DDR3-1333 Speed and Latency Shootout

Tomorrow's Bandwidth...Today!

It's often argued that DDR2 memory is certainly fast enough for today's processors, as Intel's current fastest Front Side Bus (FSB) uses a 1333 MHz data rate. But doesn't this newer FSB require "1333 MHz" memory to perform? The simple answer is no.

Intel has been using Dual Channel mode ever since the days of RDRAM on its earliest Pentium 4's to double memory bus width, simply because back then it was impossible to find memory that was both the same speed and the same bus width as its FSB. The earliest Pentium 4's used a 64-bit FSB with a 400 MHz data rate before DDR-400 became available, but two 64-bit DDR-200 (PC-1600) modules would suffice by doubling the memory bus width to 128 bits...assuming you could actually find a DDR SRAM chipset for the Pentium 4 back then. Dual-Channel technology has stuck around so that today's FSB-1333 is easily fed by two DDR2-667 (PC2-5300) modules in dual-channel mode.

Another argument can be made for running memory "synchronously" to the CPU's FSB: Isn't DDR3-1333 synchronous to FSB-1333? Again the answer is no, as Intel's FSB uses Quad Data Rate technology while the memory is only Double Data Rate. FSB-1333 runs at a 333 MHz clock rate, which is the same clock rate as DDR2-667.

Yet, many users have found small performance benefits from running RAM at up to 1.5x the CPU FSB clock, violating to concept of "synchronous clock speed" superiority. Indeed, this is the likely reason why DDR2-667 became popular long before Intel's FSB-1333 was even considered, and probably why DDR2-800 sells well even to non-overclockers.

While most PC builders won't "need" anything faster than mid-priced DDR2 for a while, DDR3 holds two key benefits over the technology it replaces: First, its maximum chip density has been extended to 8 Gb, allowing a 16-chip module to support a maximum 16 GB capacity. Second, its default voltage has been reduced to 1.50 volts from DDR2's 1.80 volts, resulting in a 30% power consumption decrease per clock speed.

To Buy Or Not To Buy?

One major factor favoring the purchase of DDR3 memory is that Intel is slowly moving all of its chipsets in that direction. The firm first introduced DDR3 support as an option on its P35 Express Northbridge, and the DDR3 market has further been expanded to around half of the newer X38 models. Motherboard manufacturers expect big spenders to be the earliest adopters of new technology, so the majority of ultra-expensive X48 chipset motherboards will likely support only this latest memory standard while it gradually works its way into lower-cost markets.

The latest technology always comes at a significant penalty in value and DDR2 is more than sufficient for most systems, so why the big push? Intel is likely preparing the desktop market for something big, specifically a move of the memory controller from the chipset to the CPU itself. As with AMD's current products, this design eliminates the bandwidth limitations of a FSB and allows future processors to receive data as fast as it can be translated.

It's up to the buyer to decide whether he or she would like to bear the burden of opening up future technology to the masses. Many of us still remember when RDRAM was pushed "needlessly" into Pentium III chipsets such as the i820 and i840 as Intel prepared its i850 Pentium 4 chipset to use the same memory format. The plan was to increase RDRAM availability prior to its necessity, but the market reacted unfavorably. Similarities of today's DDR3 push end there, as Intel isn't forcing it into any market but instead making it a performance option.

That's not to say, however, that DDR3 on Intel's FSB is without merit, as its significantly higher data rate is perfect for FSB overclocking. FSB-1600 (400 MHz FSB clock) is just around the corner, and anyone looking to push a 2.80- GHz FSB-1600 processor (400 MHz FSB clock x7) to 4.20 GHz (600 MHz FSB clock x7) will need memory capable of a 1200 MHz data rate (600 MHz DRAM clock). DDR2-1200 is a rare breed that requires extremely high voltage, top cooling and high hopes that it doesn't quit due to the fact that it's really just overstressed DDR2-800.

So while the majority of Core 2 system builders could compare DDR2-800 prices to those of various DDR3 models, overclockers can view DDR3-1333 as a faster, more economical and reliable alternative to DDR2-1200. Furthermore, as DDR3 edges closer to mainstream availability, overclockers with more restrictive budget should see it eventually reaching their own comfort levels.

Thomas Soderstrom
Thomas Soderstrom is a Senior Staff Editor at Tom's Hardware US. He tests and reviews cases, cooling, memory and motherboards.
  • dv8silencer
    I have a question: on your page 3 where you discuss the memory myth you do some calculations:

    "Because cycle time is the inverse of clock speed (1/2 of DDR data rates), the DDR-333 reference clock cycled every six nanoseconds, DDR2-667 every three nanoseconds and DDR3-1333 every 1.5 nanoseconds. Latency is measured in clock cycles, and two 6ns cycles occur in the same time as four 3ns cycles or eight 1.5ns cycles. If you still have your doubts, do the math!"

    Based off of the cycle-based latencies of the DDR-333 (CAS 2), DDR2-667 (CAS 4), and DDR3-1333 (CAS8), and their frequences, you come to the conclusion that each of the memory types will retrieve memory in the same amount of time. The higher CAS's are offset by the frequences of the higher technologies so that even though the DDR2 and DDR3 take more cycles, they also go through more cycles per unit time than DDR. How is it then, that DDR2 and DDR3 technologies are "better" and provide more bandwidth if they provide data in the same amount of time? I do not know much about the technical details of how RAM works, and I have always had this question in mind.
  • Latency = How fast you can get to the "goodies"
    Bandwidth = Rate at which you can get the "goodies"
  • So, I have OCZ memory I can run stable at
    7-7-6-24-2t at 1333Mhz or
    9-9-9-24-2t at 1600Mhz
    This is FSB at 1600Mhz unlinked. Is there a method to calculate the best setting without running hours of benchmarks?
  • Sorry dude but you are underestimating the ReapearX modules,
    however hard I want to see what temperatures were other modules at
    a voltage of ~ 2.1v, does not mean that the platinum series is not performant but I saw a ReapearX which tended easy to 1.9v(EVP)940Mhz, that means nearly a DDR 1900, which is something, but in chapter of stability/temperature in hours of functioning, ReapearX beats them all.
  • All SDRAM (including DDR variants) works more or less the same, they are divided in banks, banks are divided in rows, and rows contain the data (as columns).
    First you issue a command to open a row (this is your latency), then in a row you can access any data you want at the rate of 1 datum per cycle with latency depending on pipelining.

    So for instance if you want to read 1 datum at address 0 it will take your CAS lat + 1 cycle.

    So for instance if you want to read 8 datums at address 0 it will take your CAS lat + 8 cycle.

    Since CPUs like to fill their cache lines with the next data that will probably be accessed they always read more than what you wanted anyway, so the extra throughput provided by higher clock speed helps.

    But if the CPU stalls waiting for RAM it is the latency that matters.
  • what is on pc3-10600s "s" ?