Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Developer Transition Kit Benchmarks

Ravenflight said:
Sorry but I haven't been following Pentium develeopment as much as I apparently should have. Yonah, Conroe, Pentium -M's, what's it all mean?
Click to expand...
Pentium M was something a small group in Intel created for mobiles (actually based on the Pentium 3), which happened to be far more efficient than the Pentium 4. So now the P4 has been dumped and it'll all evolve from the Pentium M.

I found the following useful:
http://freespace.virgin.net/m.warner/Roadmap2006.htm

A quick overview of the various Pentium-M future chips
Now: Dothan (single core chip)
All following chips have SSE3 multimedia, Airport, and are 65nm
2006 Q1: Jonah / Yonah dual core Mobile processor (667Mhz FSB)
2006 Q2: Sossaman dual core desktop CPU 2Ghz speed.
2006 Q4: Merom dual core Mobile processor (800Mhz FSB)
2006 Q4: Conroe dual core desktop CPU, 64 bit.
2007 Q1: Xeon DP (Woodcrest) based on Conroe (dual core dual processor?)

Kinda fits Apple updating laptops and other G4s first in 2006, then single G5s, then in 2007 the dual G5 PowerMacs.
 
elmimmo said:
Hey, you already have plenty of official benchmarks, courtesy of Apple at http://www.apple.com/powermac/performance/

Oh, wait, no that was when intel sucked…
Click to expand...

Wait, you seem to forget something....

K sorry I don't want to use that tone.

This has been mentioned before, and the general consensus is that the reason the P4 sucks on that benchmark is the fact that it's running Windows, this is the biggest bottleneck.

And also, remember that Apple will most likely not be shipping Macs wit P4 chips in it anyway.
 
Mr Maui said:
Is it just me, or are there others who hate the fact that we can't use the term "Wintel" in the same sarcastic manner we used to. :D
Click to expand...

Yeah, I've already caught myself wanting to type "Wintel". Unfortunately my alternative "Windoze/myriad-generic-BIOS" does not have the same silky contempt dripping from it.

From now on it will be all OS all the time in the platform wars. The upside is that now that we are on essentially the same playing field with the Windoze hoards, they will have nowhere to hide from the utter and glorious superiority of Mac OS X! I guess maybe Linux can be Tiger's sidekick or something.
 
The performance is actually surpricingly well, look at pearpc, they are kinda of the same thing (x86 emulate powerpc)

And even steve jobs said:
Picture-1.jpg
 
Irresponsible

Its irresponsible for them to publish those numbers without explaining in big bold letters:

"The INTEL numbers are from Xbench under emulation under Rosetta"

Whats worse is that they compare against a native version running on a G5. Leaving people who don't understand to make a direct comparison.

Here's the truth --- The Intel Macs when the are released will wipe the floor with the current PowerMacs.
 
ogminlo said:
I guess maybe Linux can be Tiger's sidekick or something.
Click to expand...

Yea, what can Longhorn do against a Tiger and a Penguin?

You know, the score on the p4 wasn't too bad. It wasn't anything great, but it didn't suck. My laptop gets a score of about 86. Go 12" PB!
 
SPUY767 said:
Looks to me that the 2.7 pretty much wiped its ass with the intel machine. Graphics didn't suffer since they are mainly system level GPU calls. But everything else seemed to get hammered by the G5. This benchmark means bugger all however, because it's being emulated. Suffice to say, that the benchmarks fared better than they would if it was being run on PearPC.
Click to expand...

Isn't emulation supposed to be "near-native speed"? Isn't that the entire point of this vaunted "Rosetta" technology?
 
robbieduncan said:
No one seems to know what we are getting, but it's public knowledge that we are not getting Open Firmware.
Click to expand...

Whoa..... wasn't public knowledge to me :confused:

I do hope the IntelMacs will get the same good ol' startup options like "N" for NetBooting, "T" FireWire Target Disk mode, "C" for CD, "OPTION" for search for all bootable devices, ... etc.
And not "press F2 to get into BIOS options", or "press F12 for Network booting options" etc.... :rolleyes:
 
robbieduncan said:
Whilst the developer boxes are PIV the first Intel Macs are going to be laptops and lower end boxes (the Mini). As such they are going to be using the Pentium-M which is actually quite a cool running chip. It also has a sensible pipeline (shorter than the G5 iirc). The future desktop Intel Macs are going to be using the dual core 64 bit Pentium-M derivative desktop chips that Intel will be shipping next year.
Click to expand...

I actually don't think this will be the case. There is definitely going to be some transition required for an Intel-based Mac (Rosetta's not going to do it all, as has been established in this thread). That transition will be easiest if the power users have the systems first, which means starting with the high end systems.
 
Spazmodius said:
Isn't emulation supposed to be "near-native speed"? Isn't that the entire point of this vaunted "Rosetta" technology?
Click to expand...

I don't think Xbench under emulation is a good measuring stick for emulation speed, I could be wrong though.

Whats more important is for people to realize the Rosetta is only for those legacy apps that didn't, or will never be recompiled. I've gone down my list of apps, and if Adobe and MS keep thier word (and I'm certain they will), I won't really have ANY apps that need to run under Rosetta.

So no, near native speed is not the "point" of Rosetta.

Having to use Rosetta will be the exception for most users, not the rule.
 
Let's face it: For all but the most primitive Mac coding, Rosetta is Crippleware[sup]TM[/sup]. The entire reason people optimize code for the G4 and G5 is because those processors have advanced vector units and other features that allow even similarly-clocked G4 processors to stomp a G3. About the only widely-used application that won't take a serious performance hit is Office, which everybody knows is crap whore code and takes no advantage of AltiVec (or much of anything else good about the PPC platform). Everything else is going to get pounded. Not surprising Steve-O overhyped this; but what continues to surprise is the equivocation of the Mac Faithful. Folks, remember what he said? That Rosetta is the best thing since sliced cheese, this advanced emulator technology that will allow PPC apps to run on MacTels will "minimal" performance hit? This benchmark is an indicator of what those claims are worth.
 
Interesting stuff on this thread, but the benchmark results strike me a little like comparing a kiwi to a dolphin. I mean, a P4 3.6 with a G5 2.7 DP? I almost wish this wasn't posted, since the developer box is just a convenient staging platform to check software compatibility and do re-compiling. I'm going to wait until the real ones come out a year from now before making any judgements. Even then, since the Intel stuff is hitting on the low end first, a more appropriate comparison might be the final generation G4 Mini with the first generation Intel Mini, or something like that...
 
Lord Bodak said:
I actually don't think this will be the case. There is definitely going to be some transition required for an Intel-based Mac (Rosetta's not going to do it all, as has been established in this thread). That transition will be easiest if the power users have the systems first, which means starting with the high end systems.
Click to expand...
Look at Intel's chip roadmap ... link earlier in this thread. It appears the transition will have to be from the bottom up ... as confirmed by SJ at the keynote.
 
Lord Bodak said:
I actually don't think this will be the case. There is definitely going to be some transition required for an Intel-based Mac (Rosetta's not going to do it all, as has been established in this thread). That transition will be easiest if the power users have the systems first, which means starting with the high end systems.
Click to expand...
What makes you say a transition is easier if power users go first?

In favour of what you say is that Apple has apparently been discouraging developers from using SSE3, and XCode has SSE3 compiling turned off by default. Why? when all the newer Pentium-Ms will have SSE3 as of next year.

I know it's often good to have power users cut there teeth on the new stuff. But the G4s are in greater need of an update, and Intel's future standard of chips are almost ready to replace G4s, and won't be of G5 standard till a year later.

Personally I think the dual G5s have the greatest life in them, so hope they are the last to go. Partly as I think they'll be doing good work for years, partly because some Mac has to be the last to switch at the end of 1997, and if it's a G4 iBook (or any G4 machine) it'll be looking pretty old at that point.
Mr Maui said:
It appears the transition will have to be from the bottom up ... as confirmed by SJ at the keynote.
Click to expand...
Did he say this?

Greg
 
GuyClinch said:
Doom III runs much faster on those supposedly "slower" P4 chips. LMAO. It's not even close. A single P 3.6 can clean the clock of a dual 2.7 GHZ G5. And yes Doom IS multi-threaded. It's physics engine is on a different thread.

In real life the G5 just feels really sluggish. You can cry about synthetic benchmarks but in reality you Macheads are in for a treat. we are talking about an easy 30% advantage in real world use with that game. When you consider that they tested a dual processor machine on a multithreaded program it's an unbelievable thrashing for the G5.

And it's not just the g5 that sucks its the Mac memory systems which have lagged behind for years now in general.

Pete
Click to expand...

Can you post the bench marks for Doom 3 on the Mactel?!

Oh wait, you're a troll.
 
XBench is a horrible benchmark for Rosetta. Why? Simple. The vast majority of an app spends its time in operating system calls. Those are going to be native. All the drawing calls, window calls, etc.. will be native so the app will be running in a mix of emulated and native code. As Jobs said - it will run fast enough.

The real test will not be running Photoshop comparisons (since Photoshop will be native). It'll be running Quicken, Quickbooks, TurboTax, Route 66, etc...

My only real scare about these Intel Macs will be games.

Also - as others have said, the Pentium-M mobile processors beat the pants off ANYTHING Apple has to offer today in the mobile space and come near to the performance of Apple's desktop processors. The dual core stuff coming soon will spank the G5.
 
sbarton said:
I don't think Xbench under emulation is a good measuring stick for emulation speed, I could be wrong though.

Whats more important is for people to realize the Rosetta is only for those legacy apps that didn't, or will never be recompiled. I've gone down my list of apps, and if Adobe and MS keep thier word (and I'm certain they will), I won't really have ANY apps that need to run under Rosetta.

So no, near native speed is not the "point" of Rosetta.

Having to use Rosetta will be the exception for most users, not the rule.
Click to expand...
Rosetta on Intel is like Classic in OSX. It is there to ease the transition to the new system, not to make it possible for everyone to continue running old apps forever at full speed ... or like VirtualPC to allow the running of Windows apps (emulated) on a Mac. It is designed to help people through the transition period. An earlier thread talked about Classic being eliminated from the new system. Do we really still NEED to run OS9 applications 5-7 years after OSX? Time to upgrade!!! Anyone who buys a computer should realize that software packages (and their upgrades) are a necessary part of computer ownership. Computer manufacturers should not have to continue to build support for 5, 10, 15 year old apps to please those who don't want to invest money in an upgrade. If a buyer does not want to upgrade their software when they buy a new computer, they keep their old computer alongside their new computer and run the old stuff there, where it was DESIGNED to be run. How many users are not planning on upgrading their software in the next 2-3 years, but are planning on buying a new computer in that timeframe? Those who raised their hands, should still be watching black and white TV or listening to "Burns and Allen" at night on the radio.
 
Spazmodius said:
Let's face it: For all but the most primitive Mac coding, Rosetta is Crippleware[sup]TM[/sup]. The entire reason people optimize code for the G4 and G5 is because those processors have advanced vector units and other features that allow even similarly-clocked G4 processors to stomp a G3. About the only widely-used application that won't take a serious performance hit is Office, which everybody knows is crap whore code and takes no advantage of AltiVec (or much of anything else good about the PPC platform). Everything else is going to get pounded. Not surprising Steve-O overhyped this; but what continues to surprise is the equivocation of the Mac Faithful. Folks, remember what he said? That Rosetta is the best thing since sliced cheese, this advanced emulator technology that will allow PPC apps to run on MacTels will "minimal" performance hit? This benchmark is an indicator of what those claims are worth.
Click to expand...

No, the TRUTH is that only a handfull of applications, mostly Professional apps, take significant advantage of ALTIVEC. The two that I'm familure with is Photoshop and Cubase..but there are a few others. Even more to the point is that of these two apps, only the Plugins make heavy use of Altivec. It will be job for these developers yes, but its not like either of these apps were ever written from the ground up take full advantage of Altivec.

Any idiot that expects to run a pro app under Rosetta without a performance hit deserves what they get. If you run a buisness with one of these apps and for some strange reason its not ready when the Intel Macs start shipping, guess what?....DON'T BUY ONE.
 
Sharewaredemon said:
elmimmo said:
Hey, you already have plenty of official benchmarks, courtesy of Apple at http://www.apple.com/powermac/performance/

Oh, wait, no that was when intel sucked… Wait, you seem to forget something....
Click to expand...
the reason the P4 sucks on that benchmark is the fact that it's running Windows, this is the biggest bottleneck.
Click to expand...
The benchmarks state it clearly, and they state processor. Because if not, according to these other benchmarks which leave Mac OS X in the dust, a P4 with Linux on it is what, like 600% faster than running windows?

To me your "general consensus" is that only those that consent, agree. The lesson you REALLY have to pull from this is that PR people, be Apple's, Microsoft's, IBM's, Intel's, blah, blah, ALL are blatant LIARS. Not because they are evil, it's simply their job.

And also, remember that Apple will most likely not be shipping Macs wit P4 chips in it anyway.
Click to expand...
Oh, yes. So if we should take into account Apple's benchmarks, we should be calmed, because in two years an a half we will have new Mac Intels that are not slower than current PPC. They will have enough time to just be on pair…
 
OH MY GOD!!!!!!! A single processor P4 in emulation mode that isn't even intels fastest chip runs only half as fast as a dual core G5 system running natively. Apple should just die now since they are killing the platform.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back