Where to buy a bare-bones PC?

[azanon]

I understand what you were trying to clarify, and I was just trying to explain how you are wrong.   Multithreading refers to a single process with multiple threads of execution.   It is a useful concept for both single- and multi-CPU architectures.    Multiprocessing usually refers to a single application which consists of multiple *processes*, each of which can be allocated to a different CPU.

I never said otherwise.  Of course a conglomerate program that does upteen gillion things multiprocessors if  you ask said program to do 2 or more things.   I have said from the very beginning 2 separate processes (whether its one or two programs) WILL be accelerated by a dual core cpu.  Thats why i see this as semantics; what does it really matter if its two independant programs or one program that does two or more independant functions? In kind, multithreaded applications will also be accelerated by them.  I have said this from the very beginning.

So before you call me wrong, first try to figure out what i'm saying.

I believe the dual-core chips share L2 cache, which makes them a different animal than dual-CPU systems.   In some cases, sharing L2 cache may help, in other cases it may hurt vs dual-CPU.

You believe wrong.   They both have their own, independant on-die cache.  Want a link as proof or do you care to actually look into it yourself before you speak?

Yes if they had to share the cache, it might hurt performance.   Course we're only talking about in theory here, because they both have their own cache.

(edit) i know you wont believe me so i'm going to go ahead and paste a snip from page 2 of anand's article on the dual core p4.

"As we mentioned in our IDF coverage, Intel has dropped the number 4 from their naming for their dual core parts.  The new dual core desktop CPUs will simply be called the Pentium D and the Pentium Extreme Edition. 

Both the Pentium D and Pentium Extreme Edition are nothing more than two 90nm Prescott 1M dies glued together.  That means that each core has its own 1MB L2 cache, and that also means that architecturally, these chips are no different than the single core Pentium 4s that are out today - other than the obvious dual core fact. "

 

[wabmester]

I never said otherwise.

Sorry, it must have been somebody else who told me that multithreading was invented for multiple CPUs and that there was no such thing as a multiprocessing application.   

They both have their own, independant on-die cache.

That's interesting.   I just read up on the 800-series, and it looks like each core gets allocated 1MB L2 cache.   Today's 600-series single-core CPU has 2MB L2 cache.   It also looks like the dual-cores run at slower clocks, and the 2.80GHz version has hyperthreading disabled.   So, not only will most users not see a performance imrprovement, but the single core CPUs are likely to be both less expensive and faster.

 

[azanon]

Sorry, it must have been somebody else who told me that multithreading was invented for multiple CPUs and that there was no such thing as a multiprocessing application.

There was still some loss that occurred under the guise of semantics. I have said a dual core can processor two separate, independant events faster. If i understand that, then what does all this other bickering matter.

That's interesting. I just read up on the 800-series, and it looks like each core gets allocated 1MB L2 cache. Today's 600-series single-core CPU has 2MB L2 cache. It also looks like the dual-cores run at slower clocks, and the 2.80GHz version has hyperthreading disabled. So, not only will most users not see a performance imrprovement, but the single core CPUs are likely to be both less expensive and faster.

Good point. I'll be honest if it were me, i'd probably spend my bucks on a really fast single core, cause i'm a gamer, and i dont do other stuff at the same time.

 

[JB]

I have said from the very beginning 2 separate processes (whether its one or two programs) WILL be accelerated by a dual core cpu.

No, in order to see an improvment in performance the application must be designed in such a way that the load is distributed across multiple processors. (A single app can do this via parallel processing of data, or pipelining.) Take a typical simple example of 1 thread that does heavy duty processing,and another that handles UI. Although the UI may be offloaded to another processor the performance benefit is insignificant because the processor load is dominated by the other thread. Same for multiple applications. If one uses very little CPU, and the other lots the benefit will be very very small. That's why an app must be designed to expolit multiple processors. Adding threads doesn't necessarily help, it's how you add them that does.

Fixed/Float doesn't matter. What matters is CPU resource consumption (count CPU cycles.) Two different apps that swamp the CPU using fixed or float instructions will benefit provided the memory can keep up.

 

[Nords]

TH, I'd like to commend you for your admirable restraint in not jumping in between Wab & Az on the semantics of how big their multiprocessed threads are...

 

[wabmester]

TH, I'd like to commend you for your admirable restraint in not jumping in between Wab & Az on the semantics of how big their multiprocessed threads are...

Most of the pleasure I get from posting in this thread is knowing that TH must be chomping at the bit like never before. Whoaaaa, boy.

 

[Nords]

Most of the pleasure I get from posting in this thread is knowing that TH must be chomping at the bit like never before.   Whoaaaa, boy.

Watching you & Az geek-smack each other may be fun, but baiting a guy who's demonstrating good behavior is just plain mean!