[Adrenaline]

Wew, good to see you again Adrenaline. Been so long since I've seen you around.

Glad to be back - hopefully I'll have some free time in the coming weeks to get back into the SGI world.  It has been a while.

[mapesdhs]

Today I checked to see if the "timedemo 1" command works in ioQuake3 on my O2 R12K@400MHZ. And it does! ...

Just wondering, does it make any difference to your Onyx2 results if the running environment has the following added to .cshrc? It might go up a bit with O2, but it's the big iron that interests me (O2 is too fill limited for Quake3 anyway).

Code:

setenv DECOUPLE_SWAPBUF y

With regard to performance, all the Quake ports for SGIs were very rough and ready, none of them are properly optimised for SGIs, so indeed there's often heavy CPU dependency. I'm not sure if this means whether at higher resolutions the CPU dependence is made better or worse (with modern PC games it varies). I only bothered testing Quake and Quake 2, though I need to redo some of the tests with single buffered active:

http://www.sgidepot.co.uk/quake1bench.html
http://www.sgidepot.co.uk/quake2bench.html

Ian.

[Irinikus]

These are the results of running:

\timedemo 1
\demo four

on the Onyx2



It seems like a low frame rate compared to the earlier post, but the level seems to be allot more taxing than the standard levels I've shown.

Ian, the command: "setenv DECOUPLE_SWAPBUF y" didn't make a difference to the results.

These are the settings I used:



I'm far away from my Tezro at the moment, so I'd really appreciate it if someone here could run the same test, with the same settings on their Tezro or Octane2 for comparative purposes?

[Irinikus]

Here's the machine's hinv:

[mapesdhs]

Oh, in that case I guess the timedemo test is already forcing the run to be single buffered. That setting can make a big difference for other types of testing, including Quake2.

Ian.

[Irinikus]

As a matter of interest, here are the O2's results with the same settings:



here's the O2's hinv:

[mapesdhs]

Kinda figures given O2's low texel fill rate. :D  Always thought O2+ should have used Cobalt and been a proper followon with GigE & suchlike, but that's a rant from which I'd best stay far away. :}

Ian.

[Irinikus]

Kinda figures given O2's low texel fill rate.   Always thought O2+ should have used Cobalt and been a proper followon with GigE & suchlike, but that's a rant from which I'd best stay far away. :}

Ian.

I agree! The O2+ was hardly an upgrade!

[mapesdhs]

It wasn't any kind of upgrade. :}  (unless one can claim that somehow a purple case enhances brain function by channeling Aldo particles or something)  But yeah, that's a rant for another thread. :D

Ian.

[Irinikus]

Here are the Tezro's results for the same test: (The V12 definitely doesn't produce that same image quality that the IR4 system does!!!)





Here is the Tezro's hinv:

---

In due course I'll also test the dual 600MHz Octane2 and the Indigo2 R10000 Max IMPACT

---

So for now the results are the following:

Onyx2 Quad R14K@500MHz, Infinite Reality 4 - 37.6 FPS
Tezro Quad R16K@1GHz, V12 - 24.5 FPS
O2 R12K@400MHz - 3.5 FPS