| Dual Core Apple "MacBook", and RSI's Mac Intel plans [message #46940] |
Tue, 10 January 2006 11:59  |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
Apple just announced a new Intel-based, dual core "MacBook", using the
newly announced "Core Duo" chip. Check it out here:
http://www.apple.com/macbookpro/
This looks like the ultimate IDL mobile workstation. Now, the
question is, will IDL 6.3 be offered as a Universal Binary for the
Mac? Since we know RSI never really put much into Altivec (if anything at
all), it would seem to me that a port to Intel for Mac would be as simple
as clicking the "Compile Universal Binary" box. Given how many Powerbooks
I see floating around at conferences, I think they should strongly
consider ticking that box. This machine will sell.
Since they are compiling IDL with GCC, we should see similar
performance as available on similarly-equipped Linux systems (it's the
same exact compiler). Since GCC was such a woeful underperformer for
PowerPC, the speed advantages would be multiply compounded. I would
guess the new MacBook would offer IDL performance up to 5 times as
fast as the current Powerbook G4 (for problem sizes which can take
advantage of both cores). All that, and a shiny metal case too (not
to mention the built-in video-cam).
Anyone have the inside scoop?
JD
|
|
|
|
| Re: Dual Core Apple "MacBook", and RSI's Mac Intel plans [message #46959 is a reply to message #46940] |
Fri, 13 January 2006 14:29  |
Kenneth P. Bowman
Messages: 585
Registered: May 2000
|
Senior Member |
|
|
In article <pan.2006.01.13.17.56.06.516548@as.arizona.edu>,
JD Smith <jdsmith@as.arizona.edu> wrote:
> On Fri, 13 Jan 2006 15:55:32 +0000, Ricardo Bugalho wrote:
>
>> Hello,
>> a couple of points, depending on your interest: - Mac OS X for Intel won't
>> emulate Altivec. So, if IDL makes any use of it, IDL might not run. OTOH,
>> unlike SSE, Altivec never supported double precision so it's likely IDL
>> won't use it. - If you don't care about Mac OS X and just want Apple's
>> tradicional quality in laptops, you might be able to install Linux on it.
>
> It turns out IDL does use a small bit of Altivec on a very few basic
> operations. I had thought it used none, but was corrected. So this
> means that IDL likely will *not* run under Rosetta (the new
> on-they-fly PPC->Intel translator). The main question is, does IDL
> run on a G3 processor? It certainly used to, back in the day, so it
> probably still does, which means when Rosetta tells IDL it is a G3, it
> may well fall back on some slower code and not choke. SSE is another
> kettle of fish. In a year's time, probably well over 75% of IDL's
> user base will be running on the same processor architecture.
> Thinking about using SSE/SIMD is probably a good idea (and they are).
>
> JD
This is from one of the Mac rumor sites:
> In a note to clients on Friday, American Technology Research analyst Shaw Wu
> said checks with sources indicate that the upcoming release of Mac OS X
> 10.4.4 for Intel runs well, with noticeable improvements to the Rosetta
> PowerPC emulation environment that improve backward compatibility with
> AltiVec support.
>
> "We believe this will alleviate concerns that older software that hasn't been
> ported to Intel will run well without a recompile," Wu wrote.
But I suspect this is minor compared to ensuring that third party
software runs.
Ken Bowman
|
|
|
|
| Re: Dual Core Apple "MacBook", and RSI's Mac Intel plans [message #46963 is a reply to message #46940] |
Fri, 13 January 2006 11:38 |
Karl Schultz
Messages: 341
Registered: October 1999
|
Senior Member |
|
|
On Fri, 13 Jan 2006 10:56:06 -0700, JD Smith wrote:
> On Fri, 13 Jan 2006 15:55:32 +0000, Ricardo Bugalho wrote:
>
>> Hello,
>> a couple of points, depending on your interest: - Mac OS X for Intel won't
>> emulate Altivec. So, if IDL makes any use of it, IDL might not run. OTOH,
>> unlike SSE, Altivec never supported double precision so it's likely IDL
>> won't use it. - If you don't care about Mac OS X and just want Apple's
>> tradicional quality in laptops, you might be able to install Linux on it.
>
> It turns out IDL does use a small bit of Altivec on a very few basic
> operations. I had thought it used none, but was corrected. So this
> means that IDL likely will *not* run under Rosetta (the new
> on-they-fly PPC->Intel translator). The main question is, does IDL
> run on a G3 processor? It certainly used to, back in the day, so it
> probably still does, which means when Rosetta tells IDL it is a G3, it
> may well fall back on some slower code and not choke.
IDL makes a run-time check to detect Altivec. So, if the emulation layer
faithfully reports no Altivec as it should, then IDL shouldn't execute
any Altivec instructions.
If this fails, one can still use the CPU command to turn off vector
processing. Also, see !CPU.
Karl
|
|
|
|
| Re: Dual Core Apple "MacBook", and RSI's Mac Intel plans [message #46965 is a reply to message #46940] |
Fri, 13 January 2006 10:00 |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
On Tue, 10 Jan 2006 12:59:42 -0700, JD Smith wrote:
> it would seem to me that a port to Intel for Mac would be as simple
> as clicking the "Compile Universal Binary" box.
One issue I overlooked: IDL comes with many 3rd party tools, like the
license manager, the IDL Assistant, etc., which themselves would have
to be ported (or run well enough under code translation, which I expect
they indeed would). So it's not really a simple one box check and you're
done operation. Plus there turns out to be a small amount of Alitvec that
would have to be ripped out/coded around (but given they already likely
code around for pre-Altivec chips like G3, not to mention their Intel code
base, I doubt that's a major hurdle).
JD
|
|
|
|
| Re: Dual Core Apple "MacBook", and RSI's Mac Intel plans [message #46968 is a reply to message #46940] |
Fri, 13 January 2006 09:56 |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
On Fri, 13 Jan 2006 15:55:32 +0000, Ricardo Bugalho wrote:
> Hello,
> a couple of points, depending on your interest: - Mac OS X for Intel won't
> emulate Altivec. So, if IDL makes any use of it, IDL might not run. OTOH,
> unlike SSE, Altivec never supported double precision so it's likely IDL
> won't use it. - If you don't care about Mac OS X and just want Apple's
> tradicional quality in laptops, you might be able to install Linux on it.
It turns out IDL does use a small bit of Altivec on a very few basic
operations. I had thought it used none, but was corrected. So this
means that IDL likely will *not* run under Rosetta (the new
on-they-fly PPC->Intel translator). The main question is, does IDL
run on a G3 processor? It certainly used to, back in the day, so it
probably still does, which means when Rosetta tells IDL it is a G3, it
may well fall back on some slower code and not choke. SSE is another
kettle of fish. In a year's time, probably well over 75% of IDL's
user base will be running on the same processor architecture.
Thinking about using SSE/SIMD is probably a good idea (and they are).
JD
|
|
|
|
|
|
| Re: Dual Core Apple "MacBook", and RSI's Mac Intel plans [message #46972 is a reply to message #46940] |
Fri, 13 January 2006 08:57 |
Ricardo Bugalho
Messages: 22
Registered: March 2005
|
Junior Member |
|
|
Hi,
the reason is that Apple's Intel iMacs aren't PC compatible. Apple
ditched the legacy IBM PC and it's BIOS.
Instead they went EFI, a firmware standard created by Intel a few years
ago.
The only Windows versions that supports EFI are the Itanium ones (all
Itanium systems use EFI) and only Windows Vista will bring support EFI
on x86(-64) systems. Therefore, it won't boot.
On Thu, 2006-01-12 at 09:18 -0500, Matt Feinstein wrote:
> On 11 Jan 2006 08:07:11 -0800, phaccount@nycap.rr.com wro
> The killer option is a Windows/OS X dual-boot workstation, but it
> looks like that won't happen for a while. The various items I've read
> are somewhat unclear, but, apparently, dual-boot XP+OS X is not a
> possibility with the new macs because of BIOS/firmware issues. However
> Vista+OS X will be possible and, IMO, poses a major threat to linux.
|
|
|
|
|
|
|
|
|
|
| Re: Dual Core Apple "MacBook", and RSI's Mac Intel plans [message #46990 is a reply to message #46940] |
Thu, 12 January 2006 10:19 |
JD Smith
Messages: 850
Registered: December 1999
|
Senior Member |
|
|
On Thu, 12 Jan 2006 09:18:35 -0500, Matt Feinstein wrote:
> On 11 Jan 2006 08:07:11 -0800, phaccount@nycap.rr.com wrote:
>
>> I am thinking that Apple on Intel may be a demise of Linux.
>>
>> I for one, a linux and wintel user, would love to switch to Apple and
>> have the best of linux and the decent of mictel (microsoft+intel). If
>> the hardware price comes down some, a part of the linux folks may
>> switch to apple. Apple may then see a surge in computer, scientific
>> and engineering applications that partially drove linux forward.
>
> The killer option is a Windows/OS X dual-boot workstation, but it
> looks like that won't happen for a while. The various items I've read
> are somewhat unclear, but, apparently, dual-boot XP+OS X is not a
> possibility with the new macs because of BIOS/firmware issues. However
> Vista+OS X will be possible and, IMO, poses a major threat to linux.
I'd say OSX/Linux dual boot poses a threat to Windows ;). And I guarantee
we'll see Linux on these boxes faster than Windows...
|
|
|
|
| Re: Dual Core Apple "MacBook", and RSI's Mac Intel plans [message #47033 is a reply to message #46974] |
Tue, 17 January 2006 08:16 |
Wolf Schweitzer
Messages: 21
Registered: October 2001
|
Junior Member |
|
|
Gianguido Cianci wrote:
> IDL 6.3 will not be tested or supported on the
> Mac for Intel architecture.
Thanks for this update!
After an Apple developer told me 1 1/2 years back, indirectly, "not to
wait" for a fully 64-bit OS, and after one could sense the reluctance on
RSI's side, I don't regret having abandoned Apple for workstation
strength computing.
Most importantly, working on X11 isn't THAT different from working on
Linux or AIX. We talked about "hard reasons" to buy OS X based machines
recently; real (hard) reasons to us OS X are not Open Office 2.0, web
browsing, e-mail, web-based mySQL/pHp-packages, statistical software or
The Gimp (they all are available / run on Windows or Linux cheaper and
faster) - but mostly entertainment or creative sound packages such as
iTunes, Reason 3.0 Synthesizer Software. So for such, or for any other -
irrational - reason, I definitely love OS X.
As my G4 Powermac is getting extremely "old" in that it doesn't keep up
with the computing speeds that I need in IDL for some applications, I
now run IDL 6.2 on a Suse Linux 10 / AMD-workstation at home, and the
performance is simply great. Dynamic frequency scaling makes the
workstation perform with a minimum of ventilation noise otherwise as
processor frequency is scaled down on idle. I got myself that Fujitsu
Siemens Celsius v830, and it's a great machine. You really get more
hardware options (your choice of AMD Opteron processors, up to dual 280;
up to 32 GB RAM; 4 drive bays; several bays for slot/tray loading
drives) - where Apple doesn't offer nearly that much for even DOUBLE the
price on their G5 powermac platform. That RSI seems to support Linux /
AMD 64-bit in the foreseeable future was a great decision that suits me
just fine.
Conversely, no notebook of any brand really says "power programming
enabled", so all they can be used for is solving some smaller problems.
For now, that's done on existing investements without sweats
(Powerbook). For at least 2-3 years, PPC based Macs will be still widely
supported, so I quite honestly don't see why there should be a
particular big pressure "right now". Right now, one could also consider
the Centrino Duo notebooks by Dell (up to 1920 x 1200 pixels resolution,
better than any notebook Apple sells), or the Acer Travelmate 8200 (up
to 4 GB RAM, better than any notebook Apple sells).
It is really to be hoped that we will see Apple's OS X on Fujitsu
Siemens, Dell or Acer, on IBM replacing AIX, within the next 2-3 years.
If not, we hope at least for KDE to continue to be user friendly.
Wolf.
|
|
|
|
| Re: Dual Core Apple "MacBook", and RSI's Mac Intel plans [message #47076 is a reply to message #46940] |
Fri, 20 January 2006 16:36 |
Kenneth P. Bowman
Messages: 585
Registered: May 2000
|
Senior Member |
|
|
In article <1137796144.257630.54020@g14g2000cwa.googlegroups.com>,
dbowman@fullerton.edu wrote:
> I understand that emacs is the programmers' preferred way to do code,
> and that IT types inexplicably like Windows. However, IDL claims to be
> a *scientific* visualization and analysis package. If RSI really wants
> scientists to use their product, then they should be responsive to what
> *scientists* want, rather than what's popular with software engineers.
> That means things like Mac/Intel and <gasp> Direct Graphics.
>
> Honestly, the only reason I haven't switched to Matlab (which is more
> responsive to scientists) is that it would take me far too long to port
> all my legacy code. This is not a good way to keep a user-base happy!
I like his opinions *and* his last name.
Ken Bowman
|
|
|
|
| Re: Dual Core Apple "MacBook", and RSI's Mac Intel plans [message #47077 is a reply to message #46940] |
Fri, 20 January 2006 15:13 |
news.qwest.net
Messages: 137
Registered: September 2005
|
Senior Member |
|
|
<dbowman@fullerton.edu> wrote in message
news:1137796144.257630.54020@g14g2000cwa.googlegroups.com...
...
> If RSI really wants
> scientists to use their product, then they should be responsive to what
> *scientists* want, rather than what's popular with software engineers.
> That means things like ... <gasp> Direct Graphics.
...
> David Bowman
Aha, another potential member of "La Resistance".
(shhhh, we cannot talk of this here... )
-anonymous
|
|
|
|
| Re: Dual Core Apple "MacBook", and RSI's Mac Intel plans [message #47078 is a reply to message #46977] |
Fri, 20 January 2006 14:29 |
dbowman
Messages: 5
Registered: January 2003
|
Junior Member |
|
|
I've just posted the following to the RSI User Forum, but thought I'd
also share it with this group. Let the flames begin! :-)
I've been using IDL for over ten years now. My field (earthquake
seismology) is also heavily populated by mac users, and always has
been. Most of my colleagues switched from IDL to Matlab back when RSI
abandoned the mac. The subsequent port to X11 has never been truly
satisfactory. Many of the great features of IDL 5.1 never made it to
*nix - things like a quality built-in editor with syntax highlighting.
I understand that emacs is the programmers' preferred way to do code,
and that IT types inexplicably like Windows. However, IDL claims to be
a *scientific* visualization and analysis package. If RSI really wants
scientists to use their product, then they should be responsive to what
*scientists* want, rather than what's popular with software engineers.
That means things like Mac/Intel and <gasp> Direct Graphics.
Honestly, the only reason I haven't switched to Matlab (which is more
responsive to scientists) is that it would take me far too long to port
all my legacy code. This is not a good way to keep a user-base happy!
David Bowman
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
