• nukleabomb@alien.topOPB
    link
    fedilink
    English
    arrow-up
    1
    ·
    1 year ago

    For reference:

    Card name GPU Code name CUDA Cores L2 Cache (MB)
    4090 AD102-300 16384 72
    4080 Super AD103-400 10240 ??
    4080 AD103-300 9728 64
    4070Ti Super AD102-175/AD103-275 8448 48
    4070Ti AD104-400 7680 48
    4070 Super AD103-175/AD103-350 7168 48
    4070 AD104-250 5888 36
    • AzureNeptune@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      1 year ago

      AD103 only has 64M L2 so if that’s what they’ll use then the 4080S will have the same similar to the 4070 cards maxing out at 48 on AD104. If these end up replacing the originals they could be interesting but ultimately I’m not too excited

    • SkillYourself@alien.topB
      link
      fedilink
      English
      arrow-up
      1
      ·
      1 year ago

      When you put it on a table like this, 4080 Super -> 4070 Super range is getting too crowded for a 3K CUDA core spread.

      If Nvidia follows their 20-series strategy, 4080 and 4070Ti gets replaced by their Super refresh at same MSRP while the 4070S takes the slot between 4070TiS and 4070.

      • imaginary_num6er@alien.topB
        link
        fedilink
        English
        arrow-up
        1
        ·
        1 year ago

        Why don’t they just replace the 4080 and 4070Ti with higher Super prices? The performance is going up so they can claim higher pricing

        • YNWA_1213@alien.topB
          link
          fedilink
          English
          arrow-up
          1
          ·
          1 year ago

          Probably in-effect what’s going to happen. They won’t ‘discontinue’ the older cards, just never actually produce those chips for those SKUs.