Questions tagged [performance]
The speed at which a program runs
276
questions with no upvoted or accepted answers
5
votes
1
answer
903
views
veth interfaces performance problem
On a fast AWS machine (m5.2xlarge), I am creating around 600 veth interfaces, each one having a little server (with socat) running on a port.
I then start sending around 7kb/second of data per server....
5
votes
0
answers
765
views
Limiting per process dirty bytes without slowing down the whole system
Given a busy system with a really fast block device (highend NVMe raid, 2 GB/s write, 4 GB/s read) mounted at /data and an a really slow device (USB HDD with spinning 8 TB disk, 50 MB/s write, 60 MB/s ...
4
votes
1
answer
227
views
How many files in a directory is too many for ext4?
I wrote a Golang program to measure how long it takes to create 100 new files and read 100 existing files in a directory with 2k files vs a directory with 200k files:
// Create 200k files in one ...
4
votes
0
answers
1k
views
Working out what is causing my machine to freeze
Background
For a few years now, I have been having problems with my Xeon workstations freezing. For many things they have been lightning fast, but sometimes applications or even the desktop just ...
4
votes
0
answers
619
views
How to mesure the latency of thread switching
I'm learning how to test the performance of a unix-like system.
I've found a tool to measure the context switching time: http://www.bitmover.com/lmbench/lat_ctx.8.html
Now I want to know if it is ...
4
votes
0
answers
451
views
systemd slices even with low CPUShare heavily affect system responsiveness
I've created custom slice (so now I have 4 slices user, system, machine, important) and assigned huge number to CPUShares. System felt really unresponsive under high load in this slice, what seems to ...
4
votes
0
answers
126
views
Fixing terrible performance when watching/reading video and writing video files to disk
I have had this issue for quite some time and tried many times to fix it on my own and it always fails. IP schedulers, ionice and things like that always yield no result. I even went and bought an ...
4
votes
1
answer
683
views
GNU/Linux, IO performance problems
I have some problems with my system performance. As I was adviced here, I used sar:
$sar -u 5
and found out that the problems are caused by a low HDD performance, as I supposed (high %iowait).
My ...
3
votes
1
answer
1k
views
Why is DM-Integrity so slow compared to BTRFS?
I want to detect silent corruption of block devices similar to how BTRFS does that for files. I'd even like to do that below BTRFS (and disable BTRFS's native checksumming) so that I can tweak more ...
3
votes
1
answer
622
views
Why is `rpm -qa base*` so terribly slow?
I wonder:
rpm uses a (simple) database to store things like packet names.
When using rpm -qa to query that database, it is terribly slow.
For example, this command took almost a minute to complete ...
3
votes
0
answers
421
views
systemd niceness and scheduling configurations not effective in processes performance
My team inherited the care and maintenance of code for a service that runs three python scripts and creates processes which play a part in sending data from our product to another product. One process ...
3
votes
0
answers
237
views
is it possible to give priority to write more then read in disks
we have hadoop cluster , with data-nodes machines that have HDFS file system
we want to give more write to disks priority then read to disks
we read the - Using the Deadline IO Scheduler ( https://...
3
votes
0
answers
2k
views
Why is this Linux process running so slow?
I've got a Python process on my Ubuntu 19.04 system that doesn't seem to be memory starved, yet is running a lot slower than I expect. The system has 96 GB of RAM plus 96 GB of swap. This process is ...
3
votes
0
answers
392
views
What is the relation between IO wait utilisation and load average
Load average uses processes that are running or runnable or in uninterrupted sleep state. So do the processes in uninterrupted sleep state correspond with the %wa as per the top command? Both are ...
3
votes
0
answers
277
views
How important is os-variant for kvm configuration?
When searching the internet one can always read that specifying os-variant is important so KVM can optimize the guest system. Nowhere did I find an article, manual or documentation stating how ...
3
votes
1
answer
938
views
smb-share connection time increases each reboot on windows PE
We are currently developing a system that does automated BIOS updates for our computers.
The system runs in Windows 10 PE booted via PXE-Boot. One of the first steps is connecting to a SMB-Share ...
3
votes
0
answers
2k
views
Performance bottlenecks of Linux traffic shaping (tc)
I have an interesting problem.
We have a Linux server with Debian Stretch installed. The server has a single CPU E5-2680 v2 and Intel 82599ES 10G Dual Port NIC's. There are no other services (like ...
3
votes
0
answers
315
views
How can I tell if my GPU is a bottleneck?
I already know of utilities like htop that can be used to (roughly) tell if a process is primarily bottlenecking at disk read/write speeds, network IO, not being parallel, or being parallel but ...
3
votes
0
answers
115
views
Why does bash appear faster on an Ubuntu VM than on the host MacOS?
This is the laptop sitting in front of me right now:
On it is installed VMWare Fusion 8.5.3 with an Ubuntu 16.04 64 bit VM. I ran the following little test to compare performance between bash on the ...
3
votes
2
answers
475
views
Verifying IO bottleneck
We are working on a project to benchmark systems in virtual machines that share the same physical host. I think we have a bottleneck with the system hosting the virtual machines, which may ...
3
votes
0
answers
55
views
Investigating annoying short UI hangs
I just bought a new desktop machine for home (see system details below). It's supposed to be super smooth, but nevertheless I'm experiencing quite frequent UI 'stutters'. It's as in the olden days ...
3
votes
2
answers
2k
views
Tracking/Monitoring Short-Lived CPU-Intensive Process on one CPU
So I've been working with this issue for some time now, and I'm having trouble finding the tools to troubleshoot this problem, which I have to imagine exist. So here's the basics of my problem:
We ...
3
votes
0
answers
761
views
How to generate graphs from lmbench results?
I need to use the lmbench benchmarking tool to get the system performance statistics.
I am able to execute and run the results. In the results folder this is a MakeFile with this information
# ...
3
votes
1
answer
6k
views
Increasing ulimit stack size benefits?
Hello I'm new in Linux and at my work I have a server that runs a banking software.
The company that made the software said that the stack size should be set to unlimited.
Does this make the software ...
3
votes
2
answers
147
views
Optimize file layout for minimal seek in specific situations
Starting VLC is slow on my system. The reason: it reads more than 1000 different .so files. Is there a way to make it read those files faster?
I think if all the files followed immediately after each ...
3
votes
1
answer
363
views
Artificially prime buffer cache?
Is it beneficial to artificially prime the buffer cache when dealing with larger files?
Here's a scenario: A large file needs to be processed, line by line. Conceptually, it is easy to parallelize ...
3
votes
0
answers
246
views
Cgroups - memory subsystem
I want to set the memory.limit_in_bytes parameter for a control group and disable OOM, so processes freeze when the limit is reached. That said, I want to change that limit dynamically based on ...
3
votes
0
answers
2k
views
Send print job to printer immediately without queueing, aka increase print performance
I have a ruby/rails web application, one action of which sends PDF files that have been generated by the app to a Xerox 4127 printer on the network. This is accomplished via shelling out and running ...
3
votes
0
answers
493
views
router latency measured by tcpdump increases
I am using tcpdump to measure latency on a linux router, i.e. the delay experienced by packets as they pass through the router (made of Ubuntu 10). For this purpose, I captured packets in the IN and ...
3
votes
1
answer
123
views
Timescales of any task in linux kernel?
I am studying how Linux kernel works and getting interested in the timescale of anything that happens in the kernel.
Of course I know it depends on the HW, but would like to get an example.
For ...
2
votes
0
answers
63
views
What's the expected overhead of a passthrough device mapper?
I'm trying to establish a baseline throughput overhead for a passthrough device mapper; i.e. a device mapper that does nothing. Roughly following benchmarking procedures from Cloudflare, I'm measuring ...
2
votes
0
answers
39
views
A simple memcpy loop with stride of 8-16-32 Bytes, no L1 cache misses, what stalls backend cycles?
I am trying to understand the CPU cache performance in a single producer single consumer queue algorithm, but cannot pinpoint the cause of performance degradation in some cases. The following ...
2
votes
1
answer
66
views
How can a different controller explain this I/O?
My new NAS has 5 disks. They are all same model.
sde is connected to the controller on the motherboard. sda-sdd is on the raid controller.
Run 'pv /dev/sd[cde]' in parallel:
Device rkB/s wkB/...
2
votes
0
answers
224
views
How to find out what is causing decreased performance when USB-C power delivery is plugged in on a laptop running linux
I searched around and could not find an answer to this issue.
For context: I am running ArchLinux on my Tuxedo InfinityBook 14 gen7 with RTX 3050 TI, the regular power adapter that came with the ...
2
votes
1
answer
269
views
whisper.cpp takes a long time to start. Can this be avoided?
I am using https://unix.stackexchange.com/users/560724/georgi-gerganov 's whisper.cpp to transcribe recorded interviews. Many of these are short, a minuter or two. The problem is that whisper.cpp ...
2
votes
0
answers
379
views
Inode size 512 and 1024 bytes functions in ext4 and its pros and cons? (Need a official reference)
When i first trying and installing openSUSE tumbleweed, i see when formatting a ext4 i can choose the inode size 128, 256, 512, or 1024 bytes. Typically it is 256. I am thinking a larger inode size ...
2
votes
0
answers
138
views
Question about transfer data between specific physical memory and NVME
I'm trying to transfer data from specific reserved physical memory to my NVME, IIUC, there are 3 steps:
In the kernel driver, the reserved-memory (specified physical address and size in dts) is ...
2
votes
1
answer
1k
views
Why 7z not using enough CPU?
CentOS 9. 3 vCPU VM with 4Gb RAM.
I run a cron job with 7z compressing 35Gb data in 150 files 7za a -mx=9 -mmt=3 ...
RAM usage - 18%, disk queue is very small, CPU is 61% on average. Why not 100% ? ...
2
votes
0
answers
779
views
New computer running arch linux gets weirdly loud
I have a relatively new computer and I just installed Arch linux. From the moment I start the computer, it starts with a loud noise that is very similiar to a computer running on 100% of the CPU. From ...
2
votes
0
answers
58
views
Why does redirecting lines ≥1022 characters take 83× longer than redirecting lines ≥1021 characters?
I have a text file of around 10 MB containing around 50k lines. When I select all lines ≥1021 bytes and redirect the output to a regular file or pipe it to cat, it takes 0.135s. When I change this ...
2
votes
0
answers
3k
views
Why are writes so slow on ext4?
I'm trying to work out why writing to ext4 partitions on my server is so slow. I noticed this issue when I reformatted my off-site backup drives from ExFAT to ext4. When these drives were formatted as ...
2
votes
1
answer
326
views
Why my download speed is just half of the speed given by my Wi-Fi providing device in Ubuntu 20?
I had just upgraded to Ubuntu 20 (focal fossa). When I try to download anything in any browser or through terminal, the downloading speed in my PC is just half of my Wi-Fi providing device.
Please ...
2
votes
0
answers
68
views
How do I see how much CPU time my firewall uses?
I was attacked with a DNS Amplification and as a result added nearly 4,000 IP addresses to my firewall using fail2ban.
I have seen many people worried about how many rules they include in their ...
2
votes
0
answers
3k
views
What's the best practice of mergerfs with nfs?
I run mergerfs with latest Debian Testing with kernel 5.4.
Linux ches-disk0 5.4.0-4-amd64 #1 SMP Debian 5.4.19-1 (2020-02-13) x86_64 GNU/Linux
It is on a Dell R730 server with 3 disk arrays, CPU is ...
2
votes
0
answers
586
views
Which file to set isolcpus on CentOS 7?
I am using CentOS 7 and would like to isolate cores 0 and 1 for the OS/Kernel, meaning nothing will be scheduled on my other cores.
I need isolcpus
After some Googling I found this:
https://access....
2
votes
0
answers
300
views
Smooth Linux VM (VirtualBox) on Windows 10?
I need to create a Virtual Machine running a Linux-based OS. I have tried with Ubuntu 16.04 LTS and 18.04 LTS. The problem that I can't help having is that its interface is laggy and slow in terms of ...
2
votes
0
answers
54
views
dd measurement question
I am wrestling with some perceived performance issues on a 3-node HPE Apollo cluster running Linux. (The cluster supports a backup application.) These are very $$ servers. In our setup each node ...
2
votes
0
answers
105
views
Linux access windows share without loading to RAM
I have a laptop running an Arch-based Linux system and frequently need to access a windows share at my workplace. This is really annoying since it takes ages to load. Particularly, one folder is ...
2
votes
0
answers
12k
views
How to find the cause of High Load Average on Linux
I trying to find the reason why my server is getting as high as 150 in load average, and some time it get up to 200.
This is a Ubuntu Server running on a VMware ESXi virtual machine, with 8 CPU and ...
2
votes
0
answers
424
views
How can I find the deepest and most stack-hungry call chains in a program?
I have a program written in OCaml that tends to stack-overflow on platforms with small stacks.
Rather than ulimiting the program to find and solve stack overflows one by one, I'd like to get a sorted ...