#Throughput

Hacker Newsh4ckernews
2025-10-28
Hacker Newsh4ckernews
2025-10-27

An overengineered solution to `sort | uniq -c` with 25x throughput (hist)

github.com/noamteyssier/hist-rs

2025-08-07

Storage Tail Latency Matters: The Silent Killer

When we talk about storage performance, we typically think of IOPS and throughput. In the best case, we also talk about average or best-case access latency. However, tail latency, the high-percentile end of the latency distribution, is crucial for providing predictable and consistent performance. Tail latency refers to the 95th, 99th, and 99.9th percentiles on the latency distribution scale. And the 99th percentile provides the latency experienced by the worst 1% of all […]

simplyblock.io/blog/tail-laten

A chain of calls through multiple microservices levels and databases which increase the likelihood of hitting high tail latencies in one or more calls.A histogram of a database query with different numbers of clients, showing a typical distribution and 2.5x higher tail latency.
N-gated Hacker Newsngate
2025-06-28

πŸš€ Behold, the magical that promises to triple your LLM's , as if by waving a wand made of and marketing buzzwords. πŸ€–βœ¨ But wait, there's more! Experience the thrill of saving milliseconds while drowning in GitHub's relentless onslaught of you never asked for. πŸ€―πŸ™„
github.com/LMCache/LMCache

Hacker Newsh4ckernews
2025-06-28
Felix Palmen :freebsd: :c64:zirias@bsd.cafe
2025-06-16

Feels like every time I try to reduce #memory usage, I accidentally improve #throughput instead. At least THIS time, I also see reduced memory usage, nice!

#swad #coding #c #unusual #issue

swad now reaching over 31k req/s on the same hardware.
2025-05-22

Benchmark Network-Attached Storage – It’s Harder Than You Think

TLDR: Many factors influence benchmarks for network-attached storage. Latency and throughput limitations, as well as protocol overhead, network congestion, and caching effects, may create much better or worse performance results than found in a real-world workload. Benchmarking a storage solution, especially for network-attached storage, is supposed to provide us with a baseline for performance evaluation. You run the performance test, slap them into a dashboard, build a presentation, and […]

simplyblock.io/blog/network-at

With network-attached storage, there is a whole stack underneathIOPS performance comparison of Simplyblock vs. Ceph at 16K blocksize

@geerlingguy running 3 maxxed out ultra (or strix halo) in a cluster - even rsj would get a charge out of that - how many people served/how much professional work can you crank out on one of these things in one day? (I don't think people appreciate or understand) some of the work like reports and such are generated locally but more bandwidth is better(with ai lighting the way) #req/s #siege #throughput #latency

2024-12-02

ΠšΠΎΠ½Π²Π΅ΠΉΠ΅Ρ€ΠΈΠ·Π°Ρ†ΠΈΡ: ΡƒΠ½ΠΈΠ²Π΅Ρ€ΡΠ°Π»ΡŒΠ½Ρ‹ΠΉ способ ΠΏΠΎΠ²Ρ‹ΡˆΠ΅Π½ΠΈΡ пропускной способности

Π§Ρ‚ΠΎ ΠΎΠ±Ρ‰Π΅Π³ΠΎ ΠΌΠ΅ΠΆΠ΄Ρƒ ΡΡ‚ΠΈΡ€Π°Π»ΡŒΠ½ΠΎΠΉ машиной, CPU ΠΈ микросСрвисами? ВсС ΠΎΠ½ΠΈ Π²Ρ‹ΠΈΠ³Ρ€Ρ‹Π²Π°ΡŽΡ‚ ΠΎΡ‚ Β«ΠΊΠΎΠ½Π²Π΅ΠΉΠ΅Ρ€ΠΈΠ·Π°Ρ†ΠΈΠΈΒ» (Π°Π½Π³Π». pipelining). Π’ этой ΡΡ‚Π°Ρ‚ΡŒΠ΅ ΠΌΡ‹ собСрСм ΠΈΠ½Ρ„ΠΎΡ€ΠΌΠ°Ρ†ΠΈΡŽ ΠΈΠ· Ρ€Π°Π·Π½Ρ‹Ρ… сфСр, ΠΈ ΡƒΠ²ΠΈΠ΄ΠΈΠΌ, насколько унивСрсалСн ΠΏΡ€ΠΈΠ½Ρ†ΠΈΠΏ ΠΊΠΎΠ½Π²Π΅ΠΉΠ΅Ρ€ΠΈΠ·Π°Ρ†ΠΈΠΈ. ΠŸΠΎΠ»ΡƒΡ‡ΠΈΠΌ ΠΈΠ½Ρ‚ΡƒΠΈΡ‚ΠΈΠ²Π½ΠΎΠ΅ ΠΏΠΎΠ½ΠΈΠΌΠ°Π½ΠΈΠ΅ Ρ‚Π΅Ρ€ΠΌΠΈΠ½ΠΎΠ² Β«Π·Π°Π΄Π΅Ρ€ΠΆΠΊΠ°Β» (Π°Π½Π³Π». latency) ΠΈ «пропускная ΡΠΏΠΎΡΠΎΠ±Π½ΠΎΡΡ‚ΡŒΒ» (Π°Π½Π³Π». throughput). А Ρ‚Π°ΠΊΠΆΠ΅ научимся ΡƒΠ²Π΅Π»ΠΈΡ‡ΠΈΠ²Π°Ρ‚ΡŒ ΠΏΡ€ΠΎΠΏΡƒΡΠΊΠ½ΡƒΡŽ ΡΠΏΠΎΡΠΎΠ±Π½ΠΎΡΡ‚ΡŒ Π² своих бизнСс-процСссах, ΠΊΠΎΠΌΠΏΡŒΡŽΡ‚Π΅Ρ€Π½Ρ‹Ρ… систСмах, ΠΈ просто Π² повсСднСвной ΠΆΠΈΠ·Π½ΠΈ.

habr.com/ru/articles/863198/

#Π°Ρ€Ρ…ΠΈΡ‚Π΅ΠΊΡ‚ΡƒΡ€Π°_процСссоров #Π°Ρ€Ρ…ΠΈΡ‚Π΅ΠΊΡ‚ΡƒΡ€Π°_систСм #Π°Ρ€Ρ…ΠΈΡ‚Π΅ΠΊΡ‚ΡƒΡ€Π°_ΠΏΡ€ΠΈΠ»ΠΎΠΆΠ΅Π½ΠΈΠΉ #микросСрвисы #golang #pipeline #пропускная_ΡΠΏΠΎΡΠΎΠ±Π½ΠΎΡΡ‚ΡŒ #throughput #latency #ΠΊΠΎΠ½Π²Π΅ΠΉΠ΅Ρ€

Skywalker Adventure Buildersskywalkerab
2024-08-13

Streamline logistics and maximize throughput in your adventure park for enhanced customer satisfaction and higher ROI. Discover expert strategies to optimize operations and deliver unforgettable experiences. bit.ly/3XUdmzm
.
.

2u appliance would be easier to work on with upgrade options - they use embedded epyc and 2 100g intel chips to break out into 8 25g which is really good - if you diy or try to compete there is some margin allowed to save money more than likely but with the number of ports many wil be able to forego extra switches #opnsense #fw appliances #throughput #spanning port

2023-12-30

I'm writing a new big book on software performance and scalability

If you'd like to see early chapter drafts from it, and potentially give me a little feedback (or answer a few simple questions I ask of you, about your impressions), in private? please let me know

I would give you a free copy of final ebook

#programming
#software
#performance
#scale
#scalability
#latency
#throughput
#optimization
#algorithms
#datastructures
#hardware
#compute
#efficiency
#speed
#rates
#load
#cpu
#resourceusage

2023-11-12

Our very own @mtaht will be speaking at the upcoming #UnderstandingLatency 2.0 webinar series: understandinglatency.com/

Last run was a huge success, the first event of this kind put on by Domos.
Learn From Leading Experts on Network #Latency, #Bufferbloat & #Jitter.
DECEMBER 11-13, 2023 | 4 PM CET

#LibreQoS #RoundTripTime #WorkingLatency #RTT #SpeedTest #Throughput #Bandwidth #InternetServiceProvider #CDN #IXP #OpenSource #LowLatency #ISP #WISP #FWA #5G #Starlink

Client Info

Server: https://mastodon.social
Version: 2025.07
Repository: https://github.com/cyevgeniy/lmst