..

Ran this modified exe on my 386DX40 but the FPS counter is at 0 all the time...

There is already a Wolf3D benchmark program.

Look in this thread:
Performance comparison of CPU: 286-25 vs 386DX-25 vs 386SX-25

and download the benchmark pack. It's under 3D directory.

keropi wrote:

Ran this modified exe on my 386DX40 but the FPS counter is at 0 all the time...

The counter in the lower left during the 3D sequence ? First time I see this issue. It worked in DOSBOX and many other real PCs (286 and 386). Does the game run normally? Did you use the latest version ?

Can you try to run the other benchmark then : Performance comparison of CPU: 286-25 vs 386DX-25 vs 386SX-25 ?

Yes the counter on the bottom-left corner.
I am using a 386DX40 with 128kb of cache and a WD90C31 vga.
The other benchmark gives me 29fps

I have a rather nice DX25 with 8MB & a Cyrix FasMath (which I realize is irrelevant for Wolf3D) that I'm in the middle of putting together. I'll try this on the weekend. 😀

What do you want us to do for soundcard settings? Disabled?

Adlib/SoundBlaster sound to put a little bit of stress on the audio subsystem please 😀 !

kixs wrote:

There is already a Wolf3D benchmark program.

If you're referring to the one at http://thandor.net/article/10 then you can't use that, the framerate counter it implements is designed incorrectly. The one linked at the top of the thread was developed specifically for Wolf and produces correct results (verified by analyzing video captures compared against the values produced by the .EXE).

shamino wrote:

I know you meant it only for convenience, but I suggest removing the Wolf3D link and any references to that site. This thread is bound to get locked. Please note the message at the bottom of the page, it's not just a disclaimer but also a policy that is enforced strictly.

The linked executable is only for the executable and not the full game. It was compiled, with modifications, from the source code iD released publicly in 1995 on their FTP site (which is still there). Nothing is being violated here; this is just a benchmark request using a specially-compiled executable.

Disclaimer: I benchmarked several systems for this, but my 386dx-40 died before I could get to that system, hence this request of the community.

This is not the executable from http://thandor.net/article/10 . As you mentioned, that FPS counter had two flaws:

- It waited for 35 ticks to generate an fps value an multiplied it by 2. You could only get even numbers.
- It used "tics" variable which can never be less than MAXTICS (10), therefore the fps measured could never be less than 7.

This executable has a fps counter which address both these issues.

Looking forward to the book!

Do you need the benchmark of DX CPUs only ? I've got a 386SX 25MHz

I get 29 fps (lowest value) & 30 fps (highest value)
System specs
CPU: AMD 386DX 40 MHz
Motherboard: Chicony CH-386-33H/40H (UMC chipset)
System RAM: 8 MB
Graphics: Tseng Labs ET4000 with 1 MB RAM

Relevant BIOS settings have been configured for optimised performance
Once I move around a bit and started to engage the enemy, I noticed that it sometimes drops to 27 fps.

MobyGamer wrote:

Nothing is being violated here; this is just a benchmark request using a specially-compiled executable.

That isn't what I was referring to, but it's been resolved.

I'm not the target audience for this, obviously, but I have to wonder: why aren't benchmarks made to work with the shareware version of games that have one? It avoids encouraging piracy as you implicitly do in the first post, and no one is going to (or at least no one should) buy a game just to run a benchmark anyway.

This is why I haven't participated in the Doom 3 benchmark thread: I only have the demo version, and I wouldn't care to get the full version anyway.

@koverhbarc

id Software released the source code of the commercial version of Wolfenstein3D. If you attemps to build and run the executable with the shareware assets, the game will crash. This is because the assets ID were generated in .H header files and compiled in the executable. To work, the .H headers have to match the assets order.

I think NY00123 can create a Wolfenstein 3D shareware executable.

Got a solid 18 FPS on my machine now that it's up and running (which ended up being a huge pain in the butt, but that's a story for the build thread.)

System specs:
Intel 386DX/25
Cyrix FasMath CX-83D87
UMC 82C206L chipset
8MB 70ns RAM
128k(?) cache

Trident 8900cl 1MB VGA
ATI Stereo F/X CD (Soundblaster sound enabled)
Gravis Ultrasound Max 1MB
Gravis Gamepad

No HDD, running from Zip drive

I'm interested in the book too BTW. 😀

For the record I had a MAJOR brain fart during the system build. It was supposed to sport a nice zippy Tseng ET4000, but I completely didn't clue in that card was PCI until I physically had it out of the box in my hands. So I had to make do with a crummy Trident card instead. Willing to bet it lost a few FPS because of that.

^ nice one!

IMHO you are better off without the ET4000... grab a 5424+ or a WD90C30~31 and you'll have same speed but more compatibility.

fabiensanglard wrote:

id Software released the source code of the commercial version of Wolfenstein3D. If you attemps to build and run the executable with the shareware assets, the game will crash. This is because the assets ID were generated in .H header files and compiled in the executable. To work, the .H headers have to match the assets order.

Yes, that's right: this is how id did things before Doom. It's one reason modification was so difficult, though I'm sure some people must have figured out how to binary-patch the executable (or did compression make that impossible?).

So I guess you can't be blamed.