn't speed up from 600RPM (target temp 55c). CPU-Z shows 1.36v

Overclock 20% (FSB 240, CPU 3.6Ghz@1.40v in BIOS), CoolnQuiet disabled. HT 1920Mhz, NB 1920Mhz@1.25v. RAM 960Mhz@5-5-5-15-19-2T
Fans as above.
Idle: CPU 38, Core 28, MB 41 (CPU-Z @1.368v)
IntelBurnTest: CPU 56, Core 48, MB 42, CPU Fan 1480 RPM, Front Fan 610 RPM (CPU-Z: 1.416-1.424v)

So the Core has come down from 62 to 48c, which is great, but the CPU has only come down from 62 to 56, which seems a bit strange. If I'm trying to keep the CPU around 55c, it's pretty much at it's limit already. It's a bit confusing that noee's CPU temp is actually lower than his Core temp whereas my CPU is quite a bit higher, at idle and load. My idle temps haven't changed with remounting the heatsink.

IBT passed the 5 runs when overclocked but it did that before and then I got a BSOD when using Iron Portable, so obviously IBT isn't sufficient to establish stability and it may well need some more voltage but that will obviously increase temps and I don't think I can really afford to do that.

It might be because of the orientation of my CPU fan and where the "CPU" sensor is located. Instead of the fan pointing to the side (where the two exit fans are on the NSK2400), I have it sitting right on top of the memory, pointing to the rear of the chassis. Thus, much more airflow is directed past that side of the CPU, it's also where the caps are located, so they're seeing more airflow as well.

Your results with the TIM are not unusually, IME.

fwiw, I don't have any projects going on right now, so here is a shot with the machine in K10Stat "low power mode". This is how the machine is used in "everyday" mode when not encoding.

AthlonII K10Stat Low

Looking back, I wonder if it was the CPU temp that was 40c on the Biostar board or the Core. I was using the TRUE so maybe that was cooling the socket better than the Xigmatek is or maybe it was reducing the Core temps better but it seems unlikely it would get it down to 40c at 1.456v when I'm getting 48c now at 1.4v.

Even at 1.425v it's not stable at the moment at 3.36Ghz, so I might try 1.45v. Increasing the FSB from 240 to 250 and it won't boot, even with the NB and HT at 2000Mhz (RAM goes up to 1000Mhz from 960Mhz). Leaving the FSB at 240 and trying to increase the RAM from 960Mhz to the next step which is 1237Mhz and it won't boot, even if I try the 1333Mhz or 1600Mhz timings.

So my choices are go for a relatively modest overclock (3.22Ghz-3.36Ghz, whatever's stable) but with the RAM at 960Mhz or run at stock clocks and have the RAM at 1333Mhz. I want to keep this system stable for my brother and not unduly wear it out and reduce it's lifespan, so I'm going to have to be relatively conservative and not go crazy with the voltages to try and overclock it.

Whatever I do, I want to use PhenomMSRTweaker (I find it easier to understand than K10Stat) to undervolt at idle, either underclocking as well, as you've done or maybe setting the idle clock higher, say 1600Mhz but at lower voltage than it currently uses (1.10v) for 800Mhz, so it would be using less power at idle than default but running faster.

I see from your screenshot that our temps are pretty similar at idle. My Core is a little lower but my CPU fan is running faster (1040RPM vs 826RPM) than yours at idle, so that's not surprising. That's as low as the BIOS goes but I've got a beta BIOS that allows it to go down to around 800RPM, so I might use that although it's not particularly audible as it is, not above the HDDs anyway, I guess it would be more worth it if I was running from SSD, with a HDD only for storage that was asleep most of the time.

Little tip for you. If you click Configure at the bottom of the HwInfo Sensors page, you can go and change the labels so that the CPU fan is correctly identified and matches Speedfan, where I can see you've got the CPU fan correctly labelled.

OK, I think I'll give up on overclocking as this board just doesn't seem to like it and go for a nice cool and quiet underclock instead.

I'll see if I can get it to idle at 700Mhz@0.75v as you've done first. For the other end, what voltage do you think I might be able to use for 2.8Ghz? Currently it uses VID 1.40v for Vcore 1.368v. Do I need to reduce the CPU voltage in the BIOS or can I leave that on 1.4v and just use PhenomMSRTweaker to reduce it?

EDIT: Hmm, seems I can't reduce the multiplier below 4 in PhenomMSRTweaker so I can't go below 800Mhz but I've reduced the voltage from 1.1 to 1.0v and locked it in that P-state with Power Saving mode and it seems to be OK. I guess next I'll try increasing the multiplier to 8 for 1600Mhz and see if that's stable at 1.1v (default is 1.2v).

EDIT2: I've started a new thread now that I'm going to be undervolting rather than overclocking viewtopic.php?f=28&t=66511