It is pretty useless to compare the price of a single CPU with the price of an entire PC…

You're definitely right there. I put together a build with this CPU and chose the most expensive part available (except GPU because chip shortage, case because there are $5000 ATX cases for no good reason, PSU because I just got the best Seasonic one, and SSD because there are $12k enterprise ones): https://pcpartpicker.com/list/DYxhk9

So that's $3500 without a GPU, buy a $500 used GPU on eBay and you're beating Apple. And, nobody buys $1000 motherboards, so that takes $500 off. You don't need a $300 case. Etc. Basically the point of the exercise is that you can max everything out, and get a faster computer for less money, which is what the comment was trying to say.

Someone will reply and say that your time sourcing and assembling the components isn't free, or that it doesn't run OS X, etc. I get it, you don't have to say that. Just adding an actual computer that's expensive as possible that you could have right now to compare to.

I just got a Titan A200 with a Ryzen 9 5950X. This CPU is really fast, and dissipates only 105W. The Titan workstation is the quietest I have ever had. It's really incredible. Price tag: $3600.

https://www.titancomputers.com/Titan-A200-AMD-RYZEN-Professi...

According to Passmark, the 5950X is beating Intel's 12900X.

https://www.cpubenchmark.net/compare/Intel-i9-12900K-vs-AMD-...

> According to Passmark, the 5950X is beating Intel's 12900X.

As a 5950X owner I thought this sounded off. Looking at the link it made a lot more sense, it only beats the 12900K in multi-core but in single core it is significantly behind. Really it's behind for the first 8 cores, until the Intel CPU starts using efficiency cores. Compared to the M1 Ultra the 5950X is behind in both metrics at nearly double the power consumption.

Not that the 5950X is a bad CPU, particularly if you need x86 support, it's just not really the topper it used to be against these 1 year newer chips. It'll be interesting if all 3 (AMD, Apple, and Intel) manage to get the next major iterations out by the end of the year and we get a fresh comparison on more even ground.

Fair points. A direct comparison here (https://nanoreview.net/en/cpu-compare/apple-m1-ultra-vs-amd-...) with the M1 Ultra has them pretty much neck-and-neck, not considering power. Considering power, the M1 Ultra is a clear winner.

I think you’re right—we’re still on the first generation of Apple computer chips (and honestly I think the performance is still very impressive, even if overpriced), and there is a lot more motivation now on the others. We might start seeing some big improvements.

This benchmark has the 5950x going up to 194W:

https://www.tomshardware.com/reviews/amd-ryzen-9-5950x-5900x...

Throw in another 40W for the liquid cooler and it adds up.

Though you'd have a possibly noisy large black box, compared to a small quieter silver box. This is a factor in computer design - cooling is hard. A quiet small box with performance of a noisy large black box is something a lot of people would pay for (including me).

But why would you buy the M1 ultra model for single core performance?

I think it would be helpful for people to occasionally restate what they think the thesis is.

I really don't know. I suffer in both directions. I build large projects on a daily basis, so picked a Threadripper. It destroys large builds, especially C++ ones. Then I use the same computer to play games, and the CPU can only spit out 300 frames per second when my monitor can display 360, which is annoying. (It's CPU limited, not GPU limited, sadly.) So really, I want both, and nobody has both.

Single thread performance is going to be especially relevant if you are developing an older language. I'm always surprised how slow webpack is (and don't do enough frontend stuff to mandate that people switch to that Go equivalent), for example. If you have good single thread performance, you make a lot of Typescript developers happy. If you have good multi thread performance, you make a lot of C++ developers and gamers happy. So having both would be great ;)

And are you factoring in the power consumption difference ?

That can definitely add up over the years as I've experienced with my 10980xe.

Also resale value is an important factor. You will struggle to sell that PC in the years ahead which will never be a problem with the Mac.

To save the click a 12900K machine from Dell is about $3k. If you need CUDA or say SolidWorks get the PC, for video and multithreaded workloads the Mac would probably be faster, but really only benchmarks of your use case can tell you.

So made a build to essentially show how expensive an M1 Mac is compared to an intel machine but left out a critical component because it’s too expensive?

apple doesn't sell unbundled chips. Adding a motherboard and RAM would still be less than $4K for most configs.

Umm, running a 1990s fortran code that's a CFD simulation is a "real" workload. Seems relatively likely that any floating point heavy code would act similarly. Hard to say if it's the matrix multiple or the memory bandwidth that's giving apple such a large lead.

Normally I'd discount using a 28 core Intel CPU from 2019, but from what I can tell Intel hasn't improved much since then. Keep in mind that Intel has a specialized vector unit (AVX256 or AVX512 depending on the model), and the listed CPU is pretty high end (with 6 memory channels) where the normal i5/i7/i9 is only 2.

So sure it's not a video compression, gaming, or web browing benchmark, but some folks do run floating point heavy codes. Unlike CUDA, which requires a rewrite, this code wasn't specifically optimized for the M1.

Ah yes, the bot accusation comment. Everyone's favorite subcategory of HN musing.

Hey man, maybe I'm a bot too. Who knows. Blip bloop. But these are just computer chips and fresh low karma accounts with vitriol dialed to the max is a-typical of HN. By all means, M1 is the worst, why froth at the mouth about it in this manner? You explain it.

I don't have to explain anything, HN is based on the ideals of arguing in good faith and not vilifying people for their opinions. If you think they're being harsh, then say that. If you think they're wrong, then refute them. But if you're solely seeking to bash them, why even comment in the first place? How does that advance anyone's understanding of the topic, or sway anyone's belief? It's the opposite of productive conversation, if you think they're encouraging a harmful or counterproductive dialogue then you can downvote them and move on. It's not rocket science.

I specifically wrote that the comment is so harsh it reads like spam/bot. Wasn't sarcastic or trying to bash any humans, sincerely.

Well, I have lower karma than seabriez and I'm neither a bot nor I thought that I had low karma points.

For your own sake I really hope you’re trolling and not actually having your brain in a configuration where your comment makes any sense to you.