VOGONS


The Ultimate 486 Benchmark Comparison

Topic actions

Reply 260 of 280, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Also had a DX4-75 in a laptop a ways back, AST Ascentia, that one never felt very snappy either. Probably slower than DX2-66 machines, think I tried Duke3D and had to make the window real small. Doom was better, but actually playing it, it did a 1, 2, 3 FREEZE repeat.. freeze was quarter or half second, but sooooooooo annoying.

Edit: I get dizzy when we teleport in from another thread like that 🤣

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 261 of 280, by W.x.

User metadata
Rank Member
Rank
Member

Many thanks for insights, but I would like to see some real world tests, benchmarks comparsions of DX4-75 vs DX2-66. I know, they should be "similiar", it's basicaly similiar case as Pentium 75 and 66, which were benchmarked at many various places already.

Reply 262 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I suspect in synthetics and number crunching, DX4-75 would be faster, but in anything with quick moving graphics or text, the DX2-66 would dominate. It is the degree of dominance that you are after. I don't have numbers, but my guess would be 10%. If you indicate exactly which applications you are interested in, there is a higher probability that someone will come along and run the tests.

Plan your life wisely, you'll be dead before you know it.

Reply 263 of 280, by W.x.

User metadata
Rank Member
Rank
Member
feipoa wrote on 2021-12-02, 00:32:

I suspect in synthetics and number crunching, DX4-75 would be faster, but in anything with quick moving graphics or text, the DX2-66 would dominate. It is the degree of dominance that you are after. I don't have numbers, but my guess would be 10%. If you indicate exactly which applications you are interested in, there is a higher probability that someone will come along and run the tests.

Doom and Wolfeinstein 3D for games, maybe PC player benchmark, and some of those standard tests for software under DOS, they test mostly ALU, FPU, cache speeds, in some units. MDK DOS benchmark is not bad choice too. (or Quake 1 for FPU-intensive comparsion)

Reply 264 of 280, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Talking laptops though you're somewhat to very screwed by the implementation and can do little about it. If I was stuck with a DX4-75 CPU on a half decent desktop board, I'd be trying it at 100 right away, with a reasonable sink on, and if it didn't like that, try the 2x multi by board setting or socket trick and try dx2-80.

It's pretty certain it would come out over dx2-66 on synthetic tests because icomp index is higher. We've seen that DX50=DX2-66 myth thrown out though, as no benchmarks seemed to support the anecdotal data that held it as "snappier" or more responsive. On same boards with same cards it'll fall in line, but there's probably a 20% margin between best of one and crappiest of another system.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 265 of 280, by gnif

User metadata
Rank Newbie
Rank
Newbie

Just thought I would post my recent success with a Chaintech 486SOM paired with an AMD 5x86-133ADZ

This motherboard was obtained from eBay after noting that it has all the hardware for a PS/2 header except for the socket (Details on how to enable/use it), and my favourite chipset of this era, the SIS 85C496. It also uses an IMI SC464 motherboard clock chip which is designed to output various FSB clocks all the way up to 80MHz while providing a separate clock for the PCI devices.

This motherboard does not expose the additional clock speeds above 40MHz though as the configuration pin for these extra speeds is left disconnected and is pulled high internally, however with some careful soldering it's possible to tack onto pin 17 (SC2) and pull it to ground (no resistor needed as per the datasheet) enabling access to 30, 60, 80, and 66.6 MHz frequencies.

2022-10-05-091033_340x207_scrot.png
Filename
2022-10-05-091033_340x207_scrot.png
File size
11.66 KiB
Views
2466 views
File license
Public domain

Here I have done this and configured the board for a 60MHz FSB:

signal-2022-10-01-201810_002.jpeg
Filename
signal-2022-10-01-201810_002.jpeg
File size
403.38 KiB
Views
2467 views
File license
Public domain

With this change I have been able to achieve a stable 180MHz clock and I have not had to relax my cache or RAM timings and I am able to complete all benchmarks and time demos.

signal-2022-10-01-201810_004.jpeg
Filename
signal-2022-10-01-201810_004.jpeg
File size
290.3 KiB
Views
2467 views
File license
Public domain

Attempting to clock at 200MHz (66 * 3) even raising the voltage to 5V does not post. This though might be a result of the PCI bus being out of spec causing the video card to fail to operate correctly and it might be possible to resolve this by using a 74LS93 (or similar binary counter) to divide the clock down and provide a slower clock for the PCI bus (not sure, need to spend a bit more time tracing things on this motherboard first) - JP11 already sets a divide by 2 on the clock for the PCI bus, so if anything my PCI bus is under-clocked at 30MHz, however checking the datasheet for the SIS85C497 it should be possible to remove this jumper and inject a 33MHz clock signal directly using a separate clock generator.

Later when I get some more time I will take some screenshots of the benchmark results if people are interested.

Reply 267 of 280, by TrashPanda

User metadata
Rank l33t
Rank
l33t

Having had a good read of this thread I now have some idea of what 486 board to shoot for, was looking at a ShuttleHot 433 but after seeing what the Lucky Star LS-486E can do I think I will be picking one up. Not sure if its worth getting the Rev D over the Rev C2 as the C2 has 4 ram slots whereas the Rev D seems to just have the two. What else changed between the two versions ? also what is the max cacheable ram area for each revision? Rev D looks to have more cache than Rev C2 but the C2 is upgradable whereas Rev D has it soldered to the board.

This is a great Thread, so much valuable information here!

Reply 268 of 280, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++
TrashPanda wrote on 2022-10-06, 01:56:

Having had a good read of this thread I now have some idea of what 486 board to shoot for, was looking at a ShuttleHot 433 but after seeing what the Lucky Star LS-486E can do I think I will be picking one up. Not sure if its worth getting the Rev D over the Rev C2 as the C2 has 4 ram slots whereas the Rev D seems to just have the two. What else changed between the two versions ? also what is the max cacheable ram area for each revision? Rev D looks to have more cache than Rev C2 but the C2 is upgradable whereas Rev D has it soldered to the board.

This is a great Thread, so much valuable information here!

Lots of 486 performance board info in this monster too 3 (+3 more) retro battle stations

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 269 of 280, by TrashPanda

User metadata
Rank l33t
Rank
l33t
BitWrangler wrote on 2022-10-06, 02:21:
TrashPanda wrote on 2022-10-06, 01:56:

Having had a good read of this thread I now have some idea of what 486 board to shoot for, was looking at a ShuttleHot 433 but after seeing what the Lucky Star LS-486E can do I think I will be picking one up. Not sure if its worth getting the Rev D over the Rev C2 as the C2 has 4 ram slots whereas the Rev D seems to just have the two. What else changed between the two versions ? also what is the max cacheable ram area for each revision? Rev D looks to have more cache than Rev C2 but the C2 is upgradable whereas Rev D has it soldered to the board.

This is a great Thread, so much valuable information here!

Lots of 486 performance board info in this monster too 3 (+3 more) retro battle stations

I uhh seem to have even more to read !

Seems like Rev D is the one I need .. I dont have any FPM ram in my spares box but I do have nearly a gig of EDO Ram in 32mb sticks.

Reply 270 of 280, by gnif

User metadata
Rank Newbie
Rank
Newbie
Chadti99 wrote on 2022-10-05, 09:01:

Thanks for sharing, I’m def interested in more Benchmark results. This board supports 1024k cache?

I will do this tonight now I have pulled out the machine again. Yes it supports 1MB of cache, I just upgraded mine to 512K but there are markings on the PCB to show how to jumper it for 1MB.
Also I was able to get the PCI bus to run at full speed (60MHz) after identifying that my memory timings were a bit too aggressive and the cause of hangs, not the PCI bus.

Edit:

3D Bench for Fast PCs: 100.5
Chris's 3D Benchmark: 125.3 (75.2 FPS)
Chris's 3D Benchmark (640x480): 30.6 (18.3 FPS)
PC Player Benchmark: 28.2
PC Player Benchmark (640x480): 11.8
Doom min details: 387
Doom max details. 1150
Quake: 969 Frames, 51.8 seconds (18.7 FPS)

Speedsys

signal-2022-10-18-205331_002.jpeg
Filename
signal-2022-10-18-205331_002.jpeg
File size
325.71 KiB
Views
2349 views
File license
Public domain

Reply 272 of 280, by gnif

User metadata
Rank Newbie
Rank
Newbie
Chadti99 wrote on 2022-10-18, 10:58:

Very cool, can you share your memory timings/bios settings? Maybe we can help you optimize a bit further.

Sure thing, however I have iteratively tried every possible timing combination and had to settle on what I have here, anything tighter causes a hang just prior to the BIOS screen advancing to the next one when the OS starts to boot.

This is the output of a tool I wrote that dumps how the chipset registers are configured as this BIOS does not expose everything.

signal-2022-10-19-005640_002.jpeg
Filename
signal-2022-10-19-005640_002.jpeg
File size
324.56 KiB
Views
2334 views
File license
Public domain

Reply 273 of 280, by mr-spain

User metadata
Rank Newbie
Rank
Newbie
gnif wrote on 2022-10-18, 13:59:
Sure thing, however I have iteratively tried every possible timing combination and had to settle on what I have here, anything t […]
Show full quote
Chadti99 wrote on 2022-10-18, 10:58:

Very cool, can you share your memory timings/bios settings? Maybe we can help you optimize a bit further.

Sure thing, however I have iteratively tried every possible timing combination and had to settle on what I have here, anything tighter causes a hang just prior to the BIOS screen advancing to the next one when the OS starts to boot.

This is the output of a tool I wrote that dumps how the chipset registers are configured as this BIOS does not expose everything.
signal-2022-10-19-005640_002.jpeg

Those are some good numbers! I was hitting similar if not a little slower @180 here a little while back. Anyone curating an up to date list? What are your settings in the Doom demo?

Reply 274 of 280, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

I'm not your supervisor, but this thread is more about architectural comparison between various 486 board pluggable chips on directly comparable platforms, whereas all out fastest performance for any 486 is in threads like this one The World's Fastest 486

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 275 of 280, by mr-spain

User metadata
Rank Newbie
Rank
Newbie
BitWrangler wrote on 2022-10-22, 23:03:

I'm not your supervisor, but this thread is more about architectural comparison between various 486 board pluggable chips on directly comparable platforms, whereas all out fastest performance for any 486 is in threads like this one The World's Fastest 486

No worries bud I'll take it over there 😀

Reply 276 of 280, by leon22

User metadata
Rank Newbie
Rank
Newbie
amadeus777999 wrote on 2021-07-23, 18:24:

I had my Gigabyte GA-486 AM/S up and running and used a peltier element with a capable version of the 5x86. It's unfortunately not 100% stable which I can maybe change by hopping onto another board with a variable voltage regulator.
Nonetheless a nice experiment.

And which Graphic Card is in use here? Thank you!
(I was also not aware that the Am5x86-PR75 works at full speed with this Mainboard)

Thx

https://funwithretrocomputers.blogspot.com/

Reply 277 of 280, by arncht

User metadata
Rank Oldbie
Rank
Oldbie

Really nice bench… there is one thing at the conclusions, it would worth to check, which technologies were available at the specified year. Eg these highend scores based on very late 486 boards, setups, cpus - so if you want to compare, you should mention the typical use cases, which was really available. Eg:
* the p100 was available since 94, but practically just in a very low amount, so the typical setup was the p90. the p100 become more popular in 95, when the higher clocked p120/p133 was also on the market
* the typical l2 cache was until 95 the async wb cache in the pentium (socket5 boards), the pb gives +10%, but became popular in 96
* the pre-triton chipsets, boards was also slower, so if compare a p100 in a 94 setup to a 96 setup, it was a significant difference
* no proper l1 wb support before mid 95 for socket3 (practically you could buy at late 95, early 96 a mainboard with wb support) - many times you could enable it, but the realization was buggy, or simply the cpu was in wt mode. A cache bench shows the difference.
* intel released the wb based dx4 just at the very late of 95, no wb cpus on the market
* check the chipset availability - the tested umc board was probably the direct competitor of the pb caches pentiums, it was not “2 years newer”
* the support of the 5x86 in 1995 was very limited, and the cpu ran slower in these boards/bioses
* the pentium pod released just in 95

So i would compare period correct 94/95/96 etc cpus, mainboards, bioses, anyway you will compare 2-3 years old technologies, and something sounded way better in 94, than in 96, or they perform differently in a period correct setup.

* The intel dx4 wt was available in 94, the amd in 95, the 5x86 with the proper mainboards just in 96, when mainstream was much more the p100-133 with pb cache tritons.
* If you test an intel dx4 and the 5x86 in a board from 95, they will give very similar performance in fps tests on average, but it can be +-10% performance difference in the tests, etc

My little retro computer world
Overdoze of the demoscene

Reply 278 of 280, by amadeus777999

User metadata
Rank Oldbie
Rank
Oldbie
leon22 wrote on 2023-08-13, 11:45:
And which Graphic Card is in use here? Thank you! (I was also not aware that the Am5x86-PR75 works at full speed with this Main […]
Show full quote
amadeus777999 wrote on 2021-07-23, 18:24:

I had my Gigabyte GA-486 AM/S up and running and used a peltier element with a capable version of the 5x86. It's unfortunately not 100% stable which I can maybe change by hopping onto another board with a variable voltage regulator.
Nonetheless a nice experiment.

And which Graphic Card is in use here? Thank you!
(I was also not aware that the Am5x86-PR75 works at full speed with this Mainboard)

Thx

It was a Matrox Millenium II - as they usually work flawlessly at high bus speeds(including 66mhz).

Reply 279 of 280, by Siran

User metadata
Rank Newbie
Rank
Newbie

Basically bottom of the barrel for a change, but rather proud I got the 4386-VC-HD ISA board that once hosted a meager 386DX-40 fully upgraded with a 486 DX2-66 and 256K cache (inkl. Alter RAM):

20230804_132216.jpg
Filename
20230804_132216.jpg
File size
1.27 MiB
Views
1198 views
File license
Fair use/fair dealing exception
20230804_114331.jpg
Filename
20230804_114331.jpg
File size
993.02 KiB
Views
1198 views
File license
Fair use/fair dealing exception
20230804_115925.jpg
Filename
20230804_115925.jpg
File size
1.16 MiB
Views
1198 views
File license
Fair use/fair dealing exception

Quake low: 6.5fps, PC Player VGA 8.6 with a CL GD5429 ISA VGA Card.