64bit, really all that?

[Gamingphreek]

Well Nemesis... ya see there are these people who are sorta called hardcore computer gamers. And people who really care about computers. Its OUR hooby/job. If you dont care OK... thats fine, maybe it just doesn't appeal to you... but somepeople me included enjoy coputers, and the competition.

-Kevin

 

[gwag]

GOOD POST Jeff
"So really what im asking is if the 64 bit applications will just blow me away or just the average tech upgrade" simple answer = it?s just the average tech upgrade.
NO! 64bit isn?t all that, not that athlon 64's are bad. It will provide nowhere near the jump in performance as going from 16 to 32bit. When operating on floating point (decimal) numbers, a 64-bit register would seem to give added precision, as it can hold more numbers after the decimal place, but the x86 architecture, on which all current 32-bit Intel and AMD CPUs are based already provides for 64-bit floating point registers (actually 80 bits internally) so no advantage is gained. More registers are good but not gona help out gaming or word-processing rendering HTML etc. Its advantages lie in using more than 4G of ram which some 32bit processors already can. The increase in speed will come from better processor architecture and better compiled code. Look at distributed computing programs say DNETC (a simple one) for example the app has many different cores for different processors the difference in performance on an athlon using a P4 core results in 60% less performance. So if you have a 64bit beta for a game right now it is made for an A64. Quake for instance is still and may never be compiled for an athlon XP, if it was it may run 10-30% faster. 128 bit video cards still have 32 bit GPUs (cpus) and use 32bit Floating point but a 128 bit memory path. It may make a difference for folks managing huge databases, Lucas Film or Pixar but they already have been using this technology.

 

[Don66]

My question to you is..
What are you going to do with it?
If you plan on gamming then YES, my A64 Pretty much smokes my 2.8c in gamming.
If you are multitasking, then I would say go with Intel for that kind of action.

 

[Gamingphreek]

Originally posted by: gwag
GOOD POST Jeff
"So really what im asking is if the 64 bit applications will just blow me away or just the average tech upgrade" simple answer = it?s just the average tech upgrade.
NO! 64bit isn?t all that, not that athlon 64's are bad. It will provide nowhere near the jump in performance as going from 16 to 32bit. When operating on floating point (decimal) numbers, a 64-bit register would seem to give added precision, as it can hold more numbers after the decimal place, but the x86 architecture, on which all current 32-bit Intel and AMD CPUs are based already provides for 64-bit floating point registers (actually 80 bits internally) so no advantage is gained. More registers are good but not gona help out gaming or word-processing rendering HTML etc. Its advantages lie in using more than 4G of ram which some 32bit processors already can. The increase in speed will come from better processor architecture and better compiled code. Look at distributed computing programs say DNETC (a simple one) for example the app has many different cores for different processors the difference in performance on an athlon using a P4 core results in 60% less performance. So if you have a 64bit beta for a game right now it is made for an A64. Quake for instance is still and may never be compiled for an athlon XP, if it was it may run 10-30% faster. 128 bit video cards still have 32 bit GPUs (cpus) and use 32bit Floating point but a 128 bit memory path. It may make a difference for folks managing huge databases, Lucas Film or Pixar but they already have been using this technology.

What do Lucasfilm and pixar use. I know they use like server farms... but what processor and equipment do they run. Could they be running like 100's of those enterprise chips like the Itanium 2... or maybe even stuff like Crusoe and Efficeon (i know those two are like 256bit processors ... i think). Also would they be using special proprietary Video cards or would they be running a whole mess of Quadro class cards?

Your only not seeing an improvement in gaming is because there are no 64bit games. It runs in 64bit in the OS and then lags as it switches back and "emulates" 32bit. Once games have 64bit support and optimizations, then you will see a difference.

-Kevin

 

[oldman420]

where you would see the biggest boost is in processor intensive tasks like file transfers, database apps servers etc" if you have a 64 bit bus and hdwr" in day to day computing and games there is a theorhetic doubling of the bandwidth. but to take advantage of it you need a 64 bit bus pci slots etc. to me it seems like putting a huge engine in a pinto, so what, get me a drive train to match then we will play.
the world is going 64 bit as ultimatly that is the same as doubleing the clock on a 32 bit chip. but the technology for the home user is still a little young.

 

[Boomer76]

Would it not be possible to check the UT2004 64bit English Linux Demo for speed improvments over the 32bit version ?

 

[imported_oralpain]

Originally posted by: oldman420
where you would see the biggest boost is in processor intensive tasks like file transfers, database apps servers etc" if you have a 64 bit bus and hdwr" in day to day computing and games there is a theorhetic doubling of the bandwidth. but to take advantage of it you need a 64 bit bus pci slots etc. to me it seems like putting a huge engine in a pinto, so what, get me a drive train to match then we will play.
the world is going 64 bit as ultimatly that is the same as doubleing the clock on a 32 bit chip. but the technology for the home user is still a little young.

64-bit registers (wich is what makes a 64-bit cpu 64-bit) have nothing to do with any bus. 32-bit cpus are more than capable of useing a 128-bit memory controler and 64-bit PCI slots to their full capacity for example. The biggest boost form useing a 64-bit cpu and 64-bit apps would be in number crunching. Compiling, scientific, and rendering programs should see a moderate boost.

I highly doubt that there will be much more than a 10-20% speed improvement in most 64-bit programs (even fully optimized ones) over 32-bit programs, all other things being equal. If a program uses more than 4 gigs of ram then there will be a much more substantial improvement

I have plenty of 16-bit programs and their 32-bit successors. The 32-bit versions are generally no faster on the same system.

 

[imported_oralpain]

Originally posted by: Boomer76
Would it not be possible to check the UT2004 64bit English Linux Demo for speed improvments over the 32bit version ?

The 64-bit UT2004 Linux demo is significantly slower than the 32-bit one. It needs alot of work.

 

[gwag]

Originally posted by: Gamingphreek

What do Lucasfilm and pixar use. I know they use like server farms... but what processor and equipment do they run. Could they be running like 100's of those enterprise chips like the Itanium 2... or maybe even stuff like Crusoe and Efficeon (i know those two are like 256bit processors ... i think). Also would they be using special proprietary Video cards or would they be running a whole mess of Quadro class cards?


-Kevin

They have used a mixed bag of Sun ,SGI, intel in the past most big rendering done on a Sun farm "One frame in Finding Nemo, distributed across Pixar's 2,000-processor render farm, took 10 hours to render - and lasted just 1¼24 of a second on screen" have now begun switching to Mac OS X and G5 workstations for its production work soon I bet an X serve render farm.

 

[Boomer76]

Originally posted by: oralpain

Originally posted by: oldman420

I highly doubt that there will be much more than a 10-20% speed improvement in most 64-bit programs (even fully optimized ones) over 32-bit programs, all other things being equal. If a program uses more than 4 gigs of ram then there will be a much more substantial improvement.

But would not 10-20% actually be an enormous speed increase ?

I mean some people are paying 5-600 dollars more to the FX version of the A64 wich actually has much less speed increase than 20%.

The way I see it you can get a Athlon 64bit processor now for less than 200 dollars so why NOT buy it instead of the Athlon XP ones ?

 

[Vee]

"I highly doubt AMD will ever put hyperthreading on thier chips. AMDs dual core cpus are only a year away and 2 real cores/cpus is deffinatly better than 1 with hyperthreading."

They are, actually. It's awfully confused right now about interpreting what technologies belong to which core, since AMD are apparently working on both K9 and K10. But my guess is that multi threading in the core will be a feature of the K9. The gain with this (hyper) threading is that you can use the execution units better. AMD will expand their concept of OoO execution for a superscalar core, and threading in the core should make it possible to squeeze out more work from the core. K9 will also feature some massive improvement in vector instruction extensions, that will be more suitable than SSE2 for lower clocked cpus.

There is a slight possibility that the dual K8 core is designated "K9". (The rumoured new technologies then belong to K10) I don't believe that is the case though. I think K9 is a seperate new core. K9 will probably debut soon after the desktop dualcore K8. It's intended for .065 micron process, but early examples maybe will be .090micron.

 

[Vee]

Originally posted by: oralpain
I highly doubt that there will be much more than a 10-20% speed improvement in most 64-bit programs (even fully optimized ones) over 32-bit programs,

Well, the additional registers, and their flexible use, should be able to do more than 10-20%.
We'll have to see, but I think one early example, the lame encoder for Linux, improved like 100%?

Originally posted by: oralpain
If a program uses more than 4 gigs of ram then there will be a much more substantial improvement

A 32-bit programs cannot easily use more than 2GB space of addresses for code and data. And this has nothing to do with physical memory size. It's the process virtual space. And it gets fragmented!
This is the really big need for 64-bit computing. We're at limit with 32-bit. That limit is 2GB and you can compare it to the old 640KB limit of early PCs.

Originally posted by: oralpain
I have plenty of 16-bit programs and their 32-bit successors. The 32-bit versions are generally no faster on the same system.

Well, that sounds a bit strange. But it's all up to judgement. Anyway, there's no technical similarities between the 16 to 32 bit migration and the 32 to 64 bit migration. Two completely different things.

 

[Vee]

Originally posted by: LTC8K6
As long as AMD continues to produce new 32bit chips, I am not very worried about whther I should be running 64bit or not.

Depends on what you're doing. If you're into 3D games or computer 3D rendering or various engineering apps, you will Have to move to 64-bit, rather soon I think.

 

[imported_michaelpatrick33]

the sempron is for the developing country loooooooooooooow end. they are designed to match the celeron while consolidating their manufactoring lines. It will be interesting if the 754 and 939 will have disabled 64bit that can be enabled at some point or not. It does not matter about 32bit being around for awhile becuase the AMD 64 are so fast at 32 bit.

 

[imported_oralpain]

Originally posted by: Vee
Well, that sounds a bit strange. But it's all up to judgement. Anyway, there's no technical similarities between the 16 to 32 bit migration and the 32 to 64 bit migration. Two completely different things.

How are they that much different? It's basically the same idea. Going from 16-bit to 32-bit registers and address space should be quite similar to going from 32-bit registers and address space to 64-bit (though the a64/opteron only has 40-bit addressing it should not be a limit that will be reached any time soon).

I agree that the reall need for 64-bit computing are limitations with 32-bit memory addressing, but it's still going to take a while before most home users will hit those limits. Dealing with huge amounts of data will definatly be helped.

I have seen lame perform much better in 64-bit over 32-bit. Still there are many programs that show little or no improvement. Only time will tell how much of a difference we really see.

 

[sdgserv]

I have been using and Intel 3.2 for awhile. Runs strong as ever. I just built an 64 XP3400 and have been using it for about a week. Most benchmarking test are better with the AMD, that being said. In everyday usage the 3400 eats my Intel, it is quicker opening all the programs that I use. It is rock solid. I have been building systems for about 16 years and the AMD 64 impressed the heck out of me. It feels powerful like an Intel. There is nothing scientific about my observation. Just my opionion.

 

[eagle101]

well let me say this,i have a 3400/64 ulta 320 scsi hd and controllers and a prometeia cooling system im putting together sunday.....i do have a neopro k8n running fine and the music on the board is awsome
soon i will get the FX55 and Blow everybody away ill keep you posted sdgfish
...................eagle 101

 

[Accord99]

Originally posted by: Vee
Well, the additional registers, and their flexible use, should be able to do more than 10-20%.
We'll have to see, but I think one early example, the lame encoder for Linux, improved like 100%?

Which had more to do with the lousy 32-bit GCC compiler than anything else. It was not any faster than the standard 32-bit Win32 binary and a 3GHz P4 would have beaten the A64's encoding time with the 64-bit binary.

A 32-bit programs cannot easily use more than 2GB space of addresses for code and data. And this has nothing to do with physical memory size. It's the process virtual space. And it gets fragmented!
This is the really big need for 64-bit computing. We're at limit with 32-bit. That limit is 2GB and you can compare it to the old 640KB limit of early PCs.

Most typical programs don't come close to the 2GB limit. And the mainstream S745 platform generally can't even handle 3GB of memory, less than Intel's dual channel motherboards.

Well, that sounds a bit strange. But it's all up to judgement. Anyway, there's no technical similarities between the 16 to 32 bit migration and the 32 to 64 bit migration. Two completely different things.

Yes no comparison at all. 16->32 bit was an essential migration with massive benefits to all users and still took years for a complete transition. 32->64 is more niche at the moment, and not absolutely necessary for years to come.

 

[glugglug]

64-bit registers and addressing is NOT the only advantage of AMD64. It also doubles the number of general purpose registers from 8 to 16. THAT would be the biggest reason that windows API gets a boost. The # of floating point and SSE registers is doubled in 64-bit mode as well, and the FPU on an AMD64 CPU is 128-bit. The other reason is that contrary to popular belief, there are A LOT of 64-bit values handled by windows itself. File position counters have to be 64-bit or you couldn't handle files larger than 4GB. block #s and other values in NTFS are 64-bit as well. The high resolution counters used for timing in games are 64-bit as well, because with 32-bits the timing resolution tracked would roll over every 10 minutes or so. Lots of 64-bit perfmon counters as well.

 

[glugglug]

Originally posted by: Accord99
Most typical programs don't come close to the 2GB limit. And the mainstream S745 platform generally can't even handle 3GB of memory, less than Intel's dual channel motherboards.

Socket 754 board supporting 3GB
another one
and another one
still more
and more

IIRC Socket 754 max is actually 8GB, just no mobo manufacturer has put 8 DIMM slots yet. Socket 940 max is theoretically 1TB (good luck finding a mobo with 1024 DIMM sockets).

 

[Accord99]

Originally posted by: glugglug
Socket 754 board supporting 3GB
another one
and another one
still more
and more

3GB of PC2100 maybe, but not likely at PC3200 and even if it could, it is still less than "only 32-bit" dual channel P4 platfomrms. See the second last paragraph:
http://www.aceshardware.com/read.jsp?id=65000291

File position counters have to be 64-bit or you couldn't handle files larger than 4GB. block #s and other values in NTFS are 64-bit as well.

Not particularly important since the bottleneck is the hard drive.

The high resolution counters used for timing in games are 64-bit as well, because with 32-bits the timing resolution tracked would roll over every 10 minutes or so. Lots of 64-bit perfmon counters as well.

I doubt games are interested in timing precision better than milliseconds, or that they even account for a miniscule amount of time that the game spends in the CPU .

 

[caz67]

Wait until Longhorn..Then 64Bit will be mature and mainstream. It will then be worth the change to 64.

 

[Vee]

Originally posted by: Gamingphreek
Also people think 64bit...WOW!!! Some ignore the fact that there are processors that are 256bit. (Lol cant remember the name). Computing still has a long way to advance beyond 64bit. Were only at the beginning.
-Kevin

No. You are totally misunderstanding the CPU bit issue! Todays "32-bit" cpus, like the P4, are already width-wise 128-bit wide. This is really so, we have 128-bit registers and instructions operating on 128 bit wide vector fields. Also in other ways, there are even paths inside the cpu that are 256 bit wide. And this is on a "32-bit" CPU. And we are probably going to see 256-bit, maybe 512-bit registers and vector operations on AMD's K9 and Intels Conroe.
The 16/32/64-bit cpu issue has mainly to do with the length of the address field used by instructions to refer to data or code. And since 64 bits are good for a program space of 16 ExoBytes, there is certainly no need for more than 64 bits in our lifetime.
You should not understand computer code- and cpu-bits in the game console bit paradigm. Computing width is already, as I said, 128 bits wide in SSE2. And we are probably going to see computing bit-widths hit four figure numbers on "64-bit" cpus. Computing width can be anything, in a "32-bit" or "64-bit" cpu. It's not the issue.

 

[Vee]

Originally posted by: Accord99

Originally posted by: Vee
A 32-bit programs cannot easily use more than 2GB space of addresses for code and data. And this has nothing to do with physical memory size. It's the process virtual space. And it gets fragmented!
This is the really big need for 64-bit computing. We're at limit with 32-bit. That limit is 2GB and you can compare it to the old 640KB limit of early PCs.

Most typical programs don't come close to the 2GB limit. And the mainstream S745 platform generally can't even handle 3GB of memory, less than Intel's dual channel motherboards.

First of all, this is not the question of "most typical" programs today. It's the question of what we want to use the computer for.
Secondly it's about the space available to the process! Not size of RAM!
I'm already using up the 2GB space, when I'm working. That's an engineering app. But here's a few other cases that can use a larger than 2GB process space, approximately from like today: Editors and toolsets for 3D games, 3D games, 3D modelling tools and computergenerated 3D imagery and animation, cinematic editing.
Thirdly, what's the main reason for development of DDR2 and DDR3? Speed? No, signal quality enabling more RAM.

Well, that sounds a bit strange. But it's all up to judgement. Anyway, there's no technical similarities between the 16 to 32 bit migration and the 32 to 64 bit migration. Two completely different things.

Yes no comparison at all. 16->32 bit was an essential migration with massive benefits to all users and still took years for a complete transition. 32->64 is more niche at the moment, and not absolutely necessary for years to come.

16->32 bit was primarily migration from segmented addressing to linear addressing. As such it was real nice. Since the opportunity was taken to also change the OS&program model to multithreaded and with real virtual memory and memory protection, it was even nicer. ...And almost necessary. At least it was a big relief.

32->64 bit on the other hand is certainly really a necessity, since it is about expanding process space and removing a 2GB limit that we are banging our heads against. A situation that is only going to become worse, and has not much to do with the amount of available physical RAM, which everybody seem to wrongly assume. And again, the opportunity is taken to change the ISA to more flexible use of registers. Which should grant a nice performance kick. But that's just icing on the cake, not the reason for 64-bit computing. Everything else is just Intel trying to mislead the market.

 

[0roo0roo]

i mostly drool over the powernow and integrated mem controller. 64bit is bonus. http://www.silentpcreview.com/article169-page1.html