Intel 13th gen seems to outsell Ryzen 7000 everywhere so...
- Engineering support when you're designing a new motherboard
- CPU SKUs with different use conditions. Intel Industrial rated CPUs are rated for 24/7 operation, 100% duty cycle for 10 years. I'm not aware if AMD has an equivalent.
- Long product support - CPU SKUs that are "guaranteed" to be around for X number of years. As as example, some Icelake SKUs are planned to be supported for 10-15 years. Let's use an x-ray scanner as an example. The x-ray scanner mfg needs a computer inside the machine to help process info and display images. The x-ray equipment has an intended lifespan of at least 5 years and they plan to sell/service this equipment for at least 10 years. The mfg wants to use an embedded computer that's going to be commercially available for a long time so they don't have to re-engineer and re-certify their machine. This is one reason why you see some "stickiness" with Intel.
Then you have reasons associated with vendor lock in. I mean, most people buy DEVICES, not CPUs, and device manufactures are hesitant to switch, because doing so can potentially be complicated and prohibitively costly (especially if they already have deals with their current vendors). Whereas their customers will end up buying their devices regardless of whether they switch CPU vendors or not (because of reasons mentioned above)...
Ergo, as other's have already sort of alluded, it's largely because of branding, marketing, and vendor lock in.
Intel meanwhile has an established year old motherboard pipeline for their current socket that 13th gen can be dropped in to.
If this were 14th gen vs Ryzen 7000 the calculus would be different as all the Intel purchases would also include a motherboard.
Also, Ryzen 5000 is going gangbusters outselling everything else.
Which is why they just dropped their prices? (see, e.g., https://forums.anandtech.com/threads/amd-has-dropped-5000-se...)
You can see the 5600X price drop on Amazon happened Oct 26[0].
If you look at the Amazon best sellers[1], it gives you a pretty good idea of where things stand.
0: https://camelcamelcamel.com/product/B08166SLDF?context=searc...
1: https://www.amazon.com/Best-Sellers-Computer-CPU-Processors/...
Intel isn't strapped for cash to do R&D, and if they are, AMD is way worse off. Go look at profit numbers for both companies over the last 5 years. It's not even close.
To pretend like intel is going to run out of money to fund R&D is just to not even understand what's going on.
It's a bit ghoulish but Intel's saving grace may be an invasion of Taiwan.
Intel is in a good place. The question people should be asking is why its so close at all if AMD is several nodes ahead. why is 5nm and 6nm TSMC only competitive with intel and not crushing it? Their performance per watt at high load (and not at idle) is AMDs biggest differentiator. When intel goes to smaller nodes they'll gain more thermal headroom to pack more and faster cores than AMD.
The current modern tech youtubers are asking many of the wrong questions. I watched gamersnexus crowing about AMD efficiency in the highest tier chips and asked "people buying race cars care about speed more than efficiency - why are you downplaying intel's ultrafast chips on 10nm+++ vs amd's smaller and more efficient nodes, and talking about their fuel mileage for a market segment differentiated by performance?"
Theres a lot of nonsense going around.
CPU efficiency is so nonlinear at high power levels as to make comparison meaningless if they aren't matched.
CPU efficiency is an interesting number to the tech youtubers who focus on it, but is not that market differentiator at the highest end most profitable part of the market, and only matters to them. The AMD 7950x is very fast chip which is much more efficient than its intel counterpart, but at the top end of the market what matters is who performs better, not their fuel mileage. You'd be investigating a number interesting to testers but not consumers likely to buy those products - its a waste of time and effort.
I guess from a practical perspective, if you aren't modifying the chip clocks then yeah, they are talking about the right thing. But from a technical perspective, I'm more interested in the underlying architecture performance, which is to say:
If the intel chip is set to a Power Limit of 100w, and the AMD chip is set to a Power Limit of 100w, what does performance look like then? Because it's nearly meaningless to run one processor at 250w and the other at 150w, then say the 150w chip is more efficient. Processor power skyrockets at high voltages and clocks, so likely the intel chip can use 50% less power for 10% less performance, if desired - Intel merely made the (correct, imo) calculus that maximal performance is more important than efficiency for this class of chip.
I think we pretty much agree - I just think it would be interested to have a more accurate architectural efficiency/performance comparison (which would also help when choosing eg. a laptop where those numbers DO matter)
As I said before, the efficiency per watt is a youtuber number. Nobody who is buying chips really cares, or if they care they watch the youtubers who talk about it nonstop already. It's the youtubers bringing it up on top tier chips who are just out to lunch and not understanding the actual selling to a market and distorting things because of numbers they care about that class customers don't.
I don't care what a 7950x does at 100w, because if I wanted that performance I would have bought a cheaper chip. It's a stupid nonsensical number unrelated to why people buy $700 high end CPUs. You don't buy that chip to save $3 worth of electricity.
This comparison is meaningless though if the processors aren't running at the same power, however :(
https://guidehouseinsights.com/news-and-views/efficiency-is-...