OldUnreal (Engine)

...I'm not sure would be duly befitting of this forum, but it's more an observation and a question about the engine that has left me stymied ever since and that I'm wondering if anyone knows an answer to. This is about the first engine, to add.

Has anyone noticed that in the Glide, Direct3D, and OpenGL, that the lower the resolution you set the game to, even when hitting your CPU limit in frame-rates, that the faster the game gets, akin to Software Rendering (but still not quite as huge a jump)? If you run the game windowed and in questionably low resolutions, especially with sound disabled, the game gets an amazing augmentation. Does anyone know how this could be? This is more the demeanor of software rendering, isn't it?

An example:

In the Unreal Flyby, in Unreal Gold, on an AMD Athlon 64 3800+ (single core, "Venice", at 2.4 GHz), with an nVidia graphics board of recent and in the latest OpenGL renderer for Unreal Gold, at 1024x768, with sound enabled, the very first few frame-rates would be around 150FPS. If I ran the game at 320x200, it might be closer to 187FPS. If I go windowed, without sound, and lower, it's around 220FPS. And if I raise the resolution above 1024x768, it drops to 140FPS (and I know with a GeForce 9600 GT, there's no way it'd be that low). Antialiasing is disabled and the only things enabled are Anisotropic x16 filtering and trilinear filtering. V-Sync is disabled. Nothing is forced in the nVidia control panel, and the frame-rate limiter of this OpenGL renderer is not in use. My graphics card is at stock speeds too and my system is not unstable. This happens to me with any version of Unreal on any system and only with that engine, and that game (as well as Return to Na Pali, of course).

I know this isn't necessarily a problem, and might be a questionable thing to ask for some. It's just something I'm wondering about, as to think that the game isn't as fast as it could be even when maxing out the speed of the CPU at higher resolutions.... is very addling, for me. Any feedback is very appreciated in full.

The only thing I think is that you should so totally enable VSync.

Oh, of course I have V-Sync enabled. It's just a facet of the engine that flummoxes but also commands deep intrigue, for me. It's not one I get, but it's not necessarily an issue. I was wondering if this came to anyone else's notice.

It's just a facet of the engine that flummoxes but also commands deep intrigue, for me.

Lol you talk english but I don't understand that anyway.

You'd think he's talking about the fact the engine doesn't handle certain CPU cores very gracefully, especially if they have "power saving" features...

What the hell else would cause framerate to sky rocket for no reason?

I'm very sorry. Forgive me for my verbose. I'm just saying that it's an aspect of the engine that's confusing but intriguing. But it's not just certain CPU cores, as I see this happen with every system I test Unreal on (multiple or single core). But in reiteration, it's not necessarily an issue, though it does boggle me why I can't retain those frame-rates at higher resolutions when the engine is still nowhere near intense enough at current settings to be GPU-limited.

I don't understand the question or problem. The FPS relies on more than just the GPU. If increasing resolution the engine "has more to do". (all render passes, calculations etc.)
Same for disabling sound. Certain routines are skipped then, so the result is a higher FPS.

And not every routine is completely optimized (which is again, seen realistic almost impossible within a complex engine of that size).

If you enabled VSync you should have never more FPS than refresh rate (f.e. 60Hz = 60FPS), but that's only as a side note.

It maybe shouldn't differ that much anymore if you set Unreal.exe process priority to realtime (over task manager) to make sure it really uses always all CPU power available and making sure that way it really is only limited by the available CPU speed. But that's only theory.

But I wouldn't know an engine which behaves different- even if not noticeable that much maybe)

Of course it does. I'm saying, though, in every game, when a certain amount of content is rendered, the CPU can only calculate so much at a certain frequency. So if you paired a very fast GPU with a slower CPU, you'd be CPU limited for a large part, unless you ran at higher resolutions with FSAA/AF enabled. What's befuddling me here is that in this version of the Unreal engine, I've already hit that CPU limit at 1024x768. I am afar from seeing the GPU's theoretical speed of rendering the game at these settings because I'm CPU-limited. Yet when I drop the resolution, the frame-rate still increases glancingly when I've already hit my CPU limit. I hope what I'm getting at is becoming more lucid now as I don't want to perplex anyone. I find this quirk in no other engine.

just for curiosity, did you try it with 227 too?
Because Unreal as well as UT was originally compiled with vc++6 and some very old OS, maybe even Win98.
Only 227 is compiled with a recent compiler and its optimizations. Although Unreal contains ASM code for some routines (mostly for software rendering, but for Fire-effects f.e. too), most of it is of course plain c++ and lacks because of that and the old compiler it was compiled with a lot of optimizations, especially for newer CPU's and OS's.
The first time I played Unreal with was a PII-233 and this was already end of optimizations except some rare 3DNow ASM code here and there....

On my system here (with Unreal 227) I have with full AA all other eye-candy settings maxed out @ 1280x1024 a max of 20% CPU load on one of my cores with VSync enabled. Without I have ~200FPS with a CPU spike of ~40%. But my system is a Phenom 9550, so no idea if that counts

Hmm, I just had at trying build 227f....

OpenGL looks richer, but I noticed the frame rate was a bit lower for me. And I'm still encountering the same thing as detailed through this whole thread of discussion. But don't worry, it's not necessarily a problem, just something I can't find coherence in.

i'm more curious than worried. Interesting however that I'm unable to reproduce it that way...

ex member

I think that this is related to vertex input bottleneck.

-UT2k4 BSP doesn't occlude, maybe that's why they have antiportals, plus the fact that GPUs were no longer a bottleneck like in Unreal software render days.

Hmm... how would the GPUs have so much an effect in software rendering, besides driver woes? Wouldn't the CPU be what primarily helms performance in software rendering? And I'm curious as to this BSP occluding/vertex input bottleneck you speak of. It causes the frame-rate to rise when lowering the resolution even when intensely CPU limited at higher resolutions? If so, that's fascinating to me.

Spans.

I think the problem is with your cpu size. These are these we used to encounter before the more sophisticated systems came out in the last 3 years.

I'm referring to the dual core processing.  This may be a solutions to what your seeing.

___________________
Logan
[url=http://www.fxrenew.com]Forex Broker[/url]

i've noticed this in concerns of widescreen. It seems to speed up the wider the screen gets...