Is intel EMT 64 bit pure or amd 64 bit pure

[Pippero]

Sorry, but none of your arguments backs up your statement of "64bit are for sissies".
128bit addressing is used by the AS/400 but generally speaking it does not make any sense, even 64bit addressing is huge compared to the amount of memory which is going to be used in the foreseeable future.
And 128bit FP registers has nothing to do with a machine being 128bit (today, back in the 80ies this was acceptable).
Otherwise, we can say that we're all already running 128bit CPUs, because all x86 CPUs which support SSE have 128bit registers.

 

[levicki]

You can use Vista with 1GB of ram in 64-bit mode. Sure, you will loose the main advantage (>2GB RAM), but still get improved architecture bonus.

You can't use Vista with 1GB of RAM. You can toy with it. You can use Office, email and surf at the best.

I was using 64-bit Fedora on my s754 two years ago... (and it WAS faster).

Faster than what?

Actually, Conroe's EM64T performance isn't any better than Netburst's was, relative to each core's 32-bit performance.

That is not true. One of the reasons is that Conroe has 128-bit data bus as opposed to 64-bit, another one is that it does 64-bit operations instead of two 32-bit as Netburst did. Anyway, this link was given before so read it:
http://www.xbitlabs.com/articles/cpu/display/core2duo-64bit.html

I am not sure how many GPRs are missing in the EM64T implementation: SYSCFG, TOP_MEM, TOP_MEM2. Thats why the additional 8 GPRs is not the case for EM64T.

What are you referring to here? Additional 8 GPR (General Purpose Registers) are r8 to r15 and both architectures have them otherwise AMD64 register calling convention wouldn't include them as parameter passing registers. In addition both have XMM8 to XMM15 registers in 64-bit mode.

Maybe, what stoped Intel to release such feature earlier?

Lack of 64-bit OS and software? You must admit that Intel's judgment was better -- they said "no real need for 64 bits on the desktop before 2007" and they were right. We still have no decent OS and applications are scarce. To all Linux advocates: reason why I don't count Linux as a decent OS is because Linux is just a kernel anyway says Linus Torvalds.

Then, AMD came along and INVENTED AMD64

AMD didn't invent a shit. AMD64 was the same old thing as transition from 16-bit to 32-bit mode just the instruction prefix added way back was 0x66 and now it is the 0x48. Big deal. They essentially helped Microsoft to retain the monopoly by not having to port their software to new architecture.

People obviously don't understand this, as a CPU architecture AMD64 was advanced, but from a software standpoint it was nothing new. On the contrary -- it added to the legacy x86 burden we are tagging for decades now. So EPIC and IA-64 were the way to go if you wanted better software, not AMD64. Now STFU.

@Everyone:

PLEASE stop calling AMD64 and EM64T and x86-64 and x64 instructions when they aren't!!! Those are 64-bit extensions, not instructions. Not a single new instruction was introduced with AMD64, just new REX prefix for addressing mode selection.

 

[quantumsheep]

You can use Vista with 1GB of ram in 64-bit mode. Sure, you will loose the main advantage (>2GB RAM), but still get improved architecture bonus.

You can't use Vista with 1GB of RAM. You can toy with it. You can use Office, email and surf at the best.

I was using 64-bit Fedora on my s754 two years ago... (and it WAS faster).

Faster than what?

Actually, Conroe's EM64T performance isn't any better than Netburst's was, relative to each core's 32-bit performance.

That is not true. One of the reasons is that Conroe has 128-bit data bus as opposed to 64-bit, another one is that it does 64-bit operations instead of two 32-bit as Netburst did. Anyway, this link was given before so read it:
http://www.xbitlabs.com/articles/cpu/display/core2duo-64bit.html

I am not sure how many GPRs are missing in the EM64T implementation: SYSCFG, TOP_MEM, TOP_MEM2. Thats why the additional 8 GPRs is not the case for EM64T.

What are you referring to here? Additional 8 GPR (General Purpose Registers) are r8 to r15 and both architectures have them otherwise AMD64 register calling convention wouldn't include them as parameter passing registers. In addition both have XMM8 to XMM15 registers in 64-bit mode.

Maybe, what stoped Intel to release such feature earlier?

Lack of 64-bit OS and software? You must admit that Intel's judgment was better -- they said "no real need for 64 bits on the desktop before 2007" and they were right. We still have no decent OS and applications are scarce. To all Linux advocates: reason why I don't count Linux as a decent OS is because Linux is just a kernel anyway says Linus Torvalds.

Then, AMD came along and INVENTED AMD64

AMD didn't invent a ****. AMD64 was the same old thing as transition from 16-bit to 32-bit mode just the instruction prefix added way back was 0x66 and now it is the 0x48. Big deal. They essentially helped Microsoft to retain the monopoly by not having to port their software to new architecture.

People obviously don't understand this, as a CPU architecture AMD64 was advanced, but from a software standpoint it was nothing new. On the contrary -- it added to the legacy x86 burden we are tagging for decades now. So EPIC and IA-64 were the way to go if you wanted better software, not AMD64. Now STFU.

@Everyone:

PLEASE stop calling AMD64 and EM64T and x86-64 and x64 instructions when they aren't!!! Those are 64-bit extensions, not instructions. Not a single new instruction was introduced with AMD64, just new REX prefix for addressing mode selection.

Post came across as a bit angry, but i do agree with it all. Thanks for all that info, i finally feel englightened!

But was there really any need to revive this dead thread?