View Full Version : Varing Benchmark Tests (GSdx)
CKemu
June 25th, 2004, 19:30
Got a New Rig, might as well test PCSX2.
Tests done using a BETA build of GSdx.
RIG:
AMD Athlon 64 3200+
1GB DDR400 RAM
120GB SATA HDD
ASUS K8V RAID SATA Mobo
GeForce 5950u (56.64 Drivers)
PCSX2:
Current BETA Build with CVS plugins.
VUrec ON / CPUrec ON
Console OFF
SCPH10000.BIN
First Up: SCPH10000.BIN the only BIOS that goes into main menu's etc.
Note VUrec is turned OFF for the Memcard screens, as it causes PCSX2 to crash.
CKemu
June 25th, 2004, 19:49
Super Bust a Move
VUrec on or off makes no 'significant' difference to preformance, these shots where take with it off.
I'd consider this a 'playable' game in all senses as though this is not full-speed, it is certainly 'enjoyable' and suffers from no glitches 'severe' enough to make it difficult to play.
CKemu
June 25th, 2004, 19:52
Super Bust a Move Continued
CKemu
June 25th, 2004, 20:13
EverBlue
Running with vuREC off, as vuREC crashes in-game. As you can see, menu's are lightning, in-game is painfull. It's playable if your extremely patient ;) :evil:
refraction
June 25th, 2004, 20:47
did a couple on my system, with the most obvious test everybody does, FFX ;p couple of the bios too. These are purely to show the framerates it runs at in different places rather than the graphics itself.
My system:
Athlon XP 64 3400+
Radeon 9800 Pro 256Mb
300Gb & 80Gb hard drives
Onboard soundcard (crap)
1Gb DDR 500 Memory (only running at 400mhz tho :( )
PCSX2 CVS build Settings:
Recompiler Enabled
VUrec Disabled
Console Disabled
GSdx9 recent CVS
No sound plugin (tho it doesnt make much difference)
jareg
June 25th, 2004, 21:41
Hey nice shots
The PCSX 0.7 Beta is looking good and Gsdx :)
refraction
June 25th, 2004, 22:00
i must say the speed of the 2 cvs's is getting very impressive :)
Vanit
June 26th, 2004, 13:54
I think we've seen enough screenshots of PCSX2 for these to be pretty pointless. As CKEMU has said previously, and I'm sure he will agree with in this instance, only post pictures if there is a significant performance increase, which I might add is not a trait evident within any of the pictures presented so far.
CKemu
June 26th, 2004, 17:05
#1: (I fear the remarks about "oh look he's bragging again") I am a beta tester, and these shots show a significant increase in speed, as Gabest has made changes to GSdx, which have gotten a rather nice speed increase overall.
#2: They demonstrate to all those begging for blah to work on they're underpowered machines, that even titan-a-like machines which I and several others own have a hard time reaching 'full-speed'.
#3: Exactly how significant is 'significant' 2 FPS over a previous beta running at 1 FPS is still 2x the speed ;)
#4: The sheer fact that the BIOS system configuration runs at high(er) speeds is pretty significant IMO and that of the DEVs (though of course I can't speak for them completely, they did seem pleased).
#5: meh.
Raziel
June 26th, 2004, 17:18
omg he's bragging again ! :p
;)
ChaosCode
June 26th, 2004, 17:42
CKEmu is that the newest pcsx2 cvs also?
CKemu
June 26th, 2004, 17:58
Certainly is. Though the compile I am using is optimised for SSE2.
ChaosCode
June 26th, 2004, 18:04
grrrr don't tell me pcsx2 is gonna use SSE also.. Now I really need to update :oP
_E_
June 26th, 2004, 18:14
It does use SSE from the very begining....
Yours,
-Elly
ChaosCode
June 26th, 2004, 20:05
as long as it doesnt HAVE to have it Im cool with it :-p
CKemu
June 26th, 2004, 20:59
Hah, even I need to upgrade, or PCSX2 to go 64bit, even with my rig you can't go full speed ;) :p :evil:
ChaosCode
June 26th, 2004, 22:44
Right now Im not to worried about speed as much as saying hey look i can run a ps2 game even is it runs at 1fps
Darkstar
June 27th, 2004, 13:25
What a speed. Muchachachachachachacha!!!
So what is with controller bug???
hbk0259
June 28th, 2004, 09:26
well on some games, the controllers somehow stops working. for example...on pcsx2 0.5 and beta 0.6, WWF Smackdown: Just Bring it! the controls work. i can pass the memcard screen and pass the character select screen. but when the official release version 0.6 came out, the controls somehow stopped working even if the pad hack is turned on or off. in tekken tag tournament NTSC, player one doenst work only player 2 works
hbk0259
June 28th, 2004, 09:43
First Up: SCPH10000.BIN the only BIOS that goes into main menu's etc.
have you tried the USA PS2 bios with GStaris (Public Beta Release)? i tried it and i can get to the main menu of the bios
refraction
June 28th, 2004, 15:10
have you tried the USA PS2 bios with GStaris (Public Beta Release)? i tried it and i can get to the main menu of the bios
thats because GStaris "public beta release" was done by someone who isnt part of pcsx2, nor can he do any of his own code ;)
to cut a long story short GStaris is ****, you want a decent plugin, use GSdx9.
jayteemaster
August 24th, 2004, 16:53
Guys, one of the reasons why the emu does not run at a decent FPS is due to the way that the authors coded the PS2 time counters. Since they are trying only to emulate the hardware for now!
The emulator timing is based on the PS2 but one PS2 instruction takes many PC CPU cycles to execute. The Result: the emulation will always be slow despite the CPU you get.
If you increase the BIAS define (the default is 2) in the source code and recompile the code you will notice a major speed increase. At least this will work for EE side application like some demos. I got 30 fps on a textured rotatin cube (pad controlled) demo - very smooth indeed. All this in a Celeron 450 MHz and a Voodoo crap card. The problem is that for more complex demos and games the emulator hangs very easy ( i think the iop timers implementation need some work)
GiGaHeRz
August 24th, 2004, 17:46
The authors have said many times that they are programming for compatibility, and not for speed. when the emu becomes more compelte, then they will make it faster (and probably we will have better machines).
---
I don't know how the code works, but they probably have that setting for a reason.
Probably because ps2 has a lot of processors running in parallel, and if they are not correctly synchronized, then it won't work.
cooliscool
August 24th, 2004, 21:09
Maybe I should join in? oO
All these tests with recompiler on, aswell as sound via peops, and GSDX9 SSE2 version. There doesn't seem to be a great deal of difference in PCSX2 from an Athlon64 and my Celly (at least until 64bit PCSX2 ;)).
Everything is currently released (I don't beta test), just doing a little bragging. :p
bositman
August 25th, 2004, 12:52
Guys, one of the reasons why the emu does not run at a decent FPS is due to the way that the authors coded the PS2 time counters. Since they are trying only to emulate the hardware for now!
The emulator timing is based on the PS2 but one PS2 instruction takes many PC CPU cycles to execute. The Result: the emulation will always be slow despite the CPU you get.
If you increase the BIAS define (the default is 2) in the source code and recompile the code you will notice a major speed increase. At least this will work for EE side application like some demos. I got 30 fps on a textured rotatin cube (pad controlled) demo - very smooth indeed. All this in a Celeron 450 MHz and a Voodoo crap card. The problem is that for more complex demos and games the emulator hangs very easy ( i think the iop timers implementation need some work)
This setting has been tested in the past by us beta testers and it was found that it reduced compatibility by far,even though its way faster
vBulletin® v3.8.7, Copyright ©2000-2013, vBulletin Solutions, Inc.