1-25-05: 64bit < HT ?

[DAPUNISHER]

Originally posted by: Megatomic
64bit is the future.

64bit is the future for this area of usage as well.

None of that detracts from the usefulness of HT, but as it appears headed for obsolesence, and I have a dual Opteron system, I have no need for it. Subsequently, I voted for 64bit because I will still own this system when the Finalized XP Pro 64bit and supporting software are available. At which point it appears evident that significant performance gains in some areas and ability to run new software will even futher extend its useful lifespan to me.

 

[gwag]

try running a simple multi core app. like the DNET cleint. watch HT trip and stumble all over itself getting about 60% of the performance as when running in silgle core mode. also note how an athlonXP at 2.0ghz is faster by 30% than a 3.4ghz P4 with hyper threading.

 

[Megatomic]

Thanks for the link DAPUNISHER, I'll check out this software.

 

[whorush]

hey, 64 bit isnt useless if you use linux!

also, vian, check out my earlier post about HT and another thread about HT, http://forums.anandtech.com/me...adid=1504149&forumid=1

point is, i contend that HT wouldnt help AMD because AMD has a short pipeline that is getting enough data. if it were to screw up, its not too big of a bubble.

win64 is supposed to come out end of april.

 

[Megatomic]

You people calling 64 bit worthless (and similar) are going feel foolish when x64 Edition goes Gold and all the software vendors release packages for it. I'm personally dying to try out UT2004 x64.

 

[Accord99]

Originally posted by: gwag
try running a simple multi core app. like the DNET cleint. watch HT trip and stumble all over itself getting about 60% of the performance as when running in silgle core mode. also note how an athlonXP at 2.0ghz is faster by 30% than a 3.4ghz P4 with hyper threading.

Try running multiple instances of other CPU intensive applications, like Seti or Folding@Home, ScienceMark, PiFast, SuperPI and see how a 3GHz P4c handily beating FX53s and FX55s.

 

[Sc4freak]

Originally posted by: gwag
try running a simple multi core app. like the DNET cleint. watch HT trip and stumble all over itself getting about 60% of the performance as when running in silgle core mode. also note how an athlonXP at 2.0ghz is faster by 30% than a 3.4ghz P4 with hyper threading.

I've found that Hyperthreading provides a benefit. I did some simple testing, I got a 30% boost with HT while running a SMP aware program.

 

[DAPUNISHER]

Originally posted by: Accord99
Try running multiple instances of other CPU intensive applications, like Seti or Folding@Home, ScienceMark, PiFast, SuperPI and see how a 3GHz P4c handily beating FX53s and FX55s.

Please do not run 2 instances of F@H with HT because you are holding up analysis of the results! Competition is great and I know eveyone wants to have a s many points as possible but in this case it is at the expense of the science involved
Read Me Here is a quote

While it is well known that an HT CPU can run 2 work units because the OS treats it like two CPU's, and you can gain a possible 15-30 percent increase in points for that computer, it also means that each work unit is returned more slowly. In simple terms, if the project has 300 generations needed to test a theory on a protein model, and running two instances at once delays the return of work for 1 day each time, you end up with a 300 day delay. That translates to about a 10 month delay in examining your data for the final result. The bottom line is simply this: Run 1 instance for each CPU you have. An HT CPU is not two CPU's, it is one. Let's work to advance the science and spend a little less time worrying about the number of points you get.

 

[uOpt]

You can gain reliable 64 expectations from Linux who are at full 64 bit production ready for some time now.

Personally I take hyperthreading. The performance benefit is neglectable but I can do screen-intensive stuff much better when there is video encoding or compilation running in the background.

64 bits goes on my nerves because of slight unavoidable issues here and there. For example, some of the plugins for my browser are not available for 64 bits and that forces my whle browser chain into 32 bits. Pretty big pain. I didn't gain much of a code quality increase out of porting to 64 bits yet.

Hyperthreading on the other hand allows me to develop multithreaded code and have it hang earlier. Running an SMP kernel is also an advantage in itself if you fiddle with drivers and the like because you'll have a first idea whether what you did is SMP-safe.

At work 64 bits is a must, we are out of the 3 GB virtual memory. I am tired of splitting things, only to be forced to implement cross-process longjumps right thereafter ...

 

[VIAN]

try running a simple multi core app. like the DNET cleint. watch HT trip and stumble all over itself getting about 60% of the performance as when running in silgle core mode. also note how an athlonXP at 2.0ghz is faster by 30% than a 3.4ghz P4 with hyper threading.

link please, I find this very hard to believe.

hey, 64 bit isnt useless if you use linux!

also, vian, check out my earlier post about HT and another thread about HT, http://forums.anandtech.com/me...adid=1504149&forumid=1

point is, i contend that HT wouldnt help AMD because AMD has a short pipeline that is getting enough data. if it were to screw up, its not too big of a bubble.

win64 is supposed to come out end of april.

I see, well, I guess that changes everything now. HT is less than useless compared to 64-bit so, if anyone made it this far in the thread, my vote has changed to 64-bit.

 

[jamesbond007]

Originally posted by: AWhackWhiteBoy
if Intel is dropping everything and making a dash for 64bit also, then you know its going to have very large future potential.

It's all marketing hype. It doesn't have to be or have a very 'large future potential'. We didn't see AMD/nVIDIA scram to produce RDRAM-compatible chipsets or memory controllers. If the market is willing to buy it, there will surely be suppliers (Intel and AMD) that are gearing up to produce it.

I chose HyperThreading. It's useful today, benefits A LOT of applications, and it doesn't slow down applications as much or at all, when it first came out. This tends to be a very misleading 'issue' among AMD enthusiasts, which gives them a reason to bark at Intel.

In order for Windows XP64 to really shine, *ALL* drivers and software will have to be converted or rewritten to a 64-bit capable format. Remember the ATi 64-bit beta drivers? Upon release they were half, if that, of the current 32-bit drivers available.

Until everything, from soundcard drivers, NIC drivers, and everything in between is 64-bit ready and optimized, then I'll buy into it. Until then, I'm a HyperThreading man.

 

[trend]

Even though this topic seems to be a new idea.. it has been around a long time.

Personally I have a good bit of experience with this subject as I do embedded system design for a corporation. It is crazy to think how far we have gone.. for example

I remember hearing stories about the first 4bit microprocessor! Just think.. 4bit microprocessors... For all you programmers out there think about declaring a variable as a byte.. YOU COULDN"T! If you wanted to store a single letter in this proc's memory bank, you had to store 1/2 the letter at one address and the other at another address.. then keep track of this! Wow!

So next time you think about Declaring your variables... Bit, Byte, String, Variant, etc think about where we can from.. and how great it is to even be able to declare one byte at a single address...

But you ask me how does this deal with 64bit vs HT... well it deals with it a lot. moving from 32bit to 64bit creates twice as large memory banks... which is good if you need to declare LAARRRRGE variables.. but for most programs (I doni't even think I could write one that would be able to use such a large variable effeciently) these large variables are useless.. and for the companies that say they are using them.. yeah.. they probably are..but just to say they have 64bit drivers... because it is the new cool thing""
example: dim X As SuperLongVariableName (yeah Superlongvariablename would be different for diff programming languages.. but basically that takes the place of : bit,byte,string,variant,long,etc)
then x = 25666

Technically yes I am using 64bit programming.. but only because I want to say I am.. and you know what... this causes the computer to have to access a larger address and this causes slower performance....

bottom line, most applications will run slower on 64 bit procs vs 32bit.. but not noticably slower... And HT is kick butt (I think this concept is pretty easy)

----
There are three ways to measure a processors "size";

How many bits of data a processor works with
How many bits a processor uses to address memory
How many bits can move around at once

----
I copied and pasted the above from:
http://www.mackido.com/Hardware/64Bit.html

Which I definitly agree you should read if you are interested in this stuff.. and want to verify my above statements.


Well.. I probably need to get back to programming my 8bit AVR