mac performance analogue to the pc

[queasy]

Hi,

I've looking into buying a mac to perform a port. However, I'm unsure as to how performance on the mac is compared to the pc. For example, is it safe to assume that a 300mhz G3 is about the same as a 300mhz P3? Is a G4 comparable to a P4?

Any help on this matter will be greatly appreciated!

Jonathan Mak

[Raptisoft]

The mac has one mouse button, the PC has, theoretically, unlimited.

[queasy]

umm, I'm not sure if that was a joke or if there was a misunderstanding. By "performance", I meant CPU speed.

Jonathan Mak

[Sillysoft]

I would say that it's not as simple as saying 300 mhz P3 = 300 mhz G3. It used to be commonly stated (on the mac side) that the powerPC chip did more per cycle then intel chips did. I haven't heard this in a while, but I think it would still be true if you were comparing a G5 to a P4. However I think it is highly dependent on what the program is doing. Some things the powerPC chip will handle better and some thing intel chips will do better.

[Dan MacDonald]

I think that it's safe to say, that 300mhz is universally slow, regardless of platform by todays standards.

[queasy]

Hi, thanks for your replies.

Could you elaborate on "I think it is highly dependent on what the program is doing." ? What exactly does it depend on? Floating point ops? Cache-alignment? Memory access?

My game does a lot of floating point. It isn't very cache-friendly.

Thanks,
Jonathan Mak

[Mark Fassett]

Well - my 1ghz iMac, when running the normal Mac OSX interface, seems sluggish, for most tasks, compared to my wife's 950mhz athlon. However, my game runs smoother on my mac with a GF4mx than on my Athlon 2100+ with a GF fx5900.

[queasy]

Is your game cpu intensive or gpu intensive?

-j

[Abscissa]

For example, is it safe to assume that a 300mhz G3 is about the same as a 300mhz P3?
No. I realize a "true" comparison would need to take into account things like cache, instruction set, the way a program is written, etc., but I've heared that the rough general rule of thumb is that a Mac CPU is similar to a PC CPU of almost twice the Mac's clock speed. I don't know if that's been changing lately though. But in any case, a Mac CPU is almost always quite a bit faster than a PC CPU at the same clock speed. (By the same token, AMD CPU's tend to be slightly faster than same clock-speed Intels.)

Is a G4 comparable to a P4?
If you're looking at them as a group and not comparing equivilent clock speeds, then yes. Although, from what I can tell, I think the G5's are roughly comparable to some of the the newer P4's. Intel's doesn't seem to have been very interested in leaving the "Pentium 4" name lately.

Disclaimer: I'm not really that much of an expert on that stuff, so don't take my word as definite hard fact.

[Mark Fassett]

Both, really... it uses OpenGL (via PTK), but there are lots of collision detections all the time. More noticable on slower machines. I honestly don't know for sure, as I've never done any tests to measure.

[EpicBoy]

The mac has one mouse button, the PC has, theoretically, unlimited.
I suspect that you're joking, but just in case ... the Mac can use almost any USB mouse, so the button count is the same.

[Mark Fassett]

Yup. I use a MS Intellimouse on my mac. Works great, and even allows speeding up the dang cursor speed of the mac.

[queasy]

Thanks for the info guys. Abscissa, that general rule of thumb you talked about seems to correlate with what mark observed. Well, at least in that more processing can be done at the same clock speed.

-j
ps: mark, the game you're talking about is derelict right?

[princec]

Roughly it would seem that a 1GHz G5 seems to be about the same power as a 2GHz P4.

Cas :)

[Mark Fassett]

Yah - the game is Derelict.

[Evak]

its not much different than comparing a AMD64 to a Pentium 4. A G5 at the same clock speed is faster than the equivalent P4, but as the P4 clocks higher the high end ones are going to be faster.

In recent tests that I saw, wish I had the link, was a mac site comparing comparable sytems dual athlon64 v dualG5's they were pretty close.

Just found the link I posted on the blitzmax forums

http://www.barefeats.com/g5op.html

as you can see the 2 systems are pretty much neck and neck in most of the comparrisons.

Naturaly the P4 might be a little faster still, but it woudl be clocked a lot higher than either system.

[BantamCityGames]

Gotta love google (http://www.digitalvideoediting.com/2002/07_jul/features/cw_macvspc2.htm)

[Evak]

to be honest, my biggest complaint with apple isn't their systems but the misleading information they use when marketing their products. And the silly smoke and mirror tricks they use to wow their enthuisiastic userbase at the mac world expo.

Otherwise they are decent general purpose computers that are as good and sometime better than the PC at some things. OSX came a long way since OS9 where most of my bad experiences with apples computers lie.

When I decided the Amiga was dead I chose the PC for the quality of games and 3D graphics related software avaliable. And today those are probably the 2 main reasons that I have stuck with the PC. Going to be interesting to see how things change when my indie partner starts coding with Blitzmax on the mac lol.

[queasy]

Hey, yea I did try googling a lot before, but I could never find any good info. It's hard to tell with those comparisons because you don't know if adobe (or whomever) decided to provide optimized software for each OS.

I would love it if someone would just write a simple program that did some linear O(n) algorithm and compile it *as is* on both platforms. That would give some meaningful info for indie devs thinking about doing an easy port to mac.

In fact, what the heck... I should just do this myself! ;)

-j

[princec]

I would love it if someone would just write a simple program that did some linear O(n) algorithm and compile it *as is* on both platforms. That would give some meaningful info for indie devs thinking about doing an easy port to mac.Quite the opposite I think you'll find. Google for "perils" and "microbenchmarks".

Cas :)

[Evak]

Don't know but Adobe allready advertised the fact that Photoshop is overall faster in widows these days, not that its faster enough to really make a great deal of difference in what platform you use.

However I do rememer working on a 2.5D RTS where we were playing with HUGE maps sometimes over .5GB in photoshop. Thank god they decided to go with a tile based editor afterall :) and we only ended up creating unique chunks of tiles in PS.

[Diodor Bitan]

My 1.25Ghz G4 is faster than my 233Mhz P1. Emacs runs just as fast on both though. :p

[Mike Boeh]

Cosmo Bots Frams Per Second on the main title screen

PC 2.4 Ghz p4, geforce ti4400
Windowed 206 fps
Full Screen 265 fps

iMac G5, 1.6 ghz, geforce 5200
Windowed 75 fps
Full Screen 55 fps

Ouch :(

[gmcbay]

PC 2.4 Ghz p4, geforce ti4400
Windowed 206 fps
Full Screen 265 fps

iMac G5, 1.6 ghz, geforce 5200
Windowed 75 fps
Full Screen 55 fps


I'm probably asking the obvious, but are you sure that vsync on/off isn't at play here?

[Mike Boeh]

I'm probably asking the obvious, but are you sure that vsync on/off isn't at play here?
My momma didn't raise no dummy! :-)
Yes, vsync off.... As the level selection screen, with far less blitting, gets 150 fps full screen on the imac G5.....

[Red Marble Games]

Cosmo Bots Frams Per Second on the main title screen

PC 2.4 Ghz p4, geforce ti4400
Windowed 206 fps
Full Screen 265 fps

iMac G5, 1.6 ghz, geforce 5200
Windowed 75 fps
Full Screen 55 fps

Ouch :(

I guess I should know this from other posts, but: what's the underlying method of getting pixels on screen? Some paths are faster on the Mac than others, and what blasts through on the PC isn't necessarily the best path on the Mac.

[princec]

Indeed, I can't really see any reason for that gulf in difference on machines of that spec apart from doing something "wrong"..

Cas :)