close

Privacy guaranteed - Your email is not shared with anyone.

Pentium 4 vs. Core 2 Duo? Same speed?

Discussion in 'Tech Talk' started by Viper_Audio, Oct 22, 2006.

  1. Viper_Audio

    Viper_Audio

    Joined:
    Jun 4, 2002
    Messages:
    72
    Likes Received:
    0
    Location:
    Michigan
    I have a P4 2.4 GHz machine, and I am looking to upgrade the Motherbord & Processor.

    In my price range is a Core 2 Duo, 2.4 GHz processor.

    If they are both the same speed, what am I gaining by upgrading?
     
  2. NetNinja

    NetNinja Always Faithful

    Joined:
    Oct 23, 2001
    Messages:
    967
    Likes Received:
    0
    Location:
    HotLanta, GA

  3. Tennessee Slim

    Tennessee Slim Señor Member CLM

    Joined:
    Apr 14, 2004
    Messages:
    4,413
    Likes Received:
    0
    Location:
    Mucus City, USA
    Don’t take too much stock in those numbers. It’s a scale of measure that hasn’t directly represented the CPU’s processing power since before the days of the first Pentiums.

    The original IBM PCs used ‘clock speed’ as a means of advertising a computer’s processing power, and that remains the standard more than 20 years later. Originally, they were just trying to impress and amaze the customer with how fast their computers could think by publishing how many computer code instructions the processor could execute in a split second. Enter the 4.77 MHz 8086. Four-point-seventy seven million thoughts a second. Jumpin' Jehosaphat!

    But processors were much simpler in those days and could only work on solving one problem at a time. The 8086 actually could execute just 4,770,000 instructions per second. Then along came the first “Pentium” processor, the i586/P5, introduced in 1993. What was different about it was that it featured something called “Superscalar Architecture”. In simplest terms, it could think on two problems simultaneously, which they called “multithreading”.

    Even though it wasn’t strictly limited by its clock speed, Intel chose to rate the new processor’s power by the existing yardstick anyway. It was “old school” rated at 60MHz despite the fact that it could work on solving two separate problems simultaneously. The first Pentium effectively had 120 MHz worth of processing power but was marketed as a 60 MHz.

    That was the camel’s nose under the tent. From that moment forward, the old yardstick no longer applied. Incongruously, even though continued microprocessor development has made it even more irrelevant, advertising for any new PC invariably features its CPU’s clock speed. Considering that some CPUs now crunch on as many as nine problems at once, clock speed is essentially meaningless unless you’re comparing like processors from the same manufacturer running at the same front side bus speed.

    These days, the only really relevant way of comparing processors apples-to-apples is benchmarking. Benchmarking is the computerized version of the Olympic Decathalon, a series of standardized events, each testing different capabilities and tallied up so as to give a level playing field comparison of the competitors. Benchmarking doesn’t know if a CPU is multithreaded, hyperthreaded, 32-bit or 64-bit and couldn’t care less how short or numerous its datapaths are. All it cares about is how long it takes to perform a standardized task.

    Fortunately, there are a number or reputable (and I use the term loosely) online review sites who review and benchmark all the new gear, so you and I don't have to (and so you and I will buy bling-bling from their sponsors). Here’s the PCMark 2005 benchmark of all the major X86/X64 desktop processors at Tom’s Hardware: http://www23.tomshardware.com/charts8/430-450-171.png
    The 2.4GHz Core 2 Duo E6600 benchmarked at 6168 (higher is better). The wimpiest P4 in this comparison was a 2.8 GHz Prescott. It scored 3522, and that CPU is quite a bit more powerful than your 2.4 GHz P4. I browsed the 19 other benchmarks as well and the Core 2’s superiority throughout seemed to be either similar to this test’s or markedly more dramatic. Any way you look at it, the 2.4 GHz Core 2 Duo shows a heap more processing power than your 2.4 GHz P4.

    Overall, according to Tom’s, the best Core 2 Duo CPU is dramatically brawnier than any P4 or the best from AMD. Other geek sites (AnandTech, TweakTown, Sharky, HardOCP, et. al.) seem to concur: Core 2 Duo is da shiznit!
     
  4. David_G17

    David_G17 /\/\/\/\/\/\/\/

    Joined:
    Oct 7, 2002
    Messages:
    2,046
    Likes Received:
    0
    aside from shorter pipelines, and running on lower voltage and producing less heat, the core 2 duo has two cores. That is (for all practical purposes) like having two processors on one chip.* It can handle twice the amount of threads running at once (think about it like this: you can scan your hard drive with a virus scanner while working in Microsoft Word without anything slowing down).

    Also, the e6600 and above have 4 Megs of onchip cache. That's a whole heck of a lot.

    * yeah, yeah; it's not exactly like two processors (they share the same FSB), but it's close enough.
     
  5. 22/45

    22/45 Snail Driver

    Joined:
    Oct 30, 2001
    Messages:
    412
    Likes Received:
    3
    Location:
    NC
    So is there an architecture difference between my dual-core Pentium D 805 and the new Core 2 Duo chips, or did it take Intel's marketing department a while to come up with a catch phrase?

    I know the actual cores are different (Smithfield vs Conroe), but are they both at the most basic level dual-core Pentium processors?
     
  6. David_G17

    David_G17 /\/\/\/\/\/\/\/

    Joined:
    Oct 7, 2002
    Messages:
    2,046
    Likes Received:
    0

    someone else can probably explain it better, but the conroes are built on a 65 nm process vs the smithfield on a 90 nm process.

    also, the FSB is doubled (iirc).

    http://www.semiconductor-technology.com/projects/intel/

    "Intel's 65nm technology roughly doubles transistor density. The additional transistors provide dual- and multi-cores and improved cache, to innovative technologies such as virtualization and security....

    The second generation of Intel strained silicon increases transistor performance by 10% to 15% without increasing leakage. Conversely, the transistors can cut leakage by four times at constant performance compared to 90nm transistors. Transistors have a gate length of 35 nanometers and a gate oxide thickness of 1.2 nanometers. The reduced gate capacitance ultimately lowers a chip's active power. The new process also integrates 8 copper interconnect layers and a "low-k" dielectric material.

    "Sleep transistors" in the 65nm SRAM shut off the current flow to large blocks of the SRAM when they are not being used to eliminate a significant source of power consumption, especially for battery-powered devices like laptops.

    An ultra-low-power 65nm process technology under development will deliver power savings on mobile platforms and small-form-factor devices. This process addresses sub-threshold leakage, junction leakage and gate oxide leakage. Total leakage is reduced by roughly 1,000 times from Intel's standard process while maintaining about 50% of the drive current."
     
  7. 22/45

    22/45 Snail Driver

    Joined:
    Oct 30, 2001
    Messages:
    412
    Likes Received:
    3
    Location:
    NC
    I ended up asking a work buddy who follows this stuff much more closely than I, and he basically condensed the technical features you mentioned (and thanks for finding that) to a roughly 30% performance gain over near-enough "speed" rating at a lower power consumption, when comparing my old D chip to the new Core 2 Duo stuff.

    Of course, it would be a little tough for me to consider upgrading, since my 805 D (2.66ghz dual core) cost me $125 (now down to $95) and the 2.66ghz Core 2 Duo is $509 :shocked: