New Power Macs Faster than you think.

[Doctor Q]

Originally posted by shadowfax0
You sure it was 5H and 52M? My friend's single 867 get like 6-7 hours...but any details on how you ran it would be nice too But still, I'm liking that time, about ( about people, about, I swear if I hear about this someone's gonna die...) 5 workunits a day, not bad, not bad at all...

Yup, 5:52:11.2 per unit. I ran it in screensaver mode, watching the display for the first 1/2 hour (I couldn't help it - the flashing lights hypnotized me!) and then letting it turn off the display after that. I have yet to try command-line mode (no GUI), which would presumably tweak the speed still further.

 

[shadowfax0]

Oh ok, SCREEN SAVER mode, well that makes sense, I can't wait until you do it with the CLI version ::drool::

 

[MacCoaster]

Originally posted by {1984}
i guess everyone knows about the whole "MHz myth" thing...
Reason for G4 processors killing the Pentium 4 is cuz of the pipeline and i'm not gonna bother cuz everyone probably knows...

Funny that Motorola had to add a few more pipelines just to have the G4 catch up to the Pentium 4. Sorry bud, in pure performance, the Pentium 4 2.8GHz and Athlon XP 2.13GHz 2600+ has got it beat. *IF* the G4 was at ~2.x GHz, sure it'd beat the Pentium 4, but the fastest one is 1.25GHz. Dual 1.25 GHz != 2.5GHz in real world so, as pointed by the "benchmarks," Approximately 2.25 GHz G4 (1.25x2) performs just as good as a single Pentium 4 2.8GHz. There are also *A LOT* other things that contribute to performance other than just the stupid pipelines. Memory controller, bus, pretty much anything.

Right now, the G4 simply sucks. We need the G5 or the new IBM PowerPC. G4 isn't living up to its expectations unless Motorola has performed some miracle to boost the G4 to 3GHz overnight. That ain't happening, either. The best G4 Motorola has ever done was the 7410. Those Dual 533MHzs kicked other arses!

On the other hand, for productivity, the Macintosh experience is the simply best and fastest, but as a research computer, I'll take a quad Xeon running FreeBSD 5.0 for the price of a high end Power Mac G4, thank you. Otherwise, if I want to get my **** done, I'll simply buy an iMac 800MHz with the best desktop OS.

 

[Doctor Q]

Originally posted by shadowfax0
Oh ok, SCREEN SAVER mode, well that makes sense, I can't wait until you do it with the CLI version ::drool::

I am now benchmarking the CLI version of SETI@home on the Dual 1.25GHz G4. I was surprised that it was only marginally faster than the GUI version (a little under 6 hours per unit), but SETI's FAQ explains that the GUI on a dual processor uses one processor for the graphics and one for the computation, so the graphics "don't count" much. And, when running SETI CLI version, you still don't get the benefit of the dual processors (except to run something besides SETI at the same time, of course) unless you explicitly start two SETI processes at the same time, each in its own directory.
I will report my results of single-CLI runs after I determine the effect of the Energy Saver settings and the effect of the -verbose option that reports progress in the Terminal window. Then I will try double-CLI runs.

 

[Doctor Q]

I just posted my SETI@home benchmark results in Mac Forums > Mac Discussion > Distributed Computing.

 

[UnixMac]

The Solution

Lets put OS X onto the PC platform and get the best of both words.

Because while I admit that the G4 is faster than the P4 on a cycle to cycle comparison, the fact that P4 is pushing 3GHz while we are still at 1.25 is a bit sad.

Plus you are talking about 533mhz busses and RD RAM, many other hardware issues set the PC platfrom above the Mac.

I give apple an "A" for aesthetic design and a "D" for engineering design.

I give Unix and the OS X version of it an A for both design and aesthitcs.

That is my honest view. BTW I am waiting on a PB with a faster FSB and 1.0GHz because I can't bring myself to spend $3500+ on a PC133 machine!

 

[ooartist]

Proof that UNIX can scale...

To squash some WinTel people in this forum/post trying to say Windows scales better than UNIX.

Spec of a Sun machine running UNIX.

Key Specifications:


Up to 106 UltraSPARC® III Cu 900-MHz processors.
Big memory - more than 1/2 TB.
Up to 18 fifth-generation Dynamic System Domains, which are fully configurable while applications are running.
Hot-swappable Uniboard design CPU/memory boards that are common across Sun Fire server family.
Redundant, high-performance Sun[tm] Fireplane Interconnect with up to 172.8 GBps peak bandwidth.
Full redundancy of power and cooling systems.


Oh yeah! OS X is UNIX also. Hmmm.




Single CPU vs. Multi CPU. Who cares?

Macs are the whole package.

Nuff said!

 

[ddtlm]

ooartist:

While both Solaris and OSX are Unix OS's, that does not mean that OSX can scale as well as Solaris any more than it means that Solaris can run the OSX interface (which it obviously cannot).

The fact that OSX is some sort of Unix means a lot less than you seem to think. It is good because it means it adheres to certain standards, but it says little or nothing about the machinery that makes it go.

 

[MacCoaster]

Re: Proof that UNIX can scale...

Originally posted by ooartist
To squash some WinTel people in this forum/post trying to say Windows scales better than UNIX.

I never said it scaled better. In fact, I said the opposite.

However, the fact that Mac OS X's kernel is *NOT* BSD, but Mach remains. Mach is a microkernel and a freaking good one, but Mach microkernels that OS X uses has poor task switching; I'm not sure if the Mach microkernels in OS X is based on GNU/Mach (based off CMU's Mach) or the actual Mach microkernel from Carnegie-Mellon. So some of the scalability of UNIX is lost through this. But trust me, UNIX scales way beyond Windows. I've said Windows isn't for computers with more than 32 processors--that's what UNIX is for--pure science--no one needs the crap from Mac OS X and Windows to do science and mathematics research. Real men use UNIX on 8192-way systems.

 

[ddtlm]

People are throwing around "Unix" and "Windows" like they used to throw around "RISC" and "CISC". There is no reason of which I am aware why a future version of Windows cannot scale to as many processors as any version of Unix, just like the nasty ol x86 ISA has yielded top-notch processors like the P4 and Athlon.

I think that everyone here who argues otherwise is engaged in a desperate attempt to justify their worldview that equates Apple with perfection and wisdom, or perhaps equates Microsoft with evil and boundless stupidity.

 

[MacCoaster]

Originally posted by ddtlm
People are throwing around "Unix" and "Windows" like they used to throw around "RISC" and "CISC". There is no reason of which I am aware why a future version of Windows cannot scale to as many processors as any version of Unix, just like the nasty ol x86 ISA has yielded top-notch processors like the P4 and Athlon.

I think that everyone here who argues otherwise is engaged in a desperate attempt to justify their worldview that equates Apple with perfection and wisdom, or perhaps equates Microsoft with evil and boundless stupidity.

Amen. Although I think the Windows' scalability vs. Unix's scalability is a valid argument because reasons stated in my post above. It's true that there shouldn't be a reason that Windows cannot scale to as many processors, but as I've said... it's not worth it because it isn't Microsoft's market.

 

[ddtlm]

Not Microsoft's market... yet.

They want that peice of the pie. Give them time, they will try to take it.

 

[MacCoaster]

Originally posted by ddtlm
Not Microsoft's market... yet.

They want that peice of the pie. Give them time, they will try to take it.

Might not be yet, but might be never.

 

[UnixMac]

I am sorry, but you can't compare the stability of Windows (of any flavor) with OS X. I have never, repeat never, in 2 years of running OS X had a single crash. I can tell you that my office machines running 2000, are regularly down.

Sam

 

[ddtlm]

WanaPBnow:

This may come as news to you, but a single person with a single experience is nothing but a single data point. In order to have confidence in a conclusion, we need lots of data points. Data points other than yourself tend to mention OSX crashing on occasion, and tend to mention at least some versions of Windows running reliably.

Speaking for myself, I've had a small number of OSX kernel panics in a little more than a year on my home machine, dozens of kernel panics on my work Mac (which I don't even use much), and very few problems or crashes with any of the dozens of Windows NT/2k machines at work. In my experience OSX is less stable than Win2k, however I am not foolish enough to claim that my observation makes this the truth for everyone.

 

[UnixMac]

My reply to the above statements

Actually I don't think I said that MY observation was my reason for this fact. But rather the observation of my co-workers (computer engineers) at a very large multi-national company that uses windows 2000 for it's medium sized computer networks. They are now in the process of switching to IBM Unix and Solaris (where possible). The effectively have "fired" windows, except for the front office PC's.

As for OS X, I would say that every one I know with OS X (about 8 or so people) have also had a similar experience to mine, which while anecdotal, is still a damn good record.

I will defend your view that Intel's platform is ahead of the Mac (for now, it is), but windows is not the same caliber of OS, as Unix.