em64t not true x86-64?

[SpoonMuffin]

http://www.chip-architect.com/news/2...ott_part2.html

Second integer core for 64 bit processing (not for multithreading)



It is as good as sure that the second 32 bit core is exclusively used for 64 bit processing, and in a way similar to the good old bit slices. There was the 4-bit AMD 2901 that could be used to build 16, 32 or 64 bit processors. The fact that makes it possible is because the core's is limited mainly to additive and logic functions. A 64 bit staggered addition will take a total of four 1/2 cycles but you can start two of them back to back on 1/2 cycle intervals. The latency to access the cache also does not need to be increased because of the extension to 64 (48) bit addresses. The higher part of the address is only used several cycles later to check the address tags with the TLB entries and not to access the data cache itself. What will increase with one cycle is the latency from an ALU instruction to a normal speed integer instructions. This delay will increase from 2 to 3 cycles. One extra pipeline stage is needed as well, resulting in a minor increase in the branch miss prediction penalty.



The reason that we can be so sure that the second core is not used to boost the 32 bit Hyper threading capabilities is the scheduler. This unit is by far the biggest entity on the Pentium 4 die. It is larger then all the Floating Point, MMX and SSE hardware together. It is not only big but it also consist mostly out of very timing critical optimized macro cells laid out by hand. It takes a lot of time and effort to change the scheduler. We've looked to it in detail and concluded that it has mainly remained unchanged on Prescott's die. This means that the maximum uOp throughput remains six per cycle using the same dispatch ports as the Pentium 4.

http://en.wikipedia.org/wiki/EM64T#D...4_and_Intel_64

There are a small number of differences between each instruction set. Compilers generally produce binaries that target both AMD64 and Intel 64, making the differences mainly of interest to compiler developers and operating system developers.

[edit] Currently

* Intel 64's BSF and BSR instructions act differently when the source is 0 and the operand size is 32 bits. The processor sets the zero flag and leaves the upper 32 bits of the destination undefined.

* AMD64 supports 3DNow! instructions. This includes prefetch with the prefix 0F followed by opcode 0D and PREFETCHW, which are useful for hiding memory latency.

* Intel 64 lacks the ability to save and restore a reduced (and thus faster) version of the floating-point state (involving the FXSAVE and FXRSTOR instructions).

* Intel 64 lacks some model-specific registers that are considered architectural to AMD64. These include SYSCFG, TOP_MEM, and TOP_MEM2.

* Intel 64 supports microcode update as in 32-bit mode, whereas AMD64 processors use a different microcode update format and control MSRs.

* Intel 64's CPUID instruction is very vendor-specific, as is normal for x86-style processors.

* Intel 64 supports the MONITOR and MWAIT instructions, used by operating systems to better deal with Hyper-threading.

[b]* AMD64 systems allow the use of the AGP aperture as an IOMMU. Operating systems can take advantage of this to let normal PCI devices DMA to memory above 4 GiB. Intel 64 systems require the use of bounce buffers, which are slower.[b/]

* SYSCALL and SYSRET are also only supported in IA-32e mode (not in compatibility mode) on Intel 64. SYSENTER and SYSEXIT are supported in both modes.

* Near branches with the 66H (operand size) prefix behave differently. One type of CPU clears only the top 32 bits, while the other type clears the top 48 bits.

http://www.chip-architect.com/news/2...%20die%20space

Basic 64 bit integer operations

A 64 bit extension by itself does not imply that the Integer Execution Unit and the Integer Register File have to be extended to 64 bit. A minimal implementation would simply use the 32 bit integer pipeline for 64 bit integer operations. The Floating Point/MMX/SSE pipelines are already 64 bit. No need for changes here.

The dual 'Rapid Execution' Units and the 32 bit register file run a twice the frequency and are together able to handle two 64 bit operations per cycle. (The Hammer is able to do 3 per cycle but its 64 bit additions might have twice the latency) The mechanisms to decode an operation into 2 sub-operations are already available in the pipeline. The 128 bit XMM/SSE operations for example are handled in two 64 bit pieces.

It would be advantageous if the basic functional timing of the rapid executions engines can remain the same. The current ones handle 32 bit additions as two skewed 16 bit ones. the 2nd addition starts 1/2 a cycle after the first when the carry bit is available. The newer integer ALU's seems to be fully 32 bit ALU's The same trick may thus be used to handle a 64 bit addition as two skewed 32 bit ones. Hardware for a full 32 bit addition takes about 15-20% longer as that for a 16 bit addition. It seems that Intel's circuit designers have closed this gap with novel design techniques like 'forward body biasing' et-cetera.

Also Intel has 36 bits physical address size whereas AMD has 40 bits


so if i understand this correctly, intels "64bit" truely is the hackjob that many amd fans have been saying it is from the start.

weird how they got all that $ for dev and they cant even make a true 64bit core......explains why amd wins when running in 64bit mode when compared to intel

now we need g2 and k8L dam it!!!

[niko084]

The Itanium and one of the lines of the Xeon are the ONLY true 64bit Intel chips. EMT64, just means they support 64bit apps... Basically its a bunch of bs to sell new chips, and they really walk their way around that direct question.

So in turn NO EMT64 is not 64BIT!

[DanTheBanjoman]

The first EM64T CPU's apparently lacked a few instructions, as far as I know this is an old issue and has been added a long time ago.

And yes it was a hackjob, a great one though. Intel added it in like 9 months from scratch.

weird how they got all that $ for dev and they cant even make a true 64bit core......explains why amd wins when running in 64bit mode when compared to intel

They have had a true 64 bit core for many years, AMD however has non, just 64 bit extensions which is different. Next to that Intel had their own extensions which is said to be in some P4 cores but never activated. It's also speculated that it didn't work correctly yet. However because AMD launched it's 64 bit extensions while Intel said it wasn't needed yet (which they are right at since who runs 64 bit now? PAE works fine for more RAM so 64 bit is still unused) However Microsoft created Windows X64 for AMD's extensions and said to Intel that they wouldn't make a 3rd 64bit Windows, so Intel was forced to abandon their own consumer 64 bit extensions and copy AMD's extensions. So why did AMD outperform early EM64T CPU's? Because Intel was forced to adapt at lightning speed.
Long story short, it was agreat piece of marketing of AMD which gave Intel a kick in the nuts.

However it's a very old story and doesn't apply anymore. Enjoy your K8L

[DBH]

Isn't core duo 2 native 64 bit though?

[DanTheBanjoman]

Originally Posted by niko084

The Itanium and one of the lines of the Xeon are the ONLY true 64bit Intel chips. EMT64, just means they support 64bit apps... Basically its a bunch of bs to sell new chips, and they really walk their way around that direct question.

So in turn NO EMT64 is not 64BIT!

There is no true 64 bit Xeon, all of them use the very same EM64T extensions as their desktop counterparts.
And as for EM64T being marketing bullshit, very true. AMD started that, Intel had to follow the marketing hype. Intel clearly stated that 64 bit is not needed for consumers. Intel just joined AMD's game.

[SpoonMuffin]

amd64 chips are true 64bit chips when used with a 64bit os, they "mode switch" to 64bit.

ia64 is a true 64bit chip from the ground up, sparc(niagra?) is tru64bit,

xeons used em64t, hence from what this stuff says they arent "true 64bit" as defigned by x86-6 specs layed out by AMD( the ones who invented x86-64 )

intels "copy" of amd64, dubed em64t is from what i understand some weird kind of emulation in hardware that uses 32bit alu's to do 64bit work, as such its not a "true 64bit x86" chip, this is from what i can tell why intel chips loose ground against amd64 chips(k8 chips) even conroe with its "updated) em64t(also called something like em32e or something like that)

i know intel invented their own 64bit-x86 design but ms went with amd(probblybecause they where ready to run and intel wasnt in any hurry for 64bit-x86 because "nobody needs it" as intel said since k8 was announced/hit the market.

most ppl still dont use 64bit os/apps but thats due to ms not doing jack shit to promote xp x64, they didnt get hardware makers behind them, they didnt push to get drivers for 64bit avalable , they just let it languish as they worked on vista, their next planed "upgrade" to windows(and yet another bloatfeast!!!)

[SpoonMuffin]

Originally Posted by DanTheBanjoman

The first EM64T CPU's apparently lacked a few instructions, as far as I know this is an old issue and has been added a long time ago.

And yes it was a hackjob, a great one though. Intel added it in like 9 months from scratch.



They have had a true 64 bit core for many years, AMD however has non, just 64 bit extensions which is different. Next to that Intel had their own extensions which is said to be in some P4 cores but never activated. It's also speculated that it didn't work correctly yet. However because AMD launched it's 64 bit extensions while Intel said it wasn't needed yet (which they are right at since who runs 64 bit now? PAE works fine for more RAM so 64 bit is still unused) However Microsoft created Windows X64 for AMD's extensions and said to Intel that they wouldn't make a 3rd 64bit Windows, so Intel was forced to abandon their own consumer 64 bit extensions and copy AMD's extensions. So why did AMD outperform early EM64T CPU's? Because Intel was forced to adapt at lightning speed.
Long story short, it was agreat piece of marketing of AMD which gave Intel a kick in the nuts.

However it's a very old story and doesn't apply anymore. Enjoy your K8L

seems to apply since conroe as intel fans will tell you KILLS/rapes/pilleges amd in 32bit, but in 64bit they just shrug and ignore the fact that it dosnt perform as well as conroe 32bit perf would emply.

and amd/intel have cross licencing that lets each of them"Copy" eachothers instructions, its why amd chips have sse1-2-3-4 and intel chips can have em64t(amd64"support" )'

it just seems strange that if its a "true" amd64 emplimintation used in those em64t chips (conroe included) that you dont see the perf gain you would expect.

as stated, lack of 64bit os support(windows drivers fort x64 suck ass to be kind) is why nobodys really running 64bit windows other then a few crazy people who think its "better" and the people running 64bit vista.

and yes intel had been working on 64bit x86, but it wasnt working as they wanted, thats why they fuse dissabled it in prescott cores, as i understand it from people who work for intels dev dept(they order parts thru the shop i work at because its close to home) it was kinda like ia64's first chips, didnt work out as well as intel had hoped, this isnt unexpected since they where trying to adapt *shudders* netburst to alot of new uses because they wherent being allowed to work on what they really wanted to make(they wanted to work on what we now call pentium-m and core/core2 years ago...but marketing still insisted "clocks sell" )

what i think would help current chips the most would be if ms pulled their heads out of their asses and re-designed/programed windows removing all that lagacy bagege like dos and 9x game/app support, do what apple did with osx let people install a compat tool, hell make it so that it installs a "virtual pc" like setup with windows 9x or dos or whatever you need on it, thats able to take advantege of gfx acceleration and such, unlike osx it wouldnt have a huge perf hit because it would still be on a pc platform(x86/x86-64) and would also remove alot of buggs caused by supporting old lagacy apps/code in the core os, Perf could be improved massivly for newer apps, hell i have seen what can be done with older computers.

little exparment for anybody intrested.

grab vector linux soho live 5.1.2 link
install it on an old windows pc, use enlitenment desktop insted of KDE, now compare that to windows 98/98se/ME/2k/xp on the same system, compare performance as well as quility of the desktop experiance, i have a laptop thats a 233mhz p1 with 2xxmb ram, 6.2gb hdd, 4mb chips and tech video, it runs BAD to be kind under windows, slow, buggy, and cant run any current windows apps(no office for example) now look at the same system under vlsoho, its fast, responcive, has up to date office apps, also its got another kool thing, crossover is installed and guess what, ms office is installed!!!!

its used by my family as a wordprosessor and audio player, try playing audio with 2k pro on it and injoy the system studdering (sound and windows use)

im no lin lover, infact most distros are a pain, but since a few people i know dirrected me to vl i have found it to be the best distro for low end systems, specly net boxes for people who shouldnt be allowed online(spyware/addware/ARGGGGGG)

[SpoonMuffin]

Originally Posted by DanTheBanjoman

There is no true 64 bit Xeon, all of them use the very same EM64T extensions as their desktop counterparts.
And as for EM64T being marketing bullshit, very true. AMD started that, Intel had to follow the marketing hype. Intel clearly stated that 64 bit is not needed for consumers. Intel just joined AMD's game.

i know a few people using unix/linux who would dissagree with the "64bit is bullshit" part, they do gfx/renderng/recording/editiing work and the extra ram and perf they get out of using 64bit *nix does help them, not as much as it could if the apps where more mature and optimized, but again, its still a gain, and 8+gb ram really does help when your dealing with large media files.

also know of a few companys running 2003 64bit for large database/AD servers where alot of ram REALLY improves perf(talking 16-32gb on a server) pricy yes, but when it means you need 4 servers insteed of 6 to allow everybody to work without waiting for server delays its worth it.

and yes i know intel came up with a way to address more then 4gb ram years back (liket he pentium2/3 days) but it performed poorly, and requiered a custom windows kernal and apps and then normal 32bit apps couldnt take advantage of it because they can only use the first 3gb system ram

[DanTheBanjoman]

Originally Posted by SpoonMuffin

i know a few people using unix/linux who would dissagree with the "64bit is bullshit" part, they do gfx/renderng/recording/editiing work and the extra ram and perf they get out of using 64bit *nix does help them, not as much as it could if the apps where more mature and optimized, but again, its still a gain, and 8+gb ram really does help when your dealing with large media files.

Well, it was Niko's statement that it's a bunch of bullshit, which for 99% of the users is true. The few cases where it is useful aren't mostly things end users do. I personally don't have a huge databaseserver or work with tremendous files that need to be loaded in the memory all at once.

Originally Posted by SpoonMuffin

also know of a few companys running 2003 64bit for large database/AD servers where alot of ram REALLY improves perf(talking 16-32gb on a server) pricy yes, but when it means you need 4 servers insteed of 6 to allow everybody to work without waiting for server delays its worth it.

Indeed, though if pricey isn't the issue there are countless of other server class platforms that outperform x86-64 with ease. IA64 for example is great as database server. (in fact it's one of it's main uses)

Originally Posted by SpoonMuffin

and yes i know intel came up with a way to address more then 4gb ram years back (liket he pentium2/3 days) but it performed poorly, and requiered a custom windows kernal and apps and then normal 32bit apps couldnt take advantage of it because they can only use the first 3gb system ram

That's PAE, and programs can address past 4GB of RAM, Windows has a special API for it. At least some SQL servers support it. However many programs which could use it do not. Blame the programmers, not Windows.




Anyway, the Intel not being fully compatible with x86-64 is an old story and is not true anymore for newer processors. And the fact remains that home users do not need it, for the few enthusiasts that really need it there are other more advanced platforms.
Then again nobody needed 1GB either in the late 80's. So in due time every computer will be sold with 16GB of RAM and we will laugh at that because we all have 64GB. Specially because Windows 2016 requires at least 32GB of RAM to work flawlessly.

[SpoonMuffin]

by then we will also use solid state hdd's that are uber fast!!!!

may endup being no ram just uber fast drives that work as both, that would rock really!!!!! everything effectivly stored in ram!!!!!

[Tatty_One]

Originally Posted by SpoonMuffin

by then we will also use solid state hdd's that are uber fast!!!!

may endup being no ram just uber fast drives that work as both, that would rock really!!!!! everything effectivly stored in ram!!!!!

Lol thats going backwards, thats how things used to be, in the early days (early 80's) there were a number of puters on sale that effectively had no hard drives, they had a small ROM module supported by 8.....16.....32.....or 64MB RAM! many without floppys but with cassettes, ahhhhh nostalgia, matchstick men was about the best detail you would get in a game and if it had a bouncing ball then that was yer ram used up!

[niko084]

Originally Posted by DanTheBanjoman

There is no true 64 bit Xeon, all of them use the very same EM64T extensions as their desktop counterparts.
And as for EM64T being marketing bullshit, very true. AMD started that, Intel had to follow the marketing hype. Intel clearly stated that 64 bit is not needed for consumers. Intel just joined AMD's game.

The Dual Core 5000 Xeons are 64bit, along with the new Itanium line, they are full blown 64bit. If you read between all the little lines on intels site, you can figure it out. I spent about 4 hours doing it trying to figure out exactly what emt64 was.

[SpoonMuffin]

Originally Posted by Tatty_One

Lol thats going backwards, thats how things used to be, in the early days (early 80's) there were a number of puters on sale that effectively had no hard drives, they had a small ROM module supported by 8.....16.....32.....or 64MB RAM! many without floppys but with cassettes, ahhhhh nostalgia, matchstick men was about the best detail you would get in a game and if it had a bouncing ball then that was yer ram used up!

as intel prooved with p-m and core, sometimes you gotta go "back" to move forword, solidstate drives are avalable now but the price is INSAIN!!!!

honestly, in a few years when they work out the tech to give drives long life using flash type media OR just use sdram style memory with a powerbackup built in, then bring the price down it will endup being ALOT faster and cheaper then current mech hdds, also lower power draws, i saw some beta drives in a demo from seagate and hitchi, GREAT stuff, mixed flash/sd/mech units that can preload large ammounts of data into fast access memory and then access less used stuff normaly, all done without the user knowing it(no special drivers or raid type setups needed) pure flash media drives, pure sdram(ddr/ddr2/ddr3/whatever) based units(fastest but also least secure and long lived)

would rock to have 2 of the 64gb drives hitchi was demoing, they could saturate the sata2 buss(each could saturate the buss alone) put ur os on them and, damn imagin the boot times and data access times!!!!, like having ur os on one of those ddr drive cards