Go Back   Cockos Incorporated Forums > REAPER Forums > REAPER for Linux

Reply
 
Thread Tools Display Modes
Old 07-25-2020, 01:34 PM   #1
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default Any 3700X/B550 folks running Linux here?

I just popped for an Asus B550 Prime-CSM mobo, an AMD 3700X, and 16GB Corsair PC3200 DDR4 to finally build a new DAW. I know some folks here are running 3700X/B450 and 3700X/X570 systems, but I've been holding out for months on the B550 chipset, and wondering if I am going to be the first guinea pig to try this combination out with Linux and REAPER.

I specifically held out for the new fanless heatsink only B550 chipset to get a PCIe 4.0 NVMe M.2 slot and PCIe 4.0 video slot. I doubt I'll buy a 4.0 video card, but I will definitely be buying an NVMe M.2 PCIe 4.0 drive in the future. Initially, I'll be using my existing Samsung EVO SSDs.

Also, if there are any AMD 3700X specific things I should be aware of before I start my build next Thursday when the parts all arrive, any information will be appreciated. Oh, I also bought this shiny case with three slow RPM 120mm LED fans.
__________________
Glennbo
Hear My Music - Click Me!!!
--

Last edited by Glennbo; 08-02-2020 at 02:51 PM.
Glennbo is offline   Reply With Quote
Old 08-02-2020, 03:23 PM   #2
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

WOOT!!! I am up and running on the new 3700X.

Key bits of information. Tried doing this build Friday, but the 400 watt power supply from my old DAW was NOT enough to get anything but a split second flash of trying to turn on. A new 900 watt power supply fixed that problem, so I moved all my drives and video card over, and to my surprise only Kontakt went into demo mode. I ran Native Access, signed in and it instantly went back to being authorized.

I didn't have to install one single piece of software to get back to exactly where I was before building the new DAW.

Here is a HUGE difference between the Intel I just retired vs the AMD I just put into service. Every one of the 6 entries out of 9 total that say "Not affected" are vulnerabilities that on Intel processors are affected. Only three out of the nine patches for CPU flaws are needed on this new AMD.

vulnerabilities/itlb_multihit:Not affected
vulnerabilities/l1tf:Not affected
vulnerabilities/mds:Not affected
vulnerabilities/meltdown:Not affected
vulnerabilities/spec_store_bypass:Mitigation: Speculative Store Bypass disabled via prctl and seccomp
vulnerabilities/spectre_v1:Mitigation: usercopy/swapgs barriers and __user pointer sanitization
vulnerabilities/spectre_v2:Mitigation: Full AMD retpoline, IBPB: conditional, STIBP: conditional, RSB filling
vulnerabilities/srbds:Not affected
vulnerabilities/tsx_async_abort:Not affected

On the bench, almost ready to go!

Attached Images
File Type: jpg 3700X_S.jpg (51.7 KB, 1477 views)
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-02-2020, 04:09 PM   #3
JamesPeters
Human being with feelings
 
Join Date: Aug 2011
Location: Near a big lake
Posts: 3,943
Default

3700X club membership: achieved. It's a beast.
JamesPeters is offline   Reply With Quote
Old 08-02-2020, 05:10 PM   #4
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by JamesPeters View Post
3700X club membership: achieved. It's a beast.
I'm digging it, and the fact that I didn't have to spend the day installing everything from scratch and having to deal with Windows plugins and their authorizations. I expected that all my Windows plugins with copy protection would want to be reactivated, but only Kontakt came up demo mode, and a simple connect with Native Access fixed that.

Speaking of Windows . . . I ended up nuking that POS. Windows 7 (and 8 for that matter) don't see Ryzen 3700X or B550 chipset well enough to even make USB 2.0 ports work. I could do some hacks, but I just formatted that 250GB Samsung EVO to ext4 and called it a day. No more dual boot. I'm pure Xubuntu Linux now.
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-02-2020, 10:27 PM   #5
JamesPeters
Human being with feelings
 
Join Date: Aug 2011
Location: Near a big lake
Posts: 3,943
Default

Well I had decided on the X470 chipset when I set up this PC because I didn't want to wait for Linux support in the kernel to catch up to a newer chipset like yours, especially when I figured I might end up using "older" (1 year-ish old) kernels depending on the distro. But I figured Windows would be up to speed on that. Boy what a disappointment that must've been.

Anyway I don't miss Windows. Linux isn't perfect but whenever I have a gripe about it, or a friend of mine does, the running joke we'll say to each other is "oh well, back to Windows!" (as if). A bit of extra learning is required, but then Linux ends up feeling more "right" than Windows ever did.

BTW if you want the CPU fan to stay really slow: disable the "turbo" in the BIOS. It won't hit that insane top speed of 4.4 GHz, but honestly I don't need that and I doubt you do either. (It means that 3.6 GHz will be the top speed.) The CPU fan is quieter that way, and I value that more than hitting "turbo" speed. I got a Noctua fan for the CPU to help as well; it's very quiet if the CPU doesn't throttle to "turbo" speed.

Last edited by JamesPeters; 08-02-2020 at 10:33 PM.
JamesPeters is offline   Reply With Quote
Old 08-03-2020, 01:23 AM   #6
cern.th.skei
Human being with feelings
 
cern.th.skei's Avatar
 
Join Date: Sep 2007
Location: trondheim, norway
Posts: 363
Default

Quote:
Originally Posted by JamesPeters View Post
BTW if you want the CPU fan to stay really slow: disable the "turbo" in the BIOS. It won't hit that insane top speed of 4.4 GHz, but honestly I don't need that and I doubt you do either. (It means that 3.6 GHz will be the top speed.) The CPU fan is quieter that way, and I value that more than hitting "turbo" speed.
yeah.. i just recently built a 3600/b450 box (and using the default cpu fan).. this was the first thing i did.. now, i can hardly force the temperature over 40 degrees, even at pretty high cpu loads.. :-)
__________________
torhelgeskei.com
cern.th.skei is offline   Reply With Quote
Old 08-03-2020, 07:22 AM   #7
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by JamesPeters View Post
Well I had decided on the X470 chipset when I set up this PC because I didn't want to wait for Linux support in the kernel to catch up to a newer chipset like yours, especially when I figured I might end up using "older" (1 year-ish old) kernels depending on the distro. But I figured Windows would be up to speed on that. Boy what a disappointment that must've been.

Anyway I don't miss Windows. Linux isn't perfect but whenever I have a gripe about it, or a friend of mine does, the running joke we'll say to each other is "oh well, back to Windows!" (as if). A bit of extra learning is required, but then Linux ends up feeling more "right" than Windows ever did.
I held out for the B550 chipset because I wanted the PCIe 4.0 NVMe M.2 slot. I will eventually buy one of those, and since I tend to keep computers for a long time, I wanted this newest fastest SSD capability.

I didn't expect Windows 7 to work with a 3700x Ryzen, but it did boot the first time, and my mouse and keyboard worked, but then Windows found new hardware and fixed it for me so the USB controllers no longer functioned along with 8 or 9 other new "unknown" devices. Asus only has chipset drivers for Windows 10, so it was an easy decision to just format that drive ext4. I never booted to Windows anyway, other than to see if certain things acted the same as REAPER for Linux.

Quote:
BTW if you want the CPU fan to stay really slow: disable the "turbo" in the BIOS. It won't hit that insane top speed of 4.4 GHz, but honestly I don't need that and I doubt you do either. (It means that 3.6 GHz will be the top speed.) The CPU fan is quieter that way, and I value that more than hitting "turbo" speed. I got a Noctua fan for the CPU to help as well; it's very quiet if the CPU doesn't throttle to "turbo" speed.
I had massive fan noise but in my case it was a matter of the computer case having three fans, but the mobo only having two fan headers, so the side fan was plugged into a molex connector and running full speed all the time. After some research I found that the "CPU Opt" header on my mobo is for another fan (or water cooling), so I plugged the side fan into it, and things instantly got a LOT quieter.

I still had the constant speeding up and slowing down of the CPU fan though (even at idle), which was going to drive me crazy, so I looked for "Turbo" in my bios. It turns out on my Asus B550 Prime mobo they call it "Core Performance Boost", and after disabling that, the CPU fan is almost silent. Thanks for that tip!


This is what I turned off on my mobo.

__________________
Glennbo
Hear My Music - Click Me!!!
--

Last edited by Glennbo; 08-20-2020 at 07:15 AM.
Glennbo is offline   Reply With Quote
Old 08-03-2020, 12:18 PM   #8
JamesPeters
Human being with feelings
 
Join Date: Aug 2011
Location: Near a big lake
Posts: 3,943
Default

I've disabled all internal LEDs. My system looks very plain. I prefer to not really notice the computer, for sound or visuals. My PSU still makes some sound and maybe I'll replace that with a quieter one, but I didn't want to spend several times what a normal PSU costs to get one that really is as silent as possible.

Yes that "core performance boost" is what it's called in my BIOS too. The mainboard's documentation mentions that as well as a "turbo" boost/frequency setting, but I think the nomenclature depends on the CPU (brand) that's installed.

When I first powered it up, I was surprised and disappointed with the CPU fan noise. Once I realized it was always trying to throttle to "absolute maximum insanity" whenever you did anything appreciable for CPU load, I laughed. I think AMD went a bit overboard with that. For instance when I booted the PC, I heard the fan go "Vrroooooommmm!!!" Um...no, lol. I'll wait the "extra" HALF SECOND for it to boot, and not have my CPU fan sound like a jet engine, thanks. After that setting was done, the fan noise was acceptable. But I'm sensitive to background hum etc. especially lately since I have some odd hearing condition, so I got the Noctua cooler and it's silent all the time unless I'm rendering video (and even so, it's fairly quiet).

I have a "stress test" project which has multiple instances of ReaPitch (with default settings) across a number of tracks with audio items. I use it to test the system, as a basic guideline, for when I change settings (especially audio latency). On my previous PC (Core i3 6300) I can run around 50 instances of ReaPitch at low latency (around 4ms RTL) while also doing operations such as moving items, renaming them, changing track colors, and so on, with no xruns. On this PC with the 3700X, I can run 160 instances of ReaPitch at the same latency. That's with the core performance boost off. Yeah I think this PC will last me a while.
JamesPeters is offline   Reply With Quote
Old 08-03-2020, 01:58 PM   #9
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by JamesPeters View Post
I've disabled all internal LEDs. My system looks very plain. I prefer to not really notice the computer, for sound or visuals. My PSU still makes some sound and maybe I'll replace that with a quieter one, but I didn't want to spend several times what a normal PSU costs to get one that really is as silent as possible.
I don't even see all the lights because the machine is tucked under my mix desk, and pushed all the way to the back. I wanted a case with a front accessible 5.25" bay because I write to DVDs all the time for backups. I also wanted multiple 120mm slow turning fans, and by the time I had ticked all the boxes for what I wanted, there were only a handful of cases, and the one I bought was the most economical. Noise I can hear bothers me, but lights I'm not seeing don't really bug me.

Quote:
Yes that "core performance boost" is what it's called in my BIOS too. The mainboard's documentation mentions that as well as a "turbo" boost/frequency setting, but I think the nomenclature depends on the CPU (brand) that's installed.

When I first powered it up, I was surprised and disappointed with the CPU fan noise. Once I realized it was always trying to throttle to "absolute maximum insanity" whenever you did anything appreciable for CPU load, I laughed. I think AMD went a bit overboard with that. For instance when I booted the PC, I heard the fan go "Vrroooooommmm!!!" Um...no, lol. I'll wait the "extra" HALF SECOND for it to boot, and not have my CPU fan sound like a jet engine, thanks. After that setting was done, the fan noise was acceptable. But I'm sensitive to background hum etc. especially lately since I have some odd hearing condition, so I got the Noctua cooler and it's silent all the time unless I'm rendering video (and even so, it's fairly quiet).
Noise from the computer is bad juju here, because I record acoustic guitar, mandolin, and other instruments with a mic, sitting right by my machine. On the previous Intel based machine, I only had to enable Q-Fan in the bios and set it to silent. Now that I have all chassis fans being controlled by the mobo, and have disabled core performance boost, this machine is as quiet as the one it replaced.

Quote:
I have a "stress test" project which has multiple instances of ReaPitch (with default settings) across a number of tracks with audio items. I use it to test the system, as a basic guideline, for when I change settings (especially audio latency). On my previous PC (Core i3 6300) I can run around 50 instances of ReaPitch at low latency (around 4ms RTL) while also doing operations such as moving items, renaming them, changing track colors, and so on, with no xruns. On this PC with the 3700X, I can run 160 instances of ReaPitch at the same latency. That's with the core performance boost off. Yeah I think this PC will last me a while.
I've never hit the limit and exhausted my old PCs processing power or RAM limits, but I can tell this machine is much more agile, and having 10GB more memory I'll never worry about loading huge drum kits when I already have a lot of other sampled instruments loaded. The BIG plus for me is there are 6 fewer security holes needing patched on this silicon.
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-03-2020, 04:31 PM   #10
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by cern.th.skei View Post
yeah.. i just recently built a 3600/b450 box (and using the default cpu fan).. this was the first thing i did.. now, i can hardly force the temperature over 40 degrees, even at pretty high cpu loads.. :-)
I'm seeing the same thing. My temps are sitting around 38-40 now where they had been in the 50-60 range, and after stress testing with a project that used to struggle plus a bunch of other things all running at the same time, none of my five fans ever kicked up their speed or made any audible noise.

My bios has an option for it to analyze the fans min-max needed speeds and then sets a profile for them. I ran it after disabling core performance boosting, and it reduced the RPM range on every fan. I could hear them all go quiet when I saved the new profile it had generated.
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-04-2020, 05:26 PM   #11
s wave
Human being with feelings
 
Join Date: Sep 2018
Location: Colorado
Posts: 429
Default

The 3700x is on my list! My old 8 core keeps cranking... thanks for all the info....
s wave is offline   Reply With Quote
Old 08-06-2020, 10:53 AM   #12
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Last night I installed the latest lowlatency kernel from my repo, and now the temperature sensors for the mobo and CPU are functional, so I added a panel gadget for monitoring.

Then I played with getting my PC3200 DDR4 to overclock and run at 3200Mhz, but it was a no go. I updated the bios and still no go, so I clicked it down to the next slower speed of 3133Mhz and it booted. Also, overclocking the RAM to run faster does not re-enable "Core Performance Boost" and my temps are still under 40C.

I might have been able to squeeze the 3200Mhz speed by increasing voltage to the RAM but I chose to drop one click in speed instead and not heat things up inside the case more. I've messed around in REAPER with a 36 track project for more than an hour now, with other stuff also running, and it appears to be completely stable. Highest temperature I've seen with REAPER and other stuff running has been 44C. My fans don't start ramping up until 50C.
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-06-2020, 02:20 PM   #13
JamesPeters
Human being with feelings
 
Join Date: Aug 2011
Location: Near a big lake
Posts: 3,943
Default

Are you having trouble getting your RAM to the speeds it should be running it, or are you actually overclocking it?

I have 3200 MHz RAM and had to choose an "XMP profile" to make it work at that frequency (and the voltage/timing table in the specs listed for it), so it sort of seemed like I was overclocking it but I actually wasn't. It just defaults to a slower speed.
JamesPeters is offline   Reply With Quote
Old 08-06-2020, 04:40 PM   #14
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

I used DOCP, which is the Asus version of XMP. Selecting my PC3200 from a pre-configured list informed me of two advanced parameters I might also need to tweak and suggested value ranges. I didn't want to increase the voltage any more, although I'm sure from reading other forums that that would have stabilized it and let me hit the full 3200Mhz. I decided instead to keep the voltage where it was and use the one click lower in clock frequency.

Here's a brief rundown of the different variations of XMP. Note that the OC in DOCP is for OverClock and Intel themselves also refer to XMP as overclocking RAM.

https://www.intel.com/content/www/us...ofile-xmp.html

Here's a brief rundown of the different variations of XMP.

Quote:
XMP is an Intel tech introduction to automatically set DRAM to intended data rates over and above a base level (for DDR3 it applies to DRAM 1600 and up), DDR4 for DRAM 2400 and up. You enable the feature, it reads the profile off the DRAM sticks and applies it in BIOS.

DOCP (Direct Over Clock Profile), is from ASUS for AMD motherboards and came as the motherboard makers didn't want to pay royalties to Intel to implement XMP on AMD motherboards. It effectively uses the DRAM XMP profile to set up data rates and comparative timings on the AMD motherboards for a variety of data rates.

EOCP (Extended Over Clock Profiles) is Gigabytes version of the ASUS DOCP..
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-06-2020, 05:36 PM   #15
JamesPeters
Human being with feelings
 
Join Date: Aug 2011
Location: Near a big lake
Posts: 3,943
Default

I obtained the specs for my RAM from the manufacturer's site, and compared that to the values in the XMP profiles. It turns out that "profile 1" was configured for that specifically, so I just needed to select that.
JamesPeters is offline   Reply With Quote
Old 08-06-2020, 06:01 PM   #16
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by JamesPeters View Post
I obtained the specs for my RAM from the manufacturer's site, and compared that to the values in the XMP profiles. It turns out that "profile 1" was configured for that specifically, so I just needed to select that.
That's pretty much the same as how DOCP works, but they leave a couple things up to the end user. Probably so they won't get sued for copying XMP too close. I could likely have gone to 1.36v and got it to boot at 3200Mhz, but since it booted fine and has run all day in and out of REAPER with 1.35v @ 3133Mhz I'm leaving it set like that . . . For now, anyway.
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-06-2020, 09:14 PM   #17
JamesPeters
Human being with feelings
 
Join Date: Aug 2011
Location: Near a big lake
Posts: 3,943
Default

I can adjust the settings too, but I don't want to stray from what its settings are supposed to be. I was just glad that "profile 1" in XMP was set up the right way automatically so I don't have to remember everything.
JamesPeters is offline   Reply With Quote
Old 08-06-2020, 10:56 PM   #18
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by JamesPeters View Post
I can adjust the settings too, but I don't want to stray from what its settings are supposed to be. I was just glad that "profile 1" in XMP was set up the right way automatically so I don't have to remember everything.
All I can say is that this machine is totally screaming! I have yet to find a project that won't play perfectly at 64 samples and three periods for 1.4/2.9ms performance, which is exactly how I track.
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-11-2020, 12:30 PM   #19
Jack Winter
Human being with feelings
 
Jack Winter's Avatar
 
Join Date: Aug 2007
Location: Luxembourg/Spain
Posts: 1,922
Default

How much it the real life latency (tested with reainsert)?
__________________
Reaper for Linux Documentation (WIP). Software: Archlinux/KDE, Fabfilter FX, Komplete 8, Nebula, Schwa/Stillwell, T-racks Max/Amplitube/SVX, etc. Gear: i7-2600k/4700HQ/16GB, RME Multiface/Babyface, Behringer X32, Genelec 8040, etc. :)
Jack Winter is offline   Reply With Quote
Old 08-11-2020, 01:37 PM   #20
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by Jack Winter View Post
How much it the real life latency (tested with reainsert)?
I've never tried to measure the latency, but ReaInsert wanted to use 3/4 for outputs, which go to my headphone distribution amp, so I used a 1/4" stereo male to L/R RCA adapter to take a headphone out from my headphone amp, and piped that back into a front input of my Behringer UMC1820. I dunno if going through an analog headphone amp will distort the numbers but here's what ReaInsert says. Pulling my desk away from the wall to put a cable direct from an output direct to an input is a bit of a hassle as it's a rats nest of cable back there. Hopefully going through an analog device won't skew the numbers.

With a 32 sample / 3 period buffer in REAPER ReaInsert says reported latency is 96.

With a 64 sample / 3 period buffer it comes back with reported latency 192
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-11-2020, 02:52 PM   #21
JamesPeters
Human being with feelings
 
Join Date: Aug 2011
Location: Near a big lake
Posts: 3,943
Default

Reported latency isn't what Jack is asking about so much as the additional delay compensation during a loopback test. Reported latency is the same as what you see in the top right corner of Reaper (just displayed as milliseconds).

Since you're using USB for the device, your actual latency is probably higher than the reported latency. Jack just filled me in about that here.
JamesPeters is offline   Reply With Quote
Old 08-11-2020, 05:47 PM   #22
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by JamesPeters View Post
Reported latency isn't what Jack is asking about so much as the additional delay compensation during a loopback test. Reported latency is the same as what you see in the top right corner of Reaper (just displayed as milliseconds).

Since you're using USB for the device, your actual latency is probably higher than the reported latency. Jack just filled me in about that here.
Okay, I pulled my desk away from the wall, and then cabled up the physical outputs 5/6 directly into the physical inputs 1/2.

After more attempts I finally got some real unwanted loud feedback in my face, but I think I have gotten the number, and the number appears to be 196 (once I had the input trims set) in the box labeled Additional Delay Compensation after a ping.

So does that mean I'm getting 2.8/5.8ms as opposed to the REAPER display of 1.4/2.9ms?

I did figure out that the 192 samples it shows is my 64 sample selection times three periods, so do you add the 196 or roughly double that to come up with the real latency?

Either way, every single thing that I record with a mic or line input, I'll still monitor it after it's been through REAPER and some FX, just as I've done on every project since switching to REAPER 12 years ago.
__________________
Glennbo
Hear My Music - Click Me!!!
--

Last edited by Glennbo; 08-20-2020 at 07:18 AM.
Glennbo is offline   Reply With Quote
Old 08-11-2020, 06:53 PM   #23
JamesPeters
Human being with feelings
 
Join Date: Aug 2011
Location: Near a big lake
Posts: 3,943
Default

You didn't disable monitoring on the interface, I guess. Ouch.

That number is close to what I noticed of my 2i2: roughly the same amount of offset required as the reported latency (at that low a buffer setting anyway), which means that latency you thought you had, well it's actually double. It's still within reason though.
JamesPeters is offline   Reply With Quote
Old 08-11-2020, 07:21 PM   #24
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by JamesPeters View Post
You didn't disable monitoring on the interface, I guess. Ouch.
Yeah, I wasn't sure if I needed direct or through monitoring, but disabled it soon after that event.

Quote:
That number is close to what I noticed of my 2i2: roughly the same amount of offset required as the reported latency (at that low a buffer setting anyway), which means that latency you thought you had, well it's actually double. It's still within reason though.
I gave the new Ryzen 3700X a serious workout loading almost every project I have dating back to ones done in Windows from 2008. I was able to play every project I opened with 64 samples and three periods. That was not the case with my ancient i5 machine that I just retired.

So if my true round trip time is about 8.6ms, that's like standing 10 feet from your guitar amp. 10 Feet / 1100 Speed of sound = 0.009 seconds
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-13-2020, 11:32 AM   #25
Jack Winter
Human being with feelings
 
Jack Winter's Avatar
 
Join Date: Aug 2007
Location: Luxembourg/Spain
Posts: 1,922
Default

Yes, your real life latency is about double than that shown by reaper. Don't know how it breaks down as input vs output, but I'd guess that the extra latency is symmetrical. This is probably not a big thing.

What is annoying is that since reaper doesn't know about it, it can't properly compensate for the latency when overdubbing. There are preferences for adding this manually. But the real annoying thing is that this changes each time you run reaper while using an usb device on linux.. And if you increase the buffer size to something bigger say 2048 or 4096 then the discrepancy gets even bigger and might possibly become significant. So if you want to have your overdubbed audio properly aligned, you'd probably want to run reainsert and then modify the preferences every time you overdub..

I remember seeing a patch for the USB driver that made the extra latency a lot less, it was even runtime configurable. This is somewhat of a pain since you have to either patch the kernel and build it, or build the driver outside of the kernel building process. IIRC there was even an archlinux package for building it.
__________________
Reaper for Linux Documentation (WIP). Software: Archlinux/KDE, Fabfilter FX, Komplete 8, Nebula, Schwa/Stillwell, T-racks Max/Amplitube/SVX, etc. Gear: i7-2600k/4700HQ/16GB, RME Multiface/Babyface, Behringer X32, Genelec 8040, etc. :)
Jack Winter is offline   Reply With Quote
Old 08-13-2020, 06:31 PM   #26
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by Jack Winter View Post
Yes, your real life latency is about double than that shown by reaper. Don't know how it breaks down as input vs output, but I'd guess that the extra latency is symmetrical. This is probably not a big thing.

What is annoying is that since reaper doesn't know about it, it can't properly compensate for the latency when overdubbing. There are preferences for adding this manually. But the real annoying thing is that this changes each time you run reaper while using an usb device on linux.. And if you increase the buffer size to something bigger say 2048 or 4096 then the discrepancy gets even bigger and might possibly become significant. So if you want to have your overdubbed audio properly aligned, you'd probably want to run reainsert and then modify the preferences every time you overdub..
I run 64 samples and 3 periods now and never change it, so if I'm hitting about 8.6ms I can live with that real latency.

The tracks being offset by different amounts every time, could that be demonstrated by having a midi track with a clicking sticks drum VSTi that feeds another track that's recording the audio output. After laying one track down, send the midi track audio to another track and record a second time.

Then you could look at both tracks that recorded the same midi stick clicking, but in two separate recordings. Would that show the issue you are referring to?
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-14-2020, 03:30 AM   #27
Jack Winter
Human being with feelings
 
Jack Winter's Avatar
 
Join Date: Aug 2007
Location: Luxembourg/Spain
Posts: 1,922
Default

You could generate a click track and then record that through a loopback cable. Since the click is so abrupt, it makes it easy to see how much it's off.

My multiface used to have 95 samples hidden latency, but it was always the same so I just adjusted the reaper settings to add 48/47 samples and then forgot about it. Class 2.0 USB audio is much messier
__________________
Reaper for Linux Documentation (WIP). Software: Archlinux/KDE, Fabfilter FX, Komplete 8, Nebula, Schwa/Stillwell, T-racks Max/Amplitube/SVX, etc. Gear: i7-2600k/4700HQ/16GB, RME Multiface/Babyface, Behringer X32, Genelec 8040, etc. :)
Jack Winter is offline   Reply With Quote
Old 08-14-2020, 07:05 AM   #28
Soli Deo Gloria
Human being with feelings
 
Soli Deo Gloria's Avatar
 
Join Date: Oct 2013
Location: Argentina
Posts: 1,303
Default

One question : does this unreported latency generate easily perceived problems when recording and overdubbing, in any case? I've playing live with virtual instruments and recording voices and percussion on top of them, sometimes with FXs during the recording process, and haven't felt anything strange nor seen any particular offset that could be an obstacle for the sessions (I use a Native Instruments Komplete Audio 6, the old model). It's good to know if there is any potential pitfall beyond the specific measurements you're talking about, though...
Soli Deo Gloria is offline   Reply With Quote
Old 08-15-2020, 02:04 AM   #29
lucas_LCS
Human being with feelings
 
Join Date: Dec 2015
Posts: 2,100
Default

@ Glennbo
What made you choose the Asus B550 Prime-CSM over other motherboards?

::
__________________
LCS Themes
lucas_LCS is online now   Reply With Quote
Old 08-15-2020, 03:55 AM   #30
biopsin
Human being with feelings
 
Join Date: Sep 2010
Location: oslo
Posts: 142
Default

Quote:
Originally Posted by lucas_LCS View Post
@ Glennbo
What made you choose the Asus B550 Prime-CSM over other motherboards?

::
did a upgrade 6-months to a ASUS B450-ITX and MB layout/features & linux compatibility are main focus. Also longtime experience with Asus generally, no components has yet failed me, they have just become obsolete after a decade of usage.
__________________
Voidlinux_glibc / gcc_12.2. / libSwell_GDK2 - 250423 /
Reaper_6.82 / NI_KA2 / Dynaudio_BM6
biopsin is offline   Reply With Quote
Old 08-15-2020, 05:59 AM   #31
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by lucas_LCS View Post
@ Glennbo
What made you choose the Asus B550 Prime-CSM over other motherboards?

::
Two things. The CSM is an acronym for Corporate Stable Model, so unlike the gamer models, it's not trying to push anything to the limit, going instead for more stability. The other was price.

Also, I wanted the B550 chipset to get PCIe 4.0 for a future NVMe M.2 drive.
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-15-2020, 06:13 AM   #32
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by biopsin View Post
did a upgrade 6-months to a ASUS B450-ITX and MB layout/features & linux compatibility are main focus. Also longtime experience with Asus generally, no components has yet failed me, they have just become obsolete after a decade of usage.
I have three machines currently in operation that I built using Asus motherboards, and about 8 to 10 more Asus motherboards scattered around the house from previous builds.

When I was working in IT I built all the machines for our offices, all based on Asus mobos. Prolly built about 30 machines for them in the 26 years I was there.
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-24-2020, 07:38 AM   #33
Jack Winter
Human being with feelings
 
Jack Winter's Avatar
 
Join Date: Aug 2007
Location: Luxembourg/Spain
Posts: 1,922
Default

Quote:
Originally Posted by Soli Deo Gloria View Post
One question : does this unreported latency generate easily perceived problems when recording and overdubbing, in any case? I've playing live with virtual instruments and recording voices and percussion on top of them, sometimes with FXs during the recording process, and haven't felt anything strange nor seen any particular offset that could be an obstacle for the sessions (I use a Native Instruments Komplete Audio 6, the old model). It's good to know if there is any potential pitfall beyond the specific measurements you're talking about, though...
Well if you don't notice it, then it's not really a problem!

I did a small test with my babyface, and the result was a lot more consistent than I remember. I circled through buffer sizes from 128 to 2048 (with the alsa driver and 2 periods) five times, each time measuring the actual loopback latency (in samples) using reainsert. Strangely enough sometimes I had to run the reainsert test twice, as the first result was wrong.

These are the results I came up with:

128 / 280, 293, 415, 429, 293
256 / 619, 617, 617, 618, 302
512 / 122, 123, 124, 122, 123
1024 / -361, -361, -362, -362, -361
2048 / -1395, -1395, -1395, -1395, -1395

I suppose one could take an average of the measured latency discrepancy, divide it by 2, and then add that as input/output manual offset in reaper's audio->recording settings. That would at least get closer to the truth and allow reaper to better compensate for the latency when overdubbing.

But up to each of us to know if we find it significant and want to bother with making such tests. In any case, it looks like the hidden latency has become more consistent, I guess an update in either the kernel or alsa. If I remember when I'm close to our X32 and have the laptop with me, I'll make some tests with that one too. Don't know if it will give the same results, but would be interesting to know.
__________________
Reaper for Linux Documentation (WIP). Software: Archlinux/KDE, Fabfilter FX, Komplete 8, Nebula, Schwa/Stillwell, T-racks Max/Amplitube/SVX, etc. Gear: i7-2600k/4700HQ/16GB, RME Multiface/Babyface, Behringer X32, Genelec 8040, etc. :)
Jack Winter is offline   Reply With Quote
Old 08-24-2020, 10:07 AM   #34
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Quote:
Originally Posted by Jack Winter View Post
But up to each of us to know if we find it significant and want to bother with making such tests. In any case, it looks like the hidden latency has become more consistent, I guess an update in either the kernel or alsa. If I remember when I'm close to our X32 and have the laptop with me, I'll make some tests with that one too. Don't know if it will give the same results, but would be interesting to know.
So with a vocal mic in the room, I punched REAPER into record with the metronome cranked up so the vocal mic would pick it up.

Then I added a second track connected to the vocal mic, and again recorded the metronome playing through my speakers with the mic in the room and in the same spot.

Zoomed in a far as REAPER can possibly zoom, the nodes are seemingly identical, even though the two tracks were recorded at different times.
__________________
Glennbo
Hear My Music - Click Me!!!
--

Last edited by Glennbo; 09-21-2020 at 07:00 PM.
Glennbo is offline   Reply With Quote
Old 08-26-2020, 07:14 AM   #35
Jack Winter
Human being with feelings
 
Jack Winter's Avatar
 
Join Date: Aug 2007
Location: Luxembourg/Spain
Posts: 1,922
Default

I'm not sure I understand. Didn't you do the exact same thing twice? If so I'd expect to get the same result.

How do they relate to beat positions? Do they align to the click or are they offset?

Edit: I suppose they wouldn't align with the click as there is all that latency from the sound traveling through air...
__________________
Reaper for Linux Documentation (WIP). Software: Archlinux/KDE, Fabfilter FX, Komplete 8, Nebula, Schwa/Stillwell, T-racks Max/Amplitube/SVX, etc. Gear: i7-2600k/4700HQ/16GB, RME Multiface/Babyface, Behringer X32, Genelec 8040, etc. :)
Jack Winter is offline   Reply With Quote
Old 08-26-2020, 11:52 AM   #36
JamesPeters
Human being with feelings
 
Join Date: Aug 2011
Location: Near a big lake
Posts: 3,943
Default

What he did for a test is effectively (for a different purpose) the same as the test described here:

https://www.youtube.com/watch?v=HZcXZ9kEJbY

A "loopback" of sorts. Glen's test will involve some latency (physical distance to/from the mic etc.) but that's not the point in this case. The point was: will two consecutive recordings of a source--given identical start times (that Reaper determines)--record with the same offsets. In his case, it did (at least for those two recordings). It's not about whether he needs to add additional latency compensation offsets (in samples), but whether those offsets are consistent. Consistency of those offsets is more important than whether the latency is being reported correctly, especially if the latency is only out a few ms anyway (as is the case with his device).

In my case, with my USB device, using the method Kenny described in that video and the ReaInsert method that Jon described in his own video, I got inconsistent results with my offsets, sometimes from one recording pass to the next (and sometimes when restarting the project). It wasn't enough to ruin recordings (a couple ms out at most), but it was noticeable enough when I zoomed in. With higher audio buffer sizes, those offset inconsistencies grew to several ms though, which is concerning. Inputting manual offset values made no difference either; the offsets fluctuate irrespective of that.

What Glen is saying is that (at least for those two tests) his audio system is producing the same offset consistently. If that is truly consistent (if more tests produce the exact same result), then he won't have a concern about his audio being out of alignment due to variances in offset, since he has no variances like I do with my USB device.

And now I have no variances in offset since I switched to my onboard audio device. Furthermore the reported latency is within 15 samples of the actual loopback results whether using that method or ReaInsert to test. You and I were talking about this on another thread here.

Last edited by JamesPeters; 08-26-2020 at 11:57 AM.
JamesPeters is offline   Reply With Quote
Old 08-26-2020, 07:03 PM   #37
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Just as James says, I was trying to see if there was a problem with tracks lining up when the source being recorded is perfect on the beat. From what Jack said, I was expecting there to be a randomly different offset every time I pressed record, and if that were the case, I would have expected two tracks recorded at different times to expose that varying offset, so I was surprised to see the nodes line up as perfectly as they do.

That said, had it been me clicking sticks to a metronome on the first track, and me clicking sticks to the playback of the first track, the other offset issue might show up, but it would be hard to tell for sure, because I would have to hear the first track, then react to it, and click the sticks for the second track with zero delay on my part. Otherwise how would you know what was delay introduced by REAPER vs. delay introduced by human reaction time.
__________________
Glennbo
Hear My Music - Click Me!!!
--
Glennbo is offline   Reply With Quote
Old 08-27-2020, 10:33 AM   #38
Jack Winter
Human being with feelings
 
Jack Winter's Avatar
 
Join Date: Aug 2007
Location: Luxembourg/Spain
Posts: 1,922
Default

IIRC, you have to stop and restart JACK or disconnect reaper from the usb device to get this effect, or even better check by replugging the USB device. It will stay the same for the entire recording session, but might be different the next time.

Like I said I think the situation has improved, or it works better with the rme babyface instead of with the X32. But I seem to remember that I saw worse results in the past with the babyface too, so I think it's gotten better.

But as my test shows it didn't change all that often and the difference in latency wasn't as big as I saw in the past.

Still I think that doing the test with a loopback cable is probably a better idea, as one then gets a precise result in samples. Also one can then reconfigure reaper to take the hidden latency into account to better align recorded tracks. Most people might never notice, and others like to be as precise as possible!
__________________
Reaper for Linux Documentation (WIP). Software: Archlinux/KDE, Fabfilter FX, Komplete 8, Nebula, Schwa/Stillwell, T-racks Max/Amplitube/SVX, etc. Gear: i7-2600k/4700HQ/16GB, RME Multiface/Babyface, Behringer X32, Genelec 8040, etc. :)
Jack Winter is offline   Reply With Quote
Old 08-27-2020, 03:47 PM   #39
JamesPeters
Human being with feelings
 
Join Date: Aug 2011
Location: Near a big lake
Posts: 3,943
Default

Quote:
Originally Posted by Jack Winter View Post
IIRC, you have to stop and restart JACK or disconnect reaper from the usb device to get this effect, or even better check by replugging the USB device. It will stay the same for the entire recording session, but might be different the next time.
Ah! I'm using ALSA but also allowing the audio engine to stop whenever Reaper isn't in focus (and playback is inactive). That might explain why I get inconsistent offsets when I do tests like this with the USB device.
JamesPeters is offline   Reply With Quote
Old 08-27-2020, 04:51 PM   #40
Glennbo
Human being with feelings
 
Glennbo's Avatar
 
Join Date: Mar 2008
Location: Planet Earth
Posts: 9,055
Default

Sure enough, if I record one track using the metronome as before, then turn off and back on my audio device, I get a discrepancy. Looks like about 6 nodes difference.
__________________
Glennbo
Hear My Music - Click Me!!!
--

Last edited by Glennbo; 09-21-2020 at 07:01 PM.
Glennbo is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 05:59 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.