Announcement

Collapse
No announcement yet.

POWER9 Could Be A Game Changer For Cryptocurrency Mining

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • carewolf
    replied
    Originally posted by Michael View Post

    The source for what?
    The article. I was looking for a blog-post link or something. Maybe I was just blind

    Leave a comment:


  • Michael
    replied
    Originally posted by carewolf View Post
    Hmm, where is source for this?
    The source for what?

    Leave a comment:


  • carewolf
    replied
    Hmm, where is source for this?

    Leave a comment:


  • edwaleni
    replied
    It didn't say if the model they ran the test on had the NVidia CM onboard or not. IBM was showing POWER9 with the CM at the last OpenPOWER Summit, calling it the "ultimate miner".

    Leave a comment:


  • schmidtbag
    replied
    It seems a motherboard with 2x CPUs costs roughly $3000. For enterprise hardware, that price is decent. If you can use PCIe bifurcators on each of the x16 or x8 slots, and then use those PCIe x1 splitters on each bifurcated slot, that would give you 40 GPUs you could use from a single PC (5 slots * 2 bifurcators per motherboard slot * 4 spliters per bifurcated slot). With that, in addition to the already good hash rate of the CPUs, I could see this being a very viable platform for miners.

    But, it's still stupid.

    Originally posted by artivision View Post
    If it doesn't have single thread performance, eventually will fail.
    Huh? I don't get the point of your comment.
    Mining doesn't depend on single-thread performance. If you're referring to the CPU, I'm pretty sure IBM knows what they're doing more than you. POWER9 is pretty much limited to strictly servers, and servers have a very wide range of demands, which is why there are so many variations of processors to fulfill such needs.

    Leave a comment:


  • starshipeleven
    replied
    Originally posted by madscientist159 View Post
    L_A_G Thank you for sharing your view on this. On this topic I'm honestly not sure that a permanent, public ledger is a good idea for one reason only: fungibility. Right now, with cash, you don't know the history of any given bill, nor do you care or have a legal obligation to know. Bills can be take out of service and replenished as needed to ensure that one bill can be spent exactly the same as another.

    Bitcoin and other public ledger systems don't have this property. Bitcoin and other public ledger systems don't have this property. You could buy coins one day, later on have some organization link them to illegal activity that happened before you bought them, and suddenly your coins are effectively worthless (or worse, you end up being accused of things you didn't do and had no ties to).
    What? Bitcoin or any other cryptocurrency, or conventional electronic money for that matter, does not have "coins" in the sense of a physical/virtual entity with its own ID that can be traced.

    Each physical bill does have its own ID number so you can be certain that a specific bill was used in X and Y, but electronic money transfers (conventional or with cryptocurrencies) are only transactions, you can only track transactions.
    Last edited by starshipeleven; 07 November 2017, 10:42 AM.

    Leave a comment:


  • bachchain
    replied
    Originally posted by mudig View Post
    Why is POWER9 faster?
    Blast Processing

    Leave a comment:


  • pavlerson
    replied
    How can POWER9 be a game changer? It is barely as fast as a Xeon. And a GPU is 10x faster than a Xeon. I dont understand this article. Is this IBM marketing? How can a Xeon fast cpu be a "game changer"? No GPU miner will convert to POWER9 because GPUs are 10x faster. What is this?

    Leave a comment:


  • L_A_G
    replied
    madscientist159 The thing about cash is that it can at least to some extent be traced using things like fingerprints, DNA, chemical residue analysis, lightly irradiated bills, etc. not to mention that transfer of it has to be done-in person or over services like the mail, both of which are vulnerable to traditional law enforcement investigative techniques. The reason why banking has to make and store extensive transaction records is much in part to assist law enforcement.

    Tracking most crypto "currencies" is not about tracking the particular tokens, but rather the transactions themselves and in particular who's making them, so your fear of being unjustly charged for using bitcoin is pretty much completely unfounded. Tracking transactions, who's transferring how much to whom, is really what law enforcement is interested in and monero is designed to make this very difficult if not impossible. Tracking cash using the methods I described is really all about trying to get to the transactions done using it. With monero all law enforcement can do is try to look at people converting other currencies into monero and from monero to something else, making their job of figuring out transactions harder if not impossible. If you count out speculators you're more or less guaranteed that every in and out transaction is going to involve some manner of illegal activity, but conclusively proving that someone is not just a speculator can be pretty difficult.

    Because of all this monero really offers to law abiding people is to make them harder to tell apart from actual criminals. Great for criminals who can more easily hide behind law abiding people, but terrible for everyone else. So if your fears are valid someplace, them it's with monero. However even if you end up receiving tokens owned by known criminals the worst thing that can happen is that you end up under law enforcement surveillance or get taken in for questioning, neither of which is hardly life ruining.

    As I said, this level of efficiency makes is pretty clear that we're talking about hardware with hash ASICs built right into the hardware and doesn't really say much about anything else. Power, being designed for mainframe use, is known to be a very high performance architecture, but it's been a while since we saw something properly competitive from it in the desktop/server space. If I recall correctly the previous attempt at getting a power-based workstation didn't perform all that well so it's not like this is guaranteed to be all that fast either.

    When it comes to it being pre-production silicon, having worked with silicon labeled "Engineering Sample" delivered over 6 months before the product with the final silicon shipped I really wouldn't put anywhere near as much faith into massive improvements as you have. There's obviously going to be some improvements, but engineering sample performance tends to not be that far off final production silicon performance.

    Leave a comment:


  • artivision
    replied
    If it doesn't have single thread performance, eventually will fail.

    Leave a comment:

Working...
X