By using this site, you agree to our Privacy Policy and our Terms of Use. Close
Soundwave said:
Pemalite said:

Not to mention it's a theoretical number anyway.

The Tegra X2/Pascal should be roughly 50% faster than X1/Maxwell. Maybe a little more.
It could be doubled if Tegra X2/Pascal was built at 16/14nm Finfet and had the clockrates driven up more.

Xavier we don't have benchmarks for yet. Xavier has doubled the functional units over Tegra X2/Pascal.
But the real kicker is... What are the clocks?
If the clocks are lower than the Pascal based Tegra we might only see another 50% gain.

Or we could see a 150% gain. We don't know yet.

Nvidia is saying one Tegra Xavier chip at matches the Drive PX2, which is 2 Tegra X2s + 2 Pascal based GPUs (so 4 processors total) at 8 TFLOPS. 

Well, those Volta cores must be radically different then what GPUs have been packing for a long time - 512 cores with 2 operations per cycle for FP32 would need to run at insane clocks to achieve 8TFLOPS...as in 7500+ MHz insane.

For reference, made on similar, if not same 16nm proccess as advertised for Xavier, current nVidia tech needs 1920 cores running at 2000+ MHz for similar performance (overclocked GTX1070).