Hyperthreading is sweet

Yep, that is basically the same motherboard ( although built proprietarily by Intel ) for my Dell 8300. It features onboard SATA and the 800 front side bus.

As for why the new Intel Prescott "E" and "EE" processors run so hot, I think that it is a matter of what happens when you go from .13 micron like the Northwood down to .090 micron with the Prescott. In any event, read the reviews on newegg and come to your own conclusion.

You might also wish to check www.tomshardware.com for some reviews as well. They are usually pretty insightful and in-depth.

All the Best!

:)
 
The BEST thing you can do, is actually benchmark your backtesting
system on the machine you plan to buy BEFORE you buy it.

Before I buy my next computer, im gonna sneak into best buy
with a 512meg USB memory key, plug it, and run a backtest
on all the newest chips...lol :D

peace

axeman
 
After reading the third paragraph on this page:

http://www.tomshardware.com/cpu/20040201/index.html

"Deciding against designating Prescott as the next Pentium generation may be because substantial architectural changes are within store in the near future that will be more becoming of a Pentium 5 handle. After CeBIT, socket 478 will be replaced by socket 775 as well as new chipsets that support DDR II memory and PCI Express. In addition, Intel CEO Paul Otinelli recently initiated a new discussion about possible 64 bit extensions for future Intel desktop processors. "

Maybe I can keep my antique Dell running a bit longer. There is nothing wrong with it I guess. I would like to have an extra hard drive and a DVD burner. Maybe I should just do those upgrades and wait out another cycle in CPUs. I don't do backtesting like Axeman, and I don't think my P4 2ghz has ever struggled with any of my mundane tasks. Hyperthreading does sound sweet though. I guess I will wait a bit longer to upgrade.
 
Like I said, why not just grab a Dell Dimension 8300 off the Dell Outlet Site for $684.00 and then spend another $120 bucks on another 512mb of RAM to get it up to a full gig?

This way, you will be totally cutting edge for what your needs are at this time and will have the Intel 875P motherboard that can handle all sorts of "flexibility" including upgrading to the new Prescott chips or grabbing a Northwood chip that runs cooler and also does Hyperthreading.
 
Yes, I agree.
And it even comes with XP-Pro at that price.

I admit that I have a "love/hate" relationship with "tinkering" with my computer. Sometimes . . . it's just good to stay focused on trading and avoid all the hassle and learning curve of building a new machine from scratch. Hell, just trying to install all the proper drivers could take a whole day!

But then there are times when you feel a great sense of accomplishment when you install something that you can actually feel make a difference. The key I think is to identify exactly what your needs are, and then go about a prudent and time/cost effective way of satisfying those needs.
 
Quote from axeman:

The BEST thing you can do, is actually benchmark your backtesting
system on the machine you plan to buy BEFORE you buy it.
More...

Tweaking the components and testing lots of hardware is great. But what about tweaking your code to run faster in general, or tweaking it to work better on your particular hardware?

Assuming you write your own code, you can optimize it to certain processors, system designs or cache sizes. I’m not saying you have to write stuff in assembly language, which I have never done. I just use straight C and Matlab. The main thing is to reduce cache misses by choosing appropriate sized chunks of memory to operate on. A profile of your code will reveal what sections are I/O bound, memory bound, or generating cache misses. You can re-order nested loops, change your memory access patterns, change data representation, and change the proportions of floating point and integer math. Just locate your bottlenecks and figure out alternative methods.

Performance improvements can range from a factor of 2 on up. I’ve personally seen improvements around 10 times based on reducing cache misses by loop reordering, and by converting certain floating point data and operations to fixed point.
 
Quote from prophet:

Tweaking the components and testing lots of hardware is great. But what about tweaking your code to run faster in general, or tweaking it to work better on your particular hardware?

Assuming you write your own code, you can optimize it to certain processors, system designs or cache sizes. I’m not saying you have to write stuff in assembly language, which I have never done. I just use straight C and Matlab. The main thing is to reduce cache misses by choosing appropriate sized chunks of memory to operate on. A profile of your code will reveal what sections are I/O bound, memory bound, or generating cache misses. You can re-order nested loops, change your memory access patterns, change data representation, and change the proportions of floating point and integer math. Just locate your bottlenecks and figure out alternative methods.

Performance improvements can range from a factor of 2 on up. I’ve personally seen improvements around 10 times based on reducing cache misses by loop reordering, and by converting certain floating point data and operations to fixed point.
More...
Jaja,

Sooo true.

nitro
 
Quote from MrJohnGalt:

Ummm... so I should eat salads for dinner...

:confused:

I have no idea what you just wrote.
More...

and you call yourself a "programmer"?? HUMMMMPPPH! :mad:
 
You must log in or register to reply here.
Back
Top