Author Topic: Benchmarking frame rate in AH (Read 1543 times)

fradim · « **on:** March 04, 2000, 08:33:00 PM »

Hi, guys :-)

I´ve been running a site dedicated to benchmarking frame rate in iEn Warbirds, formerly avin´s framerate benchmark page: http://www.grandeavenida.com/fps/index.cfm

I have been thinking about offering the same information to Aces High players. But I have to adapt current protocol to Aces High. So I would like to know if any of you could help me with some suggestions.

fradim

hblair · « **Reply #1 on:** March 04, 2000, 10:57:00 PM »

You've done a kickbutt job filling avins shoes, I was over at the site a few days ago and was really impressed.

Quote

But I have to adapt current protocol to Aces High.

hmmm, I admit, I have no clue what you're talkin abot there, but I would be happy to give you my FPS from a certain field/plane etc.

------------------
hblair
JG 77 WB's Historical Arena & Events
The ASSASSINS WB's Main Arena, ACA
Luftwoobies Sturmgruppe Kommando Aces High

[This message has been edited by hblair (edited 03-04-2000).]

fradim · « **Reply #2 on:** March 04, 2000, 11:39:00 PM »

Thx for your support :-)

Let me give u an example why we have to adapt protocol to AH:
Warbirds benchmark protocol uses CV (field 5)to create a situation of frame rate "stress" - like the one u find in furballs.
Aces High doesn´t have a CV. So, is it possible to create such a situation of "stress"?

BTW, yesterday I have finished a new version of the page. Hope u still like it :-)

fradim

SC-GreyBeard · « **Reply #3 on:** March 05, 2000, 05:57:00 PM »

perhaps using formatrions of Buffs??

more than 3 buffs in a screen drops my FR dramatically

------------------
GreyBeard
Flight Commander, Aces High
"Skeleton Crew"
"Fly with Honor"[/i]

tshred · « **Reply #4 on:** March 06, 2000, 02:24:00 AM »

Well, I see the bigest drop in framerate when in the cockpit of a B17 or B26, add some clouds and another buff and it's bad, around 19 to 22. That was in previous versions, haven't been in the clouds in buffs with current version, but with the current version I have gained 5 to 6 more frames/sec than before, avg low 50's in 800x600 and 1024x768. I'm running the following:

celery 366 o/c 458
128 PC100 mem
Voodoo 3 3000 o/c to 180

weazel · « **Reply #5 on:** March 06, 2000, 03:28:00 AM »

Any FPS test *must* have standardized parameters that are exactly repeatable for each player that performs it. At one time I was working on a benchmarking site but my ISP won`t let me run .cgi scripts for an interactive database. The best I could come up with at the time was offline select an FW-190 at field 1,while on the runway turn off ground clutter,set viewing range to maximum distance,start engine with lowest throttle setting and record the FPS. Any plane will work but the FW seemed to be a good balance between the lower frame rates of a B-17 vs the higher FPS in some of the other fighters.

------------------

JG2 "Richthofen"

[This message has been edited by weazel (edited 03-06-2000).]

Dinger · « **Reply #6 on:** March 06, 2000, 06:03:00 AM »

Weazel, you're more or less right as to repeatability, but we want to get as low a FR number as possible, because it's in those high-stress situations that you can really see the diff. between PCs. If we make the test too easy, the fast machines will "top out".
Here's what I'd recommend:
Record your video mode
Off line
F1
B17G with 100 lb bombs. (I believe AH draws each one of those things in your bay. If it only draws some, I guess 500s) 100% fuel, guns (Don't forget 'em)
Go on the runway
Keep clutter on
Ensure viewing range is at max (shift-f4?)
turn on flight recorder
crank up motors.
Hit home key to neutralize view
When all four engines come to idle, record FR.

[This message has been edited by Dinger (edited 03-06-2000).]

Gie · « **Reply #7 on:** March 06, 2000, 08:31:00 AM »

You want FPS-stress ?
Load 100lbr into B17, set salvo 24 and drop em on runway

llbm_MOL · « **Reply #8 on:** March 07, 2000, 11:12:00 PM »

WTG Fradim!

When Ya coming over here and join the rest of us MOL's?

Anyway it looks like its going to be hard setting up a stressfull situation in AH. Wonder if Pyro or HT could help in this regard?

LLBM OUT!!!!!!!!

Dinger · « **Reply #9 on:** March 08, 2000, 04:04:00 AM »

Time to propose a benchmarking scheme:
(modifications to above: I took out the Film Recorder since it doesn't seem to affect the FR in a steady manner -- only occasional disk access that really doesn't help; clipboard does take out a couple frames.)

1. set/record resolution and color mode (16 or 32), remember to write down whether you're using HiRes terrain or not.
2. go "offline"
3. go to hangar
4. select a B-17G with 24x100 lbers, 100% fuel and MGs.
5. click on "fly"
6. start engines with idle throttle
7. bring up the clipboard (esc key)
8. Hit home key to revert to default view
9. Wait for the Frame Rate Counter to stabilize, and record FR

Here's mine:
System:
Celery 333@415Mhz, 64 Meg Ram, cheap CL TNT2 M-64 with 32 megs.
Hi res terrain
16-bit color mode
FR: 640x480: 25; 800x600: 25; 1024x768: 17

Note that going to the belly gunner gives you an even lower frame rate, but for some reason having to draw all those cockpit dials makes a big FR difference on my FE. If I add "jump to belly gunner" to the list, I get 15 FPS all the way across.

Respectfully submitted,
Dinger

[This message has been edited by Dinger (edited 03-08-2000).]

weazel · « **Reply #10 on:** March 08, 2000, 04:19:00 AM »

Using Dingers test procedure I get 31 FPS in 32 bit 1024x768 with hi-rez art package.

Celeron 466@525 mhz
256 mb CAS 3 PC-100 SDRAM
V770 TNT2 32 mb

------------------

JG2 "Richthofen"

[This message has been edited by weazel (edited 03-08-2000).]

tshred · « **Reply #11 on:** March 08, 2000, 05:45:00 AM »

Okay, here's what I get and some observations.

23 using Dingers test in 640, 800 and 1024, hi-res art installed. Note I'm limited to 16 color due to my video card.

My system:

celery 366 o/c 458
128mb cas 2 PC100
Voodoo 3 3000 not overclocked

I did notice the biggest drop however when going to belly gunner, and clipboard did not affect FR any way, but swiveling until tower and right main gear in same view and a solid 15 is the result, but 17 otherwise all other directions. This also held true for the top gunner position. I would suggest using top or bottom gunner position looking at tower the ultimate FR test to far.

ts

Dinger · « **Reply #12 on:** March 08, 2000, 06:22:00 AM »

Hmm... Clipboard is a bit of a problem now that I think about it. It does (on my machine at least) affect frame rate (I see a 2fps difference after it "settles in"), but the clipboard does not have a "Standard size", and this can slightly affect results. Still, I'm disposed to keep it.
What makes me shy away from the gunner views, is that, on my system at least, I have a huge FR difference between 1024 and 800 resolutions when in the cockpit, but that difference disappears in the belly gunner.
Looking at tower will give you a lower FR, but it isn't as precise (what if I have the tower slightly to the left?)

fradim · « **Reply #13 on:** March 08, 2000, 08:29:00 AM »

I believe we may be closer of a protocol to AH :-) thx dinger and everybody!

Remember we just don´t want to lower our framerate: above all, we need definining a set of rules that may help us comparing diferent system.

I will wait more ideas and probably prepare a first version of this protocol this weekend.

LLbm I wish I could have some help from HT but I´ve send an email a month ago and I had no reply :-( And I hope to join MOL at AH soon. My only problem is $:-)

[]

fradim

tshred · « **Reply #14 on:** March 08, 2000, 11:01:00 AM »

Dinger, I shoulda mentioned that the clipboard didn't affect FR in gunner positions only. In cockpit, it drops me down 2 as well. Also, your video card is whats limiting you to a lower framerate in 1024, where as on my system, my cpu is limiting my graphics, hence the same framerate in all resolutions(I think, don't quote me)it could also be that you are using 32 bit color where I am only using 16, so that could also be why I have the same FR in all resolutions. Someone with more knowledge could probably say what the reason is for sure, but I think that's why my FR is all the same.

But for all intents and purposes, we do need to set one standard, say 16 bit color, or have different tests appropriate for different video cards. Who's the video guru out there? Help us!

ts