6 ms·
True but the die size has also grown. So it not a 1:1 comparison. An impressive feat for sure but it is just using brute force to increase performance nowadays
by WizardAustralis 6y ago
True but the die size has also grown. So it not a 1:1 comparison.
An impressive feat for sure but it is just using brute force to increase performance nowadays.
- sbierwagen 6y agoDoing the math: GTX 1080: 42.81 mm^2 per billion transistors. 41.53 mm^2 per TFLOP. 0.97 billion transistors per TFLOP. RTX 3080: 15.55 mm^2 per billion transistors. 33.17 mm^2 per TFLOP. 2.13 billion transistors per TFLOP. They scale well for transistors against die size, and TFLOP against die size, but terribly in transistors per TFLOP? That can't be right. Checking https://www.techpowerup.com/gpu-specs/geforce-rtx-3080-ti.c3581 https://www.techpowerup.com/gpu-specs/geforce-rtx-3080-ti.c3... I now see it says "Based on speculation / placeholder", which makes me think it's all just totally made up, and not reliable evidence to declare the death of Moore just yet.
- atq2119 6y agoKeep in mind that with the 2080, they added raytracing acceleration and tensor cores. Both add area and improve performance of specific user cases, but my guess is that they don't show up in that TFlops number.
- mapleshamrock 6y agoThis is it - you can't just compare FP32 TFLOPS. The newer cards have a ton of extra precisions, custom cores for certain workloads, and on chip memory, all of which use silicon area and transistors, but none of which boost the FP32 TFLOPS metric. I could design you a chip that is nothing but FP32 multipliers and adders that has, theoretically, a ridiculous TFLOPS per mm^2, but it would be next to useless in any real workload.