Reply 100 of 225, by ih8registrations
Drop him an email to update his page. Last updated 3/10/2005 😀
Drop him an email to update his page. Last updated 3/10/2005 😀
Usually a limiting factor in the FSB with these motherboards is the on board cache. The operating frequency is that of the main bus, so in most cases you actually have to disable it to hit the really high numbers.
True. With a K6-III/2+/3+ disabling the mainboard cache won't hurt very much, but a Pentium would be rather useless without it.
And there is still the issue with the AGP/PCI clock - good chipsets got everything covered up to 100 MHZ FSB, but above that you're dealing with overclocked buses, which on SS7 boards do not have much headroom to begin with.
The "safe margin" ends around 75 MHz for AGP and 37.5 MHz for PCI, most cards will work fine below these speeds. This corresponds to a FSB overclock from 100 to about 112 MHz.
133 MHz FSB on a SS7 board will result in 89 MHz on the AGP (Intel BX users know about this issue) and 44.3 MHz on the PCI bus. At these speeds, most PC hardware connected to the buses starts doing silly things like corrupting filesystems, garbling screens or emitting clicks and screeches. 😉
Actually, some SiS 530 based SS7 boards had a 1/4 PCI divider... it switched over automatically at 124mhz, if my memory isn't failing me.
Of course, those SiS-based boards had their own set of disadvantages, the most immediately noticeable of which was the lack of an AGP slot on most (all?) of them.
But, if you wanted to try for high FSB, it might be worth a look.
Failing that, Gigabyte GA-5AA or GA-5AX than can be set to 140, then about six boards that can be set to 133.
I just saw one for sale at a local shop's junk bin. But I decided against getting it and concentrated on parts for my PIII project instead. That, and the following review:
http://active-hardware.com/english/reviews/ma … ard/ga5ax-2.htm
Interesting that so many mobo manufacturers had issues with the AGP power requirements. Some revisions of the Gigabyte GA6BXC had AGP power issues due to a bad choice of voltage regular on the board, which could also be bypassed with some soldering.
I just tried an old "flat" ceramic? P MMX 200 CPU, and set the mobo's jumpers to "max voltage" (3.5cv, 3.8io), and set the mobo's speed jumpers to 100fsb, 3x multiplier. The mobo boots OK, and reports that the CPU is 300mhz, but the machine goes "mad" soon afterwards. If I run edit.exe at the C:\ DOS prompt for example, and try and rem out lines inside autoexec.bat, garbled characters appear on the screen.
So, although the mobo POSTS and boots DOS OK, this test has failed to produce a stable system.
I am looking forward to reading prophase_j's upcoming SS7 mobo + P233 CPUs experiments! 😀 In the meantime, I'm going to remove the P200 CPU and put back the P233 CPU, and just mess about with it a bit more...
I'm looking forward to it too! Still waiting to get my heatsink and processor, but they should arrive any time now. Right now all the other components are in place:
Asus P5A rev 1.04
Pentium 233 MMX SL27S <-- still waiting 🙁
Kingston 128mb PC133 SDRAM
Diamond Viper v770U 32mb <-- TNT2 Ultra
STB Blackmagic Voodoo2 SLI
WD 6.4gb ATA66 Hard disk
Soundblaster AWE64 Gold
Some generic PCI NIC
I think it makes a nice mix, it's like a retro hot-rod! I'm interested to see if my processor will stand up to the necessary abuse 😈
No wonder I haven't found more talk about this on the internet, anyone who made it work probably knew there was little to gain, despite up to a 50% increase in clock and bus speed.
Athlon XP-M 2200+ // Epox 8KTA3
Radeon 9800xt // Voodoo2 SLI
Diamond MX300 // SB AWE64 Gold
Yeah your not going to feel that big 50% BOOST clocking your Pentium from 200 to 300. Your TNT2 and V2 SLI will outclass that little pentium chip. But should be fun, as will upgrading that system to a PIII someday. Maybe?
I have a P200 MMX and it has no issues running at 290. Can't hit 300 though, my 430TX mobo is limited to 3.5x 83. Providing you have a good cooler and ventilation it should hold up.
Also I'm thinking the P266 is a mobile part but will work in socket7 boards. At one time I was looking for one for my collection.
I've just found an old IBM 6x86MX PR233 CPU in my box of bits! 😀 I think it would be fun to see if it can be OC'd to 300. If that's possible, I could benchmark it against my OC'd 233 -> 300 Intel CPU. This IBM PR233 CPU has the following info on it -
2.5X 75MHZ-2.9V CORE
Here's one I just spotted on ebay -
Those IBM/Cyrix chips were dogs when it came to socket7 gaming, considerably worse then the K6.
I think it would be fun to see if it can be OC'd to 300.
I don't see that happening. Those chips, aside from being crappy performers, were known for heat issues and lack of scalability of the clock speed. Also remember, PR233 != 233mhz... IIRC, it actually ran at 180-190mhz, somewhere in there.
They're also buggy. You need patches for some software to work with them properly.
Just for a bit of fun, I had a quick go with the Cyrix chip, and set the mobo's jumpers to 3.1cv (default is 2.9cv), 3.6io (lowest I have is 3.5), 100fsb (default is 75), 2.5m (default).
The BIOS POSTed OK, and reported that the CPU was 366! Afterwards, the win98 splash screen appeared, but that was it - I never saw the desktop. It's interesting, because it didn't appear to crash or freeze, as the win98 splash screen's "progress bar" always kept moving, and the keyboard's CAPS LOCK key always worked when pressed. Perhaps I need some more volts?
Sounds like a competition! I have PR200 and a PR233 of the same chip, but they're Cyrix branded and not I IBM.
Well I got mine to post at PR300 but it too got stuck at the win98 screen with the bar moving across the bottom forever. Gonna try it in my P200 dos/win95 machine.
Best I could do. 209 mhz with that PR233 at 83x2.5 3.2v Chip looks just like the one in the pic i posted.
TestCPU is a neat program I just found, even has a museum section built in. Made for 90's CPUs.
Well I got mine to post at PR300 but it too got stuck at the win98 screen with the bar moving across the bottom forever.
It's interesting. BTW, I DL'd TestCPU - I like the museum section. 😀
Well the Cyrix wasn't a total slouch, at 209 it coughed up 13.1fps in DOS Quake 640x480 with a Virge card. A Pentium at 200 gave me 15.3.
Still the BIOS listed it as a PR266 and if Cyrix were selling PR266s I'm sure they would have cost more then a P200 would have back then. This is one reason why they failed, that and program comparability issues as mentioned earlier.
you should run at 320x200 so the video card doesn't affect results as much.
Quake should perform a lot better on the Pentium simply because the game is literally designed for that CPU. And because the Cyrix FPU is horrible. It's not much better per clock than a 486's. 6x86 is basically a "Winstone/Winbench" CPU.
Dark voodoo: some more overclocking 233 MMX to 300MHz fun. I haven't got any socket 7 heatsinks (to hand), so I used a socket A/462 heatsink instead. The clips wouldn't clip on to the SS7 mobo because it was the wrong shape and size, so I used my dark voodoo overclocking skillz, and got some heavy objects from around the house. It works, but I might delay mass production of this heatsink concept until the design is tweaked a bit.
I set the mobo's options to: 100 bus speed (from 66), 3 multiplier (from 3.5), and 3.2 voltage (from 2.8 ). The Intel Pentium 233MMX chip was thus magically transformed to a molten hot slab of 300MHz goodness. Amazingly, the system appears to be stable. I had to open a window 5 minutes later because of the heat!