Hacker News Reader

Comment by polskibus | Hacker News Reader

It seems that 9700K still beats AMD in many workloads. I wonder if the comparison is with spectre et. al. mitigations on Intel turned on or off?

View on HN

The thing about the 9700K (which they point out) is that it doesn't support hyperthreading. Then it even beats the 9900K by a significant amount in a lot of the same tests, even though the 9900K should otherwise be the faster one, indicating the reason is that those workloads suffer a performance detriment from enabling SMT.

It should be interesting to see the results of those tests with SMT disabled.

It probably is, though they didn't explicitly state that. Windows 1903 is used, and AMD's latest strategy to improve Windows performance is in place.

Basically, if the higher clock of the 9700K can be put to use in a single-threaded application, then that chip can still come out on top. But in enough scenarios, the Ryzen 3700X and 3900X are quite competitive.

Single threaded performance isn't really that interesting anymore except for some specific gaming benchmarks, and even in gaming, you're unlikely to be bottlenecked by single threaded performance (though memory latency could be an issue).

It is still important in compilation. That means it affect developers work many many times per day.

Compiling has generally been easier to parallelize I thought? Using make -j<whatever> has been around for a long time. If you're only compiling one file then it's harder to get an improvement, but a single file should be quick anyway so there's not a huge need to speed it up. It's not like you're doing something silly and including boost are you?

In large projects, that are multiple dependency chains, that cannot be fully parallelized. There's a critical, often a bit long path, for which single threaded is still very important. Another thing is that various platforms like nodejs, etc. may not be able to utilize parallelism much, adding to the ST importance.

I agree. Just answering the question in the thread!

Web browsing is still single threaded. Games still have a very busy main thread.

> Web browsing is still single threaded.

Only in the narrow sense that processes are not threads. See `Window > Task Manager` in Chrome.

Firefox Kraken is faster on the 3700X, for example. It'll come down more to which browser you use than the processor.

1 UI thread !== only 1 thread

Firefox and chrome use a half-dozen or so threads parsing and compiling JS code so it loads quickly. Firefox's Quantum work adds a lot of threads for rendering as well. In addition, service workers, web workers, and Wasm all spawn additional processes which serve the same purpose if you aren't sharing memory directly.

That said, the Phoronix Selenium tests show AMD ahead by quite a bit. (source)[https://www.phoronix.com/scan.php?page=article&item=ryzen-37...]

So Servo is just snake-oil?

Firefox's quantum work comes from their servo work, so I'd say it's been valuable.

From comments on the article:

" They are with Spectre and Meltdown mitigations. They are not new enough results to include anything for Fallout/ZombieLoad."

Not sure if including the latter mitigations would make AMD beat Intel in single threaded.

Yeah the benchmarks aren't quite the clear AMD win (in performance/price) I was expecting for gaming. That said, it's important to factor in all the other stuff such as motherboards and coolers. Until there is a cheaper chipset, I expect a B550, the scales are tilted a bit further towards the 9700K. The CPU prices are intersting, but for anyone upgrading the total upgrade cost is what counts. So the minimum to compare would be CPU, cooler and motherboard.

It's my understanding that AMD's stock coolers are actually quite usable and fine for most people. So that might be one less price point to consider.

But comes to C/P value, Intel is totally ruined.

Reply on news.ycombinator.com