Sign in with
Sign up | Sign in

Team Group Announces First 3GHz DDR3 Memory Modules

By - Source: Team Group | B 28 comments

Taiwanese DRAM module maker Team Group has unveiled the first 3 GHZ DDR3 device.

The 4 GB Xtreem LV PC3 24000 modules (8 GB kits sold as 2x4 GB) are offered in limited numbers in and are designed to run at CL11 13-13-35 timings on Intel's Ivy Bridge platform with 3000-series processors and Z77 chipset.

The manufacturer said that the modules comply with the XMP v1.3 specification and use bifurcated heat spreaders to guarantee a high signal quality and proper cooling.

Team Group is also offering its Xtreem LV series in 2133 MHz, 2400 MHz, 2600 MHz, 2666 MHz and 2800 MHz versions.

Display 28 Comments.
This thread is closed for comments
Top Comments
  • 23 Hide
    blazorthon , May 31, 2012 7:49 PM
    A Bad DayI was so excited, until I noticed this:I am sad.


    Timings are measured in clock cycles, so with higher clock frequencies, the same timings number is a lower real-time latency as measured in nanoseconds (or any other unit of time measurement). 13-13-35 at 3GHz is like 6-6-18 on a 1600MHz memory kit. They are very low latency. Complaining about timings like these at 3GHz is like complaining about 5-5-18 timings at 1600MHz, except with nearly the same latency, you have nearly double the bandwidth. It's a win-win for latency and bandwidth.
  • 22 Hide
    eddieroolz , May 31, 2012 7:11 PM
    Until just 4-5 years ago we were amazed at RAM that ran at 533, 677 and 800MHz. But 3000MHz? Wow!
  • 14 Hide
    Pinhedd , May 31, 2012 8:28 PM
    eddieroolzUntil just 4-5 years ago we were amazed at RAM that ran at 533, 677 and 800MHz. But 3000MHz? Wow!


    The actual DRAM module speeds haven't improved all that much. The actual chip frequency for DDR3-3000 is around 375 Megaherz. DDR3 chip frequencies are one quarter that of the IO bus frequency and there are two data transfers per IO bus clock. This corresponds to the prefetch buffer size of 8 words per access. By comparison, DDR2 had 4 words per access and DDR had 2.

    Thus, the actual memory chips haven't gotten that much faster, the IO bus has.
Other Comments
  • 1 Hide
    jacobdrj , May 31, 2012 7:09 PM
    Very cool. What is the price premium? How good is overclock-ability on other brands' high end but not as highly rated memory?

  • 22 Hide
    eddieroolz , May 31, 2012 7:11 PM
    Until just 4-5 years ago we were amazed at RAM that ran at 533, 677 and 800MHz. But 3000MHz? Wow!
  • 6 Hide
    A Bad Day , May 31, 2012 7:21 PM
    I was so excited, until I noticed this:

    Quote:
    13-13-35 timings


    I am sad.
  • -6 Hide
    jdamon113 , May 31, 2012 7:22 PM
    http://www.youtube.com/watch?v=2WNrx2jq184
  • 6 Hide
    silentdrop , May 31, 2012 7:23 PM
    Really high latencies. But the speed should compensate.
  • 4 Hide
    victorious 3930k , May 31, 2012 7:32 PM
    Corsair beat them to announcing it. Google "Corsair Dominator Platinum".
  • 23 Hide
    blazorthon , May 31, 2012 7:49 PM
    A Bad DayI was so excited, until I noticed this:I am sad.


    Timings are measured in clock cycles, so with higher clock frequencies, the same timings number is a lower real-time latency as measured in nanoseconds (or any other unit of time measurement). 13-13-35 at 3GHz is like 6-6-18 on a 1600MHz memory kit. They are very low latency. Complaining about timings like these at 3GHz is like complaining about 5-5-18 timings at 1600MHz, except with nearly the same latency, you have nearly double the bandwidth. It's a win-win for latency and bandwidth.
  • 3 Hide
    Soul_keeper , May 31, 2012 8:23 PM
    Is this using 30nm chips ?
  • 14 Hide
    Pinhedd , May 31, 2012 8:28 PM
    eddieroolzUntil just 4-5 years ago we were amazed at RAM that ran at 533, 677 and 800MHz. But 3000MHz? Wow!


    The actual DRAM module speeds haven't improved all that much. The actual chip frequency for DDR3-3000 is around 375 Megaherz. DDR3 chip frequencies are one quarter that of the IO bus frequency and there are two data transfers per IO bus clock. This corresponds to the prefetch buffer size of 8 words per access. By comparison, DDR2 had 4 words per access and DDR had 2.

    Thus, the actual memory chips haven't gotten that much faster, the IO bus has.
  • 11 Hide
    beardguy , May 31, 2012 8:51 PM
    Most of us are probably already running around 1.6 GHz RAM.

    Real world speed increase ... probably barely noticeable.
  • 6 Hide
    RealBeast , May 31, 2012 8:56 PM
    beardguyReal world speed increase ... probably barely noticeable.
    +1
  • 7 Hide
    bucknutty , May 31, 2012 9:05 PM
    But they have bifurcated heat spreaders!

    In other words the heat spreaders are forked, or divided into branches... I had to look that up.
  • 12 Hide
    blazorthon , May 31, 2012 9:09 PM
    beardguyMost of us are probably already running around 1.6 GHz RAM. Real world speed increase ... probably barely noticeable.

    Realbeast+1


    It depends greatly on the workload. In gaming, higher RAM frequency tends to not help much in the maximum FPS, but it can help in the minimums and in CPU limited games if the CPU is memory bandwidth limited (such as AMD's CPUs and Intel's older CPUs). When it comes to other workloads, such as folding, archiving, rendering, AES encryption on AVX supporting CPUs, and the like, faster RAM quickly becomes important.
  • 10 Hide
    blazorthon , May 31, 2012 9:09 PM
    blazorthonIt depends greatly on the workload. In gaming, higher RAM frequency tends to not help much in the maximum FPS, but it can help in the minimums and in CPU limited games if the CPU is memory bandwidth limited (such as AMD's CPUs and Intel's older CPUs). When it comes to other workloads, such as folding, archiving, rendering, AES encryption on AVX supporting CPUs, and the like, faster RAM quickly becomes important.


    Also, APUs (such as Llano and Trinity) benefit greatly from faster RAM.
  • 0 Hide
    beardguy , May 31, 2012 9:25 PM
    While I don't disagree, you guys seem to be mentioning the exceptions here. Specialized software and the like is not what most people are going to be running.

    I would bet money that even though this RAM is technically 100% faster than what I am currently running, I would be hard pressed to even notice much of a speed increase. Specs are cool and all, but I know from experience that they can be deceiving. Don't fall for the numbers game.
  • 7 Hide
    blazorthon , May 31, 2012 9:45 PM
    beardguyWhile I don't disagree, you guys seem to be mentioning the exceptions here. Specialized software and the like is not what most people are going to be running. I would bet money that even though this RAM is technically 100% faster than what I am currently running, I would be hard pressed to even notice much of a speed increase. Specs are cool and all, but I know from experience that they can be deceiving. Don't fall for the numbers game.


    Someone who buys 3GHz memory is either someone with a lot of money to throw around for no good reason or is someone who can use the huge performance very well. Your complaint is like complaining about a $2K 3.xGHz eight core Xeon not improving your gaming performance over a $220 i5-2500K. That isn't what it should be used for, so it's not a very valid complaint. It is helpful of you to mention that this memory won't help a lot of consumer/gaming software for most consumers who aren't using APUs, but you're then talking about using it in areas where it should not be used in.
  • -6 Hide
    beardguy , May 31, 2012 10:11 PM
    blazorthonSomeone who buys 3GHz memory is either someone with a lot of money to throw around for no good reason or is someone who can use the huge performance very well. Your complaint is like complaining about a $2K 3.xGHz eight core Zeon not improving your gaming performance over a $220 i5-2500K. That isn't what it should be used for, so it's not a valid complaint.


    My point is there is no "huge performance" gain here, but obviously you think that there is. Also you are making assumptions based on your opinion who will buy this RAM. I guarantee there will be some power users who think they need this RAM and will buy it. And that's who they will probably market it to, I mean even the name "Xtreem" is targeting this audience. I'm pretty sure Pixar isn't going to be the one buying this stuff.
  • -5 Hide
    festerovic , May 31, 2012 10:11 PM
    waste of money.
  • 7 Hide
    blazorthon , May 31, 2012 10:26 PM
    Quote:
    My point is there is no "huge performance" gain here, but obviously you think that there is. Also you are making assumptions based on your opinion who will buy this RAM. I guarantee there will be some power users who think they need this RAM and will buy it. And that's who they will probably market it to, I mean even the name "Xtreem" is targeting this audience. I'm pretty sure Pixar isn't going to be the one buying this stuff.


    Your point is wrong. For many applications, RAM performance is a bottle-neck and 3GHz memory would help greatly. I'm not making assumptions based on anything beyond what I know to be factual. Just because some people might buy this RAM for workloads where it won't help enough for the cost to be worth it doesn't mean that we should judge it based on the mistakes of misguided customers. This RAM would help RAM limited applications very much and your last post is completely denying that. For those applications, it is worth the money and people using them would be justified in buying it. Rendering, decompressing/compressing archives, some folding applications, AES encryption/decryption done on CPUs with AVX, and more would all benefit from this memory very much and to a lesser extent, APUs (although probably not enough to justify the costs).
  • -7 Hide
    beardguy , May 31, 2012 10:50 PM
    blazorthonYour point is wrong. For many applications, RAM performance is a bottle-neck and 3GHz memory would help greatly. I'm not making assumptions based on anything beyond what I know to be factual. Just because some people might buy this RAM for workloads where it won't help enough for the cost to be worth it doesn't mean that we should judge it based on the mistakes of misguided customers. This RAM would help RAM limited applications very much and your last post is completely denying that. For those applications, it is worth the money and people using them would be justified in buying it. Rendering, decompressing/compressing archives, some folding applications, AES encryption/decryption done on CPUs with AVX, and more would all benefit from this memory very much and to a lesser extent, APUs (although probably not enough to justify the costs).


    Well we agree on one thing, this RAM is a waste of money except for maybe in some extremely rare setups. Soon 3GHz RAM will be mainstream and dirt cheap anyway :) 
Display more comments