Hedon James Posted July 8 Author Posted July 8 On 7/5/2024 at 9:00 AM, Hedon James said: I seem to have eliminated my xruns, according to Cadence, by: decreasing swappiness to 10 changing cpugovernor to performance adding threadirqs and mitigations=off to kernel boot parameters changing the MIDI driver in Cadence from "Alsa-MIDI-raw" to "none"; I read this either conflicts, or adds an additional layer to a2jmidid; changing to "none" seems to have a positive effect I spent a long day in recording sessions yesterday, with no xruns; even using DrumGizmo (without tweaking) to convert MIDI to analog drums. Cadence xruns remained at 0 (this was my baseline experience in old OS and Ardour 3.5 and old KXStudio repos). But Ardour seems to randomly indicate xruns (this is a new behavior in 7.3 that didn't exist in 3.5). I don't know if Cadence is providing false zeros, or if Ardour is providing false xruns. <snipped> Just some additional notes/observations/clarifications, which may prove useful for anyone following along. Referencing the bolded statements: I forgot to mention that the JACK settings in the Cadence application were modified from buffer size of 256 (resulting in 5.8ms block latency) to 512 (resulting in 11.8ms latency), with sample rate of 44.1, which is CD quality. I see no reason to exceed that for my use. I'm sure there are use cases where 48k+ are desired, but it's overkill for me, IMO. I was tinkering with 256 buffer because I was trying to find the upper limits of capability with the Ryzen3 CPU and I'd say that limit is good for about 90% of my sessions, but occasionally spiking off an xrun. 512 seems to be a good limit 100% of my sessions, with no discernible latency issues. I think I've got JACK and Cadence sorted out, and starting to have confidence in those settings. With respect to Ardour, still getting occasional xrun indications from within Ardour (top right, next to log/msg indicator), despite Cadence/JACK indicating no issues. Having spent several hours in the studio tinkering with things this holiday weekend, I STRONGLY suspect that Ardour indications of xruns are actually "clip" notifications, when my recorded signal hit the 0.0db threshold in channel meters. I'm not 100% certain, but certainly "more likely than not"....I'm about 90% confident that is what's happening. It seems that Cadence/JACK and Ardour are using slightly different triggers to indicate an xrun has occurred. Did some additional research on side-by-side installations of Ardour (2 instances running on same machine), and it is indeed possible. Sunrat indicates the Ardour installation script installs to the /opt directory. I have no installation of Ardour7 in that directory, so I believe that repo installation occurs to the system like a typical repo binary. One caveat is that Ardour will update a "series", but won't update to the next "series". In other words, I am currently on Ardour7.3 from the repo. I may install Ardour8 from the Ardour site, which can be upgraded incrementally from 8.0 to 8.1.....to 8.9; so I can never have 2 versions of the same series at the same time....I cannot have Ardour 8.1 and Ardour 8.9, for instance, because the 8.9 installation WILL overwrite the 8.1 installation. When Ardour bumps from 8.9 to 9.0, a new installation of Ardour9 will be required for v9; and Ardour8.9 and Ardour9.0 will co-exist simultaneously in the /opt directory, each with their own launcher. So pay attention to your launchers when launching Ardour. The Ardour8 launcher will launch version 8.0-8.9; but the Ardour9 launcher will launch versions 9.0-9.9. So watch what you're clicking to open Ardour, or make sure you have the correct launcher in any docks/quick launchers. A quick mouse-over will reveal which version you're about to launch. The upshot of all this is that I'm going to follow Sunrat's suggestion and download Ardour8.9 (the most recent version) to see if that doesn't fix the many minor issues I'm experiencing in Ardour7.3. But I should do that quickly, because once they go to Ardour9.0, I KNOW there will be new and fresh bugs present in the new series that take a while to get ironed out. I'm guessing that's where I'm at right now.....someone at Debian decided Ardour7.3 is the 5-year version for Debian Bookworm Stable, even though it's a little buggy because the 3rd version in the Ardour7 series isn't as polished as say....version 7.6 or 7.9. FWIW... 1 Quote
sunrat Posted July 9 Posted July 9 1 hour ago, crp said: Is latency an issue that is a real-life problem for this? It can be for audio production. If you play a keyboard using a softsynth in a DAW, you want the note to play as quickly as possible after you hit the key. If you have a latency of, say 30ms, that's close to a 1/16th beat at 120BPM. Would be very distracting and your recording may end up unusable. Also if you do overdubs, layering fresh tracks adding to already recorded ones, it needs to be as close as possible to zero. Under 10ms is good to aim for in these cases, under 5ms even better. There are also phase considerations as well as timing, it can get quite complicated and messy. 1 Quote
Hedon James Posted August 27 Author Posted August 27 still tinkering with settings, chasing that holy grail of ZERO xruns. To recap, my former studio machine was running Lubuntu 16.04 (LXDE), with KX Studio repos, which brought in Ardour 3.5, Cadence, Jack, etc.... Ran like a champ using AMD FX-6200 CPU, 16GB RAM, and discrete GPU...Radeon 6550(?) I believe. Within past 2 years, upgraded from the FX-6200 to Ryzen 3 3200G, mainly for increased graphics ability, as that Radeon 6550 is getting pretty dated. Shortly after, I read a thread on Linux Musicians indicating that GPUs were often the "bottleneck" for performance, with a whole lot of techno-speak to back it up. But the source was Robin Gareus (x42 handle), so he's a credible guy...not just an opinion-monger. So I switched back to the discrete GPU, allowing the CPU to be relieved of graphics duty. Maybe it helped, maybe it didn't....but I noticed no change. Machine continued to run like a champ, and all was well. Fast forward to this spring, and that Lubuntu 16.04 base was out of support, and it was becoming a PITA to maintain the air-gap, but still get online long enough to swap files, stems, and tracks. So I bit the bullet and built a new workstation on the SAME hardware. But the OS was now Debian 12 (with LXQt), with Ardour 7.6 from Debian repos and KX Studio repos for other audio components, plugins, etc... So the OS changed (Lubuntu > Debian), the desktop environment changed (LXDE > LXQT), the DAW changed (Ardour 3.5 > Ardour 7.6), and my plugins changed (primarily Calf > primarily LSP Studio). With all that change, I started incurring xruns. It was very noticeable at first, on certain plugins, but I've got them massaged to a lower level....a manageable level, but still enough to BUG me. I had initially ruled out hardware, because it's the same hardware that produced ZERO xruns on the older setup. But I'm re-considering that position now....MAYBE that hardware is getting exercised a little harder to run Debian 12 vs Lubuntu 16.04; to run LXQT vs. LXDE; to run a newer version of Ardour; to run newer plugins. Software tends to "bloat" over time with newer features and patches. Maybe ALL those cumulative changes have caused the hardware to cross a threshold from "no problem" to "sometimes a problem". I rarely max the RAM, so if I consider that maybe it IS a hardware issue, I'm thinking it's likely the CPU and/or the GPU. Here's the benchmarks between the Ryzen 3 3200G and the FX-6200: https://cpu.userbenchmark.com/Compare/AMD-Ryzen-3-3200G-vs-AMD-FX-6200-Six-Core/m824486vsm2658 At first glance, substantially similar performance. I note the Ryzen 3 has 4 cores, but only 4 threads, rather than 8; and I note the cores are only running from 1257-1400 Mhz, depending on the core, based on inxi output. If the assumption is correct, and new(er) software has placed increased demands on hardware, and hardware has reached it's limits, then new(er) hardware should be solution to solve issues running more "bloated" software, correct? After much research, I've zero'd in on Ryzen 5 5600G, with comparison to current Ryzen 3 as follows: https://cpu.userbenchmark.com/Compare/AMD-Ryzen-3-3200G-vs-AMD-Ryzen-5-5600G/m824486vsm1553183 I'm specifically drawn to the 6-cores and 12 threads in the Ryzen 5, versus 4-core and 4 threads in the Ryzen 3. The triple amount of threads should be a BIG difference processing power, speed, and reduced latency, correct? I also note the BIG increase in speed AND the reduced latency, on a per-core basis. Wondering what your thoughts are, Sunrat? Coulda asked you in a PM, but seems like this information might prove helpful to other folks, as well as myself. Your notes regarding your setup on your KDE system indicate you have an Intel 5 i-core. What is your CPU? I'd be interesting in seeing the specs on your CPU and comparing benchmarks with the identified Ryzen 5. Maybe I need more CPU than the Ryzen 5 5600g? Seems like your CPU might establish a good benchmark for me to achieve? Thoughts? Quote
sunrat Posted August 27 Posted August 27 inxi -ACGSxz System: Kernel: 6.9.12-2-liquorix-amd64 arch: x86_64 bits: 64 compiler: gcc v: 12.2.0 Desktop: KDE Plasma v: 5.27.5 Distro: Debian GNU/Linux 12 (bookworm) CPU: Info: quad core model: Intel Core i5-6500 bits: 64 type: MCP arch: Skylake-S rev: 3 cache: L1: 256 KiB L2: 1024 KiB L3: 6 MiB Speed (MHz): avg: 1224 high: 1768 min/max: 800/3201 boost: enabled cores: 1: 1230 2: 800 3: 1768 4: 1100 bogomips: 25599 Flags: avx avx2 ht lm nx pae sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx Graphics: Device-1: NVIDIA GM204 [GeForce GTX 970] vendor: ASUSTeK driver: nvidia v: 535.183.01 arch: Maxwell bus-ID: 01:00.0 Display: x11 server: X.Org v: 1.21.1.7 with: Xwayland v: 22.1.9 driver: X: loaded: nvidia unloaded: fbdev,modesetting,nouveau,vesa gpu: nvidia resolution: 1: 3840x2160~60Hz 2: 3840x2160~60Hz API: EGL v: 1.5 drivers: nvidia,swrast platforms: active: gbm,x11,surfaceless,device inactive: wayland,device-1 API: OpenGL v: 4.6.0 vendor: nvidia v: 535.183.01 glx-v: 1.4 direct-render: yes renderer: NVIDIA GeForce GTX 970/PCIe/SSE2 API: Vulkan v: 1.3.239 drivers: nvidia,llvmpipe surfaces: xcb,xlib devices: 2 Audio: Device-1: NVIDIA GM204 High Definition Audio vendor: ASUSTeK driver: snd_hda_intel v: kernel bus-ID: 01:00.1 Device-2: VIA ICE1712 [Envy24] PCI Multi-Channel I/O driver: snd_ice1712 v: kernel bus-ID: 03:00.0 API: ALSA v: k6.9.12-2-liquorix-amd64 status: kernel-api Server-1: JACK v: 1.9.21 status: off Server-2: PulseAudio v: 16.1 status: active My system is old, I built it in 2016. I specifically chose this CPU because it did not have Hyperthreading - 4cores, 4 threads. I'm really not familiar with optimising AMD CPUs, haven't used one for 16 years. They should be just as good for audio performance. LinuxMusicians forum is probably still the best place to find info about system setup. There are regularly threads about building new ones and optimising for audio. Compare - https://cpu.userbenchmark.com/Compare/Intel-Core-i5-6500-vs-AMD-Ryzen-5-5600G/3513vsm1553183 I have also recently been kicking the tyres of Pipewire, trying out Debian Trixie, OpenSUSE Tumbleweed, and AVL-MXE 23.2 all of which use Pipewire. Pulled my hair out for many hours trying to work out an issue where Ardour and Mixbus would freeze after export. Turned out it was just that I chose the wrong Profile setting in pavucontrol Configuration. Ultimately Pipewire seems to work quite well after doing the same optimisations as for the old system. I still don't plan on changing to it for daily use any time soon; it's a PITA trying to use Testing or Tumbleweed due to the endless flood of updates. It will put me ahead of the pack for when Trixie eventually becomes stable though. 1 Quote
Hedon James Posted August 28 Author Posted August 28 14 hours ago, sunrat said: inxi -ACGSxz System: Kernel: 6.9.12-2-liquorix-amd64 arch: x86_64 bits: 64 compiler: gcc v: 12.2.0 Desktop: KDE Plasma v: 5.27.5 Distro: Debian GNU/Linux 12 (bookworm) CPU: Info: quad core model: Intel Core i5-6500 bits: 64 type: MCP arch: Skylake-S rev: 3 cache: L1: 256 KiB L2: 1024 KiB L3: 6 MiB Speed (MHz): avg: 1224 high: 1768 min/max: 800/3201 boost: enabled cores: 1: 1230 2: 800 3: 1768 4: 1100 bogomips: 25599 Flags: avx avx2 ht lm nx pae sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx Graphics: Device-1: NVIDIA GM204 [GeForce GTX 970] vendor: ASUSTeK driver: nvidia v: 535.183.01 arch: Maxwell bus-ID: 01:00.0 Display: x11 server: X.Org v: 1.21.1.7 with: Xwayland v: 22.1.9 driver: X: loaded: nvidia unloaded: fbdev,modesetting,nouveau,vesa gpu: nvidia resolution: 1: 3840x2160~60Hz 2: 3840x2160~60Hz API: EGL v: 1.5 drivers: nvidia,swrast platforms: active: gbm,x11,surfaceless,device inactive: wayland,device-1 API: OpenGL v: 4.6.0 vendor: nvidia v: 535.183.01 glx-v: 1.4 direct-render: yes renderer: NVIDIA GeForce GTX 970/PCIe/SSE2 API: Vulkan v: 1.3.239 drivers: nvidia,llvmpipe surfaces: xcb,xlib devices: 2 Audio: Device-1: NVIDIA GM204 High Definition Audio vendor: ASUSTeK driver: snd_hda_intel v: kernel bus-ID: 01:00.1 Device-2: VIA ICE1712 [Envy24] PCI Multi-Channel I/O driver: snd_ice1712 v: kernel bus-ID: 03:00.0 API: ALSA v: k6.9.12-2-liquorix-amd64 status: kernel-api Server-1: JACK v: 1.9.21 status: off Server-2: PulseAudio v: 16.1 status: active My system is old, I built it in 2016. I specifically chose this CPU because it did not have Hyperthreading - 4cores, 4 threads. I'm really not familiar with optimising AMD CPUs, haven't used one for 16 years. They should be just as good for audio performance. LinuxMusicians forum is probably still the best place to find info about system setup. There are regularly threads about building new ones and optimising for audio. Compare - https://cpu.userbenchmark.com/Compare/Intel-Core-i5-6500-vs-AMD-Ryzen-5-5600G/3513vsm1553183 That's what I'm looking for! Interesting enough, your i5 is only 4 core, no threading....so maybe threading isn't the issue I think it is? And your 4 cores are averaging slower speeds than my Ryzen 3, which is also a 4-core no threading CPU. In fact, your i5 is nearly identical to my Ryzen3. The biggest difference I can see is your i5 has a lower memory latency than the Ryzen 3. Not sure that's enough to explain better performance, but IF it is, the Ryzen 5 is comparable. I don't see your RAM in inxi stats. How much RAM in your box? I'm guessing by the age of your build it's DDR3, or possibly DDR4? The only other thing I've seen in the LinuxMusician Forums that I want to look into further, is isolating my USB soundcard on its own bus. Have you done that? Can you point me to a resource that explains how to do that in a scientific manner, rather than just plug into different ports and see what happens? Thanks Sunrat! 1 Quote
sunrat Posted August 28 Posted August 28 inxi -m Memory: System RAM: total: 16 GiB available: 15.58 GiB used: 2.91 GiB (18.7%) Array-1: capacity: 64 GiB slots: 4 modules: 2 EC: None Device-1: ChannelA-DIMM0 type: DDR4 size: 8 GiB speed: 2133 MT/s Device-2: ChannelA-DIMM1 type: no module installed Device-3: ChannelB-DIMM0 type: DDR4 size: 8 GiB speed: 2133 MT/s Device-4: ChannelB-DIMM1 type: no module installed I don't know much about USB sound devices. The rtcqs script will tell you if your USB bus has its own IRQ, you should see something like this: [ OK ] USB port xhci_hcd with IRQ 128 does not share its IRQ. You can list all IRQs with: cat /proc/interrupts 1 Quote
Hedon James Posted September 23 Author Posted September 23 oh boy....I really "did it" this time! went down the rabbit hole, chasing xruns, and really buggered up my system...TOASTED it! Referencing above discussions (for those following along) and ordered the Ryzen 5 5600x, with intentions to use my old Radeon 6450 GPU installed onboard rather than the all-in-one graphics of the Ryzen 5 5600g. But when I installed the Ryzen 5, all hell broke loose. I'll spare the gory details and summarize as follows. My motherboard required a BIOS update to accommodate the Ryzen 5 chipset. I believed because I was using a Ryzen 3 already, that I had that update. Almost certainly, I did not. The Ryzen 5 was installed and the computer wouldn't even POST...no beeps....NOTHING. Checked ALL connections and, after multiple attempts, believed I had a "bad cpu". But to confirm with 100% certainty, I pulled the Ryzen 5 and reinstalled the previously working Ryzen 3...with the same result. NO POST...no beeping...NOTHING. After MUCH google-searching I discovered the updated BIOS issue for my mobo, which is almost certainly the root cause. But I needed a CPU to POST before I could flash the updated BIOS. Called a computer guru buddy who does this for a living and he suggested he had an old AM4 socket CPU (an A8?) that we could install for the POST to accommodate BIOS flash. Boxed up my stuff and went over immediately. Installed his CPU (KNOWN to be good), but the same result....NO POST. At this point he delivered his diagnosis....he believes that the lack of BIOS updated somehow allowed the Ryzen 5 to corrupt the MOBO, which in turn, likely corrupted the CPU. When I pulled the R5 and reinstalled the R3, there's a very good chance the MOBO corrupted the R3 also. And the MOBO probably corrupted his AM4 A8 chip too. He didn't want to test it in a known good configuration for fear that this old $20 CPU could corrupt a $50 MOBO. So we think the lack of BIOS update on my MOBO caused the R5 to corrupt the MOBO; which then corrupted the R5; and subsequently corrupted the R3 when I re-installed to troubleshoot. DAYUMMMM.... Not sure what I could've done differently....I verified the MOBO was R5-compatible, but had no way of knowing whether the BIOS was compatible. And looking through the BIOS update/downloads, not sure which one would've provided the R5 compatibility. A true chicken & egg proposition. I had no choice but to pull my old drives (I always separate the OS and DATA and BACKUP drives for just such a scenario) and plug them into my spare system, with an antiquated AMD-FX6200 cpu. Music studio back online, but now even less speed/power than before... Quote
securitybreach Posted September 23 Posted September 23 Next time before making major changes, clone your install using clonezilla. Its pretty easy and might take 20m to complete. Download the live usb version. Ian made a tutorial years ago and not much has changed on how you use it: Quote
Hedon James Posted September 23 Author Posted September 23 And I told you THAT story ^ to make sense of this one...I've been slowly transitioning my home network from my fleet of "box tower" form factor machines to SFF, USFF, and/or MFF machines (seems like every vendor calls their machine something different); what I call "monitor-mount" form factors....machines such as Lenovo ThinkCentre and Dell Optiplex. The last "box tower" in my network was my studio machine, which was planned for conversion at a future date, once I had resolved all my xrun issues and had a few dollars to spare. That plan got accelerated tremendously, despite the unrealized criteria to trigger the event. A failed recording machine trumps all criteria. Turns out I could get my hands on one of those machines for less money than a replacement MOBO and CPU....and the refurb machine was actually better-specced than the replacement MOBO & CPU! Which made the new acquisition a no-brainer! I found a Dell OptiPlex 3000 machine, with Intel Core i5-12600 cpu and onboard Intel graphics, 256GB NVME drive, and 16GB RAM. I know from many years of experience that Dell tends to be very linux-friendly, but confirmed online that the OptiPlex 3000 specifically was linux friendly. It's a refurb from corporate enterprise, but looks brand new and is still in the Dell Support window, according to their service ID Tag?! The guy wanted $200, so I pulled the trigger! Machine specs indicated by inxi -Fxz as follows: Quote jim@OptiStudio:~$ inxi -Fxz System: Kernel: 6.8.0-45-lowlatency arch: x86_64 bits: 64 compiler: gcc v: 13.2.0 Console: pty pts/0 Distro: Ubuntu 24.04.1 LTS (Noble Numbat) Machine: Type: Desktop System: Dell product: OptiPlex 3000 v: N/A serial: <superuser required> Mobo: Dell model: 0R7HRW v: A02 serial: <superuser required> UEFI: Dell v: 1.25.0 date: 08/07/2024 CPU: Info: 6-core model: 12th Gen Intel Core i5-12600T bits: 64 type: MT MCP arch: Alder Lake rev: 5 cache: L1: 480 KiB L2: 7.5 MiB L3: 18 MiB Speed (MHz): avg: 822 high: 1040 min/max: 800/4600 cores: 1: 800 2: 800 3: 800 4: 800 5: 800 6: 800 7: 800 8: 1040 9: 834 10: 800 11: 800 12: 800 bogomips: 50688 Flags: avx avx2 ht lm nx pae sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx Graphics: Device-1: Intel Alder Lake-S GT1 [UHD Graphics 770] vendor: Dell driver: i915 v: kernel arch: Gen-12.2 bus-ID: 00:02.0 Display: server: X.org v: 1.21.1.11 driver: X: loaded: modesetting unloaded: fbdev,vesa dri: iris gpu: i915 tty: 63x17 resolution: 1920x1080 API: EGL v: 1.5 drivers: iris,swrast platforms: active: surfaceless,device inactive: gbm,wayland,x11 API: OpenGL v: 4.6 compat-v: 4.5 vendor: mesa v: 24.0.9-0ubuntu0.1 note: console (EGL sourced) renderer: Mesa Intel UHD Graphics 770 (ADL-S GT1), llvmpipe (LLVM 17.0.6 256 bits) Audio: Device-1: Intel Alder Lake-S HD Audio vendor: Dell driver: snd_hda_intel v: kernel bus-ID: 00:1f.3 Device-2: BEHRINGER GmbH UMC404HD 192k driver: snd-usb-audio type: USB bus-ID: 1-5:4 API: ALSA v: k6.8.0-45-lowlatency status: kernel-api Server-1: JACK v: 1.9.21 status: off Server-2: PipeWire v: 1.0.5 status: active Network: Device-1: Realtek RTL8111/8168/8211/8411 PCI Express Gigabit Ethernet vendor: Dell RTL8111/8168/8411 driver: r8169 v: kernel port: 3000 bus-ID: 02:00.0 IF: enp2s0 state: down mac: <filter> Device-2: Intel Wi-Fi 6E AX210/AX1675 2x2 [Typhoon Peak] driver: iwlwifi v: kernel bus-ID: 03:00.0 IF: wlp3s0 state: up mac: <filter> Bluetooth: Device-1: Intel AX210 Bluetooth driver: btusb v: 0.8 type: USB bus-ID: 1-14:6 Report: hciconfig ID: hci0 rfk-id: 0 state: up address: <filter> bt-v: 5.3 lmp-v: 12 Drives: Local Storage: total: 5.69 TiB used: 713.18 GiB (12.2%) ID-1: /dev/nvme0n1 vendor: Micron model: 2450 NVMe 256GB size: 238.47 GiB temp: 39.9 C ID-2: /dev/sda vendor: Western Digital model: WD20SPZX-22UA7T0 size: 1.82 TiB type: USB ID-3: /dev/sdb vendor: Western Digital model: WD40EZRZ-00GXCB0 size: 3.64 TiB type: USB Partition: ID-1: / size: 233.38 GiB used: 13.98 GiB (6.0%) fs: ext4 dev: /dev/nvme0n1p2 ID-2: /boot/efi size: 299.4 MiB used: 23.7 MiB (7.9%) fs: vfat dev: /dev/nvme0n1p1 Swap: ID-1: swap-1 type: file size: 512 MiB used: 415.4 MiB (81.1%) file: /swapfile Sensors: System Temperatures: cpu: 46.0 C mobo: 59.0 C sodimm: Mem C Fan Speeds (rpm): cpu: 1367 Info: Memory: total: 16 GiB note: est. available: 15.31 GiB used: 2.84 GiB (18.5%) Processes: 422 Uptime: 3d 15h 49m Init: systemd target: graphical (5) Packages: 2710 Compilers: gcc: 13.2.0 Shell: Bash v: 5.2.21 inxi: 3.3.34 That machine is an upgrade from prior hardware, so I loaded it up with OS and software, purchased an external Sabrent disk caddy and plopped my Data and Backup Drives in the 2 drive bays (my preferred configuration with these monitor-mount form factors) and have been pretty pleased so far. But I changed my software setup too....details in next post... Quote
Hedon James Posted September 23 Author Posted September 23 14 minutes ago, securitybreach said: Next time before making major changes, clone your install using clonezilla. Its pretty easy and might take 20m to complete. Download the live usb version. Ian made a tutorial years ago and not much has changed on how you use it: Not sure how Clonezilla would've helped me here. My drives (and data) were all intact. Everything "transferred" without issue. I had a hardware failure, but software/data recovery was fine. How would Clonezilla have helped? What am I missing here, SB? Quote
securitybreach Posted September 23 Posted September 23 5 minutes ago, Hedon James said: Not sure how Clonezilla would've helped me here. My drives (and data) were all intact. Everything "transferred" without issue. I had a hardware failure, but software/data recovery was fine. How would Clonezilla have helped? What am I missing here, SB? Well you could of taken the clone and restored it to another machine. Quote
Hedon James Posted September 23 Author Posted September 23 post 3 of 3 regarding my Recording Rig SNAFU: I had previously noted my "upgraded" recording rig changed EVERYTHING from a known-EXCELLENT configuration that was just too old to be upgraded anymore. I was forced to move on, despite reluctance to do so. My SUPREMELY STABLE first machine, with ZERO xruns was based on Lubuntu 16.04 and KX Studio repos of compatibility. Ubuntu uses "low-latency" kernels, which I had installed, and version 3.5(?) of Ardour. Whenever I have to upgrade, I go to the "leading edge" (not the "bleeding edge), and then hang on for the duration of LTS, possibly extending beyond that with backports. So when I rebuilt my OS, I made several changes (everything?!), LOL: I built with Debian 12 Bookworm, recommended liquorix kernel, and compatible KX Studio repos. KX Studios stopped providing DAWs, so I got the Debian version of 7.x, with updated plugins, etc... and a new batch of plugins from KX that were HIGHLY recommended by others on forums....Linux plugins, rather than Windows plugins adapted for linux use (Calf plugins). So I changed OS, kernels, DAW versions, and plugins....there were NO CONSTANTS...everything was a variable. And I had xruns like never seen before. @sunrat helped with some troubleshooting, and setup fixes, which helped TREMENDOUSLY....but I was chasing that holy grail of ZERO xruns, which I had previously experienced on older setup. Sunrat also indicated it was better to start with a dedicated audio-distro than to modify a base distro, and I trust his advice. I looked at AVLinux on Sunrat's recommendation and just couldn't come to grips with the Enlightenment WM. There was things I liked, but all in all, it was just too strange for me. I've got enough on my plate without learning to utilize a strange new interface, and I didn't want to install Enlightenment, then add another layer of DE (perhaps KDE or LXQT or Mate?) and bloat the menu with options and "dead entries". I found GeekOS DAW (based on OpenSUSE) https://geekosdaw.tuxfamily.org/en/ which seems to be an interesting distro, but we're back to "build it yourself", and I have very little experience working with RPM distros and tools...although if I was going to use an RPM distro, OpenSUSE checks all my boxes. Fedora has an audio distro, called Fedora Jam https://fedoraproject.org/labs/jam/ but the same arguments apply regarding RPM distros. KX Studio has been discontinued, along with several other audio distros. Which brought me back full-circle to Ubuntu Studio, which I had previously avoided because of all the extra packages that have NOTHING to do with audio production, but cater to desktop publishing, graphics production, etc... That's IT....that's ALL THE CHOICES for linux desktop Audio distros. Ubuntu Studio uses the XFCE DE, which I'm not a fan of, but I can live with...at least it's lightweight...and I was ready to bite the bullet and swallow the Ubuntu Studio offering. Until I saw that Ubuntu Studio also has a "studio installer" package that can be installed on ANY of the 'Buntu distros and allows you to select the packages you want, and ONLY those packages...an audio group, a graphics group, a publishing group, etc... Now THIS changes things. Wanting to get back to that holy grail configuration with ZERO xruns, I could install Lubuntu (the lightest of the 'Buntus), then install the "studio installer" package and select ONLY the audio packages, and the installer would make all the necessary tweaks "under the hood" for audio, such as low-latency kernel installation, irq tweaks, etc... The biggest drawback was the 3-year support of LTS, which is a big reason I left Ubuntu in the first place. But I also discovered (i had forgotten) that I'm an Ubuntu PRO subscriber, so I get 10-year security updates, despite the OS itself only getting 3 years. But I can live with that. I installed Lubuntu 16.04 back in 2016, with Ubuntu PRO subscription....my upgrade was forced by software/package issues with such an old version of Ardour DAW, not by security issues. I have since learned that multiple versions of Ardour can exist on the same machine, as long as they're not within the same series. (versions such as 6.x, 7.x, and 8.x can exist simultaneously; while versions 7.x1 and 7.x2 cannot). With this new knowledge, I could've gotten 2 more years out of Lubuntu 16.04, LOL! So that's where I'm at now...back on Lubuntu 24.04.1 for the recording machine ONLY; with Ubuntu Pro 'script; ubuntu-installer audio modifications; and KX Studio repos for plugins not available through Ubuntu repos. All seems to be good again, although I do experience xruns with "heavy load" such as switching kit configurations for my Drum plugins. But those xruns appear to be limited to "unloading & reloading" only....none during recording, while the transport is moving. KNOCK ON WOOD! While not ideal, I can live with this! It's been said that "if you think education is expensive, try ignorance". Don't I know THAT?! LOL! 1 Quote
Hedon James Posted September 23 Author Posted September 23 38 minutes ago, securitybreach said: Well you could of taken the clone and restored it to another machine. i just pulled my drives and plugged into my backup machine. it worked out. i suspect that was WAY quicker than restoring a Clonezilla backup. At least, in this instance. Quote
sunrat Posted September 23 Posted September 23 Wow, what a tragic story of your old system! I've had several PSUs fail, and one MOBO a month after warranty ran out, but never fried multiple CPUs. I see you have Pipewire on the new system. I have a multiboot install of Debian Trixie using Pipewire and so far it's proving to be quite good. Well, since PW 1.2.3 came out at least; 1.2.1 had a serious issue which caused export to fail in Ardour and Mixbus. I even worked out how to set up the PW native parametric EQ module which makes the room correction I need fairly simple compared to setup of EQ in JACK. Still using Bookworm as my main system though. How's Pipewire for you? Been trying to teach myself how to program MIDI synths and samplers recently. Have a lot to learn still. Good luck with the new one! Quote
securitybreach Posted September 23 Posted September 23 23 minutes ago, Hedon James said: i just pulled my drives and plugged into my backup machine. it worked out. i suspect that was WAY quicker than restoring a Clonezilla backup. At least, in this instance. Yeah, as it depends on drive speed and such. I only clone my installation drive as the other 8 drives are for storage. Cloning an ssd takes about 20 minutes or so and same to restore. You just have to make sure the drive you clone to has at least as much storage as you cloned from. It's all ncurses driven. I would clone the partition as image as its faster to do so. It only clones the data in use so if you only used 600gb on a terabyte drive, then it would make a 600gb clone. Quote
securitybreach Posted September 23 Posted September 23 I have been making a clone of my arch install every 3 months for like 12 years now. I simply overwrite the clone with the new one. Granted, I have done a restore like once or twice but I have setup my last 3 laptops with the same image. I don't have to setup anything, just clone old machine and restore to new machine. Quote
Hedon James Posted September 23 Author Posted September 23 23 minutes ago, securitybreach said: I have been making a clone of my arch install every 3 months for like 12 years now. I simply overwrite the clone with the new one. Granted, I have done a restore like once or twice but I have setup my last 3 laptops with the same image. I don't have to setup anything, just clone old machine and restore to new machine. If i was on a rolling release distro, this would be an excellent solution...maybe even my PREFERRED solution. But as a user of fixed-release distros (Debian based), I have my install media for that; and I backup my home directory and data directory (separate drive) to a backup directory. quite trivial to throw the installation usb in a machine (for an OS restore) and copy the home directory with configs/preferences to the new install. The data drive is a non-issue....create new bookmarks in the file manager. but if i was a rolling release user, your solution seems ideal for installation to new machines/drives. the only scenario still to resolve would be recovery issues....but BTRFS seems like a great idea in that case, rather than EXT4. I'm guessing that is what you do, so you can rollback to a prior? If not, what's your solution for that? Clonezilla backup doing double duty for you...disaster recovery and new installation? Quote
securitybreach Posted September 23 Posted September 23 Just now, Hedon James said: If i was on a rolling release distro, this would be an excellent solution...maybe even my PREFERRED solution. Well to be fair, the last time that I had to reinstall from something that I couldn't fix was back in 2012 with the move to systemd. Quote
Hedon James Posted September 23 Author Posted September 23 27 minutes ago, securitybreach said: Well to be fair, the last time that I had to reinstall from something that I couldn't fix was back in 2012 with the move to systemd. i hear ya...but are you running on a BTRFS for restore capabilities, or standard EXT4 format? or something else? seems like Clonezilla and BTRFS would be a sweet spot for rolling release users? Quote
Hedon James Posted September 23 Author Posted September 23 1 hour ago, sunrat said: Wow, what a tragic story of your old system! I've had several PSUs fail, and one MOBO a month after warranty ran out, but never fried multiple CPUs. I see you have Pipewire on the new system. I have a multiboot install of Debian Trixie using Pipewire and so far it's proving to be quite good. Well, since PW 1.2.3 came out at least; 1.2.1 had a serious issue which caused export to fail in Ardour and Mixbus. I even worked out how to set up the PW native parametric EQ module which makes the room correction I need fairly simple compared to setup of EQ in JACK. Still using Bookworm as my main system though. How's Pipewire for you? Been trying to teach myself how to program MIDI synths and samplers recently. Have a lot to learn still. Good luck with the new one! Pipewire threw me for a loop, as I didn't know it was there....and I still see tools and controls for "Pulse Audio". And I see options in Audacious to use ALSA, PulsaAudio, and JACK backends. It's all confusing to me. And it turns out that the Ubuntu Studio installer no longer provides "Cadence" to manage my Jack configs....that was my MAIN tool. I don't get along so well with Catia, Carla, and the other one (?)....but Cadence I know. It's gone now, unless I install from KX repo, but I don't want to do that for fear of breaking something with my Pipewire/Pulse/Also/Jack configuration. Ubuntu Studio installer provides an "audio configuration tool" that is even simpler than Cadence, so I left Cadence alone....didn't install. Used the Ubuntu Studio audio Config tool to configure audio settings....so far so good. Some of my apps default to the built-in soundcard & speaker, but I just open the PAvucontrol app and select my soundcard instead (UMC Behringer 404) and everything is good. Not sure if that will "stick" with a reboot, or I'll have to re-select again every reboot. I don't reboot very often, so might take awhile to learn that. In the meantime, I'm reluctant to tinker under the hood with sound settings until I understand the sound configuration better, vis-a-vis Pipewire vs Pulse vs Alsa vs Jack. I thought Pipeware was supposed to simplify things? Seems like they ADDED a layer, rather than simplified the layers. I also noted on the Ubuntu Studio page that Ubuntu will no longer require a "low latency" kernel, as things were merged into the main kernel and the next version 24.10 would use the regular kernel for audio. I wonder if they're referencing RT modules being merged into the kernel, that I have recently seen mentioned in articles? Your thoughts? My final observation regarding the new system...I was using your hack to change cpu performance on my old system (the Ryzen 3 configuration). CLI commands indicated the 4 cores were running "ondemand", and I switched them manually to "performance" at each boot. I note the NEW system indicates 6 cores are running in "powersave" mode. I wonder if I should be running them "ondemand"....and if so, just the 1st core, or ALL cores? (it's 2 different commands) Not sure of the pros/cons of each? And if "ondemand" is an improvement, would "performance" be even better? Or would that just result in a small-form-factor machine running too hot in that tiny box with minimal/no ventilation for cooling? Any thoughts on that? Quote
securitybreach Posted September 23 Posted September 23 21 minutes ago, Hedon James said: i hear ya...but are you running on a BTRFS for restore capabilities, or standard EXT4 format? or something else? seems like Clonezilla and BTRFS would be a sweet spot for rolling release users? I am old school, ext4 on all machines. Was using Reiser before he killed his wife. Quote
sunrat Posted September 23 Posted September 23 Pipewire basically does everything Pulseaudio and JACK do, as long as pipewire-pulse and pipewire-jack are installed. There is a lot to learn, but then there was with JACK too. And most of what you actually need to know is probably covered by that config utility ie. buffers ("quantum" in PW) and sample rate. qpwgraph is very useful for making connections too. powersave is the most conservative governor. For audio performance is best but ondemand should be sufficient for most cases if you're worried about heat. 1 Quote
securitybreach Posted September 24 Posted September 24 Yeah, I like pipewire and works just like pulse. You can even use the same apps to control it, like pavucontrol and such. Besides the package installation, this has some great info in setting it up: https://wiki.archlinux.org/title/PipeWire Quote
Hedon James Posted September 24 Author Posted September 24 13 hours ago, sunrat said: Pipewire basically does everything Pulseaudio and JACK do, as long as pipewire-pulse and pipewire-jack are installed. There is a lot to learn, but then there was with JACK too. And most of what you actually need to know is probably covered by that config utility ie. buffers ("quantum" in PW) and sample rate. qpwgraph is very useful for making connections too. powersave is the most conservative governor. For audio performance is best but ondemand should be sufficient for most cases if you're worried about heat. That's what I was thinking, but I don't know what I don't know! And considering what I just went through, I'd really like to avoid overheating the new cpu. Sounds like you would recommend the CLI incantation to change the governor to "ondemand". Should I just change the 1st core, or all cores? Or maybe try the 1st core only, and see if that resolves all xruns before switching all cores....settling on the "least amount" of governance required? What is your system set at? Quote
Hedon James Posted September 24 Author Posted September 24 1 hour ago, securitybreach said: Yeah, I like pipewire and works just like pulse. You can even use the same apps to control it, like pavucontrol and such. Besides the package installation, this has some great info in setting it up: https://wiki.archlinux.org/title/PipeWire Thanks SB! gonna take some time to digest this...a lot is above my paygrade, so will probably take awhile, but it's a good start. THANKS! 1 Quote
securitybreach Posted September 24 Posted September 24 14 minutes ago, Hedon James said: Thanks SB! gonna take some time to digest this...a lot is above my paygrade, so will probably take awhile, but it's a good start. THANKS! Well you know where to ask for assistance Quote
Hedon James Posted October 5 Author Posted October 5 On 9/23/2024 at 7:07 PM, sunrat said: Pipewire basically does everything Pulseaudio and JACK do, as long as pipewire-pulse and pipewire-jack are installed. There is a lot to learn, but then there was with JACK too. And most of what you actually need to know is probably covered by that config utility ie. buffers ("quantum" in PW) and sample rate. qpwgraph is very useful for making connections too. powersave is the most conservative governor. For audio performance is best but ondemand should be sufficient for most cases if you're worried about heat. Seems like Pipewire only allows ONE connection to my USB soundcard interface at a time. If I'm listening in Ardour, and then export to listen in music player (Audacious), Audacious will play through built-in speaker on the desktop....I CANNOT select the external usb device. But if I close Ardour (to release the connection?) and then play in Audacious, it uses the external usb device as expected. This seems like a ONE DEVICE AT A TIME configuration. Not sure if that's a characteristic of Pipewire, of the Ubuntu configuration of Pipewire, or something I have done? Does your system behave the same way? Annoying, but certainly not a deal killer. I also note that I cannot change my cpu governor to "ondemand". Can't remember the exact message, but it appears that something in the Intel CPU driver blocks the "ondemand" governor....it's "not available." So I changed the FIRST core only to "performance". I haven't noticed any difference in xruns...seems same as before....but it's not overheating either. I'm reluctant to change ALL cores to "performance" for fear of overheating. Are you familiar with the "ondemand" governor being blocked? I can run the switch command again for the exact message....but I recall something about the CPU driver and GRUB flag blocking it. Have you run into this? Other than these 2 MINOR issues, I've been getting along pretty well with the new hardware and the new OS, and the new version of Ardour (8.4?), and new plugins. A bunch of small niggly issues have been resolved....plugin windows open on top of the Ardour interface (rather than behind it), and they're not constantly resizing themselves with each new opening. The DrumGizmo vst3 interface "constantly grows" while it's open, eventually growing beyond the screen boundaries and being too big to use....but switching to the LV2 version of the same plugin doesn't have that issue. Performance doesn't seem to be affected, so I made the LV2 version a "favorite" and that problem is resolved to my satisfaction also. With all these minor issues resolved and/or fading into the background, I've been able to better focus on music production, mixing & mastering, and LISTENING to the results; rather than the distractions resultant from battling all the little stuff. It's not like I've waved a magic wand and made myself an audio engineer overnight, but I think it's helped a LOT. Back to improving my mixing & mastering skills. Here's a track completely cut with the new system: https://www.reverbnation.com/hedonmachine/song/34684890-frankensteins-monster If you have constructive criticism to offer @sunrat, hit me up in our private conversation and share your thoughts. I may never get to your level, but I'm okay with that...I'm just trying to be better tomorrow than I am today! And if anything you have to say helps, I'm gonna listen! LOL! 1 Quote
sunrat Posted October 5 Posted October 5 Different kernels are compiled with different governors. You can check available ones, this is for my current Liquorix kernel: $ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors ondemand userspace performance IIRC the default Debian kernel doesn't have ondemand but has schedutil instead which I've heard is OK. I'm not using Pipewire daily yet, I'll reboot to my test Trixie and check if I can play multiple sources. Had a quick listen to your song. Sounds good, better separation and distinction of the parts. It's late so I'll have to PM you tomorrow. 15 minutes later - I checked playing Ardour and DeadBeef together, don't have Audacious installed. If Ardour is set to ALSA out I can't play DeadBeef. However using Ardour JACK/Pipewire output and either JACK output or Pipewire output of DeadBeef, both will play at the same time. BTW, Ardour 8.8.0 came out yesterday. I upgraded already. Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.