Mark E. Dawson, Jr.
@medawsonjr
Followers
3K
Following
640
Media
30
Statuses
1K
CEO of JabPerf, Contributing Author to "Performance Analysis & Tuning on Modern CPUs" (1st & 2nd Edition available on Amazon), Blogger, and Former Amateur Boxer
Chicago, IL
Joined August 2015
My man @TanelPoder continues his quest to Make Observability Great Again! https://t.co/P0nEWya9jo
1
3
19
For those interested, here's the link to that other article I mentioned below which argues for O(N^1/2) - it's actually a 4-part article, not 3-part: https://t.co/8WjQLL5wOA
Very persuasive argument. I read another author a few years ago who argued over the course of a 3-part article for O(N^[1/2]). But this one uses a more convincing empirical justification.
1
4
13
Very persuasive argument. I read another author a few years ago who argued over the course of a 3-part article for O(N^[1/2]). But this one uses a more convincing empirical justification.
Memory access is O(N^[1/3]) https://t.co/4VRZLFhUwm
1
1
16
It looks like 0xC8B is still the IIO LLC Ways MSR on Xeon 6. But does anyone know the proper IIO "perfctrlsts_0" config register location for toggling DDIO on a per-root PCIe port basis? It's apparently different than on Skylake.
1
0
5
How did I miss this announcement from the Java Profiling corner of the Performance Community? https://t.co/F9rodJuaFS
mostlynerdless.de
Learn all about Java 25's new CPU-time profiler and why it matters in this weeks blog post from the creator himself.
0
0
5
Godbolt's promised "LLM-explanations of generated assembly" feature is now live in Compiler Explorer (still in beta). Example below: https://t.co/HyhfqE3deG
lnkd.in
This link will take you to a page that’s not on LinkedIn
0
2
29
📣 Join engineers from Netflix, Uber, ScyllaDB & Cloudflare at P99 CONF 2025: the free + virtual event focused on high-performance, low-latency systems. Talks on Rust, Linux, infra, distributed DBs, and more. ⚡ Oct 22–23 — Register now!
5
29
99
I feel like I should've known compilers are now capable of struct peeling. But I didn't know til I read the doc on AMD's AOCC compiler just now. Up to that point, I still considered it a manual optimization technique🤯
1
4
18
Microsoft's 2022 acquisition is making big strides with Hollow Core Fiber: https://t.co/PkilmIf3Bo
networkworld.com
The Azure team’s breakthrough, tested over 1,200 km of fiber, cuts transmission loss to below 0.1 dB/km and expands bandwidth, promising faster, cheaper, and more energy-efficient data networks.
1
0
1
Between these inexplicable price drops along with more local chip production (a la GlobalWafers' new Texas facility), this is a great time to snatch up a bunch of top-tier x86-64 servers: https://t.co/4WcyiXvwJy
tomshardware.com
Let the price war begin?
0
0
3
I've heard there's spotty support for EPYC CPUs in Intel's "pqos" command. But is there a true blue parallel for it on AMD systems? Or must one fiddle with /sys/fs/resctrl directly (which "pqos" conveniently obviates)?
0
0
2
Reports of Intel's death just might be greatly exaggerated. US floats the idea of taking a 10% stake? Days later, SoftBank drops $2 billion on INTC at $23/share? Is Intel the next "too big to fail" American business?
3
2
10
QUESTION: With the announcement of Intel discontinuing support for Clear Linux (CL), I'm curious - how many in our Low Latency Tribe are/were running CL in prod?
2
1
3
Excellent article from Brendan Gregg outlining what a Performance Engineering Team does, and when a company might consider hiring one: https://t.co/SxhlIYSsnw
1
5
25
Anyone notice the growing trend of tech blogs extolling the virtues of eBPF while showcasing nifty BPF coding skills to uncover within hours or days issues that could've been diagnosed using age-old, default-available Linux commands within *minutes*?🤷🏾♂️
4
3
27
Johnny's Software Lab is coming to #CppCon
#avx #performance #workshop #datacenters #embedded Find out more: https://t.co/PsOebG25GH
0
3
17