[slipjack]

Hey ya'll.

Thought I would relate a 'test' I was able to run at my local Comp USA today. They had a G5 set up and sitting right next to a G4 dualie 1.42 MDD, so I was able to boot up several programs simultaneously.

The Dual machine won hands down opening iMovie, iPhoto, and iTunes.

True, the G5 only had 256 Megs of RAM and I assume the dual had 512, but still... for those who don't need the newest and the coolest, the Dual machine looks like it might be a really good deal.

PLUS the G5 didn't seem any quieter to me than the G4, but it was hard to tell with all the background noise. They both seemed pretty quiet, actually.

I assume that anything that eats up bandwidth, the G5 would start to shine, but for mundane tasks like loading a program, the G4 dual machines can still cut it.

When is a mid-range G5 going to come out?!

[Eug Wanker]

256 MB is insufficient for OS X and those apps.

[idyll]

Indeed. 256 is the minimum if you want to run OS X.

With the same amount of ram the dual G4 and single G5 should be on par with launching applications, or so prove "real world tests."

[MindFad]

There is no "but still". Running 256MB and conducting "tests"—Photoshop, or otherwise—is like "testing" OS 9 with 32MB of RAM. It's pukey. Apple needs to start putting more than 256 in their base models, the mofos. Hopefully 512 will be a minimum come rev. B.

[Webscreamer]

I ran photoshop tests and Xbench 1.1.1 tests on 1.25ghz G4 vs. 1.6ghz G5 today.

G5 1.6 won hands down....

[Eug]

Originally posted by Webscreamer:
I ran photoshop tests and Xbench 1.1.1 tests on 1.25ghz G4 vs. 1.6ghz G5 today.

G5 1.6 won hands down....

Xbench is basically a single CPU test.

[slipjack]

Originally posted by MindFad:
There is no "but still". Running 256MB and conducting "tests"—Photoshop, or otherwise—is like "testing" OS 9 with 32MB of RAM. It's pukey. Apple needs to start putting more than 256 in their base models, the mofos. Hopefully 512 will be a minimum come rev. B.

Sure there is... 256 Megs is more than enough to run the OS and run ONE APP.... and we're just talking loading it up. Something like iTunes takes a negligible about of memory to run.

Like I said, it wasn't much of a test, I just ran iPhoto, et al., at the same time and noted which one was faster... and the Dual G4 1.42 was faster.

So... for $1699 you can get a pretty sweet dual machine. Seems like a good deal to me.

[idyll]

But it had twice the amount of ram.

[chrisutley]

Launching programs fast or apps like Photoshop and Final Cut flying, take your pick.

[Stradlater]

Originally posted by Eug:
Xbench is basically a single CPU test.

Which is all that the majority of applications takes advantage of.

[Angus_D]

Originally posted by Stradlater:
Which is all that the majority of applications takes advantage of.

If you have more than one application open at a time, the OS can intelligently schedule their threads across the processors, even if they aren't multi-threaded themselves.

[Centris650]

Correct me if I'm wrong but the G5's do not have a L3 cache. It get's it's cache from Ram. The more ram the faster the cache. Now the G4's have a 1 Meg per processor (I THINK that's right for the dual 1.25). So it seems that a G5 with 256 megs of ram WOULD be slow. Now what would be cool if it had all 4 megs. THAT would be something to see.

[Cadaver]

Originally posted by Centris650:
Correct me if I'm wrong but the G5's do not have a L3 cache. It get's it's cache from Ram. The more ram the faster the cache. Now the G4's have a 1 Meg per processor (I THINK that's right for the dual 1.25). So it seems that a G5 with 256 megs of ram WOULD be slow. Now what would be cool if it had all 4 megs. THAT would be something to see.

The (current) G5 does not support an L3 cache, that is correct. The architecture of the chip and bus however is such that it would not benefit much from one. The G4 requires an L3 cache for optimum performance because the memory subsystem and front side bus is so slow. High-speed cache RAM is very expensive, which is why you don't see large amounts being used.

Saying the G5 "gets it cache from Ram" isn't really correct usage of the terms. The G4's large L3 cache basically duplicates the contents of main RAM to speed access (in an attempt to make up for the otherwise slow bus speeds of the PowerMac G4). The PowerMac G5's architecture doesn't require this kind of a caching system to compensate.

One reason some of the 1.6GHz PowerMac G5 tests have been so underwhelming is because these machines ship with only 256MB RAM standard; this results in extensive virtual memory use, which slows the system's performance down. A G4 with only 256MB RAM would also perform poorly, regardless of the amount of L3 cache. Many stores do not seem to be upgrading the stock memory configuration on their demo units. For email and web browsing, 256MB may be acceptable. But for any more serious work, more is necessary for decent performance.

My local Apple Store has a 1.6GHz G5 on display with 768MB of RAM. It's performance was seemingly quite competitive with a Dual 1.25GHz G4 sitting next to it. Considering the dual G4 is (was) much more expensive than a 1.6GHz G5, I think the low-end G5 is a good deal.

[MusicalTone]

Originally posted by Cadaver:
Considering the dual G4 is (was) much more expensive than a 1.6GHz G5, I think the low-end G5 is a good deal.

Here in the UK the dual 1.25 costs £2,548
The 1.6 G5 £1,549.

For us the 1.6 G5 is far far the better deal.

[I Me Mine]

"Here in the UK the dual 1.25 costs £2,548
The 1.6 G5 £1,549."

But you can get a dual 1.25 for £1239 from Apple if you configure it right. Even less if you leave out the modem.

[MusicalTone]

Originally posted by I Me Mine:
[BBut you can get a dual 1.25 for £1239 from Apple if you configure it right. Even less if you leave out the modem. [/B]

Really, how do you do that? The Apple store online says £2,548 ...

[I Me Mine]

Click the £999 single processor version and Add A Processor for £240.01

[StiZeven]

I am starting to think that the current (1.x) G5 is suffering from the early P4 blues. When Intel first debuted that chip, everyone agreed that it was a far superior processor than its predecessor, but due to the new architecture of the chip, a MHz push was needed to see its true potential. In the beginning of the P4, the P3 was either the same or not too far behind in everyday real life performance. It wasn't until they went past 2GHz that you saw the real power of that chip. Perhaps (something like) this is the case with the G5 and maybe the low-end should have been clocked at 2GHz and the high-end at 2.4GHz?

Then again, everyone says that Panther will show the real power of the G5, so I guess we'll just have to wait and see. I also agree about the RAM, OS X needs at least 512MB to perform well regardless of the chip.

[sleepyrenderer]

Originally posted by StiZeven:
I am starting to think that the current (1.x) G5 is suffering from the early P4 blues. When Intel first debuted that chip, everyone agreed that it was a far superior processor than its predecessor, but due to the new architecture of the chip, a MHz push was needed to see its true potential.

Heh, that's kind of revisionist history there . The opposite actually occurred in that nearly everyone agreed that the early P4's were inferior to the P3's they were replacing and warned buyers away from wasting money on them.

It wasn't until they went through a process shrink, were coupled with better chipsets (maybe equivalent to Apple releasing the PowerMac G5 rev.2) and much of the software out there had been recompiled, that they really started to shine.

[skyman]

Originally posted by Webscreamer:
I ran photoshop tests and Xbench 1.1.1 tests on 1.25ghz G4 vs. 1.6ghz G5 today.

G5 1.6 won hands down....

You do realize that just based on Mhz the 1.6 is about 30% faster than the 1.25???

So, when you say hands down you mean much much much faster than 30%......right?

How about some test results to prove your statment??

[Shaktai]

Okay slipjack, I can't resist this.

So you compared a dual 1.42 for a total of 2.48 ghz with twice the RAM, running an optimized operating system with optimized apps, against a 1.6 ghz single G5 running a kludged operating system, and NO optimized apps.

What the heck kind of results did you expect? It is like saying that you just won at the drag races, but the guy you were racing had plugged air and fuel filters and his engine timing was off.

Wait one month and then compare against the final version of Panther. Better yet, wait 2-3 months when we start seeing G5 optimized apps and compare. Then and only then will any comparisions have merit.

If you buy a dual G4 you are buying for today.
If you buy a G5 (any version) you are buying for the future, but it is a very near future.

[slipjack]

I still think the dual 1.42 is a better deal than than the 1.6, and most likely the 1.8. This is based on my real-world experience of opening apps in both the 1.6 and the 1.42, nothing more.

No, Shaktai, I was not overly suprised that the dual beat out the 1.6, but I had started to wonder with all the hype.

If it is true that Panther is going to speed up the G5 more than the G4, than maybe in a couple months the G5 will be the better buy. But right now, you can get a great machine in the 1.42, imho, and for LESS than the 1.6 machine. (Stripped down a bit, sure.)

[idyll]

Yes, right now. But for the future the G5 will be more valuable overall. The 1.6 G5 has a faster hard drive, new motherboard, etc. The only thing that it has in common with a G4 are the regular PCI slots.

[Shaktai]

Originally posted by slipjack:
No, Shaktai, I was not overly suprised that the dual beat out the 1.6, but I had started to wonder with all the hype.

If it is true that Panther is going to speed up the G5 more than the G4, than maybe in a couple months the G5 will be the better buy. But right now, you can get a great machine in the 1.42, imho, and for LESS than the 1.6 machine. (Stripped down a bit, sure.)

I agree, that for "today" the dual 1.42 is the better deal. Like I said, buying a dual g4 though is buying for today. Buying a G5 is buying for the future.

And the hype? Well there is a lot of that.

[HodyOne]

Originally posted by slipjack:
Hey ya'll.

Thought I would relate a 'test' I was able to run at my local Comp USA today. They had a G5 set up and sitting right next to a G4 dualie 1.42 MDD, so I was able to boot up several programs simultaneously.

I hope you base your buying decision on better data than your 'test' which is basically meaningless. Besides the memory issue, did you reboot both computers before performing the test? Perhaps the dual had the programs in the cache making it an unfair comparison? How did you time the boot up of these programs simultaneously? Was the time between clicks to startup each of the programs the same (if not this could make a big difference)? The bottom line is that, while it is difficult to arrive at a good set of benchmarks to evaluate these machines, yours is misleading at best.

[I Me Mine]

Originally posted by I Me Mine:
Click the £999 single processor version and Add A Processor for £240.01

As if to make a liar out of me, Apple have changed this from today. All you can do now is pick the top of the range G4 and strip it right down to £1219.

[KidRed]

Where's the 'real world test'? The test was you launched 3 apps on 2 machines at CompUSA and the G4 had twice the amount of ram as the G5?

Ah, in my 'real world' 256 is a pipe dream and the G5 can do a bit more then open 3 apps. Also, to further the lecture, remember that none of the '3' apps you tested in the real world are optimized for the G5. So while underpowered ram wise, running a hacked 10.2 and launching 3 apps that have not been coded to take advantage of the G5, you made this 'real world' test and concluded the dual G4 was a good deal because it beat the G5 hands down?

[kjb]

Just to add to the debate, I've run a test at the local Apple store. I converted a 3.X meg QT movie from .mov to DV within iMovie. I did it using a 1.6 G5 w/ 256 megs of RAM, a Dual Gig MDD w/ 256 megs of RAM and a 1 gig. iMac w/ 256 megs of RAM. Here are the results:

G5 1.6: 25 seconds

G4 1.00x2: 31 seconds

iMac @ 1.00: 42 seconds


These results make me wonder, what if the G4 could actually scale to freqs. beyond 1.42? The G5 isn't even 25% faster than the Dual gig., much less 60%. Now, I realize that the G4 doesn't scale, but might it in the future? To me, the dual G4s that are now available seem to be the bang for your buck machines right now.

Kevin

[KidRed]

Originally posted by kjb:
Just to add to the debate, I've run a test at the local Apple store. I converted a 3.X meg QT movie from .mov to DV within iMovie. I did it using a 1.6 G5 w/ 256 megs of RAM, a Dual Gig MDD w/ 256 megs of RAM and a 1 gig. iMac w/ 256 megs of RAM. Here are the results:

G5 1.6: 25 seconds

G4 1.00x2: 31 seconds

iMac @ 1.00: 42 seconds


These results make me wonder, what if the G4 could actually scale to freqs. beyond 1.42? The G5 isn't even 25% faster than the Dual gig., much less 60%. Now, I realize that the G4 doesn't scale, but might it in the future? To me, the dual G4s that are now available seem to be the bang for your buck machines right now.

Kevin

Again, see my post above yours as to why EVERYONE needs to slow down until at least PANTHER is out before making any blanket claims.

[Cadaver]

Gentlemen, where is the debate here?

Despite the G5's performance, I think two G4 processors will frequently outpace a single G5. Think about it - Apple's previous top-end machine compared to their new bottom-end machine. Even a dual 450MHz G4 can outperform a 733MHz model quite often.

While I have little doubt the 1.6GHz G5 is a fast machine, I don't expect it to wipe the floor with a dual-processor MDD machine.

I do have high expectations for the dual 2.0, however

[slipjack]

Originally posted by HodyOne:
I hope you base your buying decision on better data than your 'test' which is basically meaningless. Besides the memory issue, did you reboot both computers before performing the test? Perhaps the dual had the programs in the cache making it an unfair comparison? How did you time the boot up of these programs simultaneously? Was the time between clicks to startup each of the programs the same (if not this could make a big difference)? The bottom line is that, while it is difficult to arrive at a good set of benchmarks to evaluate these machines, yours is misleading at best.

Yes, I did reboot both machines... just to see which booted up first... the 1.42 did...

Like I said, it wasn't much of a test, but it was enough to convince me I was right thinking I should wait for a mid-range dual G5 to come out.

Dual is the *only* way to go, imo.

[tony21]

Originally posted by slipjack:
Yes, I did reboot both machines... just to see which booted up first... the 1.42 did...

Like I said, it wasn't much of a test, but it was enough to convince me I was right thinking I should wait for a mid-range dual G5 to come out.

Dual is the *only* way to go, imo.

Dual? Every one is talking about getting the newest best machine. 6 months ago the 1.4 dual was the best. The 1.6 and the 1.8 are great machines, if you are just going to play games and stuff get a emac or imac.

[Shaktai]

Originally posted by slipjack:
Like I said, it wasn't much of a test, but it was enough to convince me I was right thinking I should wait for a mid-range dual G5 to come out.

Dual is the *only* way to go, imo.

Well, I would have to agree with that. I don't think it wil be that long of a wait. 6 months max, and maybe sooner. Dual definitely gives the most bang for the buck, not to mention by that time the OS and many apps will be optimized, which will only make it better.

[jclarkv]

Originally posted by Shaktai:
Well, I would have to agree with that. I don't think it wil be that long of a wait. 6 months max, and maybe sooner. Dual definitely gives the most bang for the buck, not to mention by that time the OS and many apps will be optimized, which will only make it better.

Great. Then I can commit my single G5 1.6 to crunching for D2OL and get a mid range dual G5 (which I can than also commit to D2OL crunching).