clueless1 wrote:

I'd guess you'd need somewhere around 800-933Mhz in order to get L1 disabled scores in the fast 386/slow 486 range.

What would a 1.5-3ghz p4 net you with caches disabled? I'm having a lot of trouble finding machines that aren't proprietary and therefore have the function locked out, but since they're everywhere and cheap I'm sure I'll find something.

Got a celeron 300 with desktop case today for $5, motherboard was shot but it actually had a 24 pin psu which is exactly what I need for a build. Plus I hear it will go up to 450 easily.

Throttle or similar tools can disable the L1 cache. I used this on an Atom netbook to run Wing Commander at a decent speed 😀

PhilsComputerLab wrote:

Throttle or similar tools can disable the L1 cache. I used this on an Atom netbook to run Wing Commander at a decent speed 😀

Throttle only brings up a percentage menu for me, at least on a 440bx chipset.

bluejeans wrote:

clueless1 wrote:

I'd guess you'd need somewhere around 800-933Mhz in order to get L1 disabled scores in the fast 386/slow 486 range.

What would a 1.5-3ghz p4 net you with caches disabled?

I tested a P4 2.53Ghz Northwood. It has only 20KB L1, with 512KB on-die full-speed L2. Setmul will work to disable L1 but because the L2 is so fast and big, it does not have much effect. So with full caches it is the fastest system on the spreadsheet. Disabling L1 drops it to 15th fastest in Doom (roughly a 50% drop). Speedsys seems to get confused and shows no slowdown with L1 disabled, but PCPBench score drops by 300%. 3dbench runs too fast either way, so there's no way to get an accurate score, as it "rolls over" 999.9.

I also tested a P3 933Mhz Intel. Setmul lets me disable L1 and L2, so I benched it at full speed, L2 disabled, L1 disabled, and L1+L2 disabled. Note that L1+L2 runs the same speed as L1. With full caches it is the 3rd fastest on the spreadsheet. With L1 disabled it performs like a fast 386. With L2 disabled, Speedsys again gets confused, Doom shows a very small drop in performance, while PCPBench score is cut in half.

When dealing with fast cpus, a lot of these benchmarks will freak out because they were built before these cpus existed.

bluejeans wrote:

PhilsComputerLab wrote:

Throttle or similar tools can disable the L1 cache. I used this on an Atom netbook to run Wing Commander at a decent speed 😀

Throttle only brings up a percentage menu for me, at least on a 440bx chipset.

And on your P4?

PhilsComputerLab wrote:

bluejeans wrote:

PhilsComputerLab wrote:

Throttle or similar tools can disable the L1 cache. I used this on an Atom netbook to run Wing Commander at a decent speed 😀

Throttle only brings up a percentage menu for me, at least on a 440bx chipset.

And on your P4?

Same on my p4 under freedos, on my netbook with an intel atom cpu, it says the chipset isn't supported (under xp, I don't really want to try under dos since there's no cd/floppy). Can I comfirm that you're using this, under dos? http://www.oldskool.org/pc/throttle/DOS

Hmm, It could have been another tool then. Maybe it was moslow or something like that?

PhilsComputerLab wrote:

Hmm, It could have been another tool then. Maybe it was moslow or something like that?

Free version of moslo can't disable cache, did you buy it?

Added some results for the VIA C3 "Samuel 2" lately. 175MHz (3.5x50) is the lowest I can go, at 165MHz (3x55) I got blue screen after few seconds and 150MHz (3x50) gave me an instant crash. I changed the multiplier before the FSB.

I discovered a Pentium MMX chip I had kicking around was TR12 capable, so I did some benchmarking to test it's range. Maximum performance is in the approximate range of a P133 as I have downclocked. In addition, I'm using the turbo switch to toggle between 50mhz and 66mhz FSB. The end result looks great as far as smooth scaling is concerned, bottoming out somewhere in early 386 territory.

See full results below.
https://docs.google.com/spreadsheets/d/1rOvIp … dit?usp=sharing

Very nice, man. Thanks for posting the spreadsheet too. Great idea to jumper the FSB to the turbo switch. What's the native speed of the cpu?

edit: your post reminded me to look at my own spreadsheet and it looks like someone mucked around with it a bit, messing with the sorting and replacing the text in the CPU column with "a:". 🤣. I restored a previous revision and downloaded a fresh backup.

I don't seem to have a very good motherboard with my pentium 100Mhz. I can only "disable system cache" which disables all the cache and there is also bios setting for setting boot speed as deturbo which does exactly same as disabling system cache. Setmul options /L1D /L2D don't seem to do anything.

for a P100, L1D should work. L2D would only work if your motherboard had L2 cache and only for AMD K6 mobile and VIA C3 cpus. But L1D *should* work. What version of DOS are you running? What are your system specs?

Using dos 6.22 for the tests. Only dos version that I have installation disks for.
It seems to be random intel pciset SB82437fx-66 socket 5 motherboard that supports 75Mhz, 90Mhz and 100Mhz pentium cpus. It has 8 KM68V257CJ-20 cache chips soldered on board.
The board has sticker with intel AA number. AA 632648-807, but I have not been able to find any more information about it.
It has 96Mb of ram and AMI bios from 1992.
Only cards I have there are SB16 and Diamond Stealth 64 DRAM PCI graphics card.
None of the setmul /L1E /L1D /L2E /L2D parameters seem to do anything. I get same benchmark results every time.

Edit: After checking a bit online, it seems my motherboard is some version of Intel Zappa

Not sure then. Like I said, it should work. When you type "setmul l1d" how does it respond? It should say "Disabling L1 Cache" in a cyan color text.

Maybe try asking in the Setmul thread?

clueless1 wrote:

Very nice, man. Thanks for posting the spreadsheet too. Great idea to jumper the FSB to the turbo switch. What's the native speed of the cpu?

edit: your post reminded me to look at my own spreadsheet and it looks like someone mucked around with it a bit, messing with the sorting and replacing the text in the CPU column with "a:". 🤣. I restored a previous revision and downloaded a fresh backup.

The original clock was 233 mhz, but this goes down to a much slower range.

Here's some benchmarks I did awhile back with my POD200 overclocked to 250Mhz. Note Ultima VII's immunity to disabling L1. And some speedsys weirdness (results not scaling like other benchmarks).

Yes, I shamelessly stole your TR12 parameter combos. 😀

Curious: was this all done with L2 enabled or disabled?

L2 enabled at all times. These particular results were taken with a motherboard whose L2 could not be disabled (SiS530 SS7). Other results I've posted in the past were on a Packard Bell socket 5 that did not have any L2. 😒

Here's the same data charted to show how L2 and FSB manipulation can be used to fill gaps in the performance capability of this chip:

This configuration covers the full range of performance from fast 386SX to P133 which is something no K6-2+ is capable of. I'm extremely impressed. Unfortunately, I'm going to be using 120Mhz as the top end due to a swap to an ASUS motherboard (the original guinea pig was a PC Chips, but I can't stand the unreliability), so it won't be completely representative of the final result.