Main Restorations Software Audio/Jukebox/MP3 Everything Else Buy/Sell/Trade
Project Announcements Monitor/Video GroovyMAME Merit/JVL Touchscreen Meet Up Retail Vendors
Driving & Racing Woodworking Software Support Forums Consoles Project Arcade Reviews
Automated Projects Artwork Frontend Support Forums Pinball Forum Discussion Old Boards
Raspberry Pi & Dev Board controls.dat Linux Miscellaneous Arcade Wiki Discussion Old Archives
Lightguns Arcade1Up Try the site in https mode Site News

Unread posts | New Replies | Recent posts | Rules | Chatroom | Wiki | File Repository | RSS | Submit news

  

Author Topic: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)  (Read 174771 times)

0 Members and 1 Guest are viewing this topic.

taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« on: October 28, 2007, 05:27:25 am »
I finally got my 4ghz Core 2 Duo built and it 100% stable  >:D, So since I promised a few people here benchmark scores on it here they are.

All benchmarks we preformed with official (I686 optimized) 0.120 MAME release & near complete set of MAME 0.120u1 ROMs.
The follow command line was run from command prompt window inside Windows XP Pro SP2 (32bit):

mamepp.exe -noautoframeskip -frameskip 0 -seconds_to_run 240 -nothrottle -nosleep -video ddraw -skip_gameinfo -effect none -nowaitvsync -noreadconfig -mt [ROM NAME]

Anyway here are the results.

gaelco3d.c
- radikalb           88.21%   52.93fps
- speedup        102.93%   61.76fps
- surfplnt           91.29%   54.77fps

Mediagx.c
- a51site4        185.95%   111.57fps

medvunit.c
- crusnusa        225.23%   128.38fps   
- crusnwld         229.70%   130.93fps
- offroadc            395.36%   225.36fps
- wargods         376.85%   214.80fps

namcos22.c
- airco22b         97.73%   58.64fps     (major graphics problems)
- alpinerd            42.36%   25.42fps   
- cybrcomm       67.79%   40.68fps
- cybrcycc           125.69%   75.41fps   
- propcycl            70.44%   42.50fps     (other than two pauses looks like it's playing at hyperspeed but still has low score  ???)
- raveracw        53.70%   32.22fps
- ridgerac            75.84%   45.50fps      
- timecris            70.84%   42.50fps

Seattle.c   
- biofreak            91.78%   55.07fps   
- blitz            122.34%   73.40fps
- blitz2k            121.94%   73.16fps
- blitz99            120.27%   68.55fps
- calspeed        146.21%   83.34fps
- carnevil            241.11%   137.43fps
- hyprdriv            140.84%   80.28fps
- mace            173.23%   98.74fps
- sfrush            144.95%   82.62fps
- wg3dh            253.19%   151.91fps

Vegas.c
- gauntdl            108.58%     61.89fps (seems slower than gauntleg due to fly overs)
- gauntleg         108.57%   61.88fps               :cheers:
- tenthdeg         62.03%   35.36fps

- xevi3dg           342.14%        205.28fps

model2.c
- Daytona         120.15%   72.06fps     (colours broken)

hornet.c
- gradius4         112.04%   67.91fps

Model3.c
- scud            40.98%   24.59fps     (graphics broken flickers like hell)

namcos21.c   
- starblad            140.06%   84.03fps

The namcos22.c driver looks to need a lot of work still, most of it roms show signs of emulation errors, so I would expect these scores to rise.

MAME performance basically scales 1 for 1 with clock speed, so say if you have a Core 2 Duo @ 3.2ghz your score will be approx 80% of these with the same setting.

If anyone has any other ROMs they would like me to benchmark on my system let me know, I'll do whole drivers if there is reason to. please no requests for pacman or donkey kong, for kicks I benched 1942 the result was 5570.54%,  3342.32fps (test lasted less than 10secs.) ;D

Also if anyone has a complied build of MAME 0.120u1 I686 optimized and MAME 0.120u1 64bit let me know, I would compile them myself but the last time I coded or compiled anything was back when I was writing Quake2 mods some 8 or 9 years ago so I'm a little rusty.  :dizzy:

I plan to benchmark these roms again under Vista 64bit in the near future, but I still have to install Vista and get it dual booting first.
For those that wish to know, the spec of my PC are below:

Intel Core 2 Duo e6850 @ 4ghz, 500fsb, 1.55v (3ghz, 333fsb, 1.35v stock )  8)
Thermalright Ultra 120 Extreme, with 96CFM 120x38mm fan, ducted, and other mods.
Gigabyte GA-X38-DQ6 (crazy cool heat plate in place)
Corsair Dominator Twin2x2048-8500C5D G (2x1GB DDR2-1066 dimms @ 2x500mhz)
XFX Geforce 8800gts 640MB (stock clock, has hit 12000+ in 3DMARK06 when overclocked)
Enermax Liberty 620watt power supply
3x Western digital 36gb Raptors in RAID-0
BLAH
Blah
blah
« Last Edit: October 28, 2007, 07:08:24 am by taz-nz »

D4RKSL4Y3R

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 24
  • Last login:February 22, 2009, 01:12:33 pm
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #1 on: October 28, 2007, 07:56:29 am »
hi i have almost the same computer as you:
Intel Core 2 Duo e6850 @  3ghz, 333fsb, 1.35v   
aDVANCED air cooling
Corsair Dominator 4GB
XFX Geforce 8800ultra 775mb
galaxy 800w power suppky
1x Western digital 36gb Raptor and a 76gb western digital Raptor.
motherboard( asus striker extream)

how the hell did you overclock so well? just good cooling? my air coolong is v good and my motherboards cooling is good as well (+i have a big case with lots of venaltion) so how far do you think i could overclock mine to without mods?

Red

  • Trade Count: (+14)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 464
  • Last login:November 07, 2017, 08:37:51 am
  • I want to build my own arcade controls!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #2 on: October 28, 2007, 12:02:54 pm »
Thanks taz.

"Seattle.c   
- biofreak            91.78%   55.07fps   
- blitz            122.34%   73.40fps
- blitz2k            121.94%   73.16fps
- blitz99            120.27%   68.55fps"

So, Blitz was fairly playable on your machine?

metahacker

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 166
  • Last login:January 17, 2024, 03:09:39 am
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #3 on: October 28, 2007, 03:16:27 pm »
can you try the same tests without -mt ?

what voltages are you running on your cpu, hypertransport, northbridge, southbridge and ram ?

the fact that blitz & gauntlet legends have such nice fps is truly inspiring.
« Last Edit: October 28, 2007, 03:18:31 pm by metahacker »

D4RKSL4Y3R

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 24
  • Last login:February 22, 2009, 01:12:33 pm
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #4 on: October 28, 2007, 03:34:19 pm »
every things on factory presets at the mo- will overclock it in bout six mounths. its just i herd that processor wasnt v good for overclocking (cutom pc magazine..)

taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #5 on: October 28, 2007, 03:46:41 pm »

- blitz            122.34%   73.40fps
- blitz2k            121.94%   73.16fps
- blitz99            120.27%   68.55fps"

So, Blitz was fairly playable on your machine?


Totally, still a little sound stutter now and then but I think that's to do with having vsync on and playing on an LCD monitor.
(I suck at this game totally so I haven't tried to play it much as yet)

The settings I used to benchmark are design to give as close to an ingame score as possible, most MAME benchmarks are CPU preformance tests which are usefull to developemer, but not to those trying to play the game.

taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #6 on: October 28, 2007, 03:48:49 pm »
For all those looking for more details of my overclocking setting, I'll post them up tonight, I'm late for work at the moment.

genesim

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 810
  • Last login:April 12, 2010, 08:18:42 pm
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #7 on: October 28, 2007, 08:54:33 pm »
Killer Instince 2 has sped up tremendously on my AMD 4400 64x2.     Wargods as well.

But the former is all I cared about.    Well...I am jealous about NFL Blitz, but at least I have the Dreamcast for that one.

u_rebelscum

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 3633
  • Last login:April 21, 2010, 03:06:26 pm
  • You rebel scum
    • Mame:Analog+
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #8 on: October 28, 2007, 08:55:58 pm »
Great to see the benchmarks. Thanks! :applaud:

FWIW, according to most mameDev, an PM optimized build is better suited for core 2 CPUs than i686 optimization.  Do you mind testing that?  (Someone said it really helped propcycl, but the site has changed since then.)

Also, as the link above hints, 0.120u1 is supposed to help some of these games go even faster.  (But note some tested slower, too.)
Robin
Knowledge is Power

tommyinajar

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 470
  • Last login:November 15, 2023, 12:23:20 pm
  • My other cab is a Cube Quest
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #9 on: October 28, 2007, 11:44:16 pm »

how the hell did you overclock so well? just good cooling? my air coolong is v good and my motherboards cooling is good as well (+i have a big case with lots of venaltion) so how far do you think i could overclock mine to without mods?

*Most* Duo Cores Overclock well, (some better than others  ;) )
with a decent Mobo.

I've got an E4300 1.8GHZ overclocked to 3.0 GHZ on air with a big heatsink. It's on 24/7 .

E6850 OC'd to 4.0 GHZ isn't a problem with the right equipment.

Here is a little help...

http://www.techarp.com/showarticle.aspx?artno=465

taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #10 on: October 29, 2007, 02:17:47 am »
Jeez... Blitz is playable?  How about the Gauntlet Legends games?


Gauntlet Legends is totally playable too, slight sound stutter now and then most noticably in voice overs, but graphics look purfect and are super smooth, I am so going to waste this week playing it.

Great to see the benchmarks. Thanks! :applaud:

FWIW, according to most mameDev, an PM optimized build is better suited for core 2 CPUs than i686 optimization.  Do you mind testing that?  (Someone said it really helped propcycl, but the site has changed since then.)

Also, as the link above hints, 0.120u1 is supposed to help some of these games go even faster.  (But note some tested slower, too.)

I'm trying to compile 0.120u1 source right now, no go on first try, I'll probably try making it a PM optized version but I got to get it to compile first. I'll post scores if I get it to work.

Killer Instince 2 has sped up tremendously on my AMD 4400 64x2.     Wargods as well.

But the former is all I cared about.    Well...I am jealous about NFL Blitz, but at least I have the Dreamcast for that one.

I'm an arcade game whore, give me some loose change and an arcade machine, no matter how old, ugly and beaten up and I'll be broke in minutes, I love playing anything from old school vertical scrollers to the lastest light gun games. I just had to have a machine that could play Gauntlet Legends, many of the games I love run with easy, some still a lot of work and sum haven't even been dumped as far as I can tell, so having a PC that can run just about anything MAME can throw at is great.





taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #11 on: October 29, 2007, 05:57:30 am »
For all of those that want to know my overclocking details:

CPU Multiplier    8x             
CPU FSB Clock   500Mhz
PCI-E Clock        Auto

Memory Multiplier 2x
Memory timing  5,5,5,15 all others auto (I've got room to go more extreme here, but I'm still tweaking)

DDR2 voltage       +0.35v
PCI-E voltage       +0.05v
FSB voltage          +0.15v
(G)MCH voltage    +0.075

CPU Voltage         1.550v  >:D   (1.520 idle in window , 1.502 underload with Vdrope)

C1E           disabled
TM2           disabled

I wrote a really long and probably boring post about overclocking, but my login timed out and I lost it, so here is the shorter version.

Hardware list for over clocking:

CPU: Core2duo E4x00 (M0 stepping) or E6x50 (G0 stepping)  33 -75% overclock possible

MB: Intel P35 or X38 chipset, nothing with Deluxe or SE in the name, just a good solid board with good cooling and plenty of voltage regulators. (I suggest the Gigabyte GA-P35-DS3 or better)

Memory: DDR2-800 or DD2-1066 match pair from good brand. (I suggest Corsair) Memory speed should be equal or greater in speed to target FSB if possible, like my system DDR2-1066 (533Mhz) target FSB of 500Mhz with 8x Multiplier to get 4Ghz, you could overclock DDR2-800 (400mhz) and run a 445Mhz FSB with a 9x multiplier to get 4ghz but this increases the risk of failure.

Power Supply: If it came in the case dump it, get a good brand name power supply one size too big you your system needs, (I suggest Enermax), cable management is good too.

Case: Plenty of front to back flow, forget anything with a dozen fans aimed in 6 different directions, you want to exchange the air out of the case not move it around, (I suggest Coolermaster Centurion 5 or CM Stacker 831)

Heatsink: Thermalright Ultra 120 Extreme, in the top 3 air coolers around and easy to come by. listed below is what I did to mine.
- Fitted 96 CFM 120x120x36mm fan, slowed to about 50-65 CFM (it was a little noisy), this is mounted a little low of the heatsink to give airflow below the heatsink fins to cool the back of the heatsink block and the north bridge heatsink.
- removed mounting studs from rear X-Brace and fitted them with washers to spread the load, allowed for fitting of heatsink while retailing gigabyte crazy cool heatsink.
- Placed 10c coin between top of heatsink block and mounting X-bracket, increased mounting force and gave a 10deg c improvement in tempature @ 100% load.
- put tape down the sides of the heatsink fins from the fan around to the half way point on the side of the heatsink to act as air duct, forcing air to pass though the whole heatsink. also taped top edge of fan on the top fin to improve air duct effect.
-arctic silver 5 paste, very very thin layer.

Time: Overclock takes time, here is a blow by blow of what I basically did.
Assembly Motherboard, CPU, RAM, Heatsink, Graphics card and PSU on desk, installed TINY XP. Set CPU Multiplier to 6x and memory to 2x, set FSB to 500mhz (CPU is still at 3ghz), tested and set memory timing, increase voltages on memory, north & south bridge, and PCI-E bus. tested for 1 hour at 100% load, increased CPU multiplier to 7x  no extra CPU voltage need yet (CPU now at 3.5ghz) tested for 1 hour 100% load. dropped FSB to 450Mhz and increased CPU multiplier to 8x and CPU voltage to 1.4v (CPU now at  3.8ghz) tested, increased CPU voltage to 1.45v to allow for Vdrope to 1.4v @ 100% load, test for 1 hour @ 100% load, increased FSB to 475Mhz, (CPU now at 3.9ghz) tested, increased CPU voltage to 1.5v to allow for Vdrope to 1.45v at 100% load, tested for 1 hour @ 100% load, increased FSB to 500Mhz with 8x CPU multiplier (CPU now at target 4ghz) post issues, tweaked FSB, & memory voltages cleared up post issues, tested increase CPU voltage to 1.55v to allow for Vdrope in windows @ 100% load to 1.5v, tested for two hours @ 100% load and tweaked little things. Set running Orthos blend test for 100% CPU load and went to bed, 10 hours later, still running 100% load no errors, shut down and spent 2 hours building system. installed fresh copy of WinXP Pro with latest drivers & updates.

Benchmarked MAME to see real world results of all that hard work.

That was still kind of long, and I sure it's full of gramma and spelling mistakes, but hopefully someone will find it interesting.

I've manage to compile a standard version of MAME 0.120 in the mean time, so now on to try a PM optimised version, and then a PM optimised version of MAME 0120u1. (benchmarks to follow in the next couple of days)

so how far do you think i could overclock mine to without mods?

I'd say 3.5-3.6ghz easy, 3.8ghz with some work, and tommyinajar said it best with:

E6850 OC'd to 4.0 GHZ isn't a problem with the right equipment.
« Last Edit: October 29, 2007, 07:51:40 am by taz-nz »

taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #12 on: October 29, 2007, 06:46:57 am »
Great to see the benchmarks. Thanks! :applaud:

FWIW, according to most mameDev, an PM optimized build is better suited for core 2 CPUs than i686 optimization.  Do you mind testing that?  (Someone said it really helped propcycl, but the site has changed since then.)

Also, as the link above hints, 0.120u1 is supposed to help some of these games go even faster.  (But note some tested slower, too.)

I've got the hang of the whole compiling bit, the basics atleast, It's late so full benchmarks will have to wait till tommorrow night, but here's a preview.

PROPCYCL (60hz)
C2D (4.00Ghz)       70.44%,      42.26fps         0120       i686 optimized.
C2D (4.00Ghz)       74.25%,      44.55fps         0120         PM optimized
C2D (4.00GHz)      100.39%,     60.23fps        0120u1     PM optimized.   :cheers: (looks like propcycle just made the playable list with minor sound stutter issues.)


Tiger-Heli

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 5447
  • Last login:January 03, 2018, 02:19:23 pm
  • Ron Howard? . . . er, I mean . . . Run, Coward!!!
    • Tiger-Heli
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #13 on: October 29, 2007, 10:06:12 am »
Just out of curiosity, how does Star Wars Trilogy do.

Great machine, btw!!!
It's not what you take when you leave this world behind you, it's what you leave behind you when you go. - R. Travis.
When all is said and done, generally much more is SAID than DONE.

FrizzleFried

  • no one listens to me anyway.
  • Trade Count: (+4)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 5144
  • Last login:May 30, 2023, 01:14:24 pm
    • Idaho Garagecade
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #14 on: October 29, 2007, 10:12:32 am »
The problem here is that I am trying to save up enough money for a CORE2DUO setup...and when I do,  that computer is going RIGHT HERE ON MY DESK...

...NOT in my MAME cabinet...it it's going to take at least another 2 generations before a CORE2DUO makes it in to my MAME cabinet.   I refuse to spend that kind of dough to play games from the 90's (only)...

When i do upgrade,  this Opteron 165 I have on my desk will go in the Horizontal MAME cabinet,  the 3200+ that is in the horizontal will go to the vertical...

>sigh<

So it will be MANY MOONS before I can play Blitz!

Visit my arcade blog at: www.idahogaragecade.com (Updated 10-28-21)

Tiger-Heli

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 5447
  • Last login:January 03, 2018, 02:19:23 pm
  • Ron Howard? . . . er, I mean . . . Run, Coward!!!
    • Tiger-Heli
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #15 on: October 29, 2007, 10:18:50 am »
The problem here is that I am trying to save up enough money for a CORE2DUO setup...and when I do,  that computer is going RIGHT HERE ON MY DESK...
Wonder how hard it would be and what kind of performance you would get if you networked the cabinet and desktop PC and ran the arcade as a client and ran Blitz off MAME on the desktop as a server?
It's not what you take when you leave this world behind you, it's what you leave behind you when you go. - R. Travis.
When all is said and done, generally much more is SAID than DONE.

FrizzleFried

  • no one listens to me anyway.
  • Trade Count: (+4)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 5144
  • Last login:May 30, 2023, 01:14:24 pm
    • Idaho Garagecade
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #16 on: October 29, 2007, 11:06:59 am »
The problem here is that I am trying to save up enough money for a CORE2DUO setup...and when I do,  that computer is going RIGHT HERE ON MY DESK...
Wonder how hard it would be and what kind of performance you would get if you networked the cabinet and desktop PC and ran the arcade as a client and ran Blitz off MAME on the desktop as a server?


Uh?!?  There is a difference between file STORAGE (what you propose) and CPU PROCESSING POWER which is what your computer does on your end when playing games, etc.  I'd have to essentially run a REALLY LONG VGA cord and USB cord from the computer in the office to the gameroom in the garage!
Visit my arcade blog at: www.idahogaragecade.com (Updated 10-28-21)

Tiger-Heli

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 5447
  • Last login:January 03, 2018, 02:19:23 pm
  • Ron Howard? . . . er, I mean . . . Run, Coward!!!
    • Tiger-Heli
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #17 on: October 29, 2007, 11:53:58 am »
Uh?!?  There is a difference between file STORAGE (what you propose) and CPU PROCESSING POWER which is what your computer does on your end when playing games, etc.  I'd have to essentially run a REALLY LONG VGA cord and USB cord from the computer in the office to the gameroom in the garage!
Well, I'm not a network guru, but wouldn't you be able to use the CPU Processing power of the desktop Core2Duo for running MAME, and just use the networked slower box in the arcade for displaying graphics and feeding inputs to the main CPU.

You could use a wireless router and wireless card if they are really far apart and you don't want to string CAT 5 through the house.
It's not what you take when you leave this world behind you, it's what you leave behind you when you go. - R. Travis.
When all is said and done, generally much more is SAID than DONE.

rdowdy95

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 609
  • Last login:July 29, 2011, 11:48:50 am
  • This is my rifle.
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #18 on: October 29, 2007, 01:18:06 pm »
This is really great news.  I will have to upgrade my computer system in 2-3 years and get some of those nice games going like Gauntlet Dark Legacy.  It is just a pain to upgrade roms and config files.  Good thing I backed everything up to a portable hard drive and I also wrote down all my instructions as well.

Mainly setting up that MameWah was a pure ---smurfette--- to do, but it does run good now with all my settings configed.

FrizzleFried

  • no one listens to me anyway.
  • Trade Count: (+4)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 5144
  • Last login:May 30, 2023, 01:14:24 pm
    • Idaho Garagecade
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #19 on: October 29, 2007, 01:27:20 pm »
Uh?!?  There is a difference between file STORAGE (what you propose) and CPU PROCESSING POWER which is what your computer does on your end when playing games, etc.  I'd have to essentially run a REALLY LONG VGA cord and USB cord from the computer in the office to the gameroom in the garage!
Well, I'm not a network guru, but wouldn't you be able to use the CPU Processing power of the desktop Core2Duo for running MAME, and just use the networked slower box in the arcade for displaying graphics and feeding inputs to the main CPU.

You could use a wireless router and wireless card if they are really far apart and you don't want to string CAT 5 through the house.

It doesn't work that way....

Visit my arcade blog at: www.idahogaragecade.com (Updated 10-28-21)

Tiger-Heli

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 5447
  • Last login:January 03, 2018, 02:19:23 pm
  • Ron Howard? . . . er, I mean . . . Run, Coward!!!
    • Tiger-Heli
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #20 on: October 29, 2007, 01:37:44 pm »
Okay, sorry!
It's not what you take when you leave this world behind you, it's what you leave behind you when you go. - R. Travis.
When all is said and done, generally much more is SAID than DONE.

divemaster127

  • Trade Count: (+60)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 1977
  • Last login:December 02, 2018, 08:05:08 pm
  • My webstore is arcadeemulator.net
    • arcadeemulator.net
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #21 on: October 29, 2007, 02:29:18 pm »
I was going to wait until next week, to post my results for the new mame machine, my main desktop is a q6600 quad core, its 200% faster than my 6600 dual core I was running, so I ordered the same machine for my mame
q6600 quad core
4 gig of ram
gigabyte motherboard
I will have the parts in on wednesday, as soon as I'm built I will list the fps on the higher end chd games
dm
I carry both ultimarc & happ items, all brand new & I ship from the united states. My online store is ARCADEEMULATOR.NET, pm if I can help in anyway.

taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #22 on: October 29, 2007, 03:39:53 pm »
I was going to wait until next week, to post my results for the new mame machine, my main desktop is a q6600 quad core, its 200% faster than my 6600 dual core I was running, so I ordered the same machine for my mame
q6600 quad core
4 gig of ram
gigabyte motherboard
I will have the parts in on wednesday, as soon as I'm built I will list the fps on the higher end chd games
dm

I'll be really interested to see how the roms that ran on 3DFX hardware do on your system, it will be interesting to see how the new MAME multithreading scales on a quad core.

« Last Edit: October 29, 2007, 03:43:19 pm by taz-nz »

metahacker

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 166
  • Last login:January 17, 2024, 03:09:39 am
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #23 on: October 29, 2007, 04:58:34 pm »
isnt mame with -mt just 2 threads ?  with the second thread just doing framebuffer stuff.

im not sure what else would be left for the other cpus to do other than 'run' disk i/o on CHDs or eat up cycles in AC97 ?  i would think the mame helper thread would not busy up the 2nd core enough to need a 3rd or 4th. 

but i really hope you do see some improvement.  it's nice to see new hardware tech making emus catch up and making stuff that was previously unusable now useful.
 

btw 500 fsb is ---smurfing--- hot.  nice work on your set up dood im impressed.

im runnin a e6600 @ 3.4 here on my desktop.  abit in9 32x max (i dont recommend it, unless u just want the cool BIOS-driven LEDs mounted under the mobo stock :))  had to crank ~1.7v into it tho, for whatever reason.  scythe infinity 120.  1 total fan in the system besides psu.  no nb/sb mods besides as5, no fans.

c2ds are mean.  reminds me of the old celeron 300A, but even better.

i hope i can get something comparable to your results with my new 2.66 6750. i dont think i'll be hittin any 4GHz tho. 

it sure is inspiring to see your results, though....

and apparently 120u1 pentium-M optimized is where the party's at....



divemaster127

  • Trade Count: (+60)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 1977
  • Last login:December 02, 2018, 08:05:08 pm
  • My webstore is arcadeemulator.net
    • arcadeemulator.net
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #24 on: October 29, 2007, 05:10:33 pm »
Good or bad, I have used pctuneup for my benchmarks up until a few months ago
I have a 6600 intel dual core...
score on pctuneup for 6600 intel 3400 w/vista
I also built a 6750 for my racer score 3650 w/vista
2 months I rebuilt my 6600 to a q6600 quad core score 7300 w/vista
I am going to run a strip down xp w/over clock i'm hoping to close to 8000 w/my quad core xp, since xp is faster than vista, its worth a try
dm
I carry both ultimarc & happ items, all brand new & I ship from the united states. My online store is ARCADEEMULATOR.NET, pm if I can help in anyway.

metahacker

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 166
  • Last login:January 17, 2024, 03:09:39 am
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #25 on: October 29, 2007, 05:50:07 pm »
hmmm, now i wonder how the 64 bit mame would compare to 32 bit

taz you are our benchmarking hero :).

u_rebelscum

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 3633
  • Last login:April 21, 2010, 03:06:26 pm
  • You rebel scum
    • Mame:Analog+
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #26 on: October 29, 2007, 07:43:49 pm »
isnt mame with -mt just 2 threads ?  with the second thread just doing framebuffer stuff.

Old news.  Aaron found a way to split some parts of the emulation into variable number of threads back in 0.119u3: :notworthy:

Code: [Select]
Added support for controlling multithreading behavior through an
environment variable OSDPROCESSORS. To override the default behavior,
set OSDPROCESSORS equal to the number of logical processors you wish
the OSD layer to pretend you have. [Aaron Giles]

And 0.120u1 improves it even more.  If you have a quad code set OSDPROCESSORS=4.  It won't be four times as fast, but I hear it will be faster for most slow games.
Robin
Knowledge is Power

taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #27 on: October 30, 2007, 05:55:35 am »
Well here they are M.A.M.E. 0.120u1 PM optimised benchmark scores:

Scores with 0.120 in front were benchmark with official I686 optimised MAME 0.120
Scores with 0.120u1 in front were benchmark with my build of PM optimised MAME 0.120u1

I use the same command line as I did for the 0.120 benchmarks but using "set OSDPROCESSORS=2"  command before tests (not even sure if I using it right but it seems to work  :dunno)

Anyway here are the results:

- 1492      
0.120      5570.54%   3342.32fps


gaelco3d.c
- radikalb   
0.120      88.21%      52.93fps          
0.120u1      117.74%      70.64fps

- speedup   
0.120      102.93%      61.76fps   
0.120u1      127.66%      76.60fps

- surfplnt   
0.120      91.29%      54.77fps   
0.120u1      112.79%      67.67fps


Mediagx.c
- a51site4   
0.120      185.95%      111.57fps   
0.120u1      186.04%      111.62fps


medvunit.c
- crusnusa   
0.120      225.23%      128.38fps   
0.120u1      220.16%      125.49fps

- crusnwld   
0.120      229.70%      130.93fps   
0.120u1      221.42%      126.21fps

- offroadc   
0.120      395.36%      225.36fps   
0.120u1      365.57%      208.38fps   

- wargods   
0.120      376.85%      214.80fps   
0.120u1      330.00%      188.10fps


namcos22.c
- airco22b
0.120      97.73%      58.64fps    
0.120u1      110.64%     66.38fps

- alpinerd   
0.120      42.36%      25.42fps   
0.120u1      65.95%      39.57fps

- cybrcomm   
0.120      67.79%      40.68fps   
0.120u1      95.78%      57.49fps

- cybrcycc   
0.120      125.69%      75.41fps   
0.120u1      157.40%      94.44fps

- propcycl   
0.120      70.44%      42.50fps   
0.120u1      100.39%      60.23fps

- raveracw   
0.120      53.70%      32.22fps   
0.120u1      81.37%      48.82fps

- ridgerac   
0.120      75.84%      45.50fps   
0.120u1      108.13%      64.88fps

- timecris   
0.120      70.84%      42.50fps   
0.120u1      101.81%      61.09fps


Seattle.c   
- biofreak
0.120      91.78%      55.07fps
0.120u1      101.70%      57.97fps

- blitz      
0.120      122.34%      73.40fps   
0.120u1      127.38%      72.61fps

- blitz2k   
0.120      121.94%      69.51fps   
0.120u1      123.73%      70.52fps

- blitz99   
0.120      120.27%      68.55fps   
0.120u1      120.26%      68.55fps

- calspeed   
0.120      146.21%      83.34fps   
0.120u1      165.08%      94.10fps

- carnevil
0.120      241.11%      137.43fps   
0.120u1      221.69%      126.36fps

- hyprdriv   
0.120      140.84%      80.28fps   
0.120u1      144.27%      82.23fps

- mace      
0.120      173.23%      98.74fps   
0.120u1      186.67%      106.40fps

- sfrush   
0.120      144.95%      82.62fps   
0.120u1      152.01%      86.64fps

- wg3dh      
0.120      253.19%      144.32fps   
0.120u1      265.32%      151.23fps


Vegas.c
- gauntdl
0.120      108.58%     61.89fps    
0.120u1      110.06%      62.73fps

-gauntleg
0.120      108.57%      61.88fps
0.120u1      111.60%      63.61fps

- tenthdeg   
0.120u1      62.03%      35.36fps   
0.120u1      67.42%      38.43fps


model2.c
- Daytona         
0.120      120.15%      72.06fps   
0.120u1      120.08%      72.05fps


hornet.c
- gradius4   
0.120      112.04%      67.91fps
0.120u1      fails instantly with memory error


Model3.c
- scud      
0.120      40.98%      24.59fps   
0.120u1      44.21%      26.53fps


namcos21.c   
- starblad
0.120      140.06%      84.03fps
0.120u1      139.95%      83.97fps

Just out of curiosity, how does Star Wars Trilogy do.

-swtrilgy (Emulation need a lot of work still major graphics issues)
0.120u1      61.66%      39.97fps


hmmm, now i wonder how the 64 bit mame would compare to 32 bit

taz you are our benchmarking hero :).

Thanks, 64bit MAME benchmarks to follow in a couple of days. Hoping to find time tomorrow night to install Vista Ultimate 64bit in dual boot with XP Pro, if that all goes well then I just have to work out how to compile 64bit versions of MAME. Once I've got those two things sorted I'll run all the benchmarks again in 64bit and any new ROMs people request.

If anyone has any other hints or trick to improving MAME preformance let me know.


« Last Edit: October 30, 2007, 07:47:18 am by taz-nz »

taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #28 on: October 30, 2007, 05:59:34 am »
isnt mame with -mt just 2 threads ?  with the second thread just doing framebuffer stuff.

You are correct but it's still good for about 5% improvement in score so I use it.

im not sure what else would be left for the other cpus to do other than 'run' disk i/o on CHDs or eat up cycles in AC97 ?  i would think the mame helper thread would not busy up the 2nd core enough to need a 3rd or 4th. 


the 2nd, 3rd and 4th core of a quad core are going to be doing nothing but play hot patato with windows services in the background with 90% of mame roms, but with the few that are Multithreaded now, the 2nd Core get a real work out and in theory the 3rd and 4th will get in on the some of the action too.

btw 500 fsb is ---smurfing--- hot.  nice work on your set up dood im impressed.

im runnin a e6600 @ 3.4 here on my desktop.  abit in9 32x max (i dont recommend it, unless u just want the cool BIOS-driven LEDs mounted under the mobo stock :))  had to crank ~1.7v into it tho, for whatever reason.  scythe infinity 120.  1 total fan in the system besides psu.  no nb/sb mods besides as5, no fans.

c2ds are mean.  reminds me of the old celeron 300A, but even better.

i hope i can get something comparable to your results with my new 2.66 6750. i dont think i'll be hittin any 4GHz tho. 

it sure is inspiring to see your results, though....

and apparently 120u1 pentium-M optimized is where the party's at....

Sounds like your not do to bad yourself on the overclocking front, but 1.7v ouch, bet that makes the core temp ramp up like hell under load.

My mate still has a Celeron 300A overclocked @ 450mhz running as a gateway server on his home network, been overclocked for so many years it doesn't run stock clock anymore.

E6750 are good to about 3.6-3.8ghz with air, you really need a good water cooling kit to take them any higher, but there are some that get lucky and make it all the way to 4.0ghz on air alone.


Tiger-Heli

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 5447
  • Last login:January 03, 2018, 02:19:23 pm
  • Ron Howard? . . . er, I mean . . . Run, Coward!!!
    • Tiger-Heli
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #29 on: October 30, 2007, 07:37:27 am »
-swtrilgy (Emulation need a lot of work still major graphics issues)
0.120u1      61.66%      39.97fps
Thanks - that's actually better than I expected!!!
It's not what you take when you leave this world behind you, it's what you leave behind you when you go. - R. Travis.
When all is said and done, generally much more is SAID than DONE.

headkaze

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 2943
  • Last login:August 14, 2023, 02:00:48 am
  • 0x2b|~0x2b?
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #30 on: October 30, 2007, 10:12:51 am »
Compiling Mame is quite easy. Check out Mame Compiler for a front end and installer for all the tools necessary to compile. It has options for compiling 64-bit versions etc.
« Last Edit: October 30, 2007, 10:15:17 am by headkaze »

u_rebelscum

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 3633
  • Last login:April 21, 2010, 03:06:26 pm
  • You rebel scum
    • Mame:Analog+
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #31 on: October 30, 2007, 01:03:34 pm »
... and in theory the 3rd and 4th will get in on the some of the action too.

set OSDPROCESSORS=4
for the 3rd & 4th get the same as the 2nd.
Robin
Knowledge is Power

divemaster127

  • Trade Count: (+60)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 1977
  • Last login:December 02, 2018, 08:05:08 pm
  • My webstore is arcadeemulator.net
    • arcadeemulator.net
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #32 on: October 30, 2007, 01:05:38 pm »
right now im running 32 bit home xp, would it be worth the jump to 64 bit xp before I rebuild to the quad core
dm
I carry both ultimarc & happ items, all brand new & I ship from the united states. My online store is ARCADEEMULATOR.NET, pm if I can help in anyway.

John IV [MameUI64]

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 83
  • Last login:August 29, 2023, 02:12:58 am
    • MameUI
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #33 on: October 30, 2007, 11:27:28 pm »
Hi there, if you have a chance please run the pure CPU test bench on the standard build using the benchmarking params here:
http://mame32qa.classicgaming.gamespy.com/Bench.htm

Aaron and I are curious about that and a similar run on a 64bit OS + 64bit Mame.  Regards -

Game Driver Arguments 0.119 .120 .120.1
crusnusa midvunit.c -str 90 -video none -nosound -nothrottle -norc -nosleep 212.24% 212.12% 178.27%
starblad namcos21.c -str 90 -video none -nosound -nothrottle -norc -nosleep 234.33% 120.83% 116.97%
gauntleg vegas.c -str 90 -video none -nosound -nothrottle -norc -nosleep 47.92% 114.97% 132.38%
blitz seattle.c -str 90 -video none -nosound -nothrottle -norc -nosleep 122.25% 87.86% 111.59%
daytona model2.c -str 90 -video none -nosound -nothrottle -norc -nosleep n/a 86.18% 87.22%
gradius4 hornet.c -str 90 -video none -nosound -nothrottle -norc -nosleep 60.36% 79.34% n/a
radikalb gaelco3d.c -str 90 -video none -nosound -nothrottle -norc -nosleep 77.46% 76.11% 95.72%
ridgerac namcos22.c -str 90 -video none -nosound -nothrottle -norc -nosleep 65.56% 64.73% 90.77%
scud model3.c -str 90 -video none -nosound -nothrottle -norc -nosleep 34.41% 32.91% 33.00%

Vista Ultimate 32 • Intel C2D E6400 3.4Ghz • 2GB Corsair DDR2 800 • ATI X1950 Pro     
Standard baseline build w/ GCC tools

taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #34 on: October 31, 2007, 06:10:57 am »
Hi there, if you have a chance please run the pure CPU test bench on the standard build using the benchmarking params here:
http://mame32qa.classicgaming.gamespy.com/Bench.htm

Aaron and I are curious about that and a similar run on a 64bit OS + 64bit Mame.  Regards -

Game Driver Arguments 0.119 .120 .120.1
crusnusa midvunit.c -str 90 -video none -nosound -nothrottle -norc -nosleep 212.24% 212.12% 178.27%
starblad namcos21.c -str 90 -video none -nosound -nothrottle -norc -nosleep 234.33% 120.83% 116.97%
gauntleg vegas.c -str 90 -video none -nosound -nothrottle -norc -nosleep 47.92% 114.97% 132.38%
blitz seattle.c -str 90 -video none -nosound -nothrottle -norc -nosleep 122.25% 87.86% 111.59%
daytona model2.c -str 90 -video none -nosound -nothrottle -norc -nosleep n/a 86.18% 87.22%
gradius4 hornet.c -str 90 -video none -nosound -nothrottle -norc -nosleep 60.36% 79.34% n/a
radikalb gaelco3d.c -str 90 -video none -nosound -nothrottle -norc -nosleep 77.46% 76.11% 95.72%
ridgerac namcos22.c -str 90 -video none -nosound -nothrottle -norc -nosleep 65.56% 64.73% 90.77%
scud model3.c -str 90 -video none -nosound -nothrottle -norc -nosleep 34.41% 32.91% 33.00%

Vista Ultimate 32 • Intel C2D E6400 3.4Ghz • 2GB Corsair DDR2 800 • ATI X1950 Pro     
Standard baseline build w/ GCC tools

Ha no problem, glad to be of some use.

All Benchmark run with Offical MAME 0.120 standard 64bit build on clean install of Vista Ultamite 64bit.

v64mame.exe -str 90 -video none -nosound -nothrottle -norc -nosleep crusnusa      271.81%
v64mame.exe -str 90 -video none -nosound -nothrottle -norc -nosleep starblad      153.87%
v64mame.exe -str 90 -video none -nosound -nothrottle -norc -nosleep gauntleg      183.42%
v64mame.exe -str 90 -video none -nosound -nothrottle -norc -nosleep blitz      164.15%
v64mame.exe -str 90 -video none -nosound -nothrottle -norc -nosleep daytona      142.72%
v64mame.exe -str 90 -video none -nosound -nothrottle -norc -nosleep gradius4      122.16%   
v64mame.exe -str 90 -video none -nosound -nothrottle -norc -nosleep radikalb      150.07%   
v64mame.exe -str 90 -video none -nosound -nothrottle -norc -nosleep ridgerac      106.77%
v64mame.exe -str 90 -video none -nosound -nothrottle -norc -nosleep scud      33.81%     :banghead:

I would have benchmarked 0.119 & 120u1 64bit builds, but I've had no joy trying to compile a 64bit build, If you can to send me a link of where I can get a 64bit build of 0.119 & 0.120u1 I'll benchmark them for you.



taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #35 on: October 31, 2007, 06:17:21 am »
Compiling Mame is quite easy. Check out Mame Compiler for a front end and installer for all the tools necessary to compile. It has options for compiling 64-bit versions etc.

Thanks for the link, great app by the way. Works great but as soon as I tick the 64bit option I get the error message below.


Deleting Object Folder...
Compiling Mame...
Using Parameters  PTR64=1
Compiling src/osd/windows/winwork.c...
cc1.exe: warnings being treated as errors
src/osd/windows/winwork.c: In function 'compare_exchange_pointer':
src/osd/windows/winwork.c:77: warning: implicit declaration of function 'InterlockedCompareExchange64'
src/osd/windows/winwork.c:77: warning: cast from pointer to integer of different size
src/osd/windows/winwork.c:77: warning: cast from pointer to integer of different size
src/osd/windows/winwork.c:77: warning: cast from pointer to integer of different size
src/osd/windows/winwork.c:78: warning: cast to pointer from integer of different size
mingw32-make: *** [obj/windows/mame/osd/windows/winwork.o] Error 1
Finished!
0 Hours 0 Minutes and 0 Seconds Elapsed.

If you've got any idea what I might be doing wrong let me know.


If anyone wants to compile a PM optimized 64bit version of MAME 0.120u1 for me & post it somewhere I can download it,  that would be great.

I did a quick test of the preformance of the standard 64bit build on MAME 0.120 under Vista 64bit and it makes a big difference, the scores not as good as the PM optimised 0.120u1 build but a lot better that the PM optimised 0.120 build, so I think there will be some killer scores from a full optimised 64 build of 0.120u1

have a look at what I mean:

C2D (4.00Ghz)   70.44%,      42.26fps   0.120 i686 optimized.
C2D (4.00Ghz)   74.25%,      44.55fps   0.120 PM optimized
CDC (4.00GHz)  100.39%,          60.23fps   0.120u1 Pm Optimized.
CDC (4.00GHz)   96.24%,      57.74fps   0.120 official 64bit build.(standard build)

« Last Edit: October 31, 2007, 03:12:24 pm by taz-nz »

John IV [MameUI64]

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 83
  • Last login:August 29, 2023, 02:12:58 am
    • MameUI
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #36 on: October 31, 2007, 05:13:19 pm »
Thanks for that 64bit run, are you fully 64bit Vista now or can you dual boot to 32bit and run an apples to apples w/ my numbers above.  You can also use my official Mame32 .120u1 as the .exe and just redirect output. Thx -

headkaze

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 2943
  • Last login:August 14, 2023, 02:00:48 am
  • 0x2b|~0x2b?
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #37 on: October 31, 2007, 10:36:13 pm »
Deleting Object Folder...
Compiling Mame...
Using Parameters  PTR64=1
Compiling src/osd/windows/winwork.c...
cc1.exe: warnings being treated as errors
src/osd/windows/winwork.c: In function 'compare_exchange_pointer':
src/osd/windows/winwork.c:77: warning: implicit declaration of function 'InterlockedCompareExchange64'
src/osd/windows/winwork.c:77: warning: cast from pointer to integer of different size
src/osd/windows/winwork.c:77: warning: cast from pointer to integer of different size
src/osd/windows/winwork.c:77: warning: cast from pointer to integer of different size
src/osd/windows/winwork.c:78: warning: cast to pointer from integer of different size
mingw32-make: *** [obj/windows/mame/osd/windows/winwork.o] Error 1
Finished!
0 Hours 0 Minutes and 0 Seconds Elapsed.

If you've got any idea what I might be doing wrong let me know.

Oh crap I never actually tested that option. After a bit of investigation I think you actually need the Windows Platform SDK and a recent DirectX SDK to compile 64-bit versions of Mame. Aparently the Windows SDK includes a VS 64-bit compiler. I will have to do another update of Mame Compiler to support this, so when I get some time I'll look into it. If MinGW can support compiling 64-bit versions of Mame, please let me know what is required to get it compiling. Thanks.

taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #38 on: November 01, 2007, 03:05:02 am »
Thanks for that 64bit run, are you fully 64bit Vista now or can you dual boot to 32bit and run an apples to apples w/ my numbers above.  You can also use my official Mame32 .120u1 as the .exe and just redirect output. Thx -

No problem glab to be of help, I'm running dual boot and will do for some time to come.

Ok apples for apples CPU bench test. (All 32bit benchmarks run in Windows XP Pro , 64bit benchcmarks run in Vista ultimate 64bit)

I downloaded the standard build of Mame 0.119 and re-named the mame.exe to mame119.exe
I downloaded the standard build of Mame 0.120 and re-named the mame.exe to mame120.exe
I downloaded the souce code fore Mame 0.120 and the 0.120u1 difffile applied the diff and compiled a standard build with mingw.
I then re-name the resulting mame.exe to mame120u1.exe (Sorry couldn't for the life of me get Mame32 0.120u1 to output to a txt file)

Then I created a batch file called "bench.bat" to make like easy:

>>>> start bench.bat <<<<
mame119.exe -str 90 -video none -nosound -nothrottle -norc -nosleep %1

mame120.exe -str 90 -video none -nosound -nothrottle -norc -nosleep %1

mame120u1.exe -str 90 -video none -nosound -nothrottle -norc -nosleep %1
>>>> end bench.bat <<<<

And call that by typing "bench [ROM NAME]" for the command prompt window.

>>>>>> UPDATED: Now including 0.120u2 results <<<<<<<
mame120u2.exe -str 90 -video none -nosound -nothrottle -norc -nosleep [ROM NAME]

Anyway here are the results:

CPU
ROM
Driver
      0.119   
     0.120   
     0120u1 
     0120u2 
   0.120(64bit)
E6400 @ 3.4ghz    crusnusa   midvunit.c
212.24%
212.12%
178.27%
   n/a
  n/a
E6850 @ 4.0ghzcrusnusa   midvuint.c
257.69%
257.32%
218.20%
214.79%
271.81%   
E6400 @ 3.4ghz    starblad   namcos21.c
234.33%
120.83%
116.97%
   n/a
  n/a
E6850 @ 4.0ghzstarblad   namcos21.c
299.30%
144.10%
144.00%
142.10%
153.57%   
E6400 @ 3.4ghz    gauntleg   vegas.c
47.92%
114.97%
132.38%
   n/a
  n/a
E6850 @ 4.0ghzgauntleg   vegas.c
68.91%
166.17%
173.07%
166.03%
183.42%   
E6400 @ 3.4ghz    blitz   seattle.c
122.25%
87.86%
111.59%
   n/a
  n/a
E6850 @ 4.0ghzblitz   seattle.c
153.43%
138.59%
143.28%
136.99%
164.15%   
E6400 @ 3.4ghz    daytona   model2.c
   n/a
86.18%
87.22%
   n/a
  n/a
E6850 @ 4.0ghzdaytona   model2.c
   n/a
106.17%
106.43%
104.94%
142.72%   
E6400 @ 3.4ghz    gradius4   hornet.c
60.36%
79.34%
   n/a
   n/a
  n/a
E6850 @ 4.0ghzgradius4   hornet.c
116.72%
164.94%
   n/a
180.32%
122.16%   
E6400 @ 3.4ghz    radikalb   gaelco3d.c
77.46%
76.11%
95.72%
   n/a
  n/a
E6850 @ 4.0ghzradikalb   gaelco3d.c
94.25%
93.97%
116.77%
129.44%
150.07%   
E6400 @ 3.4ghz    ridgerac   namcos22.c
65.56%
64.73%
90.77%
   n/a
  n/a
E6850 @ 4.0ghzridgerac   namcos22.c
80.00%
79.54%
114.89%
112.44%
106.77%   
E6400 @ 3.4ghz    scud   model3.c
34.41%
32.91%
33.00%
   n/a
  n/a
E6850 @ 4.0ghzscud   model3.c
41.21%
39.60%
39.56%
38.93%
33.81%   


If you have any other benchmarks you would like me to run let me know.

Later I'll post In-Game benchmark result for the standard build of 64bit mame in Vista 64bit.

I'm also playing around with the Microsoft SDK & Visual studio express to see if I can build a PM optimised 64bit build or Mame 0120u1.

« Last Edit: November 02, 2007, 05:48:05 am by taz-nz »

taz-nz

  • Trade Count: (0)
  • Full Member
  • ***
  • Offline Offline
  • Posts: 147
  • Last login:June 12, 2019, 10:12:13 am
  • If all else fails burn the manual!
Re: 4Ghz Core 2 Duo vs M.A.M.E. 0.120 (benchmark results)
« Reply #39 on: November 01, 2007, 07:09:17 am »
Well after running all the benchmarks again in Vista 64 with the standard build or 64bit MAME I have to say, 64bit is where it's at. If your serious about running MAME you should be running a 64bit OS. Check out the score for yourself you'll see what I mean.

Scores with 0.120 in front were benchmarked with official I686 optimised MAME 0.120
Scores with 0.120u1 in front were benchmarked with my build of PM optimised MAME 0.120u1
Socres with 0.20 64bit in front were benchmarked with official 64bit MAME 0.120 build.

- 1492      
0.120      5570.54%   3342.32fps
0.120 64bit   6503.71%   3902.23fps

gaelco3d.c
- radikalb   
0.120      88.21%      52.93fps          
0.120 64bit   134.23%      
0.120u1      117.74%      70.64fps

- speedup   
0.120      102.93%      61.76fps   
0.120 64bit   157.14%      
0.120u1      127.66%      76.60fps

- surfplnt   
0.120      91.29%      54.77fps   
0.120u1 64bit   139.72%
0.120u1      112.79%      67.67fps


Mediagx.c
- a51site4   
0.120      185.95%      111.57fps   
0.120 64bit   234.34%      
0.120u1      186.04%      111.62fps


medvunit.c
- crusnusa   
0.120      225.23%      128.38fps   
0.120 64bit   271.99%
0.120u1      220.16%      125.49fps

- crusnwld   
0.120      229.70%      130.93fps   
0.120 64bit   249.66%
0.120u1      221.42%      126.21fps

- offroadc   
0.120      395.36%      225.36fps   
0.120 64bit   425.80%
0.120u1      365.57%      208.38fps   

- wargods   
0.120      376.85%      214.80fps   
0.120 64bit   411.71%
0.120u1      330.00%      188.10fps


namcos22.c
- airco22b
0.120      97.73%      58.64fps    
0.120 64bit   124.71%
0.120u1      110.64%     66.38fps

- alpinerd     (I wish I knew why this rom is so slow, I getting sick of looking at the skiers pink arse.
0.120      42.36%      25.42fps   
0.120 64bit   53.37%
0.120u1      65.95%      39.57fps

- cybrcomm   
0.120      67.79%      40.68fps   
0.120 64bit   96.65%
0.120u1      95.78%      57.49fps

- cybrcycc   
0.120      125.69%      75.41fps   
0.120 64bit   161.28%
0.120u1      157.40%      94.44fps

- propcycl   
0.120      70.44%      42.50fps   
0.120 64bit   96.31%
0.120u1      100.39%      60.23fps

- raveracw   
0.120      53.70%      32.22fps   
0.120 64bit   68.33%
0.120u1      81.37%      48.82fps

- ridgerac   
0.120      75.84%      45.50fps   
0.120 64bit   103.33%      
0.120u1      108.13%      64.88fps

- timecris   
0.120      70.84%      42.50fps   
0.120 64bit   93.17%
0.120u1      101.81%      61.09fps


Seattle.c   
- biofreak
0.120      91.78%      55.07fps
0.120 64bit   140.22%      
0.120u1      101.70%      57.97fps

- blitz      
0.120      122.34%      73.40fps   
0.120 64bit   153.09%
0.120u1      127.38%      72.61fps

- blitz2k   
0.120      121.94%      69.51fps   
0.120 64bit   146.08%
0.120u1      123.73%      70.52fps

- blitz99   
0.120      120.27%      68.55fps   
0.120 64bit   142.65%
0.120u1      120.26%      68.55fps

- calspeed   
0.120      146.21%      83.34fps   
0.120 64bit   181.91%
0.120u1      165.08%      94.10fps

- carnevil
0.120      241.11%      137.43fps   
0.120 64bit   287.40%
0.120u1      221.69%      126.36fps

- hyprdriv   
0.120      140.84%      80.28fps   
0.120 64bit   185.50%
0.120u1      144.27%      82.23fps

- mace      
0.120      173.23%      98.74fps   
0.120 64bit   217.20%
0.120u1      186.67%      106.40fps

- sfrush   
0.120      144.95%      82.62fps   
0.120 64bit   199.68%
0.120u1      152.01%      86.64fps

- wg3dh      
0.120      253.19%      144.32fps   
0.120 64bit   342.70%
0.120u1      265.32%      151.23fps


Vegas.c
- gauntdl
0.120      108.58%     61.89fps    
0.120 64bit   123.46%
0.120u1      110.06%      62.73fps

-gauntleg
0.120      108.57%      61.88fps
0.120 64bit   153.76%            (tends to lockup)
0.120u1      111.60%      63.61fps

- tenthdeg   
0.120u1      62.03%      35.36fps   
0.120 64bit   76.21%
0.120u1      67.42%      38.43fps


model2.c
- Daytona         
0.120      120.15%      72.06fps   
0.120 64bit   157.64%
0.120u1      120.08%      72.05fps


hornet.c
- gradius4   
0.120      112.04%      67.91fps
0.120      92.69%
0.120u1      fails instantly with memory error


Model3.c
- scud      
0.120      40.98%      24.59fps   
0.120 64bit   33.57%
0.120u1      44.21%      26.53fps


namcos21.c   
- starblad
0.120      140.06%      84.03fps
0.120 64bit   146.24%
0.120u1      139.95%      83.97fps

(I'll edit this post and fill in the FPS tomorrow sometime, it's late and between the Pure CPU & In-Game benchmarks I've been at this almost 6 hours tonight.)

Well there you go, I hope to be able to post PM opitised 64bit MAME 0.120u1 benchmarks by the end of the weekend, but that all depend how things go with compiling a 64bit build.