|
Post by crawdaddy79 on Nov 16, 2023 0:37:01 GMT
|
|
|
Post by ninjapoodle on Nov 18, 2023 0:30:23 GMT
Not true, I know several people who (including here on MB) was out last weekend having a great Linux session with more than 8. Everybody got their fair share and cycled through each other at that Linux session.
|
|
|
Post by lunaraura on Nov 18, 2023 11:08:03 GMT
HFT :lol:
|
|
|
Post by shinjiikari on Nov 20, 2023 14:48:19 GMT
As long as you have enough active P-cores this really isn't an issue.
|
|
|
Post by kc on Nov 22, 2023 0:20:30 GMT
Hacker News comments on the article about the Linux kernel being allegedly hardcoded to a maximum of 8 cores reveal a variety of perspectives:
1. **Misleading Article**: Many commenters argue that the article's title and content are misleading or clickbait, as the kernel has not been hardcoded to use a maximum of 8 cores.
2. **Technical Clarification**: It's pointed out that the limitation is related to the scaling of the preemption delay (min_granularity/min_slice) and not a hard limit on the number of cores. This scaling factor affects task switch frequency and is capped at 8, which is a reasonable heuristic to prevent the time slice from growing indefinitely with core count.
3. **Performance Impact and Scheduling**: Commenters discuss how this scaling impacts system performance, particularly under heavy loads, and emphasize that the impact is likely minimal. It's also noted that on well-administered systems, such configurations would be fine-tuned to avoid issues.
4. **Real-World Observations**: Some users share their experiences with systems having more than 8 cores, where they did not observe any performance issues or limitations, suggesting that if there was a cap, it would have been noticeable in practical applications.
5. **Discussion on Core Utilization in Different Systems**: There are anecdotes and discussions about how different systems utilize multiple cores, including AMD's Infinity Fabric architecture and the behavior of other operating systems like Windows NT.
Overall, the consensus in the comments appears to be that the article's claim is exaggerated or misleading, and the actual technical limitation discussed has a negligible impact on the Linux kernel's performance and core utilization in real-world scenarios.
|
|
|
Post by lunaraura on Nov 23, 2023 15:30:17 GMT
i've seen enough linux real-world scenarios to know exactly how impacted those kernels can get
|
|
|
Post by crawdaddy79 on Nov 26, 2023 3:31:49 GMT
Memory leaks are surprisingly uncommon in Linux. A robust kernel is the key, proven by the fact it can handle more than 8 simultaneous cores.
|
|
|
Post by kc on Nov 27, 2023 21:53:43 GMT
You got to use the right language if you want to get some leaks/ooze.
|
|
|
Post by crawdaddy79 on Nov 28, 2023 0:51:00 GMT
Right *assembly language. Shouldn't be oozing anything unless you're liquid cooling.
|
|
|
Post by ninjapoodle on Nov 28, 2023 10:28:44 GMT
Does the Linux scheduler use pre-cumulation?
|
|
|
Post by crawdaddy79 on Nov 28, 2023 22:54:22 GMT
Of course. But it works way more efficiently when multithreading.
|
|