With today’s announcement that Sun Microsystems, Inc. and AMD have formed an alliance to deliver a broad range of AMD Opteron processor-based systems, Sun also announced it plans to offer its Java Enterprise System on the AMD Opteron processor and is significantly extending the reach of its Solaris Operating System and leadership in the 64-bit space. The alliance intends to provide increased market opportunities for ISVs, OEMs, and channel partners by offering an industrial strength, affordable platform that has been fully optimized and tuned for the 64-bit AMD Opteron processor.
Sun/AMD Alliance; 64-bit Java Enterprise System for Opteron
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Finally a good idea from SUN. I can see this being an awesome relationship!
If AMD’s Opteron doesn’t become the next Alpha because everyone is too frightened of using something that is not Intel. I for one hope that server administrators wake up and start beating the AMD drum, they are laying waste to the XEON line of processors. Then again, Intel always has Prescott, and MMX, SSE, SSE2, SSE3, SSE4, ad infinitum… Of course they require you to recompile every piece of software you own just to get any performance out of them… but that is Oll Korrect, right??
Quoted from: http://www.sun.com/2003-1117/feature/
…;that the SPARC processor will continue to be Sun’s flagship for extremely scalable Symmetric Multiprocessing; that Sun continues to invest heavily in the SPARC architecture and is leveraging the processor’s inherent scalability with its Throughput Computing strategy; that Sun is ready to meet customers’ needs, whether it’s raw computing power, throughput, or memory access;…
Just in case anyone wondered :^)
SPARC has a fairly bright future, now that Sun will be collaboratively developing the architecture in conjunction with Fujitsu, whose SPARC64 line has surpassed Sun’s own.
Not that suprising, after all the extreme high-end is where the sparc makes sense. The sparc might be weak on the lower end systems (like <16 cpus) but bring in a touch of opteron and the lineup is looking real good.
Gimme!
rewrite the OS strictly in LONG mode to take advantage of flat memory addressing and extra registers.
the original nextstep 486 would not run on a 386.
if not the desktop definite the server os
I agree. Even if Power PC is better, using regular hardware has always been the thing holding them back. Right now, they have the best desktop going yet few people buy it because it doesn’t run on “their” hardware.
What have I been saying?
Lets just hope everyone can forget about that whole SCO thing..
Anonymous (IP: 165.134.195.—)
apple should port osx to amd64
Why?
rewrite the OS strictly in LONG mode to take advantage of flat memory addressing and extra registers.
You’re describing architectural drawbacks of x86, a platform for which OS X is not even available. Why would they need to “rewrite” their operating system for a platform they don’t target, especially when their current platform, PPC, does not suffer from issues like segmented memory or a dearth of registers?
Dan (IP: —.riv-eres.charterpipeline.net)
I agree. Even if Power PC is better, using regular hardware has always been the thing holding them back.
Holding them back from what? Reorganizing themselves under Chapter 11 when their hardware sales plummet due to the easily availability of OS X on commodity hardware?
I’d say their current, *profitable* route is the most fiscally responsible…
Right now, they have the best desktop going yet few people buy it because it doesn’t run on “their” hardware.
Which serves as an incentive to purchase Apple hardware, which is where Apple makes money.
platform, PPC, does not suffer from issues like segmented memory or a dearth of registers?
WRONG. writing os x server to specifically take advantage of the LONG mode would be to get rid of segmented memory and dearth of registers, b/c the 64-bit mode is flat memory addressing and has additional 16 registers over 32-bit mode.
a platform for which OS X is not even available.
1- nextstep is available for x86 and
2- i suggested that it should be ported.
Reorganizing themselves under Chapter 11 when their hardware sales plummet due to the easily availability of OS X on commodity hardware?
uh apple could have a custom rom bios to only run on apple hardware. also i was speaking mostly of os x server.
profit = revenue – cost
switching to amd64 should reduce costs, so profits go up.
WRONG. writing os x server to specifically take advantage of the LONG mode would be to get rid of segmented memory and dearth of registers, b/c the 64-bit mode is flat memory addressing and has additional 16 registers over 32-bit mode.
There’s no dearth of registers on the PowerPC ISA. PowerPC processors have 32 general purpose registers, compared to 16 general purpose registers for the AMD64 architecture. Segmented memory is a drawback of IA32, not PPC. There’s no segmented memory issue for them to get rid of.
Seriously, this is one of the worst trolls ever…
PowerPC processors have 32 general purpose registers, compared to 16 general purpose registers for the AMD64 architecture.
AMD64 also has 16 SSE registers, for a total of 32 registers.
Segmented memory is a drawback of IA32, not PPC. There’s no segmented memory issue for them to get rid of.
segmented memory is not a drawback for AMD64 if written entirely in long mode.
Seriously, this is one of the worst trolls ever…
providing accurate information is trolling?
AMD64 also has 16 SSE registers, for a total of 32 registers.
PPC970 has an additional 48 rename registers, for a total of 80 registers, and that doesn’t even take into account the vector unit…
segmented memory is not a drawback for AMD64 if written entirely in long mode.
As it’s not a problem for OS X’s target ISA, why even bring it up?
providing accurate information is trolling?
You’re providing entirely irrelevant information. OS X targets the PPC ISA. What technical drawbacks of the PPC970 are so egregious as to constitute a technical advantage in moving to AMD64?
You’ve given absolutely no evidence to show that the Opteron is technically superior or more cost effective solution than the PPC970.
Jesus how many times must we repeat this over and over again till this sinks into the thick skulls of you “I wish apple would port OS-X” people !! GET THIS STRAIGHT ONCE AND FOR ALL APPLE SELLS HARDWARE USING IT’S SOFTWARE TO GET PEOPLE TO BUY THEIR PRODUCT !! Without OS-X as a Mac only product no one would buy Mac hardware ! They would dig their own graves it they would port it over to a PC arch !
PPC970 has an additional 48 rename registers, for a total of 80 registers, and that doesn’t even take into account the vector unit…
You know what rename registers are don’t you?
Intel P4 has 128 for integer and another 128 for floating point. For Athlon the numbers are 24+88 and for Opteron/Athlon64 40+120.
PPC970 is still the winner because architectural registers are far more important than rename registers.
As for segmentation, it’s not at all that bad as it’s made out to be. Intel gave it a bad rep with a bad implementation. Anyway, it doesn’t matter since only 16-bit code uses it. Both IA32 and AMD64 is flat addressed.
will solaris AMD64 work on “ordinary” Athlon 64 PCs? not just those opterons that Sun will sell as boxes?
and will Java AMD64 run on Windows, Linux and other OSes, not just the Solaris AMD64 port?
will solaris AMD64 work on “ordinary” Athlon 64 PCs? not just those opterons that Sun will sell as boxes?
Assuming that they use off the shelf Opteron components, probably using the AMD Opteron chipset, you shouldn’t have any problems. The only problems that may arise is hardware support issues for pieces of hardware via PCI/USB and so forth.
and will Java AMD64 run on Windows, Linux and other OSes, not just the Solaris AMD64 port?
I wouldn’t have a clue. I’ve only heard about Solaris and Java. Considering that they see Solaris and Opteron being on the server rather than the desktop, I would say the likelihood of an AMD64 version of Java on Linux or Windows is very low.
The fact of the matter is that does the end user really need to have a native AMD64 when the 32bit will work nicely.
From what I looks like, they’re putting Opteron + Solaris for the low end server <16, >16 will be Solaris + SPARC (which is good for large batch transactions like the good old mainframes) and for the desktop it will be Dell + JDS (Linux/GNOME/StarOffice/etc).
The AMD64 version of Java will be released on Windows and IIRC also on Linux. Remember that Sun will sell Opteron boxes with Linux as well.
Hopefully this means that Sun will optimise their JVM, e.g. hotspot for the x86-64 as they’ve done for P3/P4.
They already have…
http://www.sun.com/smi/Press/sunflash/2003-08/sunflash.20030805.9.h…
I just purchased an AMD 64 system. Would this Opteron Linux distro work with the regular AMD 64 as well? I would really like to run a Sun version of linux as I have tried out their demo cd and it’s sweet. Thanks for any feedback.
I think you would probably be happiest running SuSE Linux for AMD64. The Java Desktop from SUN runs on top of SuSE anyway. You can download Java from SUN’s site or use the version that comes with SuSE. The version compiled specifically for AMD64 will not be out until the summer of 2004. If you really wanted to try the Linux version that SUN is offering right now it wouldn’t use any special 64 bit features of the Opteron.