I'm overclocking the FSB of a motherboard by soltek based on a VIA chipset, more precisely:

---------------------------------

1-1 CPU

- Supports Intel FC-PGA 370 Celeron / Pentium III (Coppermine) CPU at
300MHz up to 750MHz or higher.
- Supports VIA Cyrix III (Joshua) CPUs.
- Supports CPU voltage auto-detect circuit.
- Supports 66 / 75* / 83* / 100 / 103* / 112* / 124* / 133* / 140* / 150* MHz
system bus speed.
- Clock multipliers up to 8x or higher.

1-2 CHIPSET

- VIA VT82C693A + VT82C596B chipset with 133MHz FSB.
- PCI Rev 2.1, 5V, 33MHz interface compliant.
- Supports 66 / 100 / 133MHz, 3.3V AGP (Accelerated Graphics Port) slot.

1-3 L2 CACHE

- Pentium II / Pentium !!! supports 512K write back cache with Pipelined
Burst SRAMs.

1-4 MAIN MEMORY

- Memory range from 8MB up to 768MB (SDRAM) with DRAM Table Free
configuration.
- Supports SDRAM with 12 / 10 / 8 / 6 ns speed.
- Supports 3pcs 168pin DIMM sockets. (3.3V Unbuffered and 4 clock type)
- DRAM supports ECC or Parity function.

1-5 BIOS

- Award Plug and Play BIOS.
- Supports ACPI and legacy APM.
- Flash Memory for easy upgrade.

1-6 SUPER I/O FUNCTON

- Integrated USB (Universal Serial Bus) controller with 2 USB ports.
- Supports 2 IDE channels with 4 IDE devices. (including ZIP / LS-120 devices)
- Provides PCI IDE Bus Master function and supports Ultra DMA33 / 66
function.
- Provides 1 floppy port.
- 2 high speed 16550A FIFO UART ports.
- 1 parallel port with EPP / ECP / SPP capabilities.
- PS/2 mouse connector and PS/2 keyboard connector.
- Built-In RTC, CMOS, keyboard controller on single I/O chip.
- Peripherals boot function. (with ATX power)

1-7 OTHER FEATURES

- ATX size 17.0cm x 30.5cm.
- 5x PCI Master slots, 1x ISA slot, 3x DIMM sockets, and 1x AGP slot.
- Provides DIP switch setting.
- Supports 66 / 100 / 133MHz Bus Clock.
- Supports WOL (Wake On LAN) function.
- Supports Keyboard Power On function.
- BIOS supports 103 / 112 / 133 / 150MHz Bus Clock.
- Provides Voice Diagnostic function for easy debug. (65FVB-X only)

---------------------------------

Basically my doubt is if im doing an massive and dangerous overclock using a Front Side Bus setting ("CPU Host Clock" in the BIOS) of 150mhz, the default value is 133, if i raise it above 150 the system even doesnt boot, i have to manually set a jumper to reset the CMOS of the mobo if i go beyond 150.  :-/

The results however, are APPARENTLY perfect, mi 866mhz coppermine processor runs at 975mhz and i seem to get an slightly overall performance gain, also no freezing problems or anything. But.. is this safe? I have a good cooler on my CPU but.. is 150mhz of FSB too much for this old mobo (2000)? what do you think?

http://www.iespana.es/grunw/bench1.gif

As you can see according to this benchmark my cpu beats in processing speed a pentium IV 1.2 ghz!  ::)
And thats without doing a "clean" benchmark, i had eMule running while this test was done...

Errr.. where do you see that it beats a PIV 1.2ghz?

Anyways, i ran my P733 at 166fsb on my previous motherboard [which got fried, ironicly enough].
But i had 166mhz ram in it, so..

In any case, if it burns - it burns, a new system based on an athlonxp 2500+ will cost you 150$, and it will be 5x faster, so who cares :)

;D Thanks for your reply DenisF (and funny "if it burns, it burns")


Quote:

Errr.. where do you see that it beats a PIV 1.2ghz?

Well, in raw MIPS, in this PRECISSE benchmark (accuracy can be discussed, of course) it does!


Quote:

Anyways, i ran my P733 at 166fsb on my previous motherboard [which got fried, ironicly enough]. But i had 166mhz ram in it, so.. In any case, if it burns - it burns

1 point more for reverting the default value of 133  ;D

Quote:

Errr.. where do you see that it beats a PIV 1.2ghz?

http://www.iespana.es/grunw/bench2.gif

Well this time there is no doubt, and take into account I have eMule downloading while running this benchmark..
I know this only means its better than a pIV 1.2 ghz in SiSoft Sandra (synthetic test), but that also means its probably better than one in real-life too. Anyway I really don't care its just kinda symbolic for me   8) (feel free to correct me if i'm wrong)

Well, the IPC count of the Pentium 3 is quite a bit higher than the P4 Willatette (which was an utter disappointment at its time).

So the Score (and performance) indeed could be a bit higher than a P4 at 1200MHz, unless SSE2 comes into play.

Anyway, be careful with your expansion Cards (the PCI and AGP Bus gets overclocked as well), and even the IDE Controller (commonly attatched to the PCI Bridge) gets its share as well. If it does not make for stable operation (e.g. several flawless Prime95 Self-Tests), it's not worth the effort.

RAM is also a factor, as it's being overclocked as well with the FSB.

Thinking about every component on my system that gets addittional stress by increasing the FSB scares me a bit.. i wouldnt like to rebuild a new system from scratch if this one gets broken. Thanks for the warning, i take it into account.

I have a rule with myself on this, if ANY sympton of unstability appears, beeing it freezings, random crashes or any other anomaly. I will revert the FSB setting to the default of 133mhz.

Surprisinlgy (?) I had this computer with the current settings up for 2 days now and no problems at all spotted. Here in spain we are on the hottest season,above 30celsius sometimes, so its the definitive test!  :)

Which scares me for other reason, maybe i will have that kind of problem that is a single one.. lets call it a forced upgrade. Undesirable situation of course.

BTW Thanks for the P4 vs P3 explanation!

btw falconfly, I have prime95 23.8.1 running here, with self-test you mean a regular benchmark or a torture test?

If you want to make sure that your System doesn't create Errors on Bit level (which can lead to Data loss on your HardDrive), Prime95 is available here :

http://www.mersenne.org/freesoft.htm

Once started, ignore Registration (it's a mathematical Distributed Science Client), and choose one of the Self-Test Options (I think it now has something like a Burn-In Option).

Let it run for an hour or two (minimum complete once Cycle of Fast Fourier Transforms) to see if there are any errors (it will immediately stop and report even single Bit errors).

I still use it frequently for stability testing after integrating new Components in my Network, before it gets onto work for science... works great :)

ok thanks for the info, will do!

burn-in sounds more like torture test than a benchmark to me :) so i guess thats it!

My cpu is RIP

Dead i guess, system doesnt boot. I'm waiting for it to get cooled and check again, but cpu or nobo seems dead, probably cpu. I'm writing this using the laptop of some1 else. Now the surprising stuff.. you may have guessed i was overclocking something when it happened dont u? I DIDNT, it was at completelly default settings for hours BUT running a long test using prime95 (i guess you will be sorry about this falconfly but dont worry, its all my responsability of course)

Next time i will take "torture test" affirmation more seriously, it certanly tortured my cpu to death

...news... (success trying to boot)
FALSE ALARM, WAS SIMPLY OVERHEAT, DONT TRY THIS AT HOME (DONT USE PRIME95 If U DoNT KNOW WHAT r U DOING)

so dont worry falconfly i dont hate you... now ;D

My conclusion:
My system doesnt pass a single test on prime95 even not overclocked at all. Beyond 1 hour and some minutes it freezes or reports an error or nearly kills cpu by overheat (maybe temp in spain currently is the cause).. . Although it seems to be perfectly stable on real-life even with FSB overclocked. And i noticed no difference on stability by overclocking FSB at all, in fact it didnt crash or freeze once when it was.

ufff...that was close

You need to sort your cooling...

Yep, looks like severe overheating of the CPU indeed.

Once it exceeds its thermal design limits, it is normal to start producing errors at some point.

That region of operation is a small thermal region at which it is very close to its death, thus it must be avoided at all costs.

(writing using laptop again...)

Well system just got frozen the same way and it was just idle (no prime95 or anything), i turned it off, i removed the case and powered it on, then i saw the cpu fan was not moving at all. So the whole problem may be the cpu fan is dying or running wrong. I cleaned the dust everywhere in the fan and heatsink, and screw the fan back on the heatsink. It now moves, but it doesnt seem to do it very fast, maybe fan is really crappity smacked up.

So the issue with prime95 could have been a coincidence or it made arise a problem with cooling on the cpu.


Quote:

You need to sort your cooling...

You seem totally right

EDIT:

I forced the move of the cooler with my hands and now it works better, moves faster

Get some lubricant spray - spray it into the fan (when turned off). ;)

This should make the fan NICE and loose! :D