elianda wrote:

The original version from Prof. Yasumasa Kanada works however in Win 3.x.

I wasn't able to get SuperPi 1.1e to run on Win3.11. Does it need the Win32's?

More results:
AMD Enhanced Am486DX4 120MHz/32mb EDO/Win95c = 50m 31.356s
Intel Pentium 75MHz (Gold Top)/FIC PA-2002 (VIA)/16mb EDO/Win95a = 45m 27.274s

Ok, my 960T wanted a shot at the title for just a minute till a intel setup comes in here and crushes it 😉

My X4 960T / Asus M3A78-T (NB:790GX/SB:750) / 8GB DDR2 1066 @ 800

First at stock 3.0GHz: 23.353s

Next at 4.4GHz: 16.286s

AdamP wrote:

elianda wrote:

Mod 1.5/XS uses kernel functions that got introduced in Win2K.

I assume that's why mine crashes?

Not sure why it crashed. Ftr, you're using 9x?

As I wrote in maddmaxtar's other thread, this version of SuperPi has 2 bugs in it.
One causes a crash (I added a fix in the zip file that maddmaxtar re-uploaded and included a readme), the other one is only cosmetic.
Afaik this version of SuperPi is the only one that will do decimals in 9x. I can't remember where I got it from though 😅

I'll build a couple other quick rigs within the next few days to do some more benchmarking.

Tetrium wrote:

Not sure why it crashed. Ftr, you're using 9x? […]
Show full quote

AdamP wrote:

elianda wrote:

Mod 1.5/XS uses kernel functions that got introduced in Win2K.

I assume that's why mine crashes?

Not sure why it crashed. Ftr, you're using 9x?

As I wrote in maddmaxtar's other thread, this version of SuperPi has 2 bugs in it.
One causes a crash (I added a fix in the zip file that maddmaxtar re-uploaded and included a readme), the other one is only cosmetic.
Afaik this version of SuperPi is the only one that will do decimals in 9x. I can't remember where I got it from though 😅

I'll build a couple other quick rigs within the next few days to do some more benchmarking.

I'm using Windows 98 original. It crashes every time, not just the first time. I also tried (and failed) to get HyperPi to run on Windows 98.

My copy came with the copy of HyperPi I have.

maddmaxstar wrote:

elianda wrote:

The original version from Prof. Yasumasa Kanada works however in Win 3.x.

I wasn't able to get SuperPi 1.1e to run on Win3.11. Does it need the Win32's?

yes, it needs win32s.

New result.
Intel Advanced/ATX (430FX), Pentium 133mhz, 64MB EDO, Windows NT 3.51 SP5.
16m 23.153s.

RogueTrip2012 wrote:

Ok, my 960T wanted a shot at the title for just a minute till a intel setup comes in here and crushes it ;) […]
Show full quote

Ok, my 960T wanted a shot at the title for just a minute till a intel setup comes in here and crushes it 😉

My X4 960T / Asus M3A78-T (NB:790GX/SB:750) / 8GB DDR2 1066 @ 800

First at stock 3.0GHz: 23.353s

Next at 4.4GHz: 16.286s

See the last page, my i7 MacBook is ahead of you 😜

elianda wrote:

maddmaxstar wrote:

elianda wrote:

The original version from Prof. Yasumasa Kanada works however in Win 3.x.

I wasn't able to get SuperPi 1.1e to run on Win3.11. Does it need the Win32's?

yes, it needs win32s.

I found my Win32s, appears to be running 32bit apps OK (oooh, Freecell), but still having issues on my 486DLC. Does SuperPi require an FPU?

Also:
System: MSI MediaLive Diva (780G+128mb Sideport)/4GB DDR2-667 Dual Chan/Win7x64
AMD Athlon 64 X2 4400+ (2.3GHz Brisbane 65nm 2x512k L2) = 39.204s

Nevermind, it's working, it just took the DLC 24min to complete the first iteration of the 256k test.

AMD K6-233
Vision Top Technology VT586TX motherboard
64MB PC-66 SDRAM
512KB PB Cache
Windows NT4.0 SP1

Result: 12m 42.204s

More to follow soon. 😁

maddmaxstar wrote:

Nevermind, it's working, it just took the DLC 24min to complete the first iteration of the 256k test.

SuperPi uses FPU, but Win 3.x brings a software emulation dll (win87em.dll or so) that catches the invalid instruction exceptions parses the FPU instruction (if it is one) and does the calculation in software.

But without FPU the result doesn't tell you much, since you basically bench the software emulation. And be aware that this can take some orders of magnitude longer...

New result. I threw together an Athlon 1400 just so I could SuperPi it 😁

Hardware used: A7V133 rev 1.05 (without the dot), 256MB PC-133(cl2), Athlon Thunderbird 1400C
2m 0.766s

So how about this: ftp://78.47.153.163/pictures/386DX40/super_pi … 0_cxFasMath.png ?

elianda wrote:

So how about this: ftp://78.47.153.163/pictures/386DX40/super_pi … 0_cxFasMath.png ?

Can't view the image, though I'm on my Macbook right now (w-Firefox). I saw it briefly until I refreshed the page - what was it, just over 6hrs for 1M on an AMD 386DX-40 with a Cyrix FasMath?

I ran a 256k Calc on my i386sx-16 with an i387-16 FPU (4MB RAM, Trident 8900D 1mb, Win3.11+Win32s): 4h 8m 49s.

I'll post a followup photo in a bit.

It's amazing just how big a difference adding an FPU makes. In comparison, before I ran the 386 test, I started a 256k on my TI486DLC 40MHz, No FPU (16mb RAM, Trident 8900D 1mb, Win3.11+Win32s) and it's still grinding away.... after 20 hours - and it's only on Loop 14!

I think my slot 1 system is doing ok:

1999 Retro Rig: P3 800 Slot 1 | ASUS P2B | 256MB PC100 - 03m 20.155s

Edit: 2nd run with closed firefox - 03m 13.706s

Here's the photos (I don't have a DOS/Win3.x screenshot utility) of my 386sx and 486DLC rigs on SuperPi 1.1e (no decimals)

i386sx-16+i387 FPU / 4mb 30pin RAM / DOS 6.22Win3.11+Win32s:
4h 8min 49s

*note: Syschk, Speedsys and PC-Config all identify as being 20MHz, but the chips are stamped as 16's. Could be possible it's overclocked but I've never touched it, and the CPU is surface mounted. The systems never run unstable and is still rock solid after 20 years, even the NiCd CMOS battery holds a charge.

And my DLC, which does not have an FPU, took far longer at over a Day just to do 256k.

Texas Instruments TI486DLC-40 (Cyrix-based, 1k L1, 8k L2)/16mb 30pin RAM / DOS 6.22+Win3.11+Win32s:
25h 8m 0s

System:

MoBo - Lucky Star 6ABX2V (440BX), RAM - 384MB PC100 SDRAM, OS - Windows 2000 SP4

Pentium 3 650MHz - 3m 7.930s
Celeron 366MHz - 6m 38.563s
Celeron 333MHz - 6m 57.971s
Pentium 2 266MHz - 7m 26.852s
Celeron 266MHz - 9m 14.117

The Celeron 266 is one of the original cacheless models. I'll do the P2 450 and 350 when I can find a suitable heatsink for them.

Heres most results all in 1 post. I probably accidentally left out a couple.
Posted for your convenience (if you find the format readable 😜 )

Core i7-2720QM 2.2Ghz - 12.854s
X4 960T 4.4GHz - 16.286s
Phenom II X6 1090T (3.2GHz+Turbo, Thuban Core, 6x512k L2, 6MB L3) = 21.434s
Core i3 M330 2.13GHz - 21.513s
PhenomII 3.2Ghz - 21.699s
X4 960T 3.0GHz - 23.353s
Athlon X2 4850e (2.5GHz Brisbane, 2x512k L2) - 36.219s
Prescott 3.0E @ 3.6Ghz - 38.750s
AMD Athlon 64 X2 4400+ (2.3GHz Brisbane 65nm 2x512k L2) = 39.204s
Phenom II X3 E840 (1.9GHz, Caspian core, 3x512mb L2 Cache) = 41.169s
Athlon 64 3200+ 2.2Ghz - 42.594s
Athlon 64 3500+ 2.2Ghz - 43.906s
Athlon 64 3200+ 2.2Ghz - 45.015s
Prescott 3.0E@3ghz - 45.094s
AMD Athlon 64 3700+ (2.2GHz 1MB L2, San Diego) = 45.312s
P4 Northwood HT 3.2Ghz - 46.734s
P4 Northwood HT 3.2GHz - 48.984s
Northwood 2.8B @ 3.15ghz - 55.57s
Athlon XP 3200+ (2.2Ghz) - 57.406s
Sempron 2600+ (1.83Ghz 333fsb) - 1m 05.438s
XP 2000+ (1.67Ghz Palomino 133fsb) - 1m 23.850s
Pentium 3 1400S - 1m50.764s
Athlon Thunderbird 1400C - 2m 0.766s
Pentium III 1400Mhz - 2m 6.150s
PIII Coppermine @ 966mhz - 2m 57.883s
Coppermine 1000 - 2m 58.616s
Pentium III 933Mhz - 3m 9.189s
P3 800 Slot 1 - 03m 13.706s
Pentium 3 650MHz - 3m 7.930s
Intel Celeron 800MHz - 3m 29.461s
AMD Athlon 700MHz - 3m 38.915s
Intel Celeron 800 - 4m 07.752s
AMD Athlon 550MHz - 4m 13.074s
Intel Celeron 466MHz - 5m 03.346s
Mobile Celeron 500MHz (Coppermine, Skt 495) = 6m 00.120s
k6/3 450 - 6m 18.131s
Celeron 366MHz - 6m 38.563s
Celeron 333MHz - 6m 57.971s
Pentium 2 266MHz - 7m 26.852s
AMD K6-2 @495mhz - 8m 7.024s
Celeron 266MHz - 9m 14.117
Mobile Pentium MMX 233MHz (Tillamook on MMC-1) - 12m 24.011s
AMD K6-233 - 12m 42.204s
Pentium 133mhz - 16m 23.153s
Intel Pentium 75MHz - 45m 27.274s
AMD Enhanced Am486DX4 120MHz - 50m 31.356s

good work tetrium bro! I was thinking of doing the same, but you beat me to it!

Yes! Good work! Would it be a good idea to use that to turn it into a global leaderboard on the first post of this thread - and update it as times get posted?

maddmaxstar wrote:

Yes! Good work! Would it be a good idea to use that to turn it into a global leaderboard on the first post of this thread - and update it as times get posted?

I've put the results in my first post.

I put them in order of fastest to slowest but left out as much information as I could (names, rigs etc. Only the CPU, it's speed and it's results).
I'm not in it for the speed contest (heck, compare my Celeron 800 to yours 😜 ), just for fun and for all of the data 😀