Pretty neat eh?
Just like I read in another post here, underclocking to 133MHz disables the L2 cache.
I do wonder though, since the L2 is bound to the CPU could it be forced on by software?
It also took a damn long time to get the P3V4X running; it did a lot of instant shut-offs, no video boots, and even then getting the PII to POST was a hassle.
I have a board with 440LX and cache stays on, independently of multiplicator setting. Maybe it is just a BIOS issue of your board?
Setting a P2 to 133 MHz is not really special though.
Swaaye did this very nicely for his architecture comparison at 133 MHz with different CPU cores.
Retronn.de - Vintage Hardware Gallery, Drivers, Guides, Videos. Now with file search
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool
On some PIIs, even 166mhz disables the L2 cache. I'm pretty sure it's not a BIOS or motherboard issue as I tested my PIIs on the same motherboard.
@elianda,
Which model do you have?
it is this mainboard:
http://mail.lipsia.de/~enigma/neu/pics/mainbo … _msi_ms6117.jpg
And the CPU was a P2 SL264, SL2HD and some P2 300 MHz I don't remember atm.
Retronn.de - Vintage Hardware Gallery, Drivers, Guides, Videos. Now with file search
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool
So far I tested about 3 three different 440BX based boards with a 133MHz Pentium-II, and they all disable the L2 at that speed.
About three other people reported the same with their 440BX boards.
Only swaaye and elianda seemingly have the L2 working on the 440LX.
L2 does not matter much in the DOS game benchmarks I did, maybe the L1 size on these P-II's is already sufficient? Also the L2 on the P-II is half speed and would be running at just 66MHz. Still the disabled L2 is annoying me somehow. Maybe a slot-1 Pin-mod would make a difference....
wrote:Setting a P2 to 133 MHz is not really special though.
Swaaye did this very nicely for his architecture comparison at 133 MHz with different CPU cores.
But that is not that long ago. About 4 years ago P-II's were considered poor DOS retro material here, But now it is well known that some can run at 2x and they are used more often.
--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul
wrote:166MHz Has L2 enabled for me.
CPU Model: SL2HE
Same here.
The model that has L2 disabled at 166mhz is a Deschutes. I think it was a SL2KA. There's more info in this huge thread.
Here is retro games 100's result of the klamath on the 440bx at 2x: no L2.
Quick question about underclocking a Pentium 2
(The Speedsys screenshot at the bottom of page 1)
wait, I will just add it here:
--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul
And here is Swaaye's Klamath benchmark at 2x on a 440FX. L2 enabled.
--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul
How can a 1995 BIOS recognise a Pentium II?
wrote:How can a 1995 BIOS recognise a Pentium II?
Probably because 440FX also supports Pentium Pro and Klamath shares a lot with the PPro.
wrote:Only swaaye and elianda seemingly have the L2 working on the 440LX
I had a troublesome 440FX board back then which I used to run my Klamath at 133. I'm not sure what happens with it on the Abit BF6 (440BX) at 133....
BTW the Celerons without L2 were liked a bit for 3D gaming because they were still beating Cyrix and AMD K6 apparently.
But what about things like MMX? Doesn't that need an appropriate BIOS?
And are the sockets/slots just interfaces, or do they also need recognition by the BIOS?
I do know this: There are plenty Pentium II-era BIOSes that cooperate fine with an unrecognized Coppermine or even Tualatin installed. Others give errors, or have a security check for some political reason. CPU's on themselves are very plug and play, and very backwards compatible towards software of course.
wrote:BTW the Celerons without L2 were liked a bit for 3D gaming because they were still beating Cyrix and AMD K6 apparently.
In page two of your 133MHz challenge I measured that one loses at most 1/3rd of the framerate when without L2.
--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul
I think my board is weird...
Coppermine 600EB @ 300MHz:
With L1 and L2 caches disabled my speedsys cpu score drops to 4.5 and the memory bandwidth drops down to ~38MB/s.
Does anyone else get a similar result?
The PII dropped down to 2.5 with both caches disabled but this board really, really doesn't like it and only posts after a long warmup time.
Once you disable the L1 cache on a pentium it is left crippled. Your board is not weird. Above we were only talking about the L2 cache.
--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul
These chips / mainbaords don't have mainboard cache (what I/we? call L2 cache on 486 / Super Socket 7 machines) anymore.
So they are either super fast (with the fast on-chip cache) or totally crippled (286, 386SX, 386DX depending on what speed, FSB, memory speed).
The FSB and memory speed (and timings) have the biggest impact with cache disabled Slot machines.
That crippling is a real treat!
Without it, many old games would not run on a P2.
Retro PC warning: The things you own end up owning you.
wrote:I do know this: There are plenty Pentium II-era BIOSes that cooperate fine with an unrecognized Coppermine or even Tualatin installed. Others give errors, or have a security check for some political reason. CPU's on themselves are very plug and play, and very backwards compatible towards software of course.
I always thought you still needed a BIOS that supported them for proper support. Take Deschutes and Katmai. These are very similar, yet the SE440BX/2 manual (I forget which) claims that Katmais won't work properly unless you update the BIOS. I once tried to use a Katmai on a PII BIOS, but it gave me problems. Though, it also gave me problems on the new BIOS because (I believe) of the newer stepping, which probably needs a later BIOS.
wrote:cache disabled Slot machines
What's the point of those? (get it? 😀)
