nghanayem covered the points well. For the N3E pitch of 23 nm, DUV SAQP would be the safest and also works well with 5 tracks.
China’s top chipmaker will ‘struggle’ to make cutting-edge chips competitively
- Thread starter Daniel Nenni
- Start date
Some of what you are saying is strange, because it's essentially repeating what I said. So, yeah, we agree on most of it.
Let's start with what we agree on, and what I stated earlier. Intel prioritizes clock speed over density and power usage. Completely agree. That was my point, that's fine for a desktop CPU, but not really good for a GPU. At all. That's my point about TSMC. They are better for what Intel needs with their GPUs, and being able to make it there gives their GPUs a boost, compared to what they would be capable of on their own nodes. Their own nodes might have a bit more clock speed, but it's almost trivial when you consider the tradeoffs in density and power use. Now they can get the best of both worlds.
Intel has had the highest clocked processors, at least compared to TSMC, I think forever. Yes, clock speed is a combination of architecture and process, so we can't say for sure. But, it's pretty close. Because you don't think AMD looks at speed paths and optimizes for them? Or they don't know how to make a high speed architecture? Look at Bulldozer, it was definitely a speed king, and definitely not a brainiac. I don't think they forgot how.
But, here's the reality, what we know. Intel makes the highest performing processor, based on single-threaded performance. This is the holy grail in computing, and the most difficult to advance. This despite it being a node "behind" the one AMD is making Ryzen 4 on. Forgetting IPC, because that is purely architecture, why does every generation of Intel processor clock higher? I mean, it's not even close now, just look at the overclocking records, no one even would consider an AMD.
So, yeah, it's pretty clear Intel is better, even a node behind, at that metric, because that just happens to be what they care most about. AMD does everything they can to improve single-threaded performance as well, including architecting the processor for good clock speeds, and eliminating bottlenecks to it; that's not just an Intel thing. But, given AMD is always behind (and so is every other processor made at TSMC), and not by a little, and given AMD has not too long ago made very high clock speed processors (which lost some clock speed when moving from 32nm to later nodes, which used higher density libraries), I don't think it's super likely this is just architecture based, and that Intel is more optimized for clock speed. It might be as well, but most of the clock speed gains Intel got on their 10nm family were from optimizations to the process, and the same could be said for 14nm.
Now if they resurrect the Pentium 4
I am not entirely sure how your remarks perfectly align with reality. While I am not familiar with the overclocking scene, I can see that the clock frequency differences between Intel and AMD processors are not that big to make such excessive quotes that you are making. And since you also mention IBM in the mix, their z14 processors were hitting far higher frequencies than any Intel processor at that time (both on 14nm class processes). So while Intel may be prioritizing higher frequencies, it not as excessive as you mention it.
In addition, I do not see architecture and process/implementation as two glass separated things. When you architect a chip, you take into consideration what the process/implementation can provide and try to make certain ....co-optimizations, while taking into consideration cost and manufacturability as another (very important) factor. Intel works totally different than AMD, because of this (or at least used to work). Furthermore, you cannot forget IPC as you mention, because it is VERY probable that an architectural decision to increase IPC will have an impact on the clock speed. And because exactly of this, not every generation of Intel processors had higher clock speed than their previous generation. You yourself mention the P4 era.... do you remember the clock rates of the last P4 chips and the clock speeds of the first Core ones? Pretty dramatic change I would say.
Concerning the power usage, this is also another story.... for how many years, did Intel processors have a very significant advantage over AMD in the laptop/mobile segment? And at the same time, why ARM seemed to prevail in the smartphone arena? There are more to the mix than just process. Architecture and software cannot be ignored the way you wrote your post.
Actually, I have mentioned IBM chips in the past, but that's also with a completely different environment. They are on super expensive machines, with very exotic cooling standard. It plays in a different market from AMD and Intel, which have essentially the same limitations.
What you aren't looking at is overclocking. Intel chips overclock to extremes very well, AMD's don't. After a certain point, AMD improves clock speed much more slowly than Intel, and gets left in the dust.
Regarding IPC, you completely misrepresented what I said. IPC is completely based on the architecture. Period. If you take the exact same design and move it on another node, it will have the exact same IPC. But, here's the point, AMD doesn't have more IPC anyway. So, it's not like AMD has a brainiac, and Intel has a speed king. They have roughly the same IPC, but Intel clocks higher. And overclocks a lot higher. So, do AMD engineers just suck? They can't make a high clock speed chip? Well, they sure did with Bulldozer/Piledriver.
Your remarks about architecture and software are pure nonsense, I never said either. If you want to debate with me, please try to be accurate, instead of trying to change what I say so you can make a good point. In other posts, I have said software is extremely important, so where does this come from? I also said clock speed is a combination of architecture and process. But, here's the rub of it, when has a TSMC processor ever had a high clock speed than an Intel made one? From any designer. So, with every generation, Intel clocks higher, regardless of CPU designer. Coincidence? Yeah, sure. But, AMD pushes IPC more, so they settle for lower clock speeds? Well, no, they don't have higher IPC. So their designers just suck? OK, no, because outside of Intel, they have the best performing processors (forget mainframes). But, but, AMD has never known how to make high clock speed processors? Well, Bulldozer clocked a lot higher than anything Intel had, when made on 32nm. Hmmmm, but it can't possibly be the process isn't as fast, so let's try to find something else.
And more to the point, even TSMC recognizes it, albeit tacitly. When they were talking about the upcoming node, I believe it was 2nm, even they said they were expecting leadership in density and power efficiency. They never mentioned performance. Fancy that ...
Oh, and the Pentium 4 is another example I mentioned. The original had 20 stages, and Prescott had 31, which was WAY more than Athlon XP/64, and Pentium III. It was a pure speed king. Look at the current processors from AMD and Intel. Do you see that divergence? No.
So, anything is possible, but everything suggests Intel has a node that will clock higher, despite "being a generation behind". I know it bothers some people, I just don't know why. The counter arguments are always "well, it can be this or that", because there's no supporting information that indicates otherwise. Now, that would be like me saying TSMC nodes aren't more power efficient, and saying that's the architecture, and at certain power levels it can be different, etc... Anything is theoretically possible, but there's little to support it.