rylando 31 minutes ago

How neat! Would be cool to see the side by sides of equivalent tier CPU’s throughout the generations

lathiat 6 hours ago

LTT Labs are writing up a lot of nice content of late.

  • fishgoesblub 6 hours ago

    Sadly posts from one of their employees, Lucas, are auto-dead and don't get seen which is sad given their effort and quality.

    • rylando 32 minutes ago

      I wonder why? Genuinely, not sarcastically

    • newsclues 11 minutes ago

      The author of the linked article is Lucas...

phire 9 hours ago

I was not expecting the CT scan to be able to identify the cores within the die.

But now I think about it, the metal layers (especially traces carrying power) are probably think enough, and different enough to show up. And even just power distribution is enough for the shape of the cores to be visible.

  • xpct 2 hours ago

    You can typically see the L1/L2 SRAM cache regions just by looking for continuous patterns, then try to divide it into the number of cores or core groups you're looking for.

    • jaen 41 minutes ago

      This is only applicable to die photographs of the silicon layer.

      For CT scans, which use X-rays, the visible structure is very likely due to the metallic (copper) interconnect, clock- and power-delivery layers, which are generally quite different from the underlying transistors.

      These layers are larger-scale, and copper also absorbs X-rays much more strongly than silicon.

      • xpct 38 minutes ago

        Ah, I see. Thank you for the correction.