Dual 300A at 558

[loophole]

After modding my case by taking out the fan grills and just having large circular holes cut out where there used to be lots of useless little holes, I thought I'd give 558 another try:

So, I set the voltage to 2.3V, and the FSB to 124 (1/4) with AGP at 2/3 (for an AGP bus at about 83), fully expecting nothing to happen (like all the other times), but to my astonishment it actually POSTed and counted through the memory, but stopped after that. Needless to say, i was astounded, but after rebooting I found nothing would happen - monitor in power saving mode and everything just as before. BUT, after turning the system off and on again, it would POST at 558 again as before.

Any suggestions - I'm pretty sure the CPUs aren't what's stopping the system getting to 558, because they barely even get hot and run at 495 (110 x 4.5) at 2.05V. Memory is all rated at PC133, and I'm pretty sure AGP is fine.

[Derek]

You *may* need more voltage. Also try booting with a PCI card - You're almost at 600Mhz!

[InactiveX]

viewtopic.php?t=879

Try using BP6FSB to set your clock speed to 558 after Windows has booted - it may work well.

[Derek: I just corrected the URL to BP6FSB)

[RRLedford]

Remember that besides the heat, E-M noise is still a OC limiting factor. Isn't there one setting that randomizes the clocking speed a few % to help lower the E-M noise factor a little.

[Derek]

Spread Spectrum is what you're referring to. It shouldn't be a limiting factor and is usually turned off. It's put there so the motherboard can pass the FCC standards.

[loophole]

Thanks for all the replies. I'll try it without any other cards in there at the same time (and anything else I can think of as well).

I just got my hard disk back for IBM (it crapped out on me, so I sent it back to Singapore), and loaded Slackware 8.1 and Win XP, so now I can run some benchmarks and more stuff like that.

After cutting the holes out of my case, 495 is now completely stable at 2.05 volts, but it would be ooooh so nice to crack the 500 MHz mark per CPU (even though the performance difference between 500 and 495 is negligible). And such a dream to get 600 each - then my RAM would actually be running at 133.

Now that I've got an OS again, I can also try and see how much VTT drops by during intesive tests. Can anybody recommend some CPU intensive tests for either Linux or Windows?

Under Windows, when I do a burn-in using Sandra, the VTT only drops to 1.47 (when the CPUs are run at 495), which to me doesn't sound so low, what does everyone else think?

And, I'll definitely try to use BP6FSB, now that I've got a Windows version again. I'll let you know how I go.

Thanks for all the help again.

[InactiveX]

A good CPU test is to run two instances of the commandline version of SETI@home. Put the program in two different folders and run them. Both your CPUs will be at 100% in task manager.

(Funnily enough, that's just what I'm doing now to see how cool my new case is.)

1.47 VTT sounds fine.

[Wolfram]

Any suggestions - I'm pretty sure the CPUs aren't what's stopping the system getting to 558, because they barely even get hot and run at 495 (110 x 4.5) at 2.05V. Memory is all rated at PC133, and I'm pretty sure AGP is fine.

The CPUs _can_ be the limiting factor. Usually, 600 Mhz is the limit for the old Celeron A. If you look at the cpu database at overclockers.com, there _are_ CPUs beyond 600, but usually _at_least_ water-cooled, if not compressor or liquid-gas-cooled.

There is a limit for overclocking even if the CPU does not get hot.

558 is very good for a 300A anyway.


Regards,

Wolfram

[InactiveX]

Also, the ones doing over 600MHz are nearly always running in single, not SMP. As far as I know, the only person to get over 110MHz FSB on a dual BP6 was Holodeck2 with his watercooled, peltier-chilled celerons at 733MHz (133MHz FSB). Whoooaahhh!!!

I think you're very lucky!

[hyperspace]

I have 2-366's running at 110 Mhz. Two Golden Orbs and two 80mm's "focused" on the RT3 area of the mainboard. Granted, the BP6 locks up randomly (maybe once a day, average) but I have not changed the EC10 capacitor, yet!

[loophole]

I guess I am pretty lucky to get to 495 dual, but them posting at 558 (and nothing else) to me suggests that they have more room to run faster and that it's the motherboard that's stopping them.

Yeah, the overclockers.com database does suggest a pattern - that to get over 600 for 300A CPUs, one needs to get something with a bit more grunt than air-cooling, but you also see a fair few people who have 300A CPUs at 558 and 600 air-cooled. 600 would max out the motherboard, so I would never be able to go past that anyway.

I'll have to run them in single mode on the BP6 and see if I can reach 558 with either one of them. The only problem is the time it takes to do all that.

After booting at 495 into Windows and using BP6FSB to change the FSB to 124, the system immediately stops responding and the monitor goes into power saving mode. Then, I have to turn the system off and turn it on again, to get it to come alive again.

[RRLedford]

OK, maybe you already have the 2nd Cele-1/366Mhz, or whatever.
But having (2) CPU fans running at high RPM - my poor ears cna't take it.
Let me just say that a 1.1GHz Cele2/128K/100 with Neo-370 adapter will set you back about $95. I've just finished moving my FSB speed up to 110MHz, and Win-XP is running just fine at 1.21GHz! Not bad for a 3-year old system.
I've just ordered (2) 1.1GHz Pentium-III's 256k/100 and will be retiring the old CPUs from both my other dual Cele1 BP6 systems (four 366 Mhz chips 2@505Mhz & 2@466Mhz). The two (used) P3s cost me $125 each & the NEO-370 adapters plus (2) premium copper CPU coolers will add another $100. So, for $175/per system I get another year of life form my dear BP6 systems.
Assuming these two CPUs will also OC to 1.2+GHz, and knowing that I can still do the DUAL-P3 mod later, is there any happier owner of (3) three-year old motherboards in the world -maybe not!

[loophole]

Thanks for all the advice guys, I'll try to overclock further when I have more time -> things have gotten pretty hectic at Uni at the moment. I'll agree that there's definitely a lot more to think about, and a lot more to try.

Hopefully one day 558 will become a reality...
I'll post back when I have some more results. If anything else pops to mind, just add it -> it's all good stuff.

[Blueneko]

Also, the ones doing over 600MHz are nearly always running in single, not SMP. As far as I know, the only person to get over 110MHz FSB on a dual BP6 was Holodeck2 with his watercooled, peltier-chilled celerons at 733MHz (133MHz FSB). Whoooaahhh!!!

I think you're very lucky!

Was that 366 @ 733 or 300A's @ 733, and was that single cpu or dual??

[loophole]

Yeah, nearly all the celerons running at 558 or 600 are only single. One day soon I'll have to borrow a friend's slot 1 motherboard and see how fast the two CPUs can go in single mode - but time is always against us -> far too much to do and too little time to do it in.

It would be so cool to get holodeck2 back again so he can tell us all about his system and the mods to get it going. I wonder if his system is still chugging away somewhere Let's hope so - it's not less than a legend here -> whenever there is a post about running at high bus speeds someone always mentions it, and everybody seems to know about it

Actually the dual 495 have more power than I particularly need anyway, so there's no great rush.

Thanks again for all your posts, and feel free to add anything else that comes to mind.

[InactiveX]

gabumon: they were dual 366MHz chips that were overclocked, what with the Celeron's fixed multiplier - 5.5 x 133 =~ 733.

The last I remember of him, he published a video how-to (part 1) about watercooling on BP6.com, and part 2 never appeared...........

[Blueneko]

Maybe while making of a #2 video, there was an accident and he somehow split some atoms making a celery go too fast and had a meltdown or something??
133 fsb bus with two celerys is almost impossible
Let me attach my squirrel cage fans to my case and think up something to go with the evercool sinks I will get.
maybe peltier coolers under the copper sinks

[Derek]

[Edited] Link Expired

[loophole]

Got my motherboard and CPUs back just recently (been playing about with Duron 1.3GHz but couldn't bare to let my BP6 go (also have to love the ability to run everything you can think of at once and still have enough power to play Solitare - I love SMP systems )) and I noticed that once I put the system all back together again it wasn't quite as stable as it was before I went AMD for about a month.

So, I went back to 2x 450 at default voltage and slowly upped the FSB until I had crashes (running Seti@Home and playing Hot Pursuit 2 - all in the name of education of course ), and then I'd go and up the core voltage a little and see if that helped. So, as I type this my BP6 is running at 486 (4.5 x 108) and 2.05 volts. I'm not far now from what I had before.

But, one thing that I noticed when I was playing around with different bus speeds and the like was that at 108 MHz bus and using WPCredit (gotta love that utility), I could achieve some incredible memory results in SiSoft Sandra 2001te Professional's memory benchmark. With my setup all tweaked out, I achieved Integer speeds of 401 MB/s and Floating point speeds of 410 MB/s. I'll post some screenshots tomorrow. It just blew me out of my boots though - to think that the good old BX chipset still has that much left in it!

One thing a little worrying is that my VTT voltage falls very low under high stress. The lowest I saw it was when I was stuffing around at 110 MHz and after running Seti and some benchmarking suites I checked MBM and found that VTT dropped to a low of 1.41. The lowest I remember getting before was 1.45 (highest hasn't changed - about 1.55). I never quite got around to replacing the EC10 capacitor on my board - it's a 100 uF, but it never really troubled me too much before. I did go and put some heatsinks on all the voltage regulators on the board though - the one right below the AGP port got incredibly hot before putting the sinking on it!

[g0fvt]

Change EC10, my BP defied me despite ludicrous peltiers and watercooling, I got too lazy to pull the Mobo out to check EC10. An early contributor to BP6.com called Somethingburning had identified a problem with VTT stability. When my BP6 died recently with duff capacitor syndrome I changed them all, I was a bit liberal with bigger value capacitors... I should of done it years ago!

BP now overclocks like hell with stability... EC10 on mine is now 4700uF low esr..

Let us know how you get on..