Apple Logic Pro /LogicExpress Discussion group photo

Yahoo Groups archive

Apple Logic Pro /LogicExpress Discussion

Index last updated: 2026-04-28 23:06 UTC

Thread

Twice the Speed? / Imac option?

Twice the Speed? / Imac option?

2005-02-02 by GAmoore@aol.com

In a message dated 2/2/05 7:03:42 AM, kamm@... writes:
> In the case of a photoshop filter, if the task is divided
> half for each processor, it can be pretty close to twice as
> fast. That "parallel processing" - which we will be seeing
> more and more.
The Apple website shows the single CPU 1.8 being 20% faster than baseline
and the Dual CPU being 66% faster than baseline. That isn't even close to
doubling the speed over the single CPU version and it never will be.


Hey we're both right. For mundane tasks, the dual 1.8 is probably exactly the same
speed as the single 1.8. It won't open a file faster, or redraw the screen faster, or
respond to a mouse click any faster at all. But when it comes to processor-demanding
task, thats where the 2nd processor comes into play In the case of a Photoshop file,
its fairly easy to divide the task into two or more pieces to do separately. With audio
I'm not so sure. Maybe bounces are similar. Or maybe if its smart, it will put some
soft synths and plugs on each processor.

Suppose you invite a friend to help you cook dinner - to bake a cake and wait an hour
while its in the oven is going to save any time. But if you were making some complicated
dishes, maybe the friend cuts some vegetables while you make a soup, and so forth, and
you can both access the kitchen counter when you need it. Depending on the dish it
might be almost the same as by yourself or maybe almost twice as fast.



The data is coming from one hard drive, that hard drive is connected to a
single PCI bus (correct me if I'm wrong)


I thought the PCI bus is separate from the hard disk bus - unless you put a SATA or
SCSI board in a PCI slot. By the way, the 2.0 and 2.5 come with PCI-X which is 5 or
6 times faster - but evidently few hardware manufacturers are taking advantage of
this. Concievably you could have some very fast disk access with a PCI-X board
I am guessing.


that can only be accessed by one
CPU or the other. Even if both CPUs could access it at the same time, it
could only respond to one at a time (unless they were both requesting the
exact same data).




The whole PowerPC architecture going back to the "G1" wich was the 601 chip
was RISC (reduced instruction set computing) - which means fewer computer
commands but done more quickly - so it might take 10 clock cycles to do something
that would have been 1 clock cycle with the Motorola 680X0 series (68000-68040).
So there is likely available clock cycles to grab a chunk of ram to start a sample, read
the next bar of midi, access another piece of the program.




The same is true of Ethernet, Audio cards, video cards and
so on. Getting the data is a big bottle neck. Hard drives are performance
dogs compared to RAM.


Good point, but there are two meters in Logic - one for disk io and the other for processor load.
Things like reverbs and soff synths have little disk requirements but big processing requirements.


> No powercore, no scsi, no graphics board upgrade, the lack of
> upgradability means you need to replace it sooner. Apple
> usually hobbles these in various ways.

Yes, to all of the above. That's were I have to decide if it is worth an
extra $1000 to get the PowerMac because it costs $999 to get that 20"
display.


I couldn't live without my 22 inch cinema and 2nd 15 in monitor.


If I put that $999 in the money market, I might have enough money
to buy another Mac in a couple of years or I could spend it on a really nice
I/O device. Right now all I have is a TASCAM US-122. Which has only two
inputs and two outputs.



I think thats a good way to look at it. With the rapid decline in computer equipment,
its best to get the minimum system that will suit your needs. If we were to follow
our emotions, the wishes of the companies and the magazine articles we all go out
and have all the newest stuff.. and be broke or if not, be wasting so much money
that you could retire or put a downpayment on a house.

I have owned a dozen or so macs over the years, and I used to be snobbish about
using the higher grade stuff - but when my titanium book turned out to be crap,
and I bought a G3 ibook (which I am typing on at this moment) which 1/3 the price
and been a pretty good machine, I changed my opinion.

Maybe you could do a little test. Create a logic song with about 20 instances of the
softsynths in Logic Express and 20 audio track files. Put them on a CD, and run them
in the store on the Imac and the Powermac, and check the load meters. In the apple
store at least, they have logic express (I saw it one a few days ago). If they are
about the same, maybe the imac is the thing for you - since its going to save you the
cost of a monitor also.

RE: [Logic_Cafe] Twice the Speed? / Imac option?

2005-02-02 by Kamm Schreiner

> Hey we're both right. For mundane tasks, the dual 1.8 is 
> probably exactly the same speed as the single 1.8. It won't 
> open a file faster, or redraw the screen faster, or respond 
> to a mouse click any faster at all. But when it comes to 
> processor-demanding task, thats where the 2nd processor comes 
> into play In the case of a Photoshop file, its fairly easy to 
> divide the task into two or more pieces to do separately. 
> With audio I'm not so sure. Maybe bounces are similar. Or 
> maybe if its smart, it will put some soft synths and plugs on 
> each processor.
> 
> Suppose you invite a friend to help you cook dinner - to bake 
> a cake and wait an hour while its in the oven is going to 
> save any time. But if you were making some complicated 
> dishes, maybe the friend cuts some vegetables while you make 
> a soup, and so forth, and you can both access the kitchen 
> counter when you need it. Depending on the dish it might be 
> almost the same as by yourself or maybe almost twice as fast.

I do think we are more or less on the same page, but I think when you are
describing getting 2x performance out of the 2 CPUs, you are describing a
technical Nirvana that simply doesn't exist in reality. If those two CPUs
have all of the code in cache and all of the data in cache, then you can get
exactly 2X performance. As soon as you need to access resources of any type,
RAM included, the performance goes down substantially. In the case of RAM,
yes, technically the CPU could be doing something worthwhile while the other
CPU is accessing memory, but it is unlikely in the extreme. The memory runs
at 400 MHz and it is CL 3 RAM which means it takes three clock cycles from
request to data available. If the 2nd CPU needs to access the RAM at that
same time, it has to wait for those full 3 clock ticks. Could it do
something else while waiting? Probably not. The decision to do something
else would, itself, consume CPU time and I can't imagine it doing anything
other than waiting. Finally, I suspect that memory accesses are likely to
come in bursts rather than a word at a time, for instance, when applying
effects to an audio track, the code is probably going to work with a chunk
of that audio at a time - 16 words, 32 words, maybe 64 words - but something
greater than a single word. Once loaded the CPU can likely perform its task
without accessing memory again until the next chunk of audio is needed.

Anyway, yes, in theory, I see your point. However, in practice, I don't
think it will ever be realized.

Kamm

Re: [Logic_Cafe] Twice the Speed? / Imac option?

2005-02-02 by GAmoore@aol.com


In a message dated 2/2/05 10:05:57 AM, kamm@... writes:


CL 3 RAM which means it takes three clock cycles from
request to data available. If the 2nd CPU needs to access the RAM at that
same time, it has to wait for those full 3 clock ticks. Could it do
something else while waiting? Probably not. The decision to do something
else would, itself, consume CPU time and I can't imagine it doing anything
other than waiting.


I think all modern CPUs are "pipeline" machines. A command doesn't just execute. It goes through 10 or 12 stages, so at any given time 10 or 12 instructions are being processed. A memory call can be made on the first cycle and read on the 4th so its not a waste of time.

RE: [Logic_Cafe] Twice the Speed? / Imac option?

2005-02-02 by Kamm Schreiner

> I think all modern CPUs are "pipeline" machines. A command 
> doesn't just execute. It goes through 10 or 12 stages, so at 
> any given time 10 or 12 instructions are being processed. A 
> memory call can be made on the first cycle and read on the 
> 4th so its not a waste of time. 

Yes. Absolutely, and your described situation can occur. However, the
pipeline in no ways *ensures* that will be the case. It only makes it much
more likely. There can be code dependencies that prevent it. In other words.
Even though the CPU is capable of performing several pipelined operations
after requesting a particular byte of memory. If the results of that memory
must be available for the currently pipelined code to go on to the next
instruction, the pipeline is potentially held up.

I'm definitely not a CPU expert, and not a pipeline architecture expert by
any means either. However, I think I can safely say that if two independent
CPUs are trying to access the same memory and that memory is running on a
slower bus, it is ultimately going to be a bottle neck. There are modern CPU
tricks that reduce the effects of the bottle neck, as you have pointed out,
but you simply can't get rid of it altogether.

Kamm

Re: [Logic_Cafe] Twice the Speed? / Imac option?

2005-02-03 by Dave Shirk

On Feb 2, 2005, at 2:22 PM, Kamm Schreiner wrote:

> I'm definitely not a CPU expert, and not a pipeline architecture 
> expert by
> any means either. However, I think I can safely say that if two 
> independent
> CPUs are trying to access the same memory and that memory is running 
> on a
> slower bus, it is ultimately going to be a bottle neck. There are 
> modern CPU
> tricks that reduce the effects of the bottle neck, as you have pointed 
> out,
> but you simply can't get rid of it altogether.
>
> Kamm
>
>

Kamm!

	I am (or was) a CPU expert.  And you are right - the path to the memory
is still a bottleneck.  But so is the path to the video card, the Hard 
drives,
the Firewire ports - etc.

	The dual G5 architecture is Extremely Elegant!  No other way to 
describe
it!  It is as efficient as current (or one year old) technology can do.
In the real world - this type of architecture should average out to 
about
a 30% improvement over a single processor - because of all the
other bottlenecks.  All things considered - it is an amazing improvement
over the past systems - and will only get better!

Just my 2 cents worth!

Sorry for the off topic post

Dave Shirk

Move to quarantaine

This moves the raw source file on disk only. The archive index is not changed automatically, so you still need to run a manual refresh afterward.