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


View All Comments

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

    Oh little Dvinnen I petty you for your ignorance

    AMD Opteron Socket 940 at 2.0GHz (9x222) 444FSB

    The opteron has the equivalent to a MEM-controller between MEM and CPU. And it is quote running at "AMD Opteron Socket 940 at 2.0GHz (9x222) 444FSB". So be that the mem will run at 444 Mhz or 222 "Dobbel-pumped" Mhz

    So even the anandtech 'beeps' don't know the architecture of the CPU... Although FSB is a wrong id of it.

    Mvh Mjello

    The fool is not the one who is deceived. But the one who wants to be deceived...

    By the way. The hypertransport-link is for peripherals.. Not memory

    Not that I really care enough to answer you little dwuuip but I am in a rare mood
  • Anonymous User - Friday, September 5, 2003 - link

    one more (hopefully constructive) suggestion - how about commentary on the benchmarks next to the appropriate graphs so we can see the #'s while we read the comments.
  • Anonymous User - Friday, September 5, 2003 - link

    quick suggestion for you guys - how come the graphs are labeled with motherboard names? shouldn't they be labeled with CPU names since this is a CPU comparison. Just skimming the article and looking at the graphs I had no idea what was being compared. If need be, put the chipset in parenthesis next to the CPU name & speed.

    excellent article though! athlon64 looks very imrpessive
  • Icewind - Friday, September 5, 2003 - link

    Wow, so far AMD looks like there on the right track, but its going to be interesting to see how Intel fires back at that with Prescott and Tejas, not to mention the change to PCI Express and what not. Next year is gonna be insane thats for sure. For now, im enjoying my 2.8.@3.2 ASUS P4C800-E, 1 Gig corsair rig and its screams. Im in looking into a 9800pro or 9800XT for my video card and im set for a good long year.
  • sprockkets - Friday, September 5, 2003 - link

    Yeah, fire hazard, that's why a 3.2P4 uses 82w of power just to sit there, and the 3200+ XP uses 62.

    Dell won't touch AMD due to Intel forcing them to.

    Intel has billions to spend on R&D, and look where it's gotten them, the loser Itanium.

    And with P4 clock speeds like a 140mhz would make the world of difference on the 54% it is behind.

    Why use 4x512MB? Don't we all need 2Gb of ram to function best?

    You're right, due to the P4 freq binge AMD did introduce its + ratings. Isn't amazing that a processor running slower memory and 1060 mhz less can still be faster?

    Funny, 20 out of 20 of my AMD systems don't have quirks in them, must be a stupid Windows user.
  • dvinnen - Friday, September 5, 2003 - link

    <<<and with 22 Mhz OC to the memory.>>>

    Wow. Did you even read the artical or know anything about The K8 line? There is no FSB to oc so you can't oc the memory that way. It gets it's external clock (aka:fsb) from the hypertransport line. So they oced the hypertransport to get the speed from there 1.8 opteron to 2.0, the shiping shipped of A64. OCing the fsb on K8 dosen't do anyithing but increase the clock becuase the HT line is about as saturated as a AGPx8 line.
  • Anonymous User - Friday, September 5, 2003 - link


    A lot of pepole play games on Linux, Q3 on Linux is actually a bit faster on my system then in w2k.

    But that's not the point. The point is to see how much there is to gain from having a 64bit OS vs. 32bit. I bet id will compile, and release, a 64bit Q3 client for linux sometimes soon. Hope there will be one for UT2003 too.
  • Anonymous User - Friday, September 5, 2003 - link

    A sad day I must say.

    Not because AMD beat Intel's best (3.2 Ghz)but because AMD only beat intels second best (3.0 Ghz)and with 22 Mhz OC to the memory.

    Anandtech is no more than a marketing machine. It's not worthy of our attention.

    It is a biased benchmark which twists reality in order to make AMD 64 look better.

    Now I must say that I would like this to be a total succes for AMD however I don't like being lied to. And that's what anandtech and AMD is doing in this review. It looks as if AMD is in to deep and might just go under because of their financial problems.
  • Anonymous User - Friday, September 5, 2003 - link

    Why not use 4x512MB memory for the P4 setup, since Anandtech's own setups have shown its best, and also, why a 3200+ Barton and 3.0GHz P4, that's not an equal comparison.

    Also, the layout seemed awful, there was a page just before the end with about 1/3 a page of text.

    And it did seem like there was a lot os ass kissin gin the article as well.
  • Anonymous User - Friday, September 5, 2003 - link


    So many lies out of your mouth.

Log in

Don't have an account? Sign up now