Question How ahead is Intel in CPU design compared to AMD?

[Adonisds]

If we could get both companies to design a CPU core now on the same process and using the same number of transistors, who would be ahead and by how much? I'm assuming Intel would be ahead because Skylake has a similar IPC to Zen 2, but Zen 2 uses a more dense node, and Skylake is from 2015 while Zen 2 is from 2019.

The both companies using the same process scenario is just to illustrate the problem and I know it will never happen, but maybe there is a smart way to compare current CPU designs and find out who is ahead and by how much. Wikichip has die sizes and die pictures of Sunny Cove and Zen 2, but I'm not sure how to interpret and compare them.

Another question: If Intel had to design future cores using the same process and the same die size as Skylake, how much further could the Skylake design be improved? Also assuming it will have to have to be x86 too and have all the same functions.

 

[mikegg]

The entire Intel Custom Foundry management waiting for manufacturing decisions because the CEO is having tea with Cook?! Sounds like fantasy corporate MMORPG. All major corp CEOs juggle their time between numerous tasks. All foundries need to allocate time and focus to make sure their clients implement designs properly (more clients, more time spent with each on their products). Arguing a standalone foundry will inherently succeed over an integrated one comes down to vertical integration as a successful business strategy. Time and time again it has been shown that vertical integration can bring great advantage for those able to master it properly.

Nobody is unstoppable, whether they seem like that or not. In movies we hack and defeat alien battleships, in reality a virus from our own planet might push us back into recession or worse.

It's just an example to make a point. Don't get too caught up in the example.

It's not a secret that a business that has a singular focus usually ends up being better at that focus than a company that has to operate in many fields. It's just business 101.

This is why one of the first things Steve Jobs did when he returned as Apple's CEO was to cut out a lot of products and focused on one or two products.

 

[Viper1j]

If we could get both companies to design a CPU core now on the same process and using the same number of transistors, who would be ahead and by how much? I'm assuming Intel would be ahead because Skylake has a similar IPC to Zen 2, but Zen 2 uses a more dense node, and Skylake is from 2015 while Zen 2 is from 2019.

The both companies using the same process scenario is just to illustrate the problem and I know it will never happen, but maybe there is a smart way to compare current CPU designs and find out who is ahead and by how much. Wikichip has die sizes and die pictures of Sunny Cove and Zen 2, but I'm not sure how to interpret and compare them.

Another question: If Intel had to design future cores using the same process and the same die size as Skylake, how much further could the Skylake design be improved? Also assuming it will have to have to be x86 too and have all the same functions.

Reminds me of the good ole' days of the CPU wars... (The days when Pentium was king, and Windows 3.11 was only a shell but still ruled.)

I remember talk that the 4 ghz, barrier could never be breached, that there was no way to deal with the heat that would be generated by the CPU.

How far we have come in so little time. I recently replaced some bad memory in a 7.2 ghz, water cooled system. (Client was a DOD contractor, and they GAVE it to her for free.) She's the go to girl for writing the Combat sims for fighters and tanks.

I have since tried to google some of the stuff I saw in that box, and I can't find ANY whitepapers on it.

I asked her to leave it to me in her will. (joking... Maybe)

 

[coercitiv]

It's just an example to make a point. Don't get too caught up in the example.

I didn't get caught in your example, merely pointed out it was wrong both as example and as business principle. A business with singular focus has both strengths and weaknesses. It's actually just business 101 for large companies to invest and diversify their portfolio in order to grow and protect themselves from future harm as some of their products or technologies may become obsolete.

Nokia was a pulp mill. Disney made cartoons. Lamborghini built tractors. Amazon sold books. Samsung was a trading company. Panasonic made light bulbs. Dell and Sony were shops. Intel made memory chips.

This is why one of the first things Steve Jobs did when he returned as Apple's CEO was to cut out a lot of products and focused on one or two products.

That's what you do when you start or recover a company. Once you're past that stage and you want to grow and/or survive, you diversify (see examples above). The Apple of today is actually the perfect counter example to your single focus argument: uses vertical integration and changed focus to a wider variety of products and services.

Edit: Since we're going OT I'll stop here, whether we agree or not.

 

[mikegg]

I didn't get caught in your example, merely pointed out it was wrong both as example and as business principle. A business with singular focus has both strengths and weaknesses. It's actually just business 101 for large companies to invest and diversify their portfolio in order to grow and protect themselves from future harm as some of their products or technologies may become obsolete.

Nokia was a pulp mill. Disney made cartoons. Lamborghini built tractors. Amazon sold books. Samsung was a trading company. Panasonic made light bulbs. Dell and Sony were shops. Intel made memory chips.

Sure, businesses diversify and use their existing business as leverage for new businesses. It does not nullify what I stated which is that a company solely focused on doing one thing well will often do it better than a company that tries to do many things well. Business 101.

In this case, TSMC does one thing and does it really well. Intel does many things and is clearly showing signs of failure in execution in areas such as mobile SoC, modem, AI, desktop processors, GPU, and node improvements.

That's what you do when you start or recover a company. Once you're past that stage and you want to grow and/or survive, you diversify (see examples above). The Apple of today is actually the perfect counter example to your single focus argument: uses vertical integration and changed focus to a wider variety of products and services.

I don't disagree with you there. Again, it still doesn't change what I said above. Of course once a business gets big enough or can use one part of of its business as leverage, it will expand.

 

[Nereus77]

I would wage we have better chances of predicting the next POTUS over the rise or fall of Intel Foundry Group in the next 5 years.

I'll take "Bernie Sanders" and "Headed into a dark trough for the next 5 years" for $50,000, Jimmy....

 

[zinfamous]

Your numbers are way off. Go look at channel sales for DIY gamers versus data center revenue, it's orders of magnitude in favor of the latter.

Uh, I never suggested that this is the bulk of their income. I made that pretty explicit, I thought. What I said that this is projected to be a $300 billion market, very soon, and being only one of two possible manufacturers in the world right now that provides one of the chief pieces of hardware that drives this market, they, like AMD, would be fools to ignore it...which is why they don't.

What did I say that complicates this statement?

What is the CPU share value of the gaming industry? $1 billion/year? Per quarter? More? I assume it is much greater.

Also, DIY builders? What a teeny tiny part of the industry. Not at all what I suggested. Consoles and OEM drive it and guess what: still the same components and still the same players.

I never suggested anything about DIY gamers. That would be silly. If anything, I argued that OEM is only going to become more valuable, with the interest in competitive RTS and low-res FPS games that, by today's standards, generally only require cheaper, mid-grade CPUs.

The OEM and console market. Obviously AMD doesn't sneeze at their console wins, as it is what kept them afloat for years. A piece of $1 billion, $5 billion, $10 billion is something that you better have a strategy to own...especially as the thread to this post suggests, it's really just an afterthought that fortunately benefits from their enterprise design. And that just makes it that much more important--it's already pretty cheap for them to address it, so it's not like it is really going to cost them that much to deploy some brain power to make sure they are addressing those billions of dollars.

"Orders of magnitude greater" than one market is irrelevant, when a billion dollars (more) in value is still a billion dollars.

 

[Viper1j]

Anybody ever wonder how much tech is being withheld from the general public, because "they" don't think we're "ready for it"?

 

[Ajay]

Anybody ever wonder how much tech is being withheld from the general public, because "they" don't think we're "ready for it"?

Uh, if it will sell, it will be sold - so nothing is being withheld 'because we're not ready'.

 

[moinmoin]

Uh, if it will sell, it will be sold - so nothing is being withheld 'because we're not ready'.

In general a lot of stuff keeps getting stuck at the R&D stage since companies don't feel there is a market for the resulting products. But I don't think Intel and AMD do much R&D beyond their existing products (and markets), if at all. Their technological fields are too highly specialized for that.

 

[lobz]

Anybody ever wonder how much tech is being withheld from the general public, because "they" don't think we're "ready for it"?

If they withold it, it's because they're not ready for it. (it = releasing commercially)

 

[moinmoin]

If they withold it, it's because they're not ready for it. (it = releasing commercially)

For AMD it's like a tradition of old to release hardware without being ready for it software wise.

 

[Ajay]

For AMD it's like a tradition of old to release hardware without being ready for it software wise.

Well, there is some truth to that, annoyingly.

 

[uzzi38]

For AMD it's like a tradition of old to release hardware without being ready for it software wise.

Yeah, they're working on that.

I can almost bet 2021 will still be a mess for them though.

 

[eek2121]

Intel is in quite a conundrum currently, and major changes are going to be required in order to right the ship:

1. 14nm is optimized for high clock speeds, similar to the Pentium 4 days. I noticed some here claimed Intel has an IPC advantage, but I have seen no benchmarks to support this. A 4 GHz Zen 2 core is faster than any currently released 4 Ghz Intel Core chip. There may be some random application out there using Intel's math libraries and/or compiler, but beyond that...(note: excluding AVX-512...for now)
2. Intel's 10nm, roughly equivalent with TSMC's 7nm, has clock speeds troubles. As nodes get smaller, clock speeds tend to fall, not rise. Prior to the Sandy Bridge era, the power savings from a die shrink helped offset this. Since Sandy Bridge, it has actually been downhill. Intel has had to optimize each shrink to boost clocks. Another challenge for Intel is the need to respond to AMD by doubling core count without a shrink. The end result is that Intel is very likely seeing lower clocks (4.0-4.2 GHz max), and without a corresponding IPC bump, they must backport all their designs until they find a compromise. This is why Intel's roadmaps are garbage.
3. Intel 7nm is likely to be similar to TSMC's 5nm. Intel was planning to be shipping at least 1 7nm Xeon chip in 2021. Coincidentally, this lines up with what we expect from AMD: it is widely speculated that Zen 4 (Ryzen 5000 series) will be on TSMC's 5nm process.
4. I expect Intel to launch 10nm desktop parts in 2021. Server and mobile will move to 7nm. We may or may not see a 10nm Xeon this year. I am willing to bet we will.

Don't count Intel out, however. It would not surprise me if they launched an entirely new CPU core, built from the ground up and optimized for 10nm. Part of me wonders what clock speeds a 10nm Pentium 4 would run at.

Likewise, don't count AMD out either. If I were a betting man, I would put $500 on AMD having at least 10% higher clock speeds with Zen 3, if not even more. AMD also changed the CPU family for Zen 3, so there are likely larger changes under the hood. While I personally expect the biggest changes to come with Zen 4, there are subtle hints that AMD's potentially final AM4/SP3 processor is going to pack quite a whallop.

I think the processor engineering teams at both AMD and Intel are equally matched in talent, with Intel only being hamstrung by middle/upper management as well as 14nm.

One last thought: I wonder how Zen 2 would perform in a 5 watt envelope. Everyone keeps saying ARM is catching up, but I think Ryzen is making huge strides in power consumption. The Snapdragon 855, for example, has a 5 watt TDP. Renoir is boosting to 4.2 GHz on 15 watts.

 

[jpiniero]

expect Intel to launch 10nm desktop parts in 2021. Server and mobile will move to 7nm. We may or may not see a 10nm Xeon this year. I am willing to bet we will.

2021 for desktop is 14 nm Rocket Lake. The only Intel 7 nm product available in 2021 is the HPC GPU but it's not clear if even that will be available to anyone other than Aurora in 2021.

 

[JasonLD]

2021 for desktop is 14 nm Rocket Lake. The only Intel 7 nm product available in 2021 is the HPC GPU but it's not clear if even that will be available to anyone other than Aurora in 2021.

I expect both the Rocket Lake and Alder Lake to be available in 2021.

 

[Markfw]

I expect both the Rocket Lake and Alder Lake to be available in 2021.

First, I am not calling anybody a liar or anything of that nature. BUT, until I see benchmarks from some respected site on any new Intel CPU's, lets just say I don't believe it. Their track record of late is ridiculous.

 

[eek2121]

2021 for desktop is 14 nm Rocket Lake. The only Intel 7 nm product available in 2021 is the HPC GPU but it's not clear if even that will be available to anyone other than Aurora in 2021.

Intel stated during a video that there would be a 7nm Xeon as well. I would have to find a link, because it was several months ago. They admittedly could have walked things back.

I highly doubt Intel will still be on 14nm at the end of 2021 for server/desktop. That would be...awkward.

 

[JasonLD]

First, I am not calling anybody a liar or anything of that nature. BUT, until I see benchmarks from some respected site on any new Intel CPU's, lets just say I don't believe it. Their track record of late is ridiculous.

Well, Mid-Late 2021 is still a long time away. If we are getting Alder Lake in 2021, then we will definitely see something later this year. Until then, we are all making wild predictions lol.

 

[Insert_Nickname]

Part of me wonders what clock speeds a 10nm Pentium 4 would run at.

Why, at 10nm, we'd have the mythical 10GHz Pentium4. Obviously.

 

[DrMrLordX]

Intel stated during a video that there would be a 7nm Xeon as well. I would have to find a link, because it was several months ago. They admittedly could have walked things back.

I highly doubt Intel will still be on 14nm at the end of 2021 for server/desktop. That would be...awkward.

On the Xeon side of things, 2021 should bring us Sapphire Rapids on 10nm. Hopefully by then they won't be reliant on anything like Cooper Lake.

 

[Ajay]

On the Xeon side of things, 2021 should bring us Sapphire Rapids on 10nm. Hopefully by then they won't be reliant on anything like Cooper Lake.

I’m confident Intel will still be selling allot of 14nm server CPUS. They only have 3 Fabs on 10nm. They would have to have high yields to have any hope of making SR a mainstream product.

 

[DrMrLordX]

I’m confident Intel will still be selling allot of 14nm server CPUS.

That will be . . . weird. Rocket Lake-SP?

 

[Ajay]

That will be . . . weird. Rocket Lake-SP?

Or, YASKLV (yet another skylake variant). Intel has built up a ton of 14nm capacity - they *have* to use it till that start converting 14nm plants over to 7nm.

 

[Staples]

As expected, AMD fans do not like this question.

I think if Intel CPUs were made on the same 7nm process and re engineered to take advantage of it, they would be 15-20% faster than their current 14nm.