[urrl5201]

My first stop today was at the CompUsa store where I rendered a 3 second dissolve in iMovie with the old 667. I was amazed to find the render was done in just 19 seconds, about 10 seconds less than my 500 Mhz. That's about 30 percent faster, quite remarkable. This is something owners of the old 667 should be very proud of.

At the Kendall Apple store I found the 800 and new 667 making only marginal improvements, nothing for me to do a backflip over, wrap one up and take one home just yet. The new 667 got 18 seconds and the 800 got 17 seconds. I know I know, a crude and unprofessional test at best, but it was enough for me to walk out of the store with the mind to be more patiently careful and wait for other independent test results from other more dependable organizations before making a purchase.
Even so the old 667 and new models ran 30 percent faster or better than my old 500, well worth an upgrade in my book. Right now for me the new 667 seems the most appealing choice, till better test results roll in. That old 667 at CompUsa cost about $300 more than what Apple sells the new 667 for. Those guys need to wake up and smell the coffee.

It would be interesting to see what the 550 owners are getting. Perform the test in this way:

Open the tutorial in iMovie.
Drag down two tutorial scenes to the storyboard.
Set the transition to dissolve.
Set the dissolve to 3 seconds.
Time how long it takes the dissolve to render.

[skyman]

Originally posted by urrl5201:
<STRONG>My first stop today was at the CompUsa store where I rendered a 3 second dissolve in iMovie with the old 667. I was amazed to find the render was done in just 19 seconds, about 10 seconds less than my 500 Mhz. That's about 30 percent faster, quite remarkable. This is something owners of the old 667 should be very proud of.

At the Kendall Apple store I found the 800 and new 667 making only marginal improvements, nothing for me to do a backflip over, wrap one up and take one home just yet. The new 667 got 18 seconds and the 800 got 17 seconds. I know I know, a crude and unprofessional test at best, but it was enough for me to walk out of the store with the mind to be more patiently careful and wait for other independent test results from other more dependable organizations before making a purchase.
Even so the old 667 and new models ran 30 percent faster or better than my old 500, well worth an upgrade in my book. Right now for me the new 667 seems the most appealing choice, till better test results roll in. That old 667 at CompUsa cost about $300 more than what Apple sells the new 667 for. Those guys need to wake up and smell the coffee.

It would be interesting to see what the 550 owners are getting. Perform the test in this way:

Open the tutorial in iMovie.
Drag down two tutorial scenes to the storyboard.
Set the transition to dissolve.
Set the dissolve to 3 seconds.
Time how long it takes the dissolve to render.</STRONG>

Thanks for the info. It would have been cool to see how the old 667 would do against the new 667 in a longer test.

The new 667 CPU bandwith is about 50% - 60% faster then the old 667. So, your test surprise me as the new 667 should have performed the dissolve much faster. 1 second is only about a 5% increase in speed which I doubt. Even the 800 only completing the test in 2 seconds less does not make much sense. But then again, the test was only a short time frame.

Any how, thanks again for the info.

[Sean Li]

I tried out the 800MHz model on Saturday; Didn't do any "formal" benchmarking, but there didn't seem to be the increase in speed I was expecting from the old 667MHz 'book I'm used to.

I also noticed a lot more heat coming from the 800 than I expected.

As urrl5201 said, it convinced me to wait for the next revision / upgrade / model.

[phobos]

Calm down guys. 5 minutes with a machine on a store can't be much help.
You don't know what the people at the stores have installed on the poor machines. You don't even (I'm pretty sure of it) have the same conditions of the 2 machines on the 2 DIFFERENT stores. So let's just wait for some official benchmarks.
And when you don't feel the machine snappy doesn't mean it isn't.
Some benchmarks on the 800model of cinema 4d (on some thread here.I asked this question) shows that powerbook800 is a lot faster than the previous models.So AGAIN let's wait for some official benchmarks.

[maffioso]

Finally some one with some common sense!

Any ideas on who will get the first benchmarks out?

Please post them here when they come out!

Thanks

[urrl5201]

"it was enough for me to walk out of the store with the mind to be more patiently careful and wait for other independent test results"

Funny, I thought I was the first one to say that. Sean Li and I are probably more cured of "buy now" fever than anything else. Your implication that neither of us has any common sense is not well taken, Chris. We may be laptop fanatics but that's beside the point.

[ 05-06-2002: Message edited by: urrl5201 ]

[maffioso]

Originally posted by urrl5201:
<STRONG>"it was enough for me to walk out of the store with the mind to be more patiently careful and wait for other independent test results"

Funny, I thought I was the first one to say that. Sean Li and I are probably more cured of "buy now" fever than anything else. Your implication that neither of us has any common sense is not well taken, Chris. We may be laptop fanatics but that's beside the point.

[ 05-06-2002: Message edited by: urrl5201 ]</STRONG>

Well as they say, you cant judge a book by its cover, and spending 5 minutes on a laptop is hardly looking into it. I would be interested to see some realworld benchmarks, from people who actually own a new PB 800Mhz, and well, about the waiting thing, I had a PB 667Mhz on order in January(In Australia), but they couldn't ship so I have waited, and well, I waited the next big thing, but Apple won't touch this laptop in any big way for at least 6 months, the only thing i can see being added is bluetooth. I have no use for bluetooth so this machine is perfect...

Any way I hope you find a laptop that fulfills your needs, even if it isn't an Apple.

[Sean Li]

Originally posted by phobos:
<STRONG>Calm down guys. 5 minutes with a machine on a store can't be much help.
You don't know what the people at the stores have installed on the poor machines. You don't even (I'm pretty sure of it) have the same conditions of the 2 machines on the 2 DIFFERENT stores. So let's just wait for some official benchmarks.
And when you don't feel the machine snappy doesn't mean it isn't.
Some benchmarks on the 800model of cinema 4d (on some thread here.I asked this question) shows that powerbook800 is a lot faster than the previous models.So AGAIN let's wait for some official benchmarks.</STRONG>

I agree that the machines may have been configured differently, however I spent some 45 minutes on the 800MHz model trying out different applications, web browsing, installing Return to Castle Wolfenstein and playing it, etc...

I'm a stronger believer in trying machines out myself than I am in benchmarks. IMHO, it wasn't as fast as I expected, and it was much hotter particularly in the corners. Benchmarks don't necessarily indicate real-world usability. Ever heard of the expression "there's lies, d*mn lies, and then there's statistics"?

Having said that, there is no denying that the Radeon 7500 has a huge impact on graphics intensive applications. It won't make much difference listening to music through iTunes, but the framerate after changing the settings so it's not capped at 30fps and making it as fast as possible is significantly faster than on the older 667MHz.

It's funny; when someone posts a report saying that it's really fast, runs very cool, blows anything else out of the water, no one asks them to do official benchmarks. Go figure

[urrl5201]

My 500 is fulfilling my needs. It took a look at the new models to truly realize that. I can judge my 500 Mhz Powerbook by its cover as well as it's internals, since I have worked with it for a year and found it to be the best laptop I have ever owned. I can see for me to upgrade makes little sense, indeed, no sense at all. I am not about to divorce her for the next quicker honey that comes along.
My 500 may take a bit longer, but she still does the job just as well and looks just as good while doing it.

But it's good we had this little chat; it helped me reconsider my buying temptations and save a lot of money, which I can spend on more practical things, like a DVD burner to finally let my wedding customers appreciate the quality of DV instead of the second rate quality of VHS tape.

Oh, and thanks again for helping me with my decision.

[ 05-06-2002: Message edited by: urrl5201 ]

[BRussell]

On a heavy Altivec task, like I imagine that dissolve would be, I'd guess you wouldn't see much improvement. But due to the L3 cache, I'd guess you'd see a more universal speed improvement than just 5%.

[urrl5201]

Actually I think you guessed right.
Like I say, it was about a 30 percent improvement. My 500 took about 29 seconds. I tested the 667's over and over and they took about 19 seconds. If anyone doubts this try it.

Go to Apple store.
Open iMovie tutorial
drag two scenes into story board
set dissolve for 3 seconds
use your wristwatch to time the dissolve render.

I am wondering is my math right however. Going from 29 to 19...is that a 30 percent or 60 percent improvement? In any case either one is a great and obvious improvement over the 500 Mhz.

I was wondering where skyman got only a 5% increase conclusion from my procedure and data, as he posted concerning 1 second, not a 3 second dissolve time. Perhaps he was refering to the old 667 time to the new 667 time which time difference was only about 1 second (18 sec vs 19). In that case I expect he would be correct.

And in spite of my method being rather elementary, and what some would have you believe, I still consider it a VERY common sense approach for me, at least until more official tests come in, since I deal mainly in video editing. All else is second priority with me.

[ 05-06-2002: Message edited by: urrl5201 ]

[phobos]

Hey guys relax!!!!!!
It's not a trivial matter like war or something....
Benchmarks in my opinion will not be baked on sites like barefeats cause it isn't linked in apple someway.
Benchmarks like the cinema4d test are the best and most real world as benchmarks can be.They test the CPU and GPU.
Also some good tests would be some photoshop and some tough games like quake.
Besides the benchmarks I'll definetely buy the new powerbook (when I have the money that is). It has all the improvements that I want and I definetely wont look back.
For one more time I managed to turn the topic off subject!!!!! Congratulate me please

[dialo]

Originally posted by urrl5201:
<STRONG>I still consider it a VERY common sense approach for me, at least until more official tests come in, since I deal mainly in video editing. All else is second priority with me.

[ 05-06-2002: Message edited by: urrl5201 ]</STRONG>

That's really interesting, considering that you are talking about video editing on a laptop. With Final Cut 3 and Avid Express DV3, it seems to me that the thing I would want most is native realtime effects. Rendering can wait until I'm done.

Why does this apply to your test? Ever notice that you can get more realtime effects with a ti500? If you don't believe me, open the rt script and see for yourself.

I'm not saying you have the same needs as me, but it just shows that the rev a/b powerbooks have their own uses.

Personally, I don't find much use for the new ones since I care about audio and realtime video editing. The rev b pbs stink at both of these.

[skyman]

Originally posted by urrl5201:
<STRONG>
I was wondering where skyman got only a 5% increase conclusion from my procedure and data, as he posted concerning 1 second, not a 3 second dissolve time. Perhaps he was refering to the old 667 time to the new 667 time which time difference was only about 1 second (18 sec vs 19). In that case I expect he would be correct.

[ 05-06-2002: Message edited by: urrl5201 ]</STRONG>

That is exactly what I was referring to. You got it right.

I am of the mind that you can take real world performance and benchmarks and average the two to get a "true" level of performance. So, while I believe that your tests are accurate, I also believe that the new 667 is significantly faster then the old 667.

We will just have to wait for more real world performance reports and benchmarks.

[ 05-06-2002: Message edited by: skyman ]

[urrl5201]

Realtime works with the 500? That's incredible. Unfortunately all I have is iMovie, so I can't apply what you are asking to my test. But thanks for the tip!!! Perhaps I WILL keep the 500 and perhaps invest in FCP3.

[xtro]

Originally posted by dialo:
<STRONG>Why does this apply to your test? Ever notice that you can get more realtime effects with a ti500? If you don't believe me, open the rt script and see for yourself.</STRONG>

Hmm, you are either lying or don't know what you are talking about. The newer 550 MHz Ti Books can't even do real time effects, so a 500 certainly could not do them.

[BRussell]

Hi.

[ 05-06-2002: Message edited by: BRussell ]

[murbot]

Hey, how are ya.

[dialo]

Originally posted by xtro:
<STRONG>

Hmm, you are either lying or don't know what you are talking about. The newer 550 MHz Ti Books can't even do real time effects, so a 500 certainly could not do them.</STRONG>

Wow, chief, you really missed the point. The mhz is not what matters here at all. Cache size and memory bandwidth. Like I said, if you don't believe me, check it out for yourself by putting them side by side, or you could do it the easy way by looking @ the rt script in fcp3 to see the cpu rating that apple itself gives it. It's posted on the web and not hard to find. Google exists for a reason.

Also, out of curiosity I did the imovie test last night on my ti500 and got 25 secs. (once 26 secs.) This isn't a huge difference, but it cuts the difference down almost 40%. You might just have something running on your machine that is holding it down.

[ 05-07-2002: Message edited by: dialo ]

[xtro]

I don't think I did. I am just saying your claims of "real time effects" are incorrect.

Ever notice that you can get more realtime effects with a ti500? &lt;snip&gt;....since I care about audio and realtime video editing.

When the PowerBook line was 550/667, Apple made it very clear that you needed the 667 to get real-time editing. The 550 doesn't have the horsepower (or cache, or memory bandwith, or whatever - it's a moot point really).

I haven't gotten into a discussion about why, I'm just stating that it can't be done on the 500. Certainly doesn't mean it's slow for video editing, that's not my point.

[DStrz]

After buying a open box Titanium 667/DVD on April 22, I returned it and purchased the new G4 Powerbook on April 29th from the Apple store in the Mall of America.

The fan only came on to the low setting. The previous unit cycled on within the first hour of use and sounded like a hoover when it kicked on to high while using Imovie.

The CD mechanism is much smoother. Perhaps the combo drives in the last build were similar. The DVD drive on the previous unit was noisy.

The graphics are very fast and sharp. So is the general feel. I did some comparisons to arrive at the decision to return the open box unit:

Open Box 667DVD: 512 Ram, DVD Drive, 30MB HDD, 6 mo warranty, scratches on the bottom, missing S-video cable, 16MB vRam, smaller screen real estate, 50W battery, Airport Card, $2169+tax

New 667DVI: 256 Ram, CDRW/DVD Drive, 30 MB HDD, 12 mo warranty, brand new clean, 32 MB Vram with brighter screen, Higher resolution screen, 55W battery. $2349+tax

Using GaugePRO, the memory bandwidth is ~250 MB/sec. The previous 667 was doing ~145MB/sec.

I can certainly feel the responsiveness and notice the brighter screen.

[dialo]

Originally posted by xtro:
<STRONG>I don't think I did. I am just saying your claims of "real time effects" are incorrect.



When the PowerBook line was 550/667, Apple made it very clear that you needed the 667 to get real-time editing. The 550 doesn't have the horsepower (or cache, or memory bandwith, or whatever - it's a moot point really).

I haven't gotten into a discussion about why, I'm just stating that it can't be done on the 500. Certainly doesn't mean it's slow for video editing, that's not my point.</STRONG>

OK



Um...it's not my claim, it's Apple's. Why don't you look into it before saying misleading things on this forum.

Disclaimer: The only reason I'm providing this explination is for the accurate info of others on this forum, esspecially those considering spending $1000 on fcp3.

Essentially, each effect has a certain cost on the cpu. The cost of each effect is listed in the RT script. The cpu class number denoted the maximum that the costs of the effects can add up to be before it stops playing in realtime.
Direct from the script:

if (dictget("switches","realtimeVideoFX","cpuclass"," value") == 42)
cpuclass = 42;
else if (cpucount &gt; 1 && cpuspeed &gt;= 800)
cpuclass = 4
else if (cpucount &gt; 1)
cpuclass = 3
else if ((cpuspeed &gt;= 500 && (l2cachesize + l3cachesize) &gt;= 1024) &#0124;&#0124; (cpuspeed &gt;= 733))
cpuclass = 3
else if (cpuspeed &gt;= 667)
cpuclass = 2
else if (cpuspeed &gt;= 550)
cpuclass = 1
else
cpuclass = 0
end if

So, as xlr8yourmac explains:

The different Mac's gets the following cpuclass tags:

CPU speed 0: If you got a G3 or a G4 below 550.

CPU speed 1: If you got a single G4 550 or above.

CPU speed 2: If you got a single G4 667 or above.

CPU speed 3: If your G4 is 500 and above AND your L2+L3 cache is 1 meg or above. If you got a single G4 733 or above. If you got more than 2 G4 Cpu's.

CPU speed 4: If there is more than 2 G4 Cpu's which run at, or above, 800 mHz

Since the ti500 has a L2 cache size of 1 MB, it gets a 3. That means that it can get 50% more realtime effects than the ti667, which can only play back effects whose combined cost adds up to 2.

Of course, apple was not going to vocalize to the world that the newer machines couldn't pull as much as the high-end rev a model; they sell more computers if you buy them in ignorance. But the company was quite obviously aware of the limitations on the newer machines.

xlr8yourmac also has an article on how the newer pbs perform more poorly in audio than the ti500. I, being a logic user, have actually performed the same tests and come up with similar results showing the inferiority of the 550/667.

I am really curious what the results for the newer machines are since they now have a cache comparable to the older powerbooks and the memory throughput issue seems to have been resolved. They will probably in line with the 500, + the extra cycles.

Blind faith, etro, can be a very bad thing. I hope we all learned something today.

[BrunoBruin]

New 667DVI: 256 Ram, CDRW/DVD Drive, 30 MB HDD, 12 mo warranty, brand new clean, 32 MB Vram with brighter screen, Higher resolution screen, 55W battery. $2349+tax

I think you did the right thing, especially as the new machine will support Jaguar and has the combo drive. But how did you get the $2,349 price? Are the stores honoring edu discounts?

[thewoof]

Originally posted by DStrz:
<STRONG>

Using GaugePRO, the memory bandwidth is ~250 MB/sec. The previous 667 was doing ~145MB/sec.

</STRONG>

GaugePro is MacOS9 only and not up to date. it does not recognize correctly the old 667 w/o L3 cache processor and indicates 400Mhz as the processor speed (of course no processor cycling enabled in the energy saving control panel). Can you trust this program as a benchmark for memory bandwidth? Also 145MB/s is not accurate as a mean value. Leave Gaugepro open for a while and you will see the bandwidth peak above 190mb/s.

[DStrz]

The apple store honored an education discount available online. They allow the generic discount that is typically not as much as a purchase made in a university apple store. The discount in the store only applies to systems. To get apple software like FCP3 for $299 , you need to order it online. Other hardware like batteries ($119) are cheaper online with the edu discount, but the shipping eats up the savings.

[BrunoBruin]

The apple store honored an education discount available online.

Thanks! I may try this route as well, I'd still have to pay tax but I could have the Ti in my hot little hands faster.

[urrl5201]

I have purchased a new version 667 and I am downloading files by firewire from my 500 at this moment. No dead pixels on the new version. I do see one pin sized white spot, could be the primer paint from the Ti's thin undercoat of paint. The Ti was incredibly, lovingly warm on display, in fact the top and bottom were. After purchasing my new Ti my salesman got my phone number; it seems he dropped his girlfriends Ti on the floor, it is not worth the cost of repair, and he wants to buy my 500 Mhz, thank God.

[xtro]

Originally posted by dialo:
<STRONG>

OK



Um...it's not my claim, it's Apple's. Why don't you look into it before saying misleading things on this forum.

Disclaimer: The only reason I'm providing this explination is for the accurate info of others on this forum, esspecially those considering spending $1000 on fcp3.

Essentially, each effect has a certain cost on the cpu. The cost of each effect is listed in the RT script. The cpu class number denoted the maximum that the costs of the effects can add up to be before it stops playing in realtime.
Direct from the script:

if (dictget("switches","realtimeVideoFX","cpuclass"," value") == 42)
cpuclass = 42;
else if (cpucount &gt; 1 && cpuspeed &gt;= 800)
cpuclass = 4
else if (cpucount &gt; 1)
cpuclass = 3
else if ((cpuspeed &gt;= 500 && (l2cachesize + l3cachesize) &gt;= 1024) &#0124;&#0124; (cpuspeed &gt;= 733))
cpuclass = 3
else if (cpuspeed &gt;= 667)
cpuclass = 2
else if (cpuspeed &gt;= 550)
cpuclass = 1
else
cpuclass = 0
end if

So, as xlr8yourmac explains:

The different Mac's gets the following cpuclass tags:

CPU speed 0: If you got a G3 or a G4 below 550.

CPU speed 1: If you got a single G4 550 or above.

CPU speed 2: If you got a single G4 667 or above.

CPU speed 3: If your G4 is 500 and above AND your L2+L3 cache is 1 meg or above. If you got a single G4 733 or above. If you got more than 2 G4 Cpu's.

CPU speed 4: If there is more than 2 G4 Cpu's which run at, or above, 800 mHz

Since the ti500 has a L2 cache size of 1 MB, it gets a 3. That means that it can get 50% more realtime effects than the ti667, which can only play back effects whose combined cost adds up to 2.

Of course, apple was not going to vocalize to the world that the newer machines couldn't pull as much as the high-end rev a model; they sell more computers if you buy them in ignorance. But the company was quite obviously aware of the limitations on the newer machines.

xlr8yourmac also has an article on how the newer pbs perform more poorly in audio than the ti500. I, being a logic user, have actually performed the same tests and come up with similar results showing the inferiority of the 550/667.

I am really curious what the results for the newer machines are since they now have a cache comparable to the older powerbooks and the memory throughput issue seems to have been resolved. They will probably in line with the 500, + the extra cycles.

Blind faith, etro, can be a very bad thing. I hope we all learned something today.</STRONG>

bah.