VOGONS


dis mobo, 486

Topic actions

Reply 20 of 23, by amadeus777999

User metadata
Rank Oldbie
Rank
Oldbie

The 486 has lots of potential which overclocking and late era boards can unleash.
The Pentium's efficiency can't be denied though and the 486s writing was on the wall as soon as '93.

Few days ago I tested a DX60(Overclocked DX50 on a LS486E@60mhz fsb(ram fastest settings/cache one notch below at 3-2-3) against a P60(standard IntelChipset + rather "slow" corporate bios/no tweaks) and latter reaches performance levels relative to former of...

130% in Doom ShareWare("486@78")
132% in 3DBench2("486@79")
220% in Blood("486@132")

Admitted, the much touted "DX4-100 boost" would be around 160%+(early Pentiums were always compared to DX4s) but this seems to mostly help with "antique" software.
Blood and Quake for example, easily annihilate the 486's clock advantage by fine tuning key-routines for the new architecture.
Also a clock-ratio of cpu2ram at 3:1 for the 486 seems to be appropriate as I could not get a meaty speedup at a 1:1 ratio between fsb and core on a "high end" board(odd cases aside).

Above stated should be more closely inspected of course as the small array of tests could strongly skew the overall picture.

Reply 21 of 23, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

I'd say what pentiums and ppro even more so, sucked at, was 16 bit code, so fire up windows and a pentium feels doggy. Even with win32s or in win95, it's like your passenger is stomping the brakes every time it meets a bit of 16bitness.... which clung on right into windows XP..... that's how the 6x86 and winchips won their spurs and their PR ratings.... business winstone.... which emulated the also 16 bit heavy office suites etc... so dx4s, 6x86, winchips, and in some aspects the K6, win over the pentium by being really really fast 286es, rather than being teh fastest on "386" 32bit code. (Though K6 had the stones there too, so had balanced performance.)

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 22 of 23, by amadeus777999

User metadata
Rank Oldbie
Rank
Oldbie

So, where would that awful "16bit-ness" occur?

Not really sold on the momentum of the 16 bit "slowdown" - more like good ol' marketing blabber.
Reminds me of how some magazines tried to paint the PentiumPro in a bad light compared to the older Pentium classic although former was a great performer despite all its supposed flaws - the weaker performance in some areas was made up more than enough in other more vital ones.

According to http://www.fermimn.gov.it/inform/materiali/ev … ir/perfchrt.htm the difference between a Pentium and the faster Cyrix processors is not really huge - in winston96 for example the Cyrix200+ scores 91.6 compared to 89.0 delivered by the Pentium 200(lovely 2%).

Of course there's more to this than some (maybe handpicked?) benchmarks but I always got the feeling that "hearsay" dictates this 16 vs 32 bit performance quagmire.

Reply 23 of 23, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

in all the dlls and exes that weren't ...32.dll or ...32.exe

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