RSS

DDR2-800 vs DDR3-1333, Does Speed Matter?

Mon, Feb 11, 2008   

(4 votes)
Loading ... Loading ...

Gaming, Technology


DDR Memory Sticks

Summary: NO (Incase you don’t want to read further)

I set out to build myself a new desktop recently and one of the decisions I had to make was “Do I go with DDR3, or stick with DDR2?”, then of course if you stick with DDR2, do you go with oldie 800 ram or shoot for the 1066 or higher but pay the premium? I wasn’t sure at the time, so I set out to some do some research, and here is what I found…

First off, DDR3 has been on the market for a while now, but you never hear about it for 2 reasons:

  1. It’s expensive as hell, usually about 2x the price of DDR2 ram or more.
  2. There is no noticeable performance improvement using it.

The reason for #2 is the same reason DDR2 wasn’t awesome right away when it first came out: the latencies are much higher and even though the bandwidth ceiling is much higher, we aren’t even maxing out what we have already… so it’s a moot point.

I was originally looking at building a machine using the Intel QX9650 and figured a quad-core beast with a 1333mhz FSB would definately need either DDR3 ram or the highest end DDR2 (1600?) RAM I could scrounge up.

Once I went looking and prices and realized that in some cases the price of 1GB of expensive DDR3 ram was the same price was 4GBs of DDR2-800 ram, I decided to look up some benchmarks to see if this mess was even worth it.

NOTE: For the folks that didn’t do a lot of computing in the 90s and building their own machines, finding “good” ram with awesome “timings” like 1 and 2 CAS latencies used to make a big difference, especially in gaming. But as computers have gotten faster and faster the difference good memory allows is completely negligible now except for overall stability… that is still important when buying good ram.

As I trapes’ed my way around online I ran across a great benchmark from Digit-Life comparing DDR2-800 to DDR3-1333… memory that couldn’t be farther apart on the performance scale; and exactly what I wanted. If there was a benefit to DDR3 this benchmark was going to make it clear.

The other great thing about this benchmark is that it used both the new, unreleased QX9770 and the QX9650 to saturate the memory with the fastest quad-core CPUs available today (and likely for the next 6 months). The results were surprising to say the least.

I immediately scrolled down to the most CPU-intensive task I could find, Video Encoding:

DDR2 vs DDR3 Video Encoding Benchmark

notice that there is absolutely no difference between the DDR2-800 and DDR3-1333 setups during this task?

Then I scrolled down to a normalized gaming score between the two and again, found almost absolutely no difference:

DDR2 vs DDR3 Gaming Benchmark

I figured with the added bandwidth provided by DDR3 that games atleast would be running faster with so much texture work to push, but apparently we aren’t even making full use of that is provided by DDR2-800 at this point… meaning something else in the PC is the bottleneck.

This is equatable to how SATA-2 connections allow for a 3GB/sec bandwidth right now, but a typical SATA hard drive will burst around 110MB/sec and sustain around 50-70MB/sec… no where near the 3GB/sec cap on the bus (unless you had some insane RAID setup).

So the good news from all of this is that you can safely buy DDR2-800 ram for $40/GB instead of the fucking crazy DDR3 ram for the forseeable future.

And don’t even bother with DDR2-1066 or higher DDR2 ram… I found benchmarks on those as well comparing to DDR2-800 that again showed almost no difference in performance.

Share This on Your Favorite Social Network:
  • Digg
  • del.icio.us
  • Facebook
  • Mixx
  • Google
  • Fark
  • Furl
  • Propeller
  • Reddit
  • Slashdot
  • Technorati
  • IndianPad
, ,

14 Comments For This Post

  1. Ralph Kradolfer Says:

    Thanks a lot for your research, you saved a lot of (my) money ;-)

  2. Riyad Kalla Says:

    Ralph, I’m glad I could help!

  3. Nelton Says:

    Dear Riyad,
    can you send me your pc specs? Maybe you also did some tests like 3D mark as well?
    I’m trying to put a game-pc together for a low price and still good enough to play recent games in good resolutions…

  4. Riyad Kalla Says:

    Nelton,
    The machine turned out being *awesome*, I am really happy with it about 70% because of it’s performance being better than I expected (the 8800 is a hell of a card) and 30% because the price was so damn cheap. I thikn you could build this machine for like $1400 today (NOTE: I didn’t need a monitor, if you need a monitor too it will be closer to $2k or 2100)

    * 3.0 Ghz Dual Core E6850
    * 4GB 800mhz DDR2 Ram (Corsair)
    * BFG 8800 GT (the new 512mb GT that is as fast as the Ultra cards, not the original 256mb one)
    * Corsair 620 watt Modular PSU
    * 2x Seagate ES.2 750 GB Drives (the enterprise-SATA ones)
    * ASUS MAXIMUS FORMULA

    All of the hardware is great but I WOULD NOT recommend the ASUS boards anymore. The computer actually hangs on bootup at the memory detection phase (I tried 4GB of Crucial and Corsair, same thing) randomly. It’s a known issue that has not and may not be fixed by ASUS, I’m relaly annoyed by that. Luckily hitting the Bios-wipe button lets me reboot… but that means I can never successfully make any modifications to the BIOS and use the computer… luckily I don’t but it’s a fucked up position ot be in.

  5. jz Says:

    Great post! But I have to wonder why. 3-4 years back I bought a P4 3.2ghz northwood system with pc3200 dual channel memory which was twice as fast as my old system. All benchmarks were about 2X. Everything was about double my P1.6ghz (oc to 2.1ghz). I knew the ram was about double the speed. The system was responsive and fast. It couldn’t be just because the cpu was changed.

    Comparing 800mhz to 1333mhz shows a 1% change in speed. But current computers are about double the speed of my system 3-4 years back with 400mhz ram.

    So I now wonder. If I could put a new Q6600 quad OC’d to 3.6ghz with the older 400mhz ram. Would the ram be a bottle neck? Would it slow things down? OR could we simply just put 400mhz ram in new systems and have the same results?

    If it’s true that 800mhz is needed today instead of 400mhz, then I would think that there must be a sweet spot for ram for these cpu’s. That’s what my guess is, but I have no way to prove it. If this is true, then it might mean that DDR3 will SOMEDAY be very useful if we get CPU’s that are maybe 1.5X to 2X faster than what we have now. Say maybe 2 years from now. It’s just a guess. But for today, I can see 1333 is a huge waste to build a new system. Thanks for the advise.

  6. jz Says:

    One other thought. Maybe those benchmarks are not improved much with faster ram. Could there be some other type of applications that would give a 66% speed improvement? Or probably not? If not, it must be that the cpu is the bottleneck somehow. hmmm.

  7. Riyad Kalla Says:

    jz,
    I’m glad the article helped. There are two factors with ram that make it “fast”, and they seem to be mutually exclusive, they are:

    * Latency
    * Bandwidth

    For some reason it seems impossible to have both, you either have 1 or the other. For example, most of these DDR2 sticks of ram have latencies of 2 and 3, but low bandwidth.

    DDR3 is ultra-high-bandwidth, but has latencies of 5, 6 and 7. Twice as much latency as the DDR2 pieces.

    So the question becomes “what will I see the biggest benefit from today?” and as you correctly stated, a lower latency stick of RAM is going to serve you better with today’s CPUs and peripherals than a higher latency ultra-fat-bandwidth stick of ram.

    In a few years where games like Crysis are the norm, and to play a game you have to stream Gigabytes off your 300 MB/sec solid state hard drive, then ultra-high-bandwidth ram that can handle streaming insane amounts of data si going to be necessary and your lower-latency ram won’t make such a difference.

    This is sort of the premise of the PS3, it doesn’t have that much ram on it, and *everything* is designed to be considered a stream… streaming audio, graphics, video, etc. but it does it as an insanely high rate of speed.

    Honestly I expect whatever Nehalem+1 to really shine and Nehalem+2 to be truely next-next gen where DDR3/4 are “Absolutely necessary”.

    So probably like 24/32 months from now. We should have the first Nehalem chips drop at the end of this year, so we’ll get a peak at what will come with the new redesign, but probably be 2010 before our pants are blown off by it.

  8. jean Says:

    thanks for this great comment on ddr2-ddr3, idont know too much about memory latency and stuff but wanted to do some research first on ram befre deciding to go for a ddr2 or ddr3 mobo

  9. Riyad Kalla Says:

    Jean,
    No problem. I really don’t expect computers to take advantage of DDR3 ram and beyond until Intel releases Nehalem chips at the end of this year and into 2009. I’m sure in 2010 CPUs will be quite a bit faster and using higher bandwidth RAM. 2009 will be a year of CPU speed growth hopefully.

  10. Ratul Says:

    The Sweet spot for core 2 duo (memory wise ) is ddr 667. Look in wikipedia for more details.

  11. Mitra Says:

    Dude… Try it on the latest Core 2 Quad.

  12. Riyad Kalla Says:

    Mitra,
    That is what the QX9770 is. Seriously, you aren’t going to start seeing performance differences until these chips are 4Ghz+ I don’t think.

    Either way Nehalem is going to be DDR3 only I believe, so we are all going to revamping our systems next year anyway.

    But if you want to build a fast budget machine, DDR2-800 all the way.

  13. roy Says:

    you chose only benchmarks that don’t need memory bandwidth, so of course you wont see any difference in performance.

  14. Riyad Kalla Says:

    Roy,

    Please clarify how Video Encoding and Games aren’t memory-bandwidth intensive.

    Besides running a straight-up memory bandwidth test, I can’t think of any other *normal* use for a computer that would use more memory bandwidth.

    After all the purpose of the article is to point out that with current CPUs on the market, DDR2-800 ram is going to give you the same performance as high-speed DDR3 ram because the other components in the computer just aren’t saturating the existing memory subsystem.

    So using a complete synthetic test that would never occur in the real world (synthetic bandwidth test) doesn’t help, because in the end everybody doing normal work with their PC are never going to realize those performance benefits.

Leave a Reply