Is intel EMT 64 bit pure or amd 64 bit pure

Toggle sidebar Toggle sidebar
rahul_cracker

rahul_cracker

Distinguished
Apr 18, 2006
105
0
18,680
hi
Please dont mind the noob question ,but i am curious to know that all intel 64 bit processors have a tag of EMT 64 bit it means the intel Extended Memory Technology for 64 bit processing ,the processors working on 64 bit are so called because they can address 2^64 memory locations but intel features extended technology is it something different from amd's 64 bit processor technology .I know these processors are based on different architectures (the memeory controller difference ).Please tell me is there a difference between EMT 64 and 64bit processors .Thanks
8O
 
hi
Please dont mind the noob question ,but i am curious to know that all intel 64 bit processors have a tag of EMT 64 bit it means the intel Extended Memory Technology for 64 bit processing ,the processors working on 64 bit are so called because they can address 2^64 memory locations but intel features extended technology is it something different from amd's 64 bit processor technology .I know these processors are based on different architectures (the memeory controller difference ).Please tell me is there a difference between EMT 64 and 64bit processors .Thanks
8O
Click to expand...

Well there's a simple answer here. They're both exactly the same. The reason for this is basically 'cos microsoft is lazy and couldn't be bothered to develop for 2 different 64bit instruction sets so they made Intel use AMDs.
 
AMD developed the x86-64 instruction set and started the 64-bit thing (not counting the Itanium). Intel finally got wise to the 64-bit shift, added a few instructions and claimed the EMT64 was different so that they didn't have to deal with copyright stuff.

So to answer your question Intel 64-bit is a (very, very) slightly expanded version of AMD 64-bit (once again not counting Itanium).

And in case you don't know, Itanium was Intel's first 64-bit proc, it wasn't compatible with 32-bit programs and had to run them through an emulator (killing performance). Itanium was intended only for the super-high end server market, and it has stayed there because of cost and incompatibility.
 
x86-64 or EM64T is basicly the same technology but they are different implemented on the both arhcitectures K8 and Netburst. The K8 has expansion of another 8 GPRs(general purpose registers) when it is running in 64bit mode, which is not the case with Netburst. The X86-64 is performing better on K8. Netburst and K8 are entirely different architectures, their differnece is not only the IMC(which belongs to the external architecture).
AMD traded x86-64 with Intel for iSSE3.
 
x86-64 or EM64T is basicly the same technology but they are different implemented on the both arhcitectures K8 and Netburst. The K8 has expansion of another 8 GPRs(general purpose registers) when it is running in 64bit mode, which is not the case with Netburst. The X86-64 is performing better on K8. Netburst and K8 are entirely different architectures, their differnece is not only the IMC(which belongs to the external architecture).
AMD traded x86-64 with Intel for iSSE3.
Click to expand...

What about Core 2 Duo's 64bit performance?
 
The reason for this is basically 'cos microsoft is lazy and couldn't be bothered to develop for 2 different 64bit instruction sets so they made Intel use AMDs.
Click to expand...

Coming from the perspective of a programmer, lazy has nothing to do with this particular decision. There would be a LOT more confusion in the PC world if they (Intel and AMD) used different extensions to support 64-bit memory access. Memory access if a very fundamental part of an operating system and there is no realistic way to "IF/THEN" the memory access portion of code depending on which CPU your operating system just happens to be running on. Which is exactly why NT was originally built the way it was (with separate HAL layers for each architecture it runs on) and why the 64-bit versions of Windows are a separate product from the 32-bit versions. There is far more to this than just picking which "64-bit library" they would use for a given CPU.
 
I havent tested it yet, but It will perfrom different from both K8 and Netburst.
So far, the benchmarks numbers are saying it is performing quite good, no matter that the on of the micro/macro fussion(I forgot which one) is not working when in long mode(64 bit).
 
There are two different things going on here. The only "pure" 64 bit system that Intel has is the IA64 architecture, which is theirs (hence IA = Intel Architecture). This is only out on the Itanium processor and will not work with IA32 software. The other is the EMT64 which was pretty much a straight up copy that I believe intel actually pays AMD for the rights to use in their more recent chips (cant remember exactly).
 
I think the two companies have some sort of antitrust agreement to not restrict Instruction Set Extensions to individual vendors. Cross-licensing, with no royalties or payments, though I'm not very sure about this.
 
AMD traded x86-64 with Intel for iSSE3.
Click to expand...

well technically they didnt trade them. i'm sure that by law they have to share all instruction set stuff such as the sse3 and x86-64 and all the other instruction sets, as intel generates most of them it only seems fair as intel would dominate the market much more if AMD didnt have access to Intels instruction sets and similarly the other way round intel wouldnt have 64 bit processing as easily if AMD hadnt given them the stuff. it has something to do with the anti-trust laws which is why intel wants amd to hand over ATI stuff when they finnaly aquire it.
 
There are (were?) only a few negligible differences between EMT64 and AMD64, and they're probably due to the fact that Intel didn't choose to license AMD64, but rather to reverse-engineer it.
But i guess that with Vista 64bit, all these small differences have already been ironed out with different CPU revisions.
 
There is a bit of royalty being paied by AMD to Intel on every processor they sell. That is part of anti-turst settlement AMD did with Intel long ago. Like wise AMD has lot of cross license with most of other companies (as they dont have much of R&D inhouse). that is the reason why AMD has lot of overhead than Intel.
 
There is a bit of royalty being paied by AMD to Intel on every processor they sell. That is part of anti-turst settlement AMD did with Intel long ago. Like wise AMD has lot of cross license with most of other companies (as they dont have much of R&D inhouse). that is the reason why AMD has lot of overhead than Intel.
Click to expand...

AMD Acquired R&D through licenses. Intel do there own R&D. Prime Examples of this are the K7 and the K8. Both use technology Licensed from Compaq (who themselves acquired it from DEC).

Hypertransport, EV6 Bus, Intergrated Memory Controller and quite a few other design patents are not in-house AMD work. They've licensed and improved upon existing technology.
 
From what I've seen AMD processors make a better conversion to 64-bit. Meaning, you'll see a greater performance jump from 32 bit to 64 bit mode with AMD than what you'll see doing the same with an Intel chip.

However...

Because most Core 2 Duo chips are faster than any AMD chip in 32-bit mode it negates the fact that you don't see as much as a performance jump.

I would imagine that Intel Core 2 Duo's would pwn AMD chips in 64-bit benchmarks, but I haven't really seen many relevant benchies.
 
I'm pretty sure Hypertransport is an entirely AMD design.
And yes in the K7 days they licensed the EV6 bus from DEC (which BTW, ended up crippling the K7 compared to the PIV), but the K7 itself was an original AMD design. (ok it can be argued that they had acquired most of the engineers which developed the DEC Alpha CPU).
Same goes for the K6, they acquired NexGen which developed the K6.. so strictly speaking that's not a licensed technology.
But what do you mean with "licensed" IMC?
I mean, an integrated memory controller can be developed in so many ways.. it is pretty much like saying "licensed L2 cache"...
 
you mentioned them aquiring engenires that used to work for other companies i hear they would most likely be doing the same with Intel employees when they laythem of unless they already have
 
EM64T and x86-64 are actually identical; the only reason Intel processors occasionally perform worse in 64-bit vs. their AMD counterparts has to do with the number of simple decoders vs. complex decoders.
 
it is strange and funny how usually AMD usually uses thing Intel has developed and always improves on them but now thet intel had the chance to imporve on something AMD made they couldnt. not very surpising really
 
Neither are 64 bit 'pure' Someone please correct me if I'm wrong, but the original AMD 64 bit processors had only a 40 bit memory address space. I believe the newer chips will/do support 48 bit memory addressing.

So no not true 64 bit in all senses.
 
The main reason they call it 64 bit is because of the ability to do 64 bit floating point operations without any 'work' arounds.

The 40/48 bit limitation is likely a hardware design shortcut to save silicon. The instruction set is likely capable of 64 bits. (That is why the new chips were released and there was little hubbub)

48 or even 48 bits of address space is more than enough for me. (I think thats something like 1024 Gbytes or 1 terabyte for 40 bit...)
 
AMD developed the x86-64 instruction set and started the 64-bit thing (not counting the Itanium). Intel finally got wise to the 64-bit shift, added a few instructions and claimed the EMT64 was different so that they didn't have to deal with copyright stuff.
Click to expand...

Actually the first 64-bit CPU was a MIPS unit, followed by the RS64, Alpha, SuperSparc, Power3, and then the Itanium. x86_64 was the latest major chip arch to go 64-bit.

So to answer your question Intel 64-bit is a (very, very) slightly expanded version of AMD 64-bit (once again not counting Itanium).
Click to expand...

AMD's x86_64 spec was implemented by Intel as sort of a reverse-engineering or clean-room type process if I heard correctly. EM64T may not be exactly similar to AMD64 in code but in implementation it is certainly compatible enough to run all of the same application binaries.

And in case you don't know, Itanium was Intel's first 64-bit proc, it wasn't compatible with 32-bit programs and had to run them through an emulator (killing performance). Itanium was intended only for the super-high end server market, and it has stayed there because of cost and incompatibility.
Click to expand...

Right. Itanium (IA64) is what Intel really wanted to bring to market after NetBurst to usher in the era of 64-bit computing. It might have worked if AMD hadn't made the Athlon 64 and its new instruction set. IA64 is actually sort of an option if you can get away with running Linux or any other *nix that is compiled for IA64 and compile your own apps. Supposedly Gentoo is pretty popular on the few people who have Itaniums at home as you compile everything from scratch and thus make your own programs to run on your CPU, making the application incompatibilities less of a problem. I'd personally not get one, but it would be neat to see just because it's an odd duck.
 
I have seen some and the Core 2s are faster than the Athlon X2s in 64-bit mode as well, although the margin is a bit less than in 32-bit. Ask Jack about his little spreadsheet on this topic. The Core 2 was made with 64-bit operation in mind like the A64, so its performance should not be a surprise. The reason that a lot of people think that it's EM64T that sucks is that the original chips it shipped with (Prescotts) had the 64-bit extensions tacked on and poorly implemented in hardware and as a result 64-bit code either ran no faster than 32-bit code or even in some cases ran slower. But that was the Prescott's fault, not the instruction set.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom