It's exactly the opposite point. It's looking like Pascal needs a huge clock advantage to be better.
That is by design, the architecture has been changed to increase clock speed, just like Intel changed the architecture of its CPUs with the Pentium 4 to increase clock speed.
The 8800gtx was a huge step and didn't require a boost.
No, the increased IPC performance meant a longer critical path and slower clock speed. If Nvidia could have increased clock speed at the same time they would have by the laws of physics prevented them.
I have not looked into gpu sizes but the architecture was a big step forward. Any how i expect Polaris to be better than gcn 1.2 and give a decent boost clock per clock where i don't think Pascal does as it's pretty similar to Maxwell.
GCN 4 is just an evolution of GCN 3, just like pascal is an evolution of Maxwell. Both AMD and Nvidia will have almost done similar amounts of architectural changes. Nvidia have made some big changes to multi-engine support and context switching, and big changes to facilitate faster clock speeds. We will have to see what AMD has done with Polaris. Nvidia have a much larger R&D budget which may come to play in this generation, but AMD's GCN needed some serious work to get real-world performance close to theoretical. From the rumors it looks like AMD is moving in the direction Nvidia has been.
As i keep repeating this does not matter much as the boosts in clocks as rumored seem to be enough to make Pascal /Maxwell a decent step on the new node. Any how this is my last post as i want this thread to get back to Polaris. \
It matters because you are continuing to fail to understand how engineers can design architectures with different IPC to achieve the same IPS, and then using your flawed understanding to suggest that AMD has made more significant changes than nvidia.
It would be great if you got back to discussing Polaris instead of taking the thread off topic.
I'll try again to get the thread back on topic.
So how would polaris not be clocked high (1.3ghz+) on 16nm I don't understand that. Surely it would overclock well if it started at 1ghz?
Clock speed depends entirely on the architecture. If AMD decided to use the smaller process to increase IPC then the instructions per a second will go up at the same clock speed
Given particular fabrication process you can either design a processor with a higher IPC and lower clock speeds, or lower IPC and higher clock speeds. There is no definitive benefit to either approach.
With a new smaller fabrication node the same Instruction length will be faster so the same IPC will result in a faster clock speed, or conversely an increase in IPC is not penalized by slower clocks due to the faster transistor switching speed.
We've seen clock speed for Polaris all over the place so i don't think anyone can say what AMD has done with Polaris.
