HyperTransport and Opteron/Athlon64 Overclocking

The first question many will have about our efforts to look at how Athlon64 will perform is how we can possibly compare an overclocked Opteron to a chip that is not overclocked. In the case of the Opteron, the comparison is more accurate than you might first think.

In normal setups (e.g. Athlon/P4), the CPU gets its clock from the FSB clock and multiplies it by the “clock multiplier” to determine how fast its internal clock should be. With a 16x multiplier, when the external clock ticks once, the CPU ticks 16 times. However, with the Athlon 64/Opteron, there is no FSB, so the CPU must get its clock from somewhere. It doesn't produce it internally; instead, it derives it from the native HT (HyperTransport) frequency, which is 200MHz, but because of the bus' nature, it runs at an effective 800MHz.

So, for our 1.8GHz Opteron 144, the multiplier is 9x, which is why raising the HT frequency to 222MHz increases the clock speed to around 2GHz. But we are increasing the HyperTransport clock in our overclocking, and not a FSB clock, which does not exist on Opteron/Athlon64. In real terms, this means our CPU overclocking has a significant impact on Performance, but it is unlikely that our increase in memory speed will have nearly as much impact on performance. Since we are nowhere near saturating the Hypertransport bus at 200 (effective 800), increasing HyperTransport to 222 (888) will not likely have much, if any, impact on overall performance. Our performance improvements, with Opteron/Athlon64, are mainly coming from increase in CPU clock — much more so than on the Pentium 4 or Athlon architectures.

Obviously, the PCI bus operates at a different frequency than the HT bus than the CPU, but they all operate based on multiples of each other, and are all derived from the HyperTransport clock.

nVidia nForce3 Chipset Performance Test Configuration
Comments Locked

79 Comments

View All Comments

  • StuckMojo - Friday, September 5, 2003 - link


    I'd like to see some benchmarks where the opteron's advanatges could really be used:

    1) some database or rendering benchmarks on workstations with more than 4GB of ram and large worksets that use more that 4G.

    x86/pentium cpus have to use a segmented memory architecture becuase of the 4G address space, so it's kind of like swapping, and is alot slower than direct access that a 64 bit chip has.

    2) how about some 64bit benchmarks on linux?

    Quake3 runs natively on linux, why have I seen none? Laziness, or lack of technical knowledge?
  • sprockkets - Friday, September 5, 2003 - link

    Intel will have a 100w toaster oven to compete with AMD, that and probably SSE3 just to make everyone recompile and distance AMD again. That and of course a 1mb cache.

    Isn't that Xenon have a 1MB but L3? That means it has 1.5MB, and still lags.
  • PointlesS - Friday, September 5, 2003 - link

    Impressive cpu but it would've been nice to see how much of an improvement the extra 200mhz made...unless I'm missing something here...does anyone have a link that has a 1.8ghz opteron and a 9800 pro?
  • Anonymous User - Friday, September 5, 2003 - link

    Minor correction on page 2. nForce was the first AMD based board to use HyperTransport IIRC, not the nForce2
  • Anonymous User - Friday, September 5, 2003 - link

    Hrm, as I am sure Intel has samples of the Opteron CPUs, its kinda surprising to see them not have anything significant ready to counter.. I highly doubt the Prescott will perform more than 10-20% better than the fastest Northwood P4.. If can't, then it'll definitely be slower than these new Opterons.. Let's see if Intel can counter, or else its gonna be a bloodshed for them..
  • Evan Lieb - Friday, September 5, 2003 - link

    #13,

    Because CPU supply right now is EXTREMELY tight. Wait until the end of the month for more info on dual Opteron/A64 performance numbers. :)

    Take care,

    Evan
  • Anonymous User - Friday, September 5, 2003 - link

    This is all very exciting stuff for AMD fans...but as a dually enthusiast, I wonder why there are no benchmarking stats for a top-end Athlon MP workstation/gaming solution? Why include the Xeon dually and not an Athlon dually? Certainly in the Content Creation areas we'd see a landslide for the good old dual MP mobos...imagine a 150 MHz FSB PC3200 2.6 GHz overclocked Barton 2500+ dually w/ the Radeon 9800...best price/performance ever IMHO...
  • Wesley Fink - Friday, September 5, 2003 - link

    As Anand and many others have been saying for months, Athlon64 is supposed to be single-channel Socket 754 and able to use unbuffered memory. Since the rumored FX Enthusiast version is said to be Socket 940 it will fit Opteron boards like the Asus SK8N, so will be dual-channel. Thus far the only ones I have seen from these aleady RELEASED motherboards have required registered memory - ECC or non-ECC - but that could of course change with later releases.
  • tazdevl - Thursday, September 4, 2003 - link

    Also, I was under the impression that the socket 939/940 boards will support unbuffered memory.
  • tazdevl - Thursday, September 4, 2003 - link

    I'd like to see a temp comparison between all the CPUs.

Log in

Don't have an account? Sign up now