“I’m not saying sheol isn’t, and I doubt the in-house forest, Lisa Su and the wonderful Zen architecture. In fact, it was already seen when 3rd gen Ryzen, made at TSMC nm, didn’t beat the current Intel 10. Generation. Just because they had 2 manufacturing advantages in Lepcsony. I’ve been skeptical so far about how much AMD’s big revival is, and how much they had a puppy with TSMC and Krzanich. “
I don’t understand your doubts. I think you put the bar high, so to speak.
The zen arrived with roughly the haswell-broadwell IPC, so it was lagging behind. Then zen2 brought in an IPC comparable to skylakes – except for games, and finally the merged CCX zen3 closed the scissors behind in gaming.
The “amazing thing about this” is that a company with a turnover of a tenth was able to do that.
At the same time, of course, this catch-up included the fact that Lisa Su, led by Lisa Su, had executed her otherwise good plans. Zen architecture isn’t amazing, it’s just plain good. (If you look at it from here, the M1 is more amazing, which brings the same power at a much lower frequency, but even there the bar vibrates to see if the 5nm – lower power consumption and higher transistor density – makes it possible) This power is not wonderful, just flawless.
And, of course, we also needed a flawless manufacturer, TSMC. Obviously, a manufacturer that was constantly struggling with slips and yield errors would not have succeeded. But AMD, along with other customers, pays this to TSMC.
On the other hand, of course, it had to be difficult for intel to make mistakes.
Zen hit them unprepared. After all, they could have done, say, the otherwise quickly released kaby lake with 6 or even 8 seeds. Of course, this may not be due to a “surprise” but a cold calculation. They don’t have a wafer for free either. Each of the later generations of coffee lake has increased the scarcity of wafer capacity.
Then, if 10nm had been operational in 2018-2019, zen2 would not have been so successful. The mistake and unpreparedness here was that there was no failsafe plan for bringing new developments to market in the event that production technology slipped. AMD was able to make zen2-> zen3 generational changes on the same manufacturing technology. Intel did not yet have a Plan B like Rocket Lake at the time.
I don’t want to read these on their heads now; frankly, I doubt AMD would have one. I’m just saying what little things Intel slipped through that could have led AMD to catch up.
The present situation also required the success of TSMC and the misfortune of Intel. But it also has its own merit for its flawless workmanship, which is not to be disputed by those like the bulldozer and Vega.
In the current situation, I don’t feel that the rise of AMD is an illusion.
On the other hand, if Intel gathers itself and develops faster or more skillfully, the brilliance may be iris-lived.
We’ll guess, and then it will turn out anyway.