A touch of instablity. New OC'd FC-PGA rig.

Klosters

Golden Member
Oct 9, 1999
1,428
0
0
My new FC-PGA 550 @733 Mhz setup is a wee bit less than 100% rock solid. My
CPU cooling setup is less than stellar. I'm using a Vantec Si5(40mm tall)HS
with a Delta 38 cfm 60mm fan. The HSF was applied with Arctic Silver, and a
copper shim. The PAL35t I bought is still in its box, because I thought I
wanted a CUBX, but bought the P3B-F instead.

These are my machine's problems. In the middle of the connection process to
both FreeI and Juno(they're not on the machine at the same time. Juno is a
jealous bastard) I get a spontaneous reboot. This happens maybe once in ten
times. I've also had the X Window snap shut twice while online in SuSE
Linux 6.2 with USWest, all right, Qwest already.

It doesn't seem to matter if the Ram Bios setting is 3-3-3-10, 2-2-2-10, or
3-2-3-10.

I've had no BSOD's, and the Dimm test from .simmtest.com indicates all is
well in the Ram department. With a mainboard temp of 26C, my CPU temp at
~idle is reported as 38, 39 and 40C. With a mobo temp of 28C, MBM 4.18(with
12C added to the raw CPU temp as "compensation." With no compensation, mobo
and CPU temps are identical) reports ~43C CPU temp.

At a mainboard temperature of 31C, CPU temp is reported as 44C, 45C or 46C,
at ~idle.
I ran Final Reality overnight about ten days ago. No problems at all, and
the highest CPU temp recorded by MBM was 63C. After adding ~250 Mb of progs
to the basic SE installation, FR isn't happy. I get a freeze and a rather
loud "horn" from the multimedia speakers(not the PC's internal speaker)
just a few seconds into the benchmark process. This necessitates the use of
the "Reset" button.

I've had the CPU at the lowest possible Core Voltage setting, 1.6v in Bios.
But the actual reported Vcc is 1.65v. I ran the QII Crusher benchmark
without any probs, many times. This AM the prog ran fine, however when I
got the console telling me the fps when the demo finished, the cursor
froze; another Reset was needed. I bumped the Core Voltage, for the first
time to 1.65 in Bios, which was reported as an actual 1.7v in MBM.

This time, the Crusher BM timedemo made with the instant "horn" from the
desktop speakers. Another Reset, and I set the Vcc back to 1.6v along the
way.

If you've continued reading this far, thanks! This is in the box:

PC Power&Cooling Silencer 275 ATX PS.

P3B-F January 2K release date, but it came with Rev. 1005 Bios.

Iwill II Slocket
P3 FC-PGA 550, at 133 Mhz FSB, Clock Speed 733 Mhz.

CL TNT PCI Vidcard

CL SB PCI-512 Soundcard(with both the latest driver as well as the Liveware
update(for PIIIE compatibility)

3 64 Mb CAS3 PC-133 Samsung GA Dimms
1 64 Mb CAS3 PC-133 Samsung CT-75, or TC-75 Dimm. Sorry, I can't remember
if it's "TC" or "CT." This odd man out is in Dimm Bank 0, which is the slot
closest to the CPU.
All Dimms are single-sided.

*From what I understand, even though I always get a proper Ram count-up in
Bios, and Windows and Linux always report 256 Mb, no mainboard can really
run correctly with four Dimm slots filled. The RAM is fine, but that's not
the issue. ATX, PC98, PC-100 Sdram, Jedec or whatever specs make provision
for 3 Dimm slots, and ONLY three Dimm slots.*
Is this a true story?

Better 42.5 80mm's are on the way, and I'll order an Alpha PEP66 today too.

All help in troubleshooting this little nagging problem is most welcome.
Thank you for your time and effort!








 

AdamK47

Lifer
Oct 9, 1999
15,760
3,594
136
Has anyone noticed that the top of this thread contains a post that shouldn't be there, or am I just seeing things?
 

lupin

Platinum Member
Oct 11, 1999
2,944
0
0
YEs, it's there.

Uniden UniPRO - http://www.uniden.com/docs/product/prdetail.cfm?product=PC100

4MB of RAM, WinCE 2.0, Voice recorder, *BUILT-IN* 28.8 v34. modem, CompactFlash slot. Barely used. Comes with docking cradle, case, and software. Asking $95 shipped (or trade).


I Want:
64 megs of PC66
CD-ROM
I'll pay $105 Shipped for a Nomad 2

-------------------------
This is me typing.
 

BUTCH1

Lifer
Jul 15, 2000
20,433
1,769
126
Do you have the same problems at default CPU speed?
Maybe try more volts, but monitor temps close...
 

Klosters

Golden Member
Oct 9, 1999
1,428
0
0
Thank you for your reply, BUTCH1! Everyone else on this thread is not concerned about my post. I dunno why the OT posting at the top of this thread is there either, but I can't do anything about it.

I don't think raising the Vcc will help. As I wrote, doing this earned me the "horn" and a system freeze at the very start of the Crusher timedemo in QII.
I haven't backed the FSB down to 100 Mhz. I've read posts by folks that state that they NEVER could get the CPU to run at its former OC'd FSB after backing the FSB down to "stock." Call me paranoid.

When the new HSF and case fans arrive, I'll remove the Dimm from slot 1. Hopefully, this and better case and CPU cooling will earn me a rock solid mosheen. Thanks again for your on-topic response.
 

Eug

Lifer
Mar 11, 2000
24,120
1,766
126
Hey, I've noticed an off topic message at the top. Why do you think that is? (Just kidding. ;))

"I haven't backed the FSB down to 100 Mhz. I've read posts by folks that state that they NEVER could get the CPU to run at its former OC'd FSB after backing the FSB down to "stock." Call me paranoid."

OK, you are paranoid. If you can't run it the higher FSB after backing down, it probably wasn't ever stable at the higher speed anyway, or something else has changed.

Run it at stock speed and make sure everything is fine first.

Also, make sure everything is seated correctly. I've gotten spontaneous reboots when cards or the CPU weren't perfectly stable in the slots. (However, that may not apply, if you're getting consistent reboots with specific software. Sure it's not driver related?)

Didn't reply before because your message was so d@mn long. :p
 

Klosters

Golden Member
Oct 9, 1999
1,428
0
0
I appreciate your reply, Eug. Thanks! This "I backed the FSB down, when I reset it to the higher speed, she just won't run. Even after raising the Core Voltage" syndrome isn't an isolated phenomenon. I've seen this problem reported about five times, and three of the reports were from hardware Web sites. This seems to be a feature of FC-PGA's. If the box was probably was unstable anyway, why did it boot into the higher FSB and run basically ~OK for a few hours or days the first time? I don't think I'm overly fearful about this. And I haven't seen a fix for this problem posted anywhere, either.

I don't overclock CPU's for the thrill of it. I OC because I can't afford a faster CPU. The "gotta have it" illness bit me. The rub is that a PIII FC-PGA 700 now costs what the 550 did two months ago. Live and learn--maybe!

You do post important questions. The URM's supplied with the P3B-F are much easier to use than the one's that came with the P2B were. However, they sure as hell do not provide the rock solid support for the Iwill Slocket II that the old ones did. I'm so looking forward to removing them and the mainboard as well, and replacing them with older URM's that need two screws installed from the bottom, or threading permanent tie-wraps through the URM mounting holes, from the bottom of the mobo. NOT!!

Oh well. It's always something! Thanks for your response!
 

jose

Platinum Member
Oct 11, 1999
2,079
2
81
Hello Klosters ,

Just an opinion, I had problems w/ Iwill2 slocket on a P2B-F w/ a c550. The same processor worked w/ an Asus slocket. I would try 1.7v w/ the Artic Silver paste & the Alpha. But try using only 2 dimms. If everything fine then add an addl. dimm.

BTW I go back & forth all the time. 100 - 133fsb.

Keep us posted, hope this helps.

Regards,
Jose
 

Nevin

Senior member
Oct 29, 1999
292
0
0
The temps you are reporting are too high for your current cooling setup. The Asus board you have is capable of reading and displaying the actual CPU temperature as measured by the processor's internal thermal diode so I am confused when you say,"With a mainboard temp of 26C, my CPU temp at ~idle is reported as 38, 39 and 40C. With a mobo temp of 28C, MBM 4.18(with 12C added to the raw CPU temp as "compensation." With no compensation, mobo and CPU temps are identical) reports ~43C CPU temp.

Is there is a chance that the copper shim is too tall and is actually lifting the heatsink off the CPU core or that even with the shim, the heatsink is not flush to the core? Most of the problems you are reporting do sound like they could be heat related problems.

There is no scientific reason why setting your bus speed back to default would prevent the CPU from being overclockable in the future anymore than driving your car across town at 30 MPH would prevent it from ever going 70 MPH on the freeway again. A phenomena of the internet is that one person will post his experience to 5 or 10 different forums and it will appear that 5 or 10 different people had the same experience where it was actually only one.

Nevin
 

Klosters

Golden Member
Oct 9, 1999
1,428
0
0
Thank you for your suggestions, jose and nevin! The Iwill Slocket II worked perfectly with my P2B, running a C366@550. The same Slocket and CPU combo worked fine on the P3B-F. The Iwill seems fine with the PIIIE 550@733 too. The Slocket isn't held in the Slot 1 as firmly as I would like, as I mentioned in my earlier post. The new Asus URM's that came with the P3B-F make installing and removing the Slocket much easier. The P2B's pre-installed URM's made it a beatch to remove the Slocket, no joke.

It's possible that the HSF(it's a weakly little 40mm tall Vantec Si5 HS, with a 38 cfm Delta on top of it) isn't aligned properly. I eyeballed the CPU/HSF joint carefully. The copper shim's from 3dfxcool, whatever that's worth, and I did use Arctic Silver optimised for a small area. The important point is that if you have a small HS, a wind tunnel fan doesn't mean poop.

I removed the 64 Mb Dimm from Slot 0. Now 1,2,3 are populated with identical 64 Mb single sider's.


This has helped stability, crossed fingers. Thanks a bunch for your help on this!
 

jose

Platinum Member
Oct 11, 1999
2,079
2
81
Hey Klosters after reading your post, I check my temps.

700mhz 1.65v = 53C / 127F
933mhz 1.65v = 58C / 136F

I think I have the Hotest System around. Literally ;)

I think I better get me a Alpha & some artic silver real fast.

Regards,
Jose
 

Klosters

Golden Member
Oct 9, 1999
1,428
0
0
It depends if your reported CPU temps are at ~idle, or while running a QIII or suchlike benchmark. With a mainboard temp of 30C/86F, MBM reports a CPU temp of ~42C/107.6F, at "idle." But running the QII Crusher Timedemo peaks the CPU at 67C/152.6F. Lowering my FSB from 133 to 124 Mhz had no effect on the lower CPU temp figure. At load was a different story. I got 62C in the Crusher BM, instead of 67C.

My new case fans got here today. Yaay! With luck, the PEP66 will be here tomorrow or Saturday. Happy and cool OCing to you!