amadeus777999 wrote:

could you be more specific on the sloped surface issue in Duke?

build apparently uses a couple of fpu instructions for slopes, so it's no surprise it would chug on a system lacking an fpu.

Nice, I gotta admit I dreamed about owning one of these back in the day.

auron wrote:

amadeus777999 wrote:

could you be more specific on the sloped surface issue in Duke?

build apparently uses a couple of fpu instructions for slopes, so it's no surprise it would chug on a system lacking an fpu.

Yery interesting info, maybe the integer version can be patched back in there?

there is a reference to nexgen in dn3dhelp.exe and 3drealms' website, did you try these steps already?

http://legacy.3drealms.com/tech/allverfixes.html wrote:

NexGen/586 CPU's

If you have a NexGen, you should first try going into your CMOS settings, and looking for a setting called Alternate Instructions. Make sure this is on. If this still doesn't work, visit their WWW site at http://www.nexgen.com and download the Q87 Floating Point Emulator. This should get you going.

actually i wonder how q87 would perform on quake, nexgen did claim to have tested it with a few other games in their guide.

auron wrote:

there is a reference to nexgen in dn3dhelp.exe and 3drealms' website, did you try these steps already?

http://legacy.3drealms.com/tech/allverfixes.html wrote:

NexGen/586 CPU's

If you have a NexGen, you should first try going into your CMOS settings, and looking for a setting called Alternate Instructions. Make sure this is on. If this still doesn't work, visit their WWW site at http://www.nexgen.com and download the Q87 Floating Point Emulator. This should get you going.

actually i wonder how q87 would perform on quake, nexgen did claim to have tested it with a few other games in their guide.

Any FP emulator will do actually, though they are slow. I don't think the fps are anywhere close to playable. Those alternate instructions are a few 486 specific instructions (CMPXCHG, XADD, BSWAP) emulated through hypercode loaded from the BIOS chip into the L2 cache on start-up. Nx586 is a very advanced 386 processor in fact.

Opti Mainboard Pentium Speed 😳

It isn't OPTI. The NxVL chip was designed by NexGen and manufactured by Fujitsu. 82C206 / 83C206 is a standard peripheral controller manufactured by hell knows how many vendors including OPTI.

Their PCI mainboard is based on two NexGen designed chips (VL82C500PSFC and VL82C501PSFC), Symphony SL82C105 PCI IDE controller and SMC FDC37C665GT floppy controller.

Both mainboards also make use of a standard 8042 compatible keyboard controller.

auron wrote:

amadeus777999 wrote:

could you be more specific on the sloped surface issue in Duke?

build apparently uses a couple of fpu instructions for slopes, so it's no surprise it would chug on a system lacking an fpu.

Ok, interesting, thanks.

amadeus777999 wrote:

auron wrote:

amadeus777999 wrote:

could you be more specific on the sloped surface issue in Duke?

build apparently uses a couple of fpu instructions for slopes, so it's no surprise it would chug on a system lacking an fpu.

Ok, interesting, thanks.

I`m aware of that, i just mean its like a OPTI based Pentium 60... sloooow as hell 😀

i really would like a Nexgen System 😁

With Duke3D: Using Q87 I get 11 fps in the first scene and without 13 dps. So loading Q87 is actually slower than the integrated FPU Emu.
(Also when I install Q87 to Win95 is locks up on boot)

Have not tried Q87 or any FPU emulator on my VLB system - but willing to give it a try and post here. P90 is my default CPU - small heatspreader. Have a P100 big heatspreader that also works nicely on the VLB board, but my P110 big heatspreader doesn't - think is not supported.

Any suggestions for FPU emulators?

Well, there is an FPU emulator included if you run Win3.x

Q87 works in protectmode ?

matze79 wrote:

Q87 works in protectmode ?

From Q87.TXT

> Q87 runs with DOS 5 or 6, all DOS Extender applications,
all games, Windows 3.x, and Windows 95.

elianda wrote:

I finished? my Nx586 sytem and you can read about it here: http://retronn.de/imports/nexgen_nx586.html […]
Show full quote

I finished? my Nx586 sytem and you can read about it here: http://retronn.de/imports/nexgen_nx586.html

Here is a preview pic:

Now dust off your NexGen systems for some comparison scores 😀.

Dammit! Now I'm going to have to build one of mine.

elianda wrote on 2016-11-06, 01:29:

.......

I mean Duke3D was recompiled for 386, maybe the compiler has Nx586 as target as well? Could be worth a try.
Also the Nx586 can be switched to run native RISC86 code.

At this point I must consider this mythical ability of Nx586 to be a an urban legend. I've scoured over every bitsavers 5x86 datasheet and Usenet post and could not find a single reference to somebody successfully executing a native Risc86 instruction.

From http://landley.net/history/mirror/collate/cpu3.html

The NexGen/AMD Nx586 (early 1995) is unique by being able to execute its micro-ops (called RISC86 code) directly, allowing optimised RISC86 programs to be written which are faster than an equivalent x86 program would be, but this feature is seldom used. It also features two 16K I/D L1 caches, a dedicated L2 cache bus (like that in the Pentium Pro 2-chip module) and an off-chip FPU (either separate chip, or later as in 2-chip module).

"Seldom", hehehe.

From https://groups.google.com/g/comp.arch/c/8Wqdj … /m/jKtPNTJ94RIJ

One of the latest bullshitting concerning microcoding was successfully implemented by NexGen in their line of Nx586 CPUs, where […]
Show full quote

One of the latest bullshitting concerning microcoding was successfully
implemented by NexGen in their line of Nx586 CPUs, where the x86
instruction set was implemented in the microcode, called RISC86. And
they (NexGen) promised to open this option for programmers (just using
one instruction to turn ON/OFF the internal microcode core, responsible
for x86 instr. set interpretation. Unfortunately, NexGen is the property
of AMD, and (as far as I know) AMD plans to stop this line of
bullshitting.

By the way, Nx586 runs very nicely with 1Mb of cache. Much better than with 256Kb of which 64Kb reserved for microcode.

This mod is still on my list. The ICs have been sitting in a bin; I'd have to take the motherboard out of the case though. Did you just replace those 8 chips, or was there more to the mod? Did you use solder paste, or solder the pads one-by-one with a contact-based soldering iron?

When you say that 1024K is better than 256K, did you run any numbers to demonstrate the comparison?

Are you running the PF110 CPU? My system has the PF110, so I was concerned that perhaps the higher FSB might cause issues with 1024K vs. 256K using the same memory timings. This is an issue I've witnessed on socket 3 boards.

This is in my todo list also, chips are ready but there is no time to deal with it...

feipoa wrote on 2022-08-17, 18:51:

Are you running the PF110 CPU? My system has the PF110, so I was concerned that perhaps the higher FSB might cause issues with 1024K vs. 256K using the same memory timings. This is an issue I've witnessed on socket 3 boards.

Isn't the cache on a separate bus? Or is that still dependant on the FSB?