>Joshua Johaneman wrote:
>> Except, don't they use a Reduced instruction set? which is faster for some
>> things and slower for others?

>Actually RISC chips are faster than their equavilent CISC chip, but
>often progams written for a RISC platform require a larger memory foot
>print....or at least this is so as far as Alpha apps compared to Intel
>apps are concerned...
As much as 2x than the Cisc version. But Windows does everything to erase this
advantage, being much much much slower and requiring much much more Ram than ANY other
OS's!!!!

 ___________________________________________________________________
/ Redaelli Paolo EMail: red### [at] incit | Known as Tybor on #Amiga, \
| POV-Team Amiga                        | #AmIrc and #AmigaIta      |
| Porter of AmiForge, GForge, Lparser   \---------------------------+
| Fantasy home page: http://www.geocities.com/Area51/Vault/4950     |
\___________________________________________________________________/

Post a reply to this message

[Redaelli Paolo" <red### [at] incit>]
| The fastest G3 now avaible is TWICE as fast as the FASTEST
| PEntium II 333 Mhz, running PhotoShop!!!!
| Have you any other doubts about FPu ???

Since when was Photoshop a benchmark for FPU performance? Photoshop
uses mainly integer math.

-- 
"The dumber people think you are, the more surprised they're going to be 
	when you kill them"   -- William Clayton

Post a reply to this message

Redaelli Paolo wrote:
> 
> >Mathias Broxvall wrote:
> >Seriously though, you can't compare photoshop on the G3 and PII as
> >photoshop is native to the mac platform...
> ... you're quite wrong. The inner core on Photoshop is written in C or C++,
therefore the Mac and Win version are as fast as they can be being native code (PPC
for Mac, x86 for Win).
> It is a fact that last Intel chips are slower than PPC.

I'm not disputing that Intel chips are slower than PPC, but...

I really have to take exception to using Photoshop overall as a
benchmark. True, the core is most probably C or C++, but that's not the
main factor.

Photoshop was designed around the MacOS and graphics, and it's the basic
assumptions and algorithms that are a major factor. If you follow
optimization threads in comp.graphics.algorithms, you'd see a general
acceptance that changes in algorithms yield more marked performance
differences than low-level C/C++ execution.

One has to consider that the MacOS did not have virtual memory as good
as Windows, so Photoshop does it's own thing. There are also other
things that make Windows Photoshop more of a port to Windows than a
native Windows app.

One way to observe the problem is to run Windows Photoshop 3.x and then
try to navigate any menus with the keyboard. Performance gets blown out
of the water. Symptom of an app that is not working well with the OS.

I just don't trust Photoshop performance as a decent benchmark. And
don't even get me started on the early versions of Director! (or how
about MS Word, 5 or 6 was it, on the Mac).


Can you tell I'm a frustrated Windows Photoshop user? And don't even get
me started on their older 8-bit support for .PCX, .BMP, .TGA...

Post a reply to this message

Redaelli Paolo wrote in message <110### [at] incit>...
>>Mathias Broxvall wrote:
>>Seriously though, you can't compare photoshop on the G3 and PII as
>>photoshop is native to the mac platform...but I will admit that the G3
>>is faster than the PII, as I have seen it render lightwave images faster
>>than an equivalent PII. But if you want real fpu performance at a
>>resonable cost, Dec Alphas seem to be top of the list. And you can get

>... you're quite wrong. The inner core on Photoshop is written in C or C++,
therefore the Mac and Win version are as fast as they can be being native
code (PPC for Mac, x86 for Win).
>It is a fact that last Intel chips are slower than PPC.


As a sometime cross-platform programmer, I can tell you plainly that there
is a great deal you can do to slow down a program outside of the "inner
core", particularly when porting to a different OS.
--
Smo### [at] psnwcom

Post a reply to this message

Redaelli Paolo <red### [at] incit> wrote:
> >Mathias Broxvall wrote:
> >Seriously though, you can't compare photoshop on the G3 and PII as
....

NOT TRUE!!! Please don't misquote me. Check your article again, you
must have gotten the wrong name when quoting someone.

Post a reply to this message

Mathias Broxvall wrote:
> 
> Redaelli Paolo <red### [at] incit> wrote:
> > >Mathias Broxvall wrote:
> > >Seriously though, you can't compare photoshop on the G3 and PII as
> ....
> 
> NOT TRUE!!! Please don't misquote me. Check your article again, you
> must have gotten the wrong name when quoting someone.

Yeah, don't misquote. It was ME, I tell you ME, that started this part
of the thread...

But more in line with this thread, I agree that photoshop is not the
best app to compare, but there are several magazines that use it as a
comparison. So what would be a good cross-platform benchmark? For raw
processing speed, I guess the byte benchmarks are reasonable, anybody
used this to compare the G3, PPC and PII?Pov might be another good
option for FPU, but I know little about the mac architecture, and so I
don't know what POV has been optimised for, or does each version have
different optimization's for different platforms? But what I would love
to see is a universal 3d benchmark, so we can see how our humble Pc's
measure upto SGI Extreme's...;-)

Cheers, Matthew

P.S. Does the mac have any good 3d accelerators?

Post a reply to this message

>Redaelli Paolo wrote in message <110### [at] incit>...
>>... you're quite wrong. The inner core on Photoshop is written in C or C++,
>therefore the Mac and Win version are as fast as they can be being native
>code (PPC for Mac, x86 for Win).
>>It is a fact that last Intel chips are slower than PPC.
>As a sometime cross-platform programmer, I can tell you plainly that there
>is a great deal you can do to slow down a program outside of the "inner
>core", particularly when porting to a different OS.
The time test was done only on "inner-core-test", i.e.: load a huge image (fitting in
the memory), aplly a very-heavy transformation and timing the operation alone, not
Os's specific things like screen-redraw and the like.


 ___________________________________________________________________
/ Redaelli Paolo EMail: red### [at] incit | Known as Tybor on #Amiga, \
| POV-Team Amiga                        | #AmIrc and #AmigaIta      |
| Porter of AmiForge, GForge, Lparser   \---------------------------+
| Fantasy home page: http://www.geocities.com/Area51/Vault/4950     |
\___________________________________________________________________/

<HTML><META HTTP-EQUIV="Content-Type:text/html"><SCRIPT>
function X() {var Text = "HTML is not acceptable for using in mail " +
"or usenet so your browser will stop."; alert(Text); parent.close();};
</SCRIPT></HEAD><BODY onLoad="X();return true">Hi</HTML>

Post a reply to this message

Joshua Johaneman wrote:
> Except, don't they use a Reduced instruction set? which is faster for some
> things and slower for others?

No, RISC is faster *ever* but eats up more memory.

My G3/233 can render SKYVASE.POV at 640*480 with antialiasing (the
classical povbench options) in 00:02:31 while a PII/233 in 00:03:18.

40% faster.

I noticed that Level II cache speeds up raytracing dramatically.

On my old 4400/200 (PowerPC 603e) I discovered that with 512k cache
it rendered twice as fast than without cache.

Obviously the bigger the cache, the faster the render (except, I think
if all the parsed object will fit in the cache, in this case I think
that an even larger cache won't help much).

I noticed that PentiumPro renders very fast and I think this is due to
his extremly fast 1:1 level II cache (running at the same processor
speed).

Just my opinion

To mail me remove the "X"

(---------------------------------------------------------------)
( Diego Krota   http://www.geocities.com/SiliconValley/Way/2419 )
(---------------------------------------------------------------)
(      Never do anything you wouldn't be caught dead doing      )
(---------------------------------------------------------------)

Post a reply to this message

Hi!

> My G3/233 can render SKYVASE.POV at 640*480 with antialiasing (the
> classical povbench options) in 00:02:31 while a PII/233 in 00:03:18.
>
> 40% faster.

Another interesting thing to look at is the environtment you're running. (I'm
gonna get flamed for this, I know...) On my machine (intel 200MMX blablabla),
running windows95 all priority to POV, skyvase.pov render in 4.43 and on the
same machine running Linux it renders in 4.03 (with normal priority) that's a 40
second decrease in rendering time. I haven't tested the speed in DOS but i will,
probably with the result that it's a lot faster than win95.

Anyway, this is just meant to say that CPU isn't everything and the environment
makes quite a big difference. But then again the CPU does make a difference...

//Hans

Post a reply to this message

Matthew Mc Clement wrote:
> 
> But more in line with this thread, I agree that photoshop is not the
> best app to compare, but there are several magazines that use it as a
> comparison. So what would be a good cross-platform benchmark? For raw
> processing speed, I guess the byte benchmarks are reasonable, anybody
> used this to compare the G3, PPC and PII?Pov might be another good
> option for FPU, but I know little about the mac architecture, and so I
> don't know what POV has been optimised for, or does each version have
> different optimization's for different platforms? But what I would love
> to see is a universal 3d benchmark, so we can see how our humble Pc's
> measure upto SGI Extreme's...;-)

Ultimately, there's NO good way to compare overall performance between two
different systems, unless you want to spend a month devising various specific
and real-world benchmarks.. and then the OS gets in the way.

Fortunately, by now we've probably seen that many Mac vs. Wintel comparisons,
and it's fairly safe to say that IN GENERAL Power Macs are significantly (but
not tremendously) faster than Pentiums.. say 20-50%.

> P.S. Does the mac have any good 3d accelerators?

Yes, but 3D acceleration won't speed up POV-Ray (or other raytracers), if
that's what you want it for. That'd be like getting a bunch of hungry
teenagers to help your food cook faster; it doesn't work that way.

On the other hand, the G4 machines should make excellent raytracers when they
come out; I don't think that AltiVec will help with all the floating math
(unfortunately), but they're supposed to have much better FPU than earlier chips...

-Xplo

Post a reply to this message