Dual 533 Mhz vs 667 DVI TiBook

Re: Dual 533 Mhz vs 667 DVI TiBook

Originally posted by tpjunkie
I went home from college for the weekend last week, and was totally floored by the fact that my dual 533 Mhz digital audio G4 tower ran OS X markedly better than my 667 DVI ti book, which is no slouch, and as far as I had used it at school, never had any speed issues, even when running photoshop, or imovie. But when I got home and actually used the old dualie, everything just seemed faster, more fluid and responsive. They have the same bus speeds, the dual 533 does have 384 MB RAM whereas the tibook has 256. The tibook has 1 MB L3 cache that the tower doesn't, and the tower still has the stock geforce 2 MX video card in it. Anyway, I was wondering if anyone out there with a 1 Ghz tibook (or the 17 albook if its shipped by now) has any comparisons to either the 500 mhz or 533 dualies, to see how the latest portables stack up against some of the older g4s out there.

I really wasn't aware of how much the dual processors made a difference until this, and i was floored that such an old machine (well comparitvely, its 2 years) dominated a six month old machine.

Click to expand...

The other performance factor between your Powermac and Powerbook is that the Powermac has faster disks in it.

A 1 GHz Mac should be as quick as a Dual 533 if not faster as long as the disk accesss speed is the same. And if the system bus is at the same speed.

A faster system bus and a faster disk can make a slower Processor seem faster when compared to a faster processor that has slower system bus and disks.

Also, it depends on what application(s) you run. However,in general with Mac OS X a Dual processor at a slightly slower speed will always win out if everything else is the same (cache, disk speed, system bus speed)

Re: Re: Dual 533 Mhz vs 667 DVI TiBook

Originally posted by Bear

A 1 GHz Mac should be as quick as a Dual 533 if not faster as long as the disk accesss speed is the same. And if the system bus is at the same speed.

Click to expand...

a 1Ghz will smoke the 533 Dual in single processor tasks.

i wouldnt be surprised at all with the comparisson of the ti and the dual. dont forget the graphics card. that speeds up performance with quartz extream. your power book can surely use more ram though. especially if your running large apps like photoshop.

Single PM 733 vs. Dual PM 533

Another nice comparison: Single PM 733 (QS) and Dual PM 533 (Digital Audio).
Both same grfx card (GeForce 2 MX), both 133 bus, both 40 GB 7200 rpm disk, both 512 MB RAM.
Using Aqua, i.e. browsing the HD and using some graphical features in the finder, the Dual 533 seems (almost) twice as fast! Playing Nascar the dual 533 actually gets one-and-a-half times the framerate using the exact same graphics options. Using single proc apps, I hardly could spot any difference in speed, probably because even if the app only uses one processor, OS X tries to maximaize load ballance anyway...
Mac OS 9 is a totaly different matter, ofcourse.

Good to know that Mac OS X really does use the two procs to its maximum

tpjunkie:

my dual 533 Mhz digital audio G4 tower ran OS X markedly better than my 667 DVI ti book

Click to expand...

Of course the 533 is faster... it's the old G4 core with the short pipe. Those things kicked butt. The new G4 was as weak compared to them as a P4 was compared to a P3, but just like happened with the P4, eventially the new G4's clocked out of reach and were faster overall... however any new-core G4 at low clock is still easy pickings for an old-core G4 (in many but not all things).

The tibook has 1 MB L3 cache that the tower doesn't

Click to expand...

The tower has 1MB of L2 per chip, and this backside L2 is essentially identical to the "baskside" L3 on the laptop. They are both more or less the same type of memory and more or less the same speed. The fact at one is L3 does not make it better.

Originally posted by ddtlm
tpjunkie:


Of course the 533 is faster... it's the old G4 core with the short pipe. Those things kicked butt. The new G4 was as weak compared to them as a P4 was compared to a P3, but just like happened with the P4, eventially the new G4's clocked out of reach and were faster overall... however any new-core G4 at low clock is still easy pickings for an old-core G4 (in many but not all things).


The tower has 1MB of L2 per chip, and this backside L2 is essentially identical to the "baskside" L3 on the laptop. They are both more or less the same type of memory and more or less the same speed. The fact at one is L3 does not make it better.

Click to expand...

Eh. The fact that it is L3 means that there is ALSO L2, and that is definitely better. Also, the 7400/7410 did NOT have a big advantage over the longer pipelined 744x and 745x processors except on tasks that had huge numbers of hard to predict branches (where the longer pipeline hurts the later processors) or were dependent on cache size (and not dependent on cache latency), since the earlier processors had larger off chip L2 caches, while the later ones had small on chip caches. Certainly they got slightly better IPC, but the slower caches and worse Altivec dispatch scheme more than made up for that slight gain. I would say that comparing the 7410 to the P3 and the 7450 to the P4 is quite an exaggeration (especially the Willamette P4s, which just sucked).

Catfish_Man:

I stand by my claims. I've seen plenty of benchmarks showing the old G4 being much faster per clock, and at only 25% higher clock speed the new one is often going to find itself in trouble.

processors except on tasks that had huge numbers of hard to predict branches

Click to expand...

And on poorly optimized code where the short and stout pipeline made the old G4 much more tolerant of data hazards as well as heavy branching.

I would say that comparing the 7410 to the P3 and the 7450 to the P4 is quite an exaggeration (especially the Willamette P4s, which just sucked).

Click to expand...

And I would say that its not. Even the orginal P4's had plenty of things they could do faster than P3's.

The HUGE bloat of Quartz sucks the performance out of single cpu G4s, also the 667Mhz G4 (745x) isn't as efficient as the 533Mhz G4 (7410).

Think about it for a second, all Quartz extreme does is map the windows and desktop onto a polygon as textures and then fire it to the screen, that's the final stage of the rendering process. PDF, the dog slow vector renderer ( essentially a RIP) apple chose to use is going to eat cpu time for breakfast. Just look at how acrobat reader renders even simple, 72 dpi pages of images and text, you can't say apple have some miraculous optimisations that are going to make that any faster than it in already is, adobe have had 5 versions of acrobat so far, the reader is getting slower at rendering pages, not faster!

A dual cpu model can palm off the whole quartz engine aswell as running the other elements of the GUI like reading the names, sizes and filetypes of files as you open a folder etc... on the second cpu, leaving 100% of the first cpu purely and the remaining processing power of the second cpu for running you're applications.

A single cpu on the otherhand has to do all this is at once and even the fastest graphics card in the world isn't going to accelerate PDF rendering because it's 100% cpu bound.

On top of this we have the comparison of the CPUs

TiBook 667:

7 stage pipeline

2 x 32K L1
256K 1:1 ratio on chip L2
1Mb 4:1 ratio backside L3 (or not if it's the one from the 550,667Mhz range)

Dual 533Mhz G4

4 stage pipeline
2 x 32K L1
1Mb 2:1 ratio backside L3

not mention 2 cpus obviously.

Re: good feedback

Originally posted by tpjunkie
it's good to hear about your comparison macsrg8, i wish apple would hype the performance boost of the dual processors more than they do, although i guess it would be hard to compare, seeing as they don't make single processor models of the two higher speed power macs...

Click to expand...

Indeed.
And by hyping the performance of the duals, the iMac and single G4 tower would be "downgraded" too much.... (which is true, in my opinion)