Running JVM on Intel 64-bit xeon

[Owen M. Astley]

Hi,

Has anybody tried writing/running java on a 64-bit Intel Xeon processor
under Windows, and can they answer the following questions:

1. Does the standard Sun JVM work?

2. Is there a 64-bit JVM for this processor? Would a JVM for the AMD64
work?

3. If I link into external code via JINI, with that code in a dll, do I
need to compile the external code to match the JVM? ie can I use a
32-bit JVM with a 64-bit compiled dll?

Thanks,
Owen Astley

 

[Roedy Green]

Has anybody tried writing/running java on a 64-bit Intel Xeon processor
under Windows, and can they answer the following questions:

1. Does the standard Sun JVM work?

You can find out. See http://mindprod.com/jgloss/jdk.html
Since a Xeon has 32 bit Pentium mode, I don't see why not.

2. Is there a 64-bit JVM for this processor? Would a JVM for the AMD64
work?

No AMD is for an Athlon, not a Xeon. AMD stole Intel's thunder in the
64-bit world.

3. If I link into external code via JINI, with that code in a dll, do I
need to compile the external code to match the JVM? ie can I use a
32-bit JVM with a 64-bit compiled dll?

You mean JNI.

I presume you need to use the C JNI library to match your JVM. I have
not experimented to see what happen if you try to use 32 bit code.

With JAWS you can provide several different JNI implementations and it
picks the correct one.

 

[Mark Thornton]

You can find out. See http://mindprod.com/jgloss/jdk.html
Since a Xeon has 32 bit Pentium mode, I don't see why not.



No AMD is for an Athlon, not a Xeon. AMD stole Intel's thunder in the
64-bit world.

My understanding is that the JVM for AMD64 DOES work with 64 bit Xeons.
It does not, of course, work with Itanic processors.

Mark Thornton

 

[Timo Stamm]

My understanding is that the JVM for AMD64 DOES work with 64 bit Xeons.
It does not, of course, work with Itanic processors.

Yes, I'm running the AMD64 JVM under debian on a Xeon server.

Roedy is right, AMD really did steal Intel's thunder in the 64-bit
world, but Intel introduced an instruction set that is compatible to AMD64.

From the wikipedia:

| Due to a severe lack of success with Intel's Itanium and Itanium 2
| processors, the 90 nm version of the Pentium 4 (Prescott) was built
| with support for 64-bit instructions (called EM64T by Intel, though it
| was much the same as AMD's AMD64 instruction set), and a Xeon version
| codenamed Nocona was released in 2004.


Timo

 

[Roedy Green]

My understanding is that the JVM for AMD64 DOES work with 64 bit Xeons.
It does not, of course, work with Itanic processors.

Shall we try the experiment? That would be great if it did. The last
time I looked was a few months back. They referred to it as AMD, no
mention of Xeon. That's a funny thing to do.

Is this true?

Itanium: 64 bit, wildly different from anything else. No 32 bit
support?

AMD Athlon: basically a 32-bit Pentium with add on 64-bit support of
AMDs devising, Pentium like.

AMD Opteron : basically a 64 bit Athlon with tacked on 32 bit support.

Xeon: pentium with AMD 64 bit? Intel 64bit extension?

to run in 64 bit mode, do you need a 64 bit OS, or do 32-bit OSes
allow some 64-bit tasks?

 

[Luc The Perverse]

No AMD is for an Athlon, not a Xeon. AMD stole Intel's thunder in the
64-bit world.

Which as a side note was Intel's fault.

 

[christian.jean]

1. Does the standard Sun JVM work?

Yes it will work. But be warned that the 64-bit implementation on Xoen
processors do not work all that efficiently. In many bencharmks the
32-bit code runs faster than the 64-bit. Intel was rushed into
delivering its 64-bit product so they didn't have time to optimize it.
The 64-bit versions comming out now though are much, much faster!

Use the standard AMD 64-bit JVM!

I've installed a Debian linux on my AMD Opteron. I installed
absolutely everthing in native 64-bit (kernel, drivers, JVM, database,
etc), it was easy and it works like a charm.

If you set it up like I have, when you compile your C code, it will
compile it all in 64-bit transparently.

As a side note:

Our company wanted to go 64-bit, but I refused the Xeon processor and
recommended the Opteron instead.

There are several reasons for this:

1) AMD is true to its road-map, while Intel is all over the place (due
to FUD). I'll be able to upgrade my Opteron process with a quad core
in about 12 months time and literaly triple my server performance.
Intel on the other hand, the roadmap did not allow me to do this.

2) AMD is tue native 64-bit and runs flawlessly! Intel's 64-bit is at
times slower than 32-bit mode. Processor runs alot hotter (I'm using a
U1) and didn't want to take the chance.

---

And for those Intel fans, AMD did not steal 64-bit from Intel...
engineers said it was impossible to do, so Intel didn't want it and
they didn't do it. They gave it away, so AMD just gladly took it.

And if you know your history, AMD did not steal x86 instruction set,
they were given a license by Intel when IBM told them to. So they have
all the legal rights to it. It doesn't mean that AMD inovated a much
more powerful way to process such x86 instructions than Intel that they
are to be blamed.

Either way, you WILL get on the bandwagon... sooner or later!

 

[Roedy Green]

And if you know your history, AMD did not steal x86 instruction set,

I think an instruction set should be no more patentable than a
keyboard layout, or that steering wheels are on the left side of the
car, or the order of the pedals.

This is a user interface. It has to be standard to be convenient for
users.

The companies compete on how they implement the instructions.

Patenting a instruction set is a bit like letting a word processing
company patent the format of its data files so that no one but them
can use the data.

The value lies in the code written in the language of some instruction
set. The true owners are the composers of that code and they should
be allowed to use it as they see fit.