C2D did o/c, now won't

"Howard Goldstein" <hgoldste RemoveThis @mpcs.com> wrote in message
news:1191603756.1310@news.queue.to...
> On Fri, 05 Oct 2007 13:38:48 GMT, Howard Goldstein <hgoldste RemoveThis @mpcs.com>
> wrote:
> : Memory is solid at 2.1v DDR2-1066 with the relaxed factory 5-5-5-15 2t
> : timings
>
> At the risk of this becoming a stream of consciousness blog on my
> problems the RAM issue could be significant and I know at least one
> other person here had RAM issues before. It says 'solid at 2.1v' but
> that's not the whole story.
>
> This was two Corsair 1024-8500C5D dominators rated at 2.1v but they
> were solid only in the B1 and B2 slots and flunked memtest86+ in test
> 2 almost right away if in A1 and A2.
>
> After the 13:38 post I pulled the two 1024-6400C4s out of the P5W
> board and tossed them in the Striker B1 and B2 set to their rated
> 2.2v. And they too were solid at their rated speeds, but no
> improvement in overclocking.
>
> So I went to pull them out and they were hotter than hell. WTF? Same
> VDIMM as on the P5W DH, better cooling the Striker and they're
> cooking. Didn't notice this with the massive heatsinks on the dominators.
>
> **Well to make a long story short does anyone have any idea where on the
> Striker I can actually get a VDIMM measurement to see what's really
> there? ***
>
> With the VDIMM set to anywhere <=1.9v the 8500C5Ds run at 1066 in A1 and
> A2. Grrrrr
>

First of all, are you trying to use NTune? If so, get rid of it first and do
your settings in the bios. I have found that NTune is just a mess and it, in
and of itself, caused CTDs and lockups. My E6600 ran very well at 3.4Ghz on
the same board. Do use the NB fan that Asus included with the board..... It
works. One other thing is that when you do your OC project, take the memory
out of the equation by setting it much lower than rated speed or use
'Linked' and 'Sync Mode' to keep it from causing any issues when you
overclock the processor. After you find your best stable OC, then work on
getting your memory up to par. At least then, you will know whether the
memory or your overclock is causing the issues. Running with the Vcore set
to auto, I am now running a Q6600 at 3.35Ghz very stable and cool on the
same Striker Extreme.

Ed
>
 >> Stay informed about: C2D did o/c, now won't 

This message is not archived
 >> Stay informed about: C2D did o/c, now won't 

> I couldn't get Ntune to startup - it doesn't even opena window - and
> set that aside. The other one, what was it Nmonitor? That freezes
> the system for me in 30 seconds. All of the things I'm doing are via
> bios 1303
>
> I can't fit the asus fan on the NB heatsink since the Ultra 120 is in
> the way so I put it instead on the mosfet heatsink across the
> processor from it and use a cable tie to tie a 40mm on the funky
> northbridge combo heatsink. Under load I see about 118F on the bottom
> side of the heatsink. I think I degraded the heatsink though as when
> I took it off to replace the compound I left too many scratches. This
> afternoon I'm going to pull it off again, lap the NB and SB (do you
> lap them? If so, how do you safely get the sticky insulator/cushion
> off of the sink so it can be resused?). Why it would help when OCing
> *only* the processor I don't know, but when I put them back on I'll
> secure the pipes with nuts and bolts maybe that'll help get more heat
> off of a power fet or something (?)
>
> I set the processor OCing aside momentarily and dropped back to the
> memory after posting the above. FWIW the version 2.2 Corsair 8500C5Ds
> (apparently with Qimonda instead of Micron memory) are solid at DDR2
> 800 with 4-4-3-3 1T when fed the rated 2.1V (measured with a vom,
> requires setting 2.075 in bios as VDIMM is 30mV high).
>
> I relaxed the timing to 4-4-4-12 2T and in Sync that yields DDR2 800
> and started back on the processor overclocking.
>
> I can't even get partways into windows safe mode boot at 400x9. 377x9
> gets me into prime95 or orthos for 10 seconds before the system hard
> resets - not even a bloody error, just a system reset. It's very
> disturbing.
>
I took the SE out of the box and installed it. I haven't done any mods or
anything to it. 118f(-40ishC) is not bad for that reading. Set your OCing to
Manual and use the FSB/Memory adjustments. I run my Q6600 FSB at 1490 on
Auto for Vcore for a bit over 3.35Ghz. Yes, it is a G0 and the E6600 I had
before was the B0 stepping. You might try and set your memory even lower,
like around 700-750 while working on the processor OC. You just want to
eliminate any memory issues while OCing the processor. I can make you go in
circles....... Even my Mushkin Extreme 9200 will not come close to it's
rated speed even with very relaxed timings. The best I have gotten is 966
with relaxed timings and at 833 with tighter timings I see no difference in
memory performance in Sandra XI.

Ed
>
> Is your Q6600 a G0? My E6600 is a B2 and maddeningly this same
> processor had only thermal issues in the P5W DH at 400x9
 >> Stay informed about: C2D did o/c, now won't 

This message is not archived
 >> Stay informed about: C2D did o/c, now won't 

> You might be making someone who does this all the time go in circles,
> but you're not making me go in circles. Remember that I'm new at this
> OCing stuff. I have a fair background in embedded systems that isn't
> really relevant to this (Vcc < 5v is still freaky weird). Since that's
> all
> obsolete now the repetition is good, it'll sink in better that way.
>
> The 118F is a tad low after lapping the heat pipe assembly, close
> though. The heatsink has a convenient little cubbyhole by the heat
> pipes that the temp sensor can be pushed into. With lapping its
> 125F. I don't know how hot the southbridge is but man, that is hot,
> hot like can't keep your finger on it for 5 seconds wihtout wincing.
> That can't be normal (?). At any rate I surrendered to the urge and
> ordered a couple of chipset sinks and a few packages of the wee copper
> mosfet sinks in case that'll help.
>
> I need to try 1490. I revisited close by at 1506 and was rewarded
> with orthos taking a dump after about 5 seconds...but maybe 1490 is a
> better divider on the boards? Where I'm tapped out now is 1599FSB
> or around 400x8. Memory I've taken down to 667 and manually sliding in
> 5-5-5-27 2T (may not be 27, it was whatever it maxed out at) to try to
> remove that from the equation. I should try lower.
>
> Am inclined to pull the Q6600 G0 out of the p5w to try to rule out
> weirdness on my SE.
>
> Do you remember what voltage settings you used to get your E6600
> rocking?

LOL.........I actually made a typo there.....should have read "it" can
make you go around in circles. My wife says I maker her go around in
circles, but that is another matter...... We have many new folks to OCing
come here, so welcome.... You cannot change the divider on the E6600 from
9x, so that is not going to help you. It is fixed and not changeable even
though your bios may 'say' it has changed. I use liquid cooling, so I have
pretty much gone to the max on the Vcore before damaging the processor. With
air cooling, you would probably want to stay in the 1.4-1.5v range and watch
your temperatures closely. I use 2.1-2.15v for my memory, even though it is
rated at 2.35v because it seems more stable there. If I remember correctly,
I used about 1.45v to get to 3.5Ghz on my particular processor. I have the
Nvidia 680i chipset (Asus Striker Extreme) so you will have to experiment a
bit to see how your MB chipset reacts to voltage and frequency changes. If
you are stable at 1599, remember you are still at 9x and not 8x. That should
be 3.6Ghz in my book and that is a very good OC on the E6600, especially
with air cooling. Try and make sure your temps stay safe. I try and keep my
maximum temps with air cooling under 65C if possible. I usually don't worry
a lot about idle temperatures once I know they are good at default speeds.
Stress (Orthos or Prime 95) temps are much more important. We generally use
'C' rather than 'F' for temps because that is what we have been used to for
the last 10yrs or more........

Ed
 >> Stay informed about: C2D did o/c, now won't 

This message is not archived
 >> Stay informed about: C2D did o/c, now won't 

> Thank you Ed. I've had a great welcome here and I'm really enjoying
> picking everyone's brain here, even the passive pluckings from the
> other threads. You're one of the fortunate ones who making a typo
> continues to make sense. I wish I had that happen to me.
>
> Running wprime for a benchmark I have one hint that my 6600 is actually
> unlocked for downward multiplier changes. Taking it down to x7 from
> x8 increased the benchmark timing from about 28 seconds to about 32
> seconds. I'd really like to throw in the towel but I have this weird
> compulsion to keep moving on until I can get it back down to the appx
> 24 seconds I was getting on the P5W DH with 400x9.
>
> It so feels like a power supply issue, if I can boot it at all above
> 3.2 any combination of FSBs or multipliers that result in > 3.2 causes
> a hard reset, within 10-20 seconds of starting up orthos. But I can
> do it on two different brands of power supplies so, oh I don't know.
> Tonight's try is my last shot in case it's thermal. I'm putting that
> heat pipe assembly away and tossing a Noctua NC-U6 on the NB, a
> Thermalright HR-05 on the SB, and a bunch of little copper swiftech
> ram sinsk on the FETs.
>
> Your voltages are close to what I need for 3.2. I have to feed this
> 1.43 for 400x8 stability. 400x9 at any vcore up to the step below 1.7
> won't even get me past the i/o initialization in xp.
>
> Re: F vs C - I'm trying man! I'm older than dirt, it's hard to get
> the feel for it. But you're right everyone uses it and should use
> it so I better turn everything into C and let time work its wonders,
> get me trained.

I am a lot closer to 60 than 50yrs old, so join the Old Farts Geek
Club........ Keep us posted on your progress.


Ed
 >> Stay informed about: C2D did o/c, now won't 

This message is not archived
 >> Stay informed about: C2D did o/c, now won't 

"Howard Goldstein" <hgoldste RemoveThis @mpcs.com> wrote in message
news:1192405519.22216@news.queue.to...
> On Sun, 14 Oct 2007 13:33:45 GMT, Ed M. <ed RemoveThis @edmedlin.com> wrote:
> : I am a lot closer to 60 than 50yrs old, so join the Old Farts Geek
> : Club........ Keep us posted on your progress.
> :
>
>
> So you walked to school in the snow uphill both ways too. No wonder I
> can understand you. My striker passed away on Friday night/early
> Saturday and took the painstakingly lapped E6600 with it, may they
> both rest in peace. That'll teach me to lap an in-warranty CPU before
> I've stabilized the system it's destined to live in. I suppose I
> should make a nerdy makeup compact mirror with it for Mrs. hgoldste (like
> she'd be caught dead with it)
>
> I've got an E6750 and a DFI lan party UT P35 T2R coming in to keep me
> occupied while waiting the 5 years it takes Asus to RMA the Striker.
> I'll post how that works out.
>
> So I'm back in the Q6600 G0 stepping boat with you for a while. Mine
> runs *hot*. 1.3125V for 333x9 and I'm idling at 68-71C per coretemp
> on the P5W DH. It seems kind of high, doesn't it?. The Arctic Freezer 7
> isn't the best cooler but it sure isn't the worst.

There is something amiss there and I suspect CoreTemp has issues with quads
on the Striker Extreme. Using the Asus monitor and checking in the bios my
liquid cooled Q6600 lists at idle about 26-29C. In CoreTemp, it reads closer
to 40C and I have removed, cleaned and emory-clothed the water block.......
I did everything possible even turning the radiator fan up to full blast and
the idle temps just won't change in CoreTemp. The max temps are pretty close
to what I expect running Orthos or Prime at about 50C (the cores change a
lot). Orthos only does 2 cores max, but Prime95 does all four. I have had
this issue since I flashed the bios to the new (1305??) and then back to
1303. I also updated CoreTemp to the newer version and since all that I have
had weird idle temps. Even when I first turn the system on the idle temps
will be at 40C immediately which is not really possible with this liquid
cooling system. It would usually take at least 10-15mins to reach maximum
idle temp even with the radiator fan on low. The voltages I am using to get
to 3.5-3.7ghz are not high at all at under 1.3v so my idle temps should not
be much more than 30C max. I am not going to lose sleep over it.........

Ed
Yea, I walked 5mi to school in knee-deep snow barefooted........
 >> Stay informed about: C2D did o/c, now won't 

This message is not archived
 >> Stay informed about: C2D did o/c, now won't 

> My bad, 70C in coretemp on the P5W DH with the Q6600, not the SE.
> With almost no variation during prime95 -- something amiss in the way
> ASUS wrote the interface or in coretemp I'm agreeing with your
> suspicion here, particularly in light of your results with water
> cooling. That's a relief. Yeah, unless you're refilling your cooler
> with water out of the hot water heater every time you turn it on or it
> taking 20 minutes to boot it's hard to imaging it *bam* booting
> right up to 40C. I bet when you touch the cooling block, as when I
> touch the copper base of the heatsink, it doesn't feel anywhere near
> as hot as what coretemp's reporting...
>
> (1303 was where I stopped with BIOSs on the SE...)
>
> Anyway I got the E6750 in yesterday (idles at 35C) and know why
> they're cheaper than the E6600s -- its multiplier tops out at x8.
> (*$%&*($% s hould have researched it more it's worth $20 for another
> multiplier step

Yea, the 6750 has less overhead than the 6600 for OC'ing. It just seems that
you should get a bit more than 1303fsb, but I haven't had any experience
with that processor myself so I really don't know. I know my water cooling
is working because there is warm air coming out of the radiator and the
incoming hose is warm. The top of the waterblock is just slightly warmer
than when the computer is off....... It is the Swiftech block with all the
diamond shaped facets in it so there isn't a lot of heat transferred to the
rest of the block, just to the liquid. With the E6600 CoreTemp read fine. It
always read at idle about 3-4deg cooler than the bios. Now with the Q6600 it
reads 10-12degs hotter than the bios at idle....... Something is amiss
there.


Ed
 >> Stay informed about: C2D did o/c, now won't 

This message is not archived
 >> Stay informed about: C2D did o/c, now won't 

"Howard Goldstein" <hgoldste.DeleteThis@mpcs.com> wrote in message
news:1192733263.51859@news.queue.to...
> On Thu, 18 Oct 2007 12:35:36 GMT, Ed M. <ed.DeleteThis@edmedlin.com> wrote:
> : Yea, the 6750 has less overhead than the 6600 for OC'ing. It just seems
> that
> : you should get a bit more than 1303fsb, but I haven't had any
> experience
> : with that processor myself so I really don't know. I know my water
> cooling
> : is working because there is warm air coming out of the radiator and the
> : incoming hose is warm. The top of the waterblock is just slightly
> warmer
> : than when the computer is off....... It is the Swiftech block with all
> the
> : diamond shaped facets in it so there isn't a lot of heat transferred to
> the
> : rest of the block, just to the liquid. With the E6600 CoreTemp read
> fine. It
> : always read at idle about 3-4deg cooler than the bios. Now with the
> Q6600 it
> : reads 10-12degs hotter than the bios at idle....... Something is amiss
> : there.
>
> Not sure if this will help but the beta coretemp 0.95.4 deals with my
> E6750 correctly where the coretemp relase (.95? didn't make a note of
> it) didn't. I haven't tried .95 on my Q6600 G0 and not sure if it's
> going to talk to the quad properly, can't run coretemp in safe mode and
> still trying to stabilize the quad...
>
> I have this weekend set aside for playing with the 6750. teh google
> turns up many posts of folks who've pushed this thing to 470 on air
> with the P35 but haven't seen any reports of it going that far on my
> DFI. I'll be mostly satisfied if I can get 420x8 out of it as that'll
> be better than I ever managed with my SE and it'll keep me satiated
> while waiting for the 45mms
>
> Speaking of which guru3d has a piece today on I think it was a X9850
> (don't remember) pushed to beyond 5Ghz on an X38 board at a Taiwan
> expo in the last week or so. (take with salt grains?)
>
It is probably true. I haven't seen that particular article, but some of the
cooling solutions they use, like liquid nitrogen, are pretty impressive.

> Did you ever post what your liquid setup is and are you cooling the
> power FETs with water? Maybe it's time to stick my toe in the water.
> If you could look beneath my kitchen sink though you'd want me to stay
> far away from plumbing stuff. Given that I probably shouldn't take
> the plunge just yet.
>
Mine is a Swiftech Quiet Power water cooled case. It is based on the Antec
P180. I only use the CPU block for now, but may go a bit further and do the
chipsets later.

> How much Vcore are you pushing into your Q6600? I was almost stable
> at 347x9 with 1.3875 but woke up to find prime95 lost two cores after
> 6 hours. Don't really want to run it at 1.4 for 24/7 which is running
> now (nb., this is on a P5W DH)(the SE replacement could actually get
> back here next week thanks to newegg for going the extra mile when I
> was 2 days beyond the 30 day replacement period)

Right now at 1488fsb I am using about 1.3v and it is solid.


Ed
 >> Stay informed about: C2D did o/c, now won't