noshutdown wrote:

first, this demo requires doom2 v1.9 to run, its commercial software so i won't provide it for obvious reason, you must find one […]
Show full quote

first, this demo requires doom2 v1.9 to run, its commercial software so i won't provide it for obvious reason, you must find one by yourself. 😒

and then you need to download 3 files:
the map(either link will do):
http://www.doomworld.com/hr2/hr2final.zip
http://doomedsda.us/lmps/586/hr2final.zip
the demo:
http://doomedsda.us/lmps/586/1/h232x747.zip
and you need a slightly hacked doom2 executable(enter download page):
http://sourceforge.net/projects/prboom-plus/f … m2-plus/1.92.6/
the original doom2.exe can only display 128 sprites at a time, and this hack simply removes the limit so that all sprites (monsters, projectiles, items etc.) can be displayed. the original doom2.exe can also play the demo, but in a "cheating" fashion with many monsters not being displayed.

when you have downloaded the files, extract and run:
doom2p -file hr2final.wad -nosound -timedemo h232x747

my test results:
k6-2+366, via vp3 motherboard, geforce2mx
33fps with hacked doom2 executable(playing the demo right)
50fps with original doom2.exe(with many monsters disappearing).
so this system of year 1997 spec can just barely run doom.

I didn't know there is a k6-2+366. Did you down clock it?

lazibayer wrote:

I didn't know there is a k6-2+366. Did you down clock it?

yeah i did, because the board has official support for only up to 66fsb anyway.

noshutdown wrote:

lazibayer wrote:

I didn't know there is a k6-2+366. Did you down clock it?

yeah i did, because the board has official support for only up to 66fsb anyway.

Did you try setting the jumpers to x2? K6-2+ should take it as x6 anyway.
Did you check the data sheet of the clock generator on your board? Lots of boards have hidden settings that can be figured out by checking that data sheet.

lazibayer wrote:

Did you try setting the jumpers to x2? K6-2+ should take it as x6 anyway.
Did you check the data sheet of the clock generator on your board? Lots of boards have hidden settings that can be figured out by checking that data sheet.

of course i could set it to 2x and get it to run at 6x, but 6x clock results in degraded memory performance as many know, so 5.5x is the better sweetspot.
and the vp3 chipset was rated at 66fsb with only 1/2 pci divider anyway. there is a 75fsb option on the board but not guaranteed to be stable, and i don't want to bother testing it.

noshutdown wrote:

lazibayer wrote:

Did you try setting the jumpers to x2? K6-2+ should take it as x6 anyway.
Did you check the data sheet of the clock generator on your board? Lots of boards have hidden settings that can be figured out by checking that data sheet.

of course i could set it to 2x and get it to run at 6x, but 6x clock results in degraded memory performance as many know, so 5.5x is the better sweetspot.
and the vp3 chipset was rated at 66fsb with only 1/2 pci divider anyway. there is a 75fsb option on the board but not guaranteed to be stable, and i don't want to bother testing it.

Sorry for creating a deviation in your thread. Honestly I didn't know 6x will degrade memory performance, and I will try to confirm that tomorrow.

The h232x747.zip demo is really intense. Never seen so many hellspawn before in a level 😮

Here are my test results:

Pentium 166MMX, Asus TX97-XE Socket 7 motherboard, 64mb EDO ram, Matrox Millennium II 4mb PCI
Doom2p.exe
Timed 16726 gametics in 35935 realtics = 16.3fps
Doom2.exe
Timed 16726 gametics in 17971 realtics = 32.6fps

Not really surprised the fps are so low, there's just way too much going on in the demo for the P166MMX to handle. Run in DOS 6.2, clean boot.

Pentium 4 2.4GHz Northwood, Soyo P4I-845PE Socket 478 motherboard, 512mb DDR-333 ram, NVIDIA GeForce 4 Ti 4200 128mb AGP
Doom2p.exe
Timed 16726 gametics in 11571 realtics = 50.6fps
Doom2.exe
Timed 16726 gametics in 9476 realtics = 61.8fps

Now this is much better. The P4 Northwood has the muscle to handle a Doom 2 demo like this. Run in DOS 7.1, clean boot.

Finally, no overclocking, timing or voltage changes. All test results were run with BIOS defaults.

noshutdown wrote:

lazibayer wrote:

Did you try setting the jumpers to x2? K6-2+ should take it as x6 anyway.
Did you check the data sheet of the clock generator on your board? Lots of boards have hidden settings that can be figured out by checking that data sheet.

of course i could set it to 2x and get it to run at 6x, but 6x clock results in degraded memory performance as many know, so 5.5x is the better sweetspot.
and the vp3 chipset was rated at 66fsb with only 1/2 pci divider anyway. there is a 75fsb option on the board but not guaranteed to be stable, and i don't want to bother testing it.

Out of curiosity may I ask if it only affects VP3 mobos? Just did an experiment on my K6-3+450/P5A-B and couldn't find the difference. 😵

lazibayer wrote:

Out of curiosity may I ask if it only affects VP3 mobos? Just did an experiment on my K6-3+450/P5A-B and couldn't find the difference. 😵

the phenomenon has been known by many. i've confirmed it on both mvp3 and ali5 boards, but only at 100fsb. there is about 5% drop in sisoft memory score, in everest test its write and copy that drops. in other tests like superpi and quake, there is only around 2% increase compared to 9% higher clock.

noshutdown wrote:

lazibayer wrote:

Out of curiosity may I ask if it only affects VP3 mobos? Just did an experiment on my K6-3+450/P5A-B and couldn't find the difference. 😵

the phenomenon has been known by many. i've confirmed it on both mvp3 and ali5 boards, but only at 100fsb. there is about 5% drop in sisoft memory score, in everest test its write and copy that drops. in other tests like superpi and quake, there is only around 2% increase compared to 9% higher clock.

I tried on my 3 K6-3+/450 builds with BIOS set to "setup default". The first one is an APZ @ GA-5SMM, 2.0V

The second is an APZ @ P5A-B, 2.1V

The third is an ACZ @ P5S-VM, 2.2V

I couldn't find any ram performance drop on x6 multiplier. Maybe the drop is unique to Win9x? Or maybe it's a side effect induced by overclocking? I can safely assume by setting clock to 100x6 everyone is overclocking their chips since their is no official release of SS7 chip beyond 570MHz, and some chips may show unexpected behavior while being stretched, even they could pass the stress tests.

an update of result on following test platform:
via vp3 motherboard(512kb onboard disabled because it can only cache for 128mb anyway)
256mb sdram
idt winchip running at 50*2(well i know cyrix6x86-50 may be slower, but the board wouldn't boot with that 6x86, either cpu is dead or board is defective)
rendition v1000
windows98se
on this platform i got 4.2fps!

and i've found another doom demo(and map) which puts even higher load on the computer. its played on a missionpack named "speed of doom"(SOD), which unfortunately can't be run with original doom executable because the maps are too large and too complex. the missionpack was designed for a modern win32 sourceport named prboom instead.
with the prboom engine the h232x747 demo played at 8.9fps on the same platform, or more than twice the speed of the original doom engine, yet the SOD demo played at only 2.7fps with prboom. so we can assume that if it would ever run with the original doom executable, it would be around 1.3fps for a 586 class computer!