BGA Venting Thread ;)

Discussion in 'Sager and Clevo' started by FredSRichardson, Nov 29, 2016.

  1. Meaker@Sager

    Meaker@Sager Company Representative

    Reputations:
    6,669
    Messages:
    40,517
    Likes Received:
    9,054
    Trophy Points:
    931
    It being the exact same part has never stopped Intel charging more if they want to ;)

    It's just no one would solder a load of extreme editions to their boards and risk all that wastage.
     
    Mr. Fox likes this.
  2. iunlock

    iunlock 7820HK @ 4.7GHz [LHz]

    Reputations:
    755
    Messages:
    3,143
    Likes Received:
    3,051
    Trophy Points:
    331
    I think the next mystery will be whether the smaller die size would be able to support high TDP's like we have now....

    It's going to be interesting...as we are seeing it on laptops that are way too thin, I think we've already reached that point where 14nm is way too small for the die size.

    Unless there is a major break through in technology...it would be nothing short of quantum physics if we get a die size smaller than 10nm that can handle high TDP...

    We should put up a TUTO on how to shave off the pins on the Ryzen to make it compatible with the LGA socket. Of course add in to LM the entire socket first to ensure thorough coverage and contact.

    :p

    For those of you into the Food World / Foodies, it's like when ice burg lettuce was once something restaurants would throw away and discarded....

    Now it has become a delicacy and served in 5 star restaurants...a little ranch, blue cheese and topped with bacon bits....$15 for an appetizer of what once used to end up in the rubbish.

    BGA-CPU's like the horrific HQ are along the lines of the same thing...it's that ice burg lettuce, except that it needs to be topped with marketing hype and ignorant consumers who follow the flashing lights, walking into hypnosis willing to pay full price.

    Also to add, since this is a completely different jurisdiction from the brand specific forums, for the record, I don't pay full price for any of my bga machines lol...not even close. I wouldn't in a million years pay that price tag. One would be shocked at the deals that are possible and trust me... even the most extreme bga advocate would take one for what I pay for them...in case some of you were wondering, "why all the bga machines?" LOL...sensible purpose and practicality for the right sensible price.

    There's a difference between ignorant / naive and spending top dollar on something that is not worth top dollar VS being aware and spending only a fraction of the cost on something and not ever feeling a buyers remorse for a purchase that wasn't worth it.

    .
     
  3. Mr. Fox

    Mr. Fox Undefiled BGA-Hating Elitist

    Reputations:
    17,057
    Messages:
    27,532
    Likes Received:
    29,151
    Trophy Points:
    931
    I have not been a fan of die shrinks for a good while now. The talking heads make it sound like something special, but history paints a slightly different picture for us. I think die shrinks will continue be more and more problematic for those that value performance and enjoy competitive overclocking. A low TDP CPU cannot provide amazing performance, at least in the context of the enthusiast arena, and small die size has trouble handling high TDP. We have already seen some evidence that die shrinkage is taking us the opposite direction of where some of us want to go, and I doubt that is going to do anything except worsen. This is where the bang for buck and cost per clock arguments start to unravel. It doesn't matter if a particular processor offers equal or better cost per clock performance ratio when it lacks the capacity to compete with something that has theoretically unlimited TDP and more overclocking headroom. The fact that it can compete until it reaches the point that it peters out and gets left behind becomes irrelevant (a lame excuse) and the rationale that it is just as good goes right out the window.

    Now we are at a point when it is not even possible to offer cost-to-performance ratio rationale because that's also all jacked up. When there is no cost savings (as we see today) and the inferior part has limitations the better part does not, but sells for the same price as the better part, I view that as tragedy and failure. Instead of shrinking dies, they should leave them alone, or make them larger, and pack more awesomeness into them... higher TDP, more cores, etc. Making them smaller "just because" or based on conventional wisdom (AKA popular stupidity) is kind of retarded.
     
    Papusan, TBoneSan and iunlock like this.
  4. iunlock

    iunlock 7820HK @ 4.7GHz [LHz]

    Reputations:
    755
    Messages:
    3,143
    Likes Received:
    3,051
    Trophy Points:
    331
    I agree. It'd be nice if the die size stopped right at 14nm so that they would have to revert going the other way lol.

    It would also be nice if somehow the ODM's were forced to go the other way as well...
     
    Papusan and Mr. Fox like this.
  5. Stooj

    Stooj Notebook Evangelist

    Reputations:
    92
    Messages:
    498
    Likes Received:
    323
    Trophy Points:
    76
    Wow..

    Die shrink leads to 2 very big things:
    1. Increased transistor density
    2. Transistor switching with less current
    These things literally translate into more performance per watt. And what history are you even talking about? GPUs are a prime example of die shrink benefits at work. Every shrink increases transistor count significantly and performance has jumped massively every time. Just look at the Nvidia top tier cards over the last few shrinks and perf/watt has scaled almost linearly each time. AMD has seen the same benefits (albeit less frequently). More transistors = more switching = more calculations = more work done.

    On the CPU side the benefits are less pronounced for consumers because Intel have actually made every jump on a more regular basis. Furthermore most pc performance "enthusiasts" are actually just gamers who don't even see the big picture.

    Look at the server world where core-count matters and things are VERY different. Each die shrink has allowed huge jumps in core count and the perf jumps are there as well. That's not to say that single-core performance hasn't jumped though, literally every generation has increased performance relative to the process shrink. Literally every CPU generation has been faster than the last....

    o_O There are real physical limits to how fast a transistor can switch and you can't just keep pumping voltage into infinity to make it switch faster.

    "Theoretically unlimited TDP" is possibly one of the silliest things I've ever read. It would require "theoretically unlimited power" for starters, which is just absurd.


    As far as overclocking, it's rather simple. As the process shrinks, so do the tolerances. As tolerances shrink, so do the variances in each chip. With fewer variances they can come out of the factory far closer to their maximum potential. A "highly overclockable" part, when you actually think about it, implies a larger variance in manufacturing which is actually a bad thing. This is why the "silicon lottery" will become less of a thing. Pascal GPUs really put this into perspective for many.

    I certainly get the "tuner" mentality of getting more out of your chip. But to suggest that the chip itself should be purposely made inferior (which is exactly what is required for overclocking to be meaningful) is patently ridiculous.
     
  6. Tanner@XoticPC

    Tanner@XoticPC Company Representative

    Reputations:
    134
    Messages:
    1,395
    Likes Received:
    1,087
    Trophy Points:
    181
    What, your laptop isn't tokamak powered? Literally feces.

    Am I doing that right?
     
    Stooj, Galm and Ionising_Radiation like this.
  7. Galm

    Galm "Stand By, We're Analyzing The Situation!"

    Reputations:
    879
    Messages:
    4,991
    Likes Received:
    2,270
    Trophy Points:
    231
    I think the shrinking is definitely good for overall performance. You're right that it hurts enthusiasts who want to OC somewhat, but I'd rather have higher stock performance as time goes on. And that has been increasing.

    I do agree I'd rather have a large die, that would be a large boost.
     
    FredSRichardson and iunlock like this.
  8. Meaker@Sager

    Meaker@Sager Company Representative

    Reputations:
    6,669
    Messages:
    40,517
    Likes Received:
    9,054
    Trophy Points:
    931
    I don't think anyone would want a 180nm 1080, it would take up most of the system lol.
     
    JaredJ likes this.
  9. Tanner@XoticPC

    Tanner@XoticPC Company Representative

    Reputations:
    134
    Messages:
    1,395
    Likes Received:
    1,087
    Trophy Points:
    181
    Never know what people will buy, just have to market it right. :)
     
    JaredJ likes this.
  10. Meaker@Sager

    Meaker@Sager Company Representative

    Reputations:
    6,669
    Messages:
    40,517
    Likes Received:
    9,054
    Trophy Points:
    931
    When it only clocks to 400mhz and generates 2kw of heat then they would not want it over a modern design :p
     
    dm477 likes this.
Loading...

Share This Page