To say that AMD has been uncharacteristically quiet lately would be an understatement of epic proportions. The company that had been so vocal about their K8 architecture in the past will hardly say anything at all about future products, extending even to its forthcoming AM2 platform. In just two months AMD is scheduled to officially unveil its first DDR2 platform (Socket-AM2), but we've heard virtually nothing about performance expectations.

Back in January we sought to discover for ourselves what AMD's Socket-AM2 platform would have in store for end users. You'll remember that when Intel made the shift to DDR2 it basically yielded no tangible performance improvement, and we were all quite afraid that the same would be true of AM2. When we finally tested the AM2 samples that were available at the time, performance was absolutely dismal. Not only could AMD's AM2 not outperform currently shipping Socket-939 platforms, but due to serious issues with the chip's memory controller performance was significantly lower.

Given that AMD was supposed to launch in June at Computex, the fact that AM2 was performing so poorly just five months before launch was cause for worry. Despite our worries, we elected not to publish benchmark results and to give AMD more time to fix the problems. We're not interested in creating mass panic by testing a product that's clearly premature.

In February we tried once more, this time with a new spin of the AM2 silicon, but performance continued to be lower than Socket-939. Luckily for AMD, the performance had improved significantly, so it was slower than Socket-939 but not as much as before.

The next revision of the AM2 silicon we received sometime in March, and this one finally added support for DDR2-800, which is what AM2 will launch with supposedly at Computex. With the launch only three months out, we expected performance to be at final shipping levels, and we were left disappointed once more. Even with DDR2-800 at the best timings we could manage back then, Socket-AM2 was unable to outperform Socket-939 at DDR-400.

That brings us to today; we're now in the month of April, with less than two months before AMD's official unveiling of its Socket-AM2 platform at Computex in June, and yes we have a brand new spin of AM2 silicon here to test. We should note that it's not all AMD that's been holding AM2 performance behind. The motherboard makers have of course gone through their fair share of board revisions, not to mention the various chipset revisions that have changed performance as well. Regardless, according to internal AMD documents, AM2 CPUs are going to start being sold to distributors starting next month, leaving very little time for significant changes to the CPU to impact performance. We feel that now is as good of a time to preview AM2 performance and put things into perspective as we're likely to get before the official launch.

What's AM2?
Comments Locked


View All Comments

  • Calin - Monday, April 10, 2006 - link

    DDR2-800 at 4-4-4 should be equivalent to DDR-400 at 2-2-2. Also, DDR2-800 at 6-6-6 would be the same (latency-wise) as DDR-400 at 3-3-3.
  • Furen - Monday, April 10, 2006 - link

    Not quite, DDR2-800 at 4-4-4 is the equivalent of DDR-400 at 4-4-4 because the memory cells run at 200MHz on both modules. Like I said above, though, module latency is not the only factor affecting the total latency, so perhaps DDR2 memory controllers help mitigate this huge latency hit. One of the main reasons why DRAM manufacturers love DDR2 is because their yields are much higher than they are on higher-clocked, aggresively-timed DDR1 due to the higher prefetch (lower operating clock) and the looser timings.
  • defter - Monday, April 10, 2006 - link


    Not quite, DDR2-800 at 4-4-4 is the equivalent of DDR-400 at 4-4-4 because the memory cells run at 200MHz on both modules.

    That's not true. "Cas latency" values are relative to the 400MHz clock instead of 200MHz base clock that DDR2-800 has.
  • Furen - Monday, April 10, 2006 - link

    The 400MHz clock is the clock the IO operates at, while the memory arrays operate at half the IO clock, so 200MHz in this case (so yes, DDR2 ram operates at a sort of quad data rate). Since a Column Access Strobe is a memory array operation it is, naturally, measured in memory array clocks. The "base clock" for DDR2 is actually 400MHz because it is the external clock.
  • menting - Monday, April 10, 2006 - link

    defter is correct,
    time delay on memory is calculated by the clk speed that the memory takes in * latency
    think of it as a black box operation.
  • MrKaz - Monday, April 10, 2006 - link

    OK. Didnt know that.

    I always tought that DDR1 2-2-2 was always better than higher DDR2 numbers...
  • Furen - Monday, April 10, 2006 - link

    It is, the main factor affecting latency is the memory cell clock, which runs at the same clock on both modules. So you can do a straight comparison between the two latencies to see which will yield you a better MODULE latency. Of course, module latency is just one part of the whole latency equation, the memory controller being the other big part. Perhaps AMD made the controller more latency-friendly by sacrificing maximum bandwidth, which would explain the abnormally-low usable bandwidth.
  • ozzimark - Monday, April 10, 2006 - link

    just something to keep in mind. same 1.8ghz cpu clock:

    200mhz at 2-2-2 = 51.5ns
    300mhz at 3-3-3 = 43.8ns

    mhz wins over timings when it comes to comparing absolute latency
  • Furen - Monday, April 10, 2006 - link

    That only applies when comparing the same type of memory.

    DDR2 memory cells run at 1/4 the "Effective clock," so DDR2 800 runs at 200MHz, which is the same as DDR 400.
  • ozzimark - Monday, April 10, 2006 - link

    true, but you notice the latency that is in the review. seems that what i say holds true to an extent

    btw, the timings are in signal clocks last i checked, not cell clocks, which runs at 1/2 the speed of the double data rate signal. kinda confusing, but oh well. point of the matter is that ddr400 at 2-3-2 is higher latency than ddr2-800 at 4-5-4

Log in

Don't have an account? Sign up now