Welcome to the MacNN Forums.

hunkhuang · Sep 20, 2003, 10:00 AM

Hey hah,

My boss just got me a G5 last week and we returned to Apple Centre because after testing the G5, it was only 25% faster than G4, what a piece of crap... and after we have discussed about this matter, i think the main problem is, G5 has to run in 64 bit OS then only it can perform better.

So my new iMac, currenly running Jaguar (32 bit).. quite smooth... But when i install the Panther (64-bit), will it slow down my iMac??

Developer · Sep 20, 2003, 10:09 AM

Nothing in Panther will be 64 bit on your iMac.

MindFad · Sep 20, 2003, 10:16 AM

iMac over a new G5? Nice.

It will probably speed things up a bit, but you'll probably also wish you had that G5, too.

barbarian · Sep 20, 2003, 10:37 AM

>it was only 25% faster than G4, what a piece of crap..

His tests must have been seriously flawed (for example he might have been testing something that only taxed the GPU and not the processor).

1. G5s are significantly faster than iMacs at doing just about everything.

2. The speed will even be better in Panther.

3. Panther is not 64 bit although it can access the full range of memory in a 64 bit G4 machine (extra memory bandwidth is the main advantage of a 64 bit processor)

4. From what I've seen Panther is significantly faster than Jaguar on all Macs including old ones.

VEGAN · 12:22 PM

Originally posted by hunkhuang:
My boss just got me a G5 last week and we returned to Apple Centre because after testing the G5, it was only 25% faster than G4, what a piece of crap...

Albert Einstein -- Two things are infinite: the universe and human stupidity; and I'm not sure about the universe.

Spheric Harlot · Sep 20, 2003, 10:55 AM

Originally posted by barbarian:
4. From what I've seen Panther is significantly faster than Jaguar on all Macs including old ones.

ESPECIALLY old ones.

G3 Macs apparently benefit the most from speed increases in Panther.

My iMac G3 400 was amazingly more responsive when I briefly installed the first Panther preview a few months ago.

G5s will reap hefty benefits from Panther as well, as Jaguar is in no way optimized for the G5.

Question would be, though: What G4 was your boss comparing to which G5? The dual 1.42 GHz G4 is not worth dumping for a single 1.8 GHz G5. But if he was comparing ANY G4 to a dual 2.0Ghz G5 and only saw a 25% increase, there is something very, VERY wrong.

Other question: did he have the clear plastic panel on the side removed? G5s will run at something like half speed when this panel is missing because the cooling system depends upon it. You can remove the metal side panel, but not the clear plastic.

-s*

Xeo · Sep 20, 2003, 11:48 AM

Wow, I can't believe you gave up the G5 like that. I'm sure after your "extensive" testing you must have known what you were doing...

hunkhuang · Sep 20, 2003, 12:09 PM

i'm not so sure about the plastics thing maybe i'll do the double check tomorrow..

But i'm just worried about my G4 iMac. Will it be slower running Panther compare to Jaguar?

lookmark · Sep 20, 2003, 12:14 PM

Good lord. No. It will be faster.

OK?

oVeRmInD911 · Sep 20, 2003, 12:16 PM

Originally posted by hunkhuang:
i'm not so sure about the plastics thing maybe i'll do the double check tomorrow..

But i'm just worried about my G4 iMac. Will it be slower running Panther compare to Jaguar?

No. It will be faster. Much faster.

hunkhuang · Sep 20, 2003, 12:27 PM

Oh yes yes.. yippie...

Kenneth · Sep 20, 2003, 12:29 PM

Panther should perform better on your iMac.
Meanwhile..

NVIDIA GeForce4 MX, 64MB DDR video memory

CharlesS · Sep 20, 2003, 12:40 PM

I can't believe that your boss bought a G5 for you and you turned it down. Wow.

Mrjinglesusa · Sep 20, 2003, 01:04 PM

unbelievable.

macvillage.net · Sep 20, 2003, 02:25 PM

Anyone run panther on a beige g3? Is it faster on one of those as well?

I know it's not supported, but would be nice if it speeds things up.

THE MAC GOD · Sep 20, 2003, 02:49 PM

Oh my god... 7b68 runs like a dream on myh REV> A IMAC... so I can't see it bogging down any machine that has power. Snappy as hell.... hmmmm

OptimusG4 · Sep 20, 2003, 02:57 PM

Originally posted by VEGAN:
Albert Einstein -- Two things are infinite: the universe and human stupidity; and I'm not sure about the universe.

Esp. since he said Jaguar was 32-bit and Panther was 64-bit. Passing down a free G5 to get an iMac G4, well I have to say that you my friend are intelligent. Way to go.

bartman00 · Sep 20, 2003, 09:11 PM

Lay off guys.. I meen after his extensive testing of emailing, surfing porn, and holding down the W key in Word found the G5 to be almost the same speed as his imac who can blam him for saving the company some coin?!

hunkhuang · Sep 21, 2003, 12:06 AM

don't say like that, pal...
We were just testing the speed whether is it really work well because it is so expensive.

We returned it for the check-up and fixing.. my boss will add somemore RAM for me. that's all.. we stil need G5 as we are doing visual effects stuffs for movies and commercial.

Thanks !!

Adam Betts · Sep 21, 2003, 12:53 AM

Originally posted by hunkhuang:
don't say like that, pal...
We were just testing the speed whether is it really work well because it is so expensive.

You have to be joking..

G5 IS FRICKING CHEAP COMPARING TO MOST HI-PERFORMANCE COMPUTER!

YOU CAN'T FIND A BETTER COMPUTER WITH EXCELLENT PRICE!

I agree 100% with what vod[k]a quoted. It works so well in this thread.

hunkhuang · Sep 21, 2003, 12:58 AM

Yeah it's pretty cheap.. but not in Asia Countries.

jock · Sep 22, 2003, 06:03 AM

hunkhuang,
There are numerous sites that testify to the speed of the new G5, go and have a look as on an ROI level, you will not find ANYTHING that comes close. As to the speed of Panther I can say that it made a huge difference to the two machines that I have seen it on, one a G4 powerbook 12inch and the other was a G4 1.42, they really were like new machines. I use a G4 1.42 on Jag for final cut on a near daily basis, and running Panther just blew it away.

philski · Sep 28, 2003, 11:08 PM

hunkhuang isn't lying. however there are some caveats that need to be stated in order to justify the actions taken.

the g5 that was kindly lent to us for testing was a preproduction model that had seen a lot of activity around the region and was badly damaged from a recent trip to india. now the state of the case was in no way held responsible for the performance of the box. in another thread i noticed that there was mention of performance deterioration if the plastic window was removed. this could have been a factor - there was no plastic window on this model. also, as previously stated, this was a preprod model and that in itself could have biased the performance. however, these were not the determining factors for our rejection of that particular model.

the tests were based on the following:
a short shake script with an output of only eighteen frames.
no custom plugins, just regular flavor shake effects.
media and output to reside on the local disk.
(a second pass where media and output resided on the xserve raid)
the shake script to be run via the terminal as opposed to the more lengthy display via the flipbook.

the test was conducted on the following.
a dual 1.4 ghz g4, 1gb ram, 120gb drive, os 10.2.6
a dual 1.7ghz xeon, 1gb ram, 120gb drive, win2k
(also tested using media from network drive)
external dual p3 renderers, network drive (ultra scsi raid), win2k
preprod dual 2ghz g5, 1gb ram, 120gb drive, os 10.2.7
shake 2.5.1 / tremor 2.5.1

the conclusions that we drew from this test, which is by no means meant as an exhaustive test, were that the g5 was faster in most instances but not as fast as we had hoped and been led to believe they would be. they are however indicative of the kind of performance benchmarks that would most likely affect us as this is the kind of work we do - very little biomechanics or genetic decoding going on here!

the g5 only managed to perform 25% better than the g4. it was only 20% faster than the xeon.
the dual p3s were the slowest performers of all but that was expected.

the factors that we decided could not be accounted for were operating sytem and software version. we hope that the new 10.3 release will play a significant part in acclerating the results of our simple homebrew test. the other clincher will be a new optimised version of shake. again, this will be somewhat inconclusive as apple no longer release shake for windows and tremor is a legacy product in its own right. also, we have no way of benchmarking a linux flavor of shake in house as we have no linux installations and no licenses. there are other sources that can testify to the speed of linux over mac os or windows. again these are inconclusive to us as we run no linux and plan to drop windows support in favor of a mac os and irix based compositing environment.

the most interesting aspect of our tests involved the xserve raid which is a truly remarkable storage device. tests indicated that there was no performance deterioration between locally stored media and xserve raid media. in fact, standard software indicated that the read write performance was double that of a homebrew ultra160 raid (four 15000rpm cheetahs and an atto ultra160 card). again, it would have been a good idea to test against an ultra320 setup but it wasn't available...
we tested two raid sizes on two preprod xserve raids. a 320gb raid5 and 1.2tb raid5. the 1.2tb raid was, as expected, marginally faster. both drives were tested on one channel. we look forward to implementing a 2tb, dual channel xserve raid. for anyone with enough to splurge, i at least recommend the purchase.

we are now awaiting delivery of a production dual 2ghz g5 and anxiously await a release of panther and a revised version of shake. the plan is to implement four dual g5 compositing workstations and an xserve raid running through a fibre switch (probably a brocade). we will insist on more conclusive testing as and when the opportunity arises. for now, it would seem at first glance, that the g5 is unwarranted expenditure. the xserve raid already plugs into our pipeline and will be implemented soon. it is a big blow that the g5's performance was not as any of us had hoped but as most apple fans, we will wait for it to get better...

kindest regards

phil man
international pixel bitch and professional nomad