Hacker News new | past | comments | ask | show | jobs | submit login

You seem to be misinformed.

Everyone loves to hate Intel (and rightfully so) but Tiger Lake has better IPC and implicitly better single threaded performance than Ryzen. In real world tasks (web browsing, opening an IDE) it often comes on top of AMD.

It only loses to AMD in multi threaded benchmarks because of the lower core count (everyone loves to flex Cinebench scores online, but I buy my laptop to use, not run benchmarks).

Also, the iGPU in Tiger Lake is more modern and more powerful than the outdated Vega iGPU in Ryzen and it also has AV1 codec hardware decoding unlike Ryzen which is stuck at VP9.

Add in Thunderbolt support and using Tiger Lake makes perfect sense for this form factor. Ryzen shines best in gaming laptops with discrete GPUs.




> In real world tasks (web browsing, opening an IDE)

My real-world task is not opening an IDE, it’s using that IDE. Modern C++ compilers are using all available CPU cores just fine.

> It only loses to AMD in multi threaded benchmarks

These are the only workloads I care about. Not just compilation, many other things as well.

You only need single-threaded performance for 2 things, for the stuff that’s inherently serial like gzip, or to run programs made more than 5-10 years ago.

I don’t normally play games on a laptop, but even videogames use multiple cores for decades now, since the Xbox 360 / PS3 generation.


> You only need single-threaded performance for 2 things, for the stuff that’s inherently serial like gzip, or to run programs made more than 5-10 years ago.

You must not be grand strategy game players. Any titles from Paradox and some other popular games are all limited by single core IPC. They are a great example of the limits of multi-threading, some processes and problems cannot be adapted to take advantage of it. If I get such a laptop I'd spend 50% of the time I use it playing such games.


> They are a great example of the limits of multi-threading

These are examples of lazy programmers and PC-only games. Consoles have many cores for decades and these cores are slower than PCs, developers of cross-platform titles have embraced multithreading for quite some time now.

Multithreading does have limits. Some things are borderline impossible to parallelize, examples include gzip, streaming encryption algorithms, or parsing long streams of HTML and JavaScript. However, vast majority of the CPU-bound stuff found in videogames scales just fine with CPU cores.


For gzip, try pigz. It's the parallel version of it.

I used to want high single core performance for my python code. I've since switched to using VSCode with devcontainers which run remotely on my 5950x. That makes it fast enough ;-).


Also web is still single threaded.


Indeed, but web is not computationally expensive. Even old computers like my laptop with i3-6157u CPU handle web just fine.


What do you mean by that?

Firefox and Chrome both have gpu hardware acceleration, and different threads/processes for UI and webpages.


Javascript is parsed and executed on one core.


If you have just one tab, with a simple webpage with no iframe, yes. But this use-case is handled even by a first-gen Raspberry Pi.

Once you start having several tas open, or web pages with different iframes, modern browsers put those in different processes.


I am informed by my personal experiences. I had an core i5 laptop from 2019 vs r4800 from 2020 and the experience was night and day running arch linux. Just smoother running everything on the AMD. And my power usage was objectively less and my fans turned on way less subjectively.

A single browser tab is single threaded, but you REGUARLY are forced to run several electron apps, and have many tabs open. Each requiring processes and threads and ryzen blows it out of the water in responsiveness. It was not uncommon for some language-server running to lag out because some chrome shit was blowing up my cpu/memory, which drove me away from vscode again.

The intel is maybe better for gaming or other inherently single program stuff. But at the cost of using WAY more energy and sucking for everything else I do regularly.

Further the m1 macs make them both look fairly power hungry and laggy.


> the iGPU in Tiger Lake ... has AV1 codec hardware decoding

Oh neat, I wasn't aware there were any AV1 decoders in hardware yet, that was always something somebody was complaining about on HN threads (probably MPEG employees HA). Tiger lake was released late last year and AV1 was finalized in 2018; that's got to be a record turnaround time from new codec to hardware decoders in top-selling chips.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: