Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?members/tanj.16713/recent-content
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Recent content by Tanj

  1. T

    Explosions are never good, 2025 still going to happen ?

    Waste sulfuric acid tank, not fuel tank. And who manufactured a relief valve that could explode out sending a man 20ft? This points neither to the union nor TSMC. Start with who made that device and why was it unsafe? It is unlikely that any worker would have the means to have found that fault.
  2. T

    Intel secures high-NA EUV kits over rivals

    Why would they not? DUV and pitch splitting have been the starting points for fins, gates, and low level metals for a while. It is the cuts where they would switch to EUV. And possibly vias if those are not 193i multiple exposure combined with self-aligned structures, which may beat some EUV...
  3. T

    Where is all the power going to come from for AI/ML

    But the peaks are always the problem, and if you add to the baseline the peaks become more painful, often to different people than cause the problem. Data center contracts generally stipulate they get priority, leaving others to scramble. They are learning to be more flexible on that - for...
  4. T

    Where is all the power going to come from for AI/ML

    Context is everything. Sometimes ideas fail in context, not because the idea could not work in the right context (or era - lots of good things today are similar to things that failed miserably in the past). Tesla tried battery swap for a market that turned out to be happy to charge at home...
  5. T

    What technologies will replace shrink?

    Shrink has not completely ended, but let's just postulate 14A is the end of the line for logic, including SRAM. Channel length cannot get any shorter, spacing between devices cannot be narrower, and CFET actually works. But there are still things that keep changing. DRAM has yet to go true...
  6. T

    Where is all the power going to come from for AI/ML

    SMR will happen first outside of the USA, likely Chinese engineering both in China and in areas of the world where their trade and tech is welcomed. SMR is an ideal format for it to be an exporting industry. Data center planning is way more sophisticated than us in this forum. I used to work...
  7. T

    Where is all the power going to come from for AI/ML

    Since trip miles > 700 are a large part of the graph the data will not line up exactly to travel time between rests or refuel points. I assume those longer routes would use a two person crew to keep running, so a battery swap would still be a preferred option.
  8. T

    Where is all the power going to come from for AI/ML

    Swapping batteries on a truck make a lot more sense for long haul. And would make lighter tractors fungible for both local and long haul routes. You could cut the battery pack in half, simplify electricity provisioning at recharge stations (since they would run at a steady rate), and even...
  9. T

    Where is all the power going to come from for AI/ML

    If you own the status quo, hiring a few people to spread misinformation that protects the status quo is an excellent investment.
  10. T

    Die Photographs

    Try messaging Scotten Jones. His company obtains many die photos, he may have some contacts who can help you.
  11. T

    Alternatives to Shrink for Performance?

    Already in use. More would be nice but if it was easy they already would. Shrinking the dimensions does not make such materials easier. I wonder if such diverse-circuit tactics will help much for AI chips where mostly they are grinding out low precision tensors and moving data. Sounds like...
  12. T

    Alternatives to Shrink for Performance?

    Really challenging to try tricks like negative capacitance ferroelectrics or tunnelling combined with the complex fabrication of ribbon or CFET. These complex shrinks reward conservative devices. Whether the construction is ever mastered well enough to start adding gate trickery - maybe, but...
  13. T

    Alternatives to Shrink for Performance?

    K change makes no difference, since gate capacitance is set by need to control the field in the gate. The point of high-K was to increase thickness to avoid tunneling while applying the same field to the gate. GAA shortens the gate, which is nice for CPP reduction, but it adds a 4th side so C...
  14. T

    Alternatives to Shrink for Performance?

    Yes, and AI computation is already at the thermal limit with close to ideal layouts. It can gain a bit from functional diversity with denser logic (separate pipelines for int8, FP8, BF16, etc.) but as more of the modelling is simply learning to use the fastest units - FP4/Int8 which are...
Back
Top