VOGONS


486 mobo + 586 chip

Topic actions

Reply 60 of 148, by retro games 100

User metadata
Rank l33t
Rank
l33t

The SX/33 chip and Windows 98 works at 66 FSB! To set this mobo to 66 FSB, your FSB jumper settings need to look like this -

--
oo
--

Looking at Sandra's MFLOPS value of 2, I wonder if I swapped out the SX/33 for a DX/33, I'd get a much better score. I'm not sure why Sandra thinks that the CPU and FSB are 95 MHz. Before booting in to Win98, I booted in to DOS, and ran ChkCPU and it told me that the clock speed was 67.something. That must be the accurate reading!

sx33.jpg

Reply 61 of 148, by retro games 100

User metadata
Rank l33t
Rank
l33t

The SX/33 chip overclocks up to 66 FSB, but a DX/33 won't POST at 66 or 60 or even 50. I wonder if a weaker type of "DX style" processor would overclock better, not made by Intel. I've got an AMD DX-4 120 chip (3x multi, 40 FSB) somewhere, but unfortunately not immediately available. I wonder if the FSB could be overclocked OK to 50?

Reply 62 of 148, by Old Thrashbarg

User metadata
Rank Oldbie
Rank
Oldbie

The SX/33 chip overclocks up to 66 FSB, but a DX/33 won't POST at 66 or 60 or even 50.

I was kinda figuring you'd run into that... I imagine the lack of an FPU on the SX allows it to have significantly more tolerance for high clocks. I'm not sure it'd be much different for chips from other manufacturers... the fact is, you were running a 100% overclock on that SX/33, and that's pretty well out of the norm for any 486. If you can find that DX4-120, there's a decent chance it'd run at 2x multiplier, 66fsb. It might be possible to run 3x50, but I wouldn't count on it... again, 150mhz is pretty high for a 486.

And I've developed a bit of a hunch on how your board works... I suspect it has three PCI dividers, automatically set. 1:1 at 33mhz, 3:4 at 40-60mhz, and 1:2 at 66mhz. That would help explain why it works at 66mhz but you were having trouble getting a POST at 60mhz, because a 3:4 divider would put the PCI speed at 45mhz, which a lot of video cards won't handle (but Nvidia cards tend to be a little more tolerant). That'd also mean you'd be very likely to run into IDE problems with the 60mhz setting.

Reply 63 of 148, by 5u3

User metadata
Rank Oldbie
Rank
Oldbie
Old Thrashbarg wrote:

The SX/33 chip overclocks up to 66 FSB, but a DX/33 won't POST at 66 or 60 or even 50.

I was kinda figuring you'd run into that... I imagine the lack of an FPU on the SX allows it to have significantly more tolerance for high clocks.

Not only that, the SX-33 came two years later than the DX-33.

DX-33:
L_Intel-A80486DX-33.jpg

SX-33:
L_Intel-A80486SX-33.jpg

Reply 64 of 148, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

Very nice die shots! I'm quite positive that the newer a cpu is, the better chance you'll have at overclocking it.
And my, a 100% overclock! Well done RG100!

Old Thrashbarg wrote:

And I've developed a bit of a hunch on how your board works... I suspect it has three PCI dividers, automatically set. 1:1 at 33mhz, 3:4 at 40-60mhz, and 1:2 at 66mhz. That would help explain why it works at 66mhz but you were having trouble getting a POST at 60mhz, because a 3:4 divider would put the PCI speed at 45mhz, which a lot of video cards won't handle (but Nvidia cards tend to be a little more tolerant). That'd also mean you'd be very likely to run into IDE problems with the 60mhz setting.

I've been doing some research today with little success due to me having been very busy last week.
Seeing as only some boards have the magic jumper it may well be possible that many other boards either have this option in the BIOS (though not very likely) or it's an option native to the chipset that the board makers simply didn't implement.

It gave me an idea of a possible benchmark, a direct way to compare the POD and DX2 clock for clock.
I know the POD works with a x2.5 multi, but I've read that removing it's fan will force it into lowering it's multi to just x1.
Remove it's fan (perhaps place a fan next to it to help cool it), set the pci at 33Mhz and set the magic jumper and you have a POD running with the exact same settings as a DX2-66!

I wonder if any of the technical data of these UMC and sis chipsets still exist on the net somewhere.

Edit: Anyway RG100, you could perhaps test the 60Mhz with an ISA card instead of PCI? Perhaps you'll have a greater chance of it booting?
It would enable you to bench it so we can know for sure about the pci divider

Reply 65 of 148, by retro games 100

User metadata
Rank l33t
Rank
l33t

Thanks a lot for great info, people.

Old Thrashbarg wrote:

I suspect it has three PCI dividers, automatically set

+

Tetrium wrote:

or it's an option native to the chipset that the board makers simply didn't implement

IMHO, that's the most plausible reason. Perhaps the mobo manufacturers considered that these "built in to the UMC chipset" higher FSB settings (60 and 66, and I notice that even 50 is omitted from the writing on the PCB on this UMC chipset based board) were too risky.

Tetrium wrote:

... perhaps test the 60Mhz with an ISA card instead of PCI

Unfortunately, I don't have any ISA video cards. However, I do have some PCI video cards that will boot and work OK at 60 and 66 FSB. The only one that seems to struggle is the Dell S3 GX Nitro. The Riva 128 works OK in DOS, and the S3 Virge 325 works OK, even using a Win98 driver. BTW, unfortunately I have no POD either.

Reply 66 of 148, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

I finally did a text string search for ½ inside th99free, hoping to find more mobo's with the magic jumper. Unfortunately it turned up only 3 relevant results: The 2 boards already listed and an Opti Pentium board with VLB.
Apparently we'll have to either find boards that set the pci divider automatically, or start trying out undocumented jumper settings.
A 3rd and 4th possible way could be perhaps to reflash with a bios that does support pci dividers (risky) or start modding the chipsets themselves (extremely risky).

I'd say...the hunt is on! 😉

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 67 of 148, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

I was on one of the other hardware forums today and found a thread interesting to mention in this topic. Apparently someone in Germany has actually build a Socket 3 running an AMD 5x86 at 200Mhz with the help of some modern PCI devices. He doesn't mention how he actually did it, but seeing he either manages to get the PCI bus running at 66Mhz (very very unlikely) the only other option is he's also making use of the PCI divider.
Link: http://www.planet3dnow.de/vbulletin/showthread.php?t=388817
The forum's in German so for the people who can't read it, better use Google translate 😉

Theres a pretty good chance that theres a narger number of 486 boards out there that have this divider 😀

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 68 of 148, by retro games 100

User metadata
Rank l33t
Rank
l33t

^ In his hardware set up description, he mentions an "AMx586 160 @ 200 MHz" processor, and the screenshot shows "AMD-X5-S CPU". I wonder - is this an AMD chip that runs at a stock speed of 160 MHz, which has been OC'd by 40 MHz, to reach an operating speed of 200 MHz? What is an X5 chip? Maybe this is a realistic overclock, for such a chip? Perhaps this chip is different (better/faster) to the one that I was using, which was a 133 MHz CPU?

Reply 69 of 148, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
retro games 100 wrote:

^ In his hardware set up description, he mentions an "AMx586 160 @ 200 MHz" processor, and the screenshot shows "AMD-X5-S CPU". I wonder - is this an AMD chip that runs at a stock speed of 160 MHz, which has been OC'd by 40 MHz, to reach an operating speed of 200 MHz? What is an X5 chip? Maybe this is a realistic overclock, for such a chip? Perhaps this chip is different (better/faster) to the one that I was using, which was a 133 MHz CPU?

I believe it is. Looking at the post screen it mentions a sis 496/497, which is Socket 3.
It's the AMD 5x86-160 (4x40).
I still believe a late production date 5x86 (1997/1998) should be able to run at this speed though.
Edit:I assume he's running the 5x86 chip @ 3x66Mhz

Edit2: I forgot to add to the thread I found out one of my boards has the PCI divider described in the manual. It's quite likely that many other 486 boards have this setting in the BIOS setup screen.

Edit3: Btw, RG100 where do you live?

Reply 71 of 148, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

England? Nice... I lived in Leicester for 3 years once and thoroughly enjoyed it. Great pubs 🤣

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 72 of 148, by retro games 100

User metadata
Rank l33t
Rank
l33t

BTW, I got this mobo [please see my original post] to work with the AMD P75 chip. That's using a 3x multi, and a 60 MHz bus speed = 180 MHz real clock speed. I used the ADW variant. The ADZ variant would not work. Also, I got the mobo to POST at 200 MHz, but it would not boot up DOS. I may try this again tomorrow, using different compact flash drives...

I noticed that this particular mobo seems to ignore/switch off L2 cache, if the bus speed is set to 60MHz.

Reply 74 of 148, by retro games 100

User metadata
Rank l33t
Rank
l33t
Tetrium wrote:
retro games 100 wrote:

I noticed that this particular mobo seems to ignore/switch off L2 cache, if the bus speed is set to 60MHz.

🙁

However, it doesn't seem to affect benchmarking scores. The BIOS POST display says something about "now using accellerated cache". It's a bit odd.

Reply 75 of 148, by retro games 100

User metadata
Rank l33t
Rank
l33t

AMD P75 5x86 133 MHz overclocked to 200 MHz!

Speedsys = 75! The BIOS timings are loose. The mobo's jumpers are set to 50 MHz bus speed, and 4x multi. The secret to getting this working was to -

1) inside the BIOS, set the I/O recovery from 2 to 8.

2) use a compact flash device plugged directly on to the mobo's integrated IDE pins, rather than use a CF device attached to a cable. Also, I am using a 512MB SanDisk card. They seem to be good cards.

200B.jpg

ChkCPU
PICT2049.JPG

Cachechk(7)
PICT2053.JPG

Reply 76 of 148, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
retro games 100 wrote:

AMD P75 5x86 133 MHz overclocked to 200 MHz!

Cheers! So how did you finally manage to pull it off? Found a better 5x86 chip? Or did you put it on ice? 😜

Reply 77 of 148, by retro games 100

User metadata
Rank l33t
Rank
l33t

I've got the AMD 133MHz CPU + UMC 486 mobo seen in my original post stable in Windows 98. Behold! That's 50MHz bus, 4x multi = 200 MHz CPU clock speed. The BIOS timings are set to about "medium strength". The mobo's cache chips are only 15ns, and so the BIOS cache timings are set to "slowest".

If you've got a similar set up, please experiment with the I/O recovery time in the BIOS. I changed this setting from 2 to 8, and that allowed me to do this outrageous overclock.

s4.jpg

Edit: Quake timedemo demo1 in DOS 6, I get 17.7 fps, and that's Full screen!

Reply 79 of 148, by udam_u

User metadata
Rank Member
Rank
Member

AMD P75 5x86 133 MHz overclocked to 200 MHz!

You set a new record to beat. Amazing result! It seems that your CPU is magic. (: What is the production date of your CPU (inscription under AMD-X5-133ADW)?
Have you got Voodoo2? If so, can you retest Quake1 with openGL patch?

2) use a compact flash device plugged directly on to the mobo's integrated IDE pins, rather than use a CF device attached to a cable. Also, I am using a 512MB SanDisk card. They seem to be good cards.

Are you sure that it makes any difference?

Regards! (: