Jump to content

Jim Roseberry

Members
  • Posts

    1,128
  • Joined

  • Last visited

  • Days Won

    9

Posts posted by Jim Roseberry

  1. On 2/16/2019 at 10:27 PM, panup said:

    DPC Latency Checker reported normal Windows 10 values: around 1100 us with no spikes.  It is much more than in XP or 7 but still OK.

     image.png.3d9a6eb09ae3b39dff61699df45a38f4.png

     

    And here's Task Manager / CPU Performance while Cakewalk is playing full mixed 80 track, heavily produced mix:

    image.png.6b7abe268ad82d5a4bbcbe89f83f6af4.png

    Disk usage (audio is on M drive) is 2% (!) and CPU usage 12%.  Obviously the mix needs more tracks and plugins. ;)

     

     

     

    Don't use DPC Latency Checker in Win10... use Latency Mon instead (accurate and lists more specific information).

    Regarding M.2 drives and SATA ports:

    Many Z370/Z390 motherboard have two M.2 slots.

    Typically, if you run the second M.2 slot using 4 PCIe lanes, the motherboard will disable the last pair of SATA ports.

    If you don't have two M.2 Ultra SSDs, put the M.2 SSD in the first M.2 slot... as most motherboard will run this slot with 4 PCIe lanes... and not disable the last pair of SATA ports.

    • Thanks 1
  2. 32 minutes ago, Noel Borthwick said:

    @Jim RoseberryCan the processor threads be clock speed locked with earlier intel processors than the I9?

    Hi Noel,

    Great to have you here!

     

    You can certainly disable CPU throttling.  That should be the case (already) with your machine (assuming the BIOS settings haven't changed).

    If throttling is happening, let me know... and I'll PM settings to eliminate it.

     

    Whether you can lock all cores at the highest TurboBoost frequency depends on the specific CPU.

    ie:  The latest i9-9980xe is pushing the limits of that design/architecture.  You may not be able to lock all 18 cores at 4.5GHz (and achieve 100% stability).

    Same with earlier designs

     

    Designs with higher core counts typically can't be clocked as high.

     

    You've got adequate cooling.

    In the BIOS, you can change the TurboBoost multiplier to 35 (for all cores).

    If there's any instability, go back into the BIOS and drop that number down.

    • Like 1
  3. 3 hours ago, Daniel Vrangsinn said:

    If 550 is enough, I will of course buy smaller and save some money

    If you're building a DAW with a i9-9900k, a 550w power-supply is definitely on the lean side.

    You want to leave yourself the opportunity to add a GTX video card (video editing), more internal drives, use of bus-powered MIDI controllers, etc.

    • Like 1
    • Not every process in a DAW can be multi-threaded.
    • Performance increase is not a 1:1 ratio when adding cores.  (Doubling the number of CPU cores doesn't double performance)

    Thus, clock-speed is king when choosing a CPU for a DAW.

    What you don't want to do is sacrifice significant clock-speed for more cores.

    Xeon CPUs often have significantly slower clock-speed... and can thus result in a significant performance hit.

     

    In a perfect scenario, you want highest clock-speed... and maximum number of CPU cores.

    This is why the i9-9900k is such a great choice for DAW purposes.

    • 8 cores, 16 processing threads that can all be locked at 5GHz
    • With the right air-cooler, it runs near dead-silent

     

    To best the i9-9900k, you're talking high-end socket-2066 i9 CPU... which is significantly more expensive.

    Higher-end socket-2066 i9 CPUs require water-cooling with large radiator.

    ie:  Noctua air-coolers (NH-U14s, NH-D15s) can't dissipate enough heat from an i9-9980xe when all cores are under heavy load.

    More expensive CPU, Motherboard, Cooler, Power-Supply (it adds up quickly)

     

     

     

    • Like 1
  4. 3 minutes ago, mettelus said:

    I think that is the crux of the OP... a 1TB drive should be used, especially if fast, otherwise is extreme overkill (and waste of money) to leave it mostly empty. In that situation a 200GB "C" drive partitioned to an 800GB "Sample Drive" would make a lot of sense. Imaging the C partition would be small and quick, and the rest of the drive could be used for "fast read" material.

    I completely agree with using a smaller boot drive.  😉

    Just mentioning the size of the drive itself (unused space) won't affect backup size/time

     

    Rather than using a single large drive partitioned Boot/Samples, the best coarse of action is to use two separate drives.

    With two separate drives, you've literally doubled the performance (vs. a single partitioned drive).

    • Each of two HDs would sustain 190MB/Sec
    • Each of two SATA SSDs would sustain 540MB/Sec
    • Each of two M.2 Ultra SSDs would sustain 3400MB/Sec
    • Like 2
  5. 57 minutes ago, synkrotron said:

    I still have some homework to do regarding the M." drive. I am struggling to find out if my chosen motherboard, ASUS ROG STRIX Z390-E will actually use a Samsung 970 EVO M/2 NVMe card to its fullest. It has two M.2 ports that support "PCIe 3.0 x4 mode" but I have no idea what that means...

    M.2 Ultra SSDs use 4 PCIe (x4) to achieve full performance.

    If the motherboard has two M.2 slots, running the second using 4 PCIe lanes almost always results in the motherboard's last two SATA ports being disabled.

    You can work around this...

    Put the second M.2 Ultra SSD on a PCIe host card... and place that card in a PCIe slot that has 4 or more PCIe lanes.   😉

    • Like 1
  6. 1 hour ago, mettelus said:

    A massive OS drive makes disk imaging problematic, both in time and space. That is one situation where I would consider using a partition. My OS drive is only 256GB, and I use junctions to keep it under 150. Images of a mostly full 1TB drive would take an hour or so, and need another massive drive to store the images to.

    You want to keep the OS drive clean/lean.

    If you're using a 1TB SSD for boot drive... and only using 200GB of that 1TB of space, backup won't be slow/tedious.

    Unused drive space won't increase the size of backups.

  7. To clarify about M.2:

    M.2 drives come in two varieties

    • SATA - speed is same as 2.5" SATA SSD
    • NVMe (M.2 Ultra) - speed is ~3400MB/Sec for the best drives

     

    Conventional HDs sustain ~190MB/Sec

    SATA SSD sustains ~540MB/Sec

    M.2 Ultra SSD sustains ~3400MB/Sec

     

    IMO, NVMe (M.2 Ultra) is overkill for a boot drive.

    A machine is going to boot fast... and apps open quickly using a standard SATA SSD.

    Use M.2 Ultra drives "strategically"... if/when necessary.

    If you have a particular sample library or libraries where you need massive polyphony, M.2 Ultra SSD is a great solution.

    If you have a particular library that loads slow, put that library on a M.2 Ultra SSD... and it'll load much faster.

    ie:  I find HALion 6 to be a bit sluggish when loading samples... so I put the library on a M.2 Ultra SSD (now loads much faster).

     

    You could run a single/large M.2 Ultra SSD... and partition it for OS/Audio/Samples... but that's not ideal.

    From a performance standpoint, you're still best having separate physical drives for OS/Audio/Samples.

    If you're making heavy use of Samples, I'd want those on multiple SSDs (scaled based on your disk-streaming polyphony needs).

    We have some clients (mostly doing huge orchestral mock-ups for scoring video games) who want to be able to achieve 4000 stereo voices of disk-streaming polyphony.

    That takes multiple SSDs (including M.2 Ultra)...

    • Thanks 1
  8. 5 hours ago, Soundwise said:

    You can do this in ReValver and Bias Amp. So, there's nothing revolutionary. But the ability to use Kemper profiles is the main reason why I purchased the TH-U upgrade.

    Tone-Matching is not the same thing as Kemper style "Profiling".  

    Tone-Matching uses FFT to match the frequency-response of the Amp/Cab/Mic "Rig".

    Tone-Matching (Impulse-Responses) can't capture/represent non-linear (distortion) components.

    None of the source amp's distortion characteristics are captured/reproduced.

     

    You can actually "Tone-Match" with any AmpSim that allows loading IRs.  

    The FFT process has to be done separately with your DAW.

    • Use an FFT to match the frequency content
    • Shoot an IR using that resultant FFT
    • Load this IR into your AmpSim

     

    Bias Amp and ReValver sound nothing close to the realism of a Kemper.

    • Thanks 2
  9. Cab IRs capture the frequency response of a mic'd guitar/bass cabinet.

    You can then play an Amp signal (real or amp-sim) thru the Cab IR... and it sounds very close to the real Amp/Mic'd Cab.

     

    If you've heard of the Kemper Profiling Amp, it sends a series of test signals thru a real mic'd Amp/Cab...

    The result of what the Kemper receives back allows it to very accurately reproduce the whole mic'd Amp/Cab "Rig".

    The Rig Player in TH-U has all the parameters that are present on the Kemper.

    The only thing missing (vs the Kemper); the ability to create profiles yourself.  I'm assuming Overloud is doing this for legal reasons.

    If you could create your own Profiles (and assuming they're as accurate as the Kemper), it's very much a game-changer.

    The ability to very accurately reproduce the sound of any mic'd amp/cab (with a $200 plugin) would be revolutionary.

     

    • Like 2
  10. Even similar guitars can have significantly different output levels.

    I've got two guitars with humbuckers in the bridge (one covered and the other open-coil)

    Using exact same preamp settings,  the guitar with open-coil is ~5dB hotter.

     

    BTW, This is the reason that all presets are terrible.

    If they weren't created with a very similar guitar, they're going to sound "wrong" because there's so much variance between instruments/pickups.

    • Like 2
  11. Even similar guitars can have significantly different output levels.

    I've got two guitars with humbuckers in the bridge (one covered and the other open-coil)

    Using exact same preamp settings,  the guitar with open-coil is ~5dB hotter.

  12. 1 hour ago, Tobias said:

    Oh is that right? I have mine on High ( I believe ) but my guitar is all single coil. Maybe that is why many presets are way too "hot", will have to try that. 

    Use the input-meter to adjust your gain going into TH-3 or TH-U.

    Peak/loudest playing should be in the green area of the meter.

    • Like 2
  13. Don't like the stock Cabs (don't like them in Helix, HeadRush, and most other AmpSims both hardware and software).

    Using my preferred Cab IRs, TH-U is sounding pretty good.

    Comparing it to Helix Native and HeadRush (hardware) side-by-side, it's holding up pretty well.

     

    Overloud has a set of "British" profiles available for purchase for the Rig Player.

    The included Profiles are (IMO) a little on the lower-end side.  Just enough to show what's possible... but leaves you wanting better Amp/Cab/Mic "Rig" profiles.

     

    CPU use is pretty low (even when running at 96k)

    On higher-gain setting, background noise is a little lower than some other AmpSims.  Working with guitar/bass pickups in close proximity to a computer (EMI), that's always a welcome thing.

     

    • Like 1
    • Thanks 2
  14. 50 minutes ago, SomeGuy said:

    1. Not possible on Windows Home. Can only defer (not disable) on Pro (feature updates for a year, and a month for security uodates).

    2. Only do this on a laptop if it never leaves the charger - in which case why not just get a desktop 😛

    3. See above. 

    4.  This can cause issues sometimes, but typically with software with a very old code base. Never had an issue with it (or with Cakewalk). I've never turned this off, personally.

    5.  This is the one thing to look over. But, with Windows Home, you are very limited in what you can accomplish. Security headline are 85% FUD, anyways. Irrelevant if the PC isn't internet connected. Registry entries are vestigial in Windows SKUs that don't support the feature. They exist, but do nothing. So what websites tell you to do are often ineffectual. Microsoft is smarter. 

    ---

    Things I recommend:

    - Disable Background Apps for all but staples (like Microsoft Edge, Groove Music, etc.) - this affects UWP apps only. 

    - Enable Long Paths

    @SomeGuy, I've been building DAWs professionally for 25 years.  😉

    I'm well aware of not being able to disable Windows Updates with the Home version of Win10.  That's why we recommend the Pro version.

    If you have a laptop running off battery-power... and you allow full Performance and Power throttling, it's DAW performance will be terrible.

    • CPU throttling causes high DPC Latency
    • CPU core parking (at the wrong time) can cause glitches/dropouts
    • Power-Management shutting down USB Root Hubs can cause attached MIDI controllers and Audio interfaces to lose connection.

    Regarding Fast Startup, apparently you've not seen the bizarre issues it can cause (corrupt and lost data - especially when dual-booting - even when the OS installs are completely isolated).  We've disable this setting on hundreds of machines... and it's never cause an issue. 

    Apps constantly running in the background (especially numerous apps)... and "phoning-home" information to Microsoft is not what you're looking for in a well-configured clean/lean DAW.

    • Thanks 1
  15. 49 minutes ago, John said:

    Hi Jim thanks for that. I am casually looking for a new graphics card. It is not a priority. 

    What was funny is I have Davinci  Resolve which checks ones graphics card and it said it OK but it wont be good for so many things that Resolve can do. LOL 

    In using it I found it makes my videos look really good where Vegas Video Pro is, shall we say, not as good looking. 

    So I have a Nvidia GTX 550 ti. hooked to two wide screen monitors.  1 gig mem ddr5. I want to up that to 8 gig. and a better card.  I don't want to spend a lot of money on it. LOL When I got it it was a middle of the road card. Each year it goes further down the line becoming lower and lower. I don't need 4k but.........😀

    Hi John,

    Either the GTX-1070 or a Vega-64 (AMD's offering that's about equal to a GTX-1080Ti) would both do well.

    • Neither will cause high DPC Latency
    • Both have a 0dB fan mode (noiseless when working with audio).

     

  16. If you're into AmpSims, Overload has now released TH-U (update to TH-3).

    https://www.overloud.com/products/th-u-full

     

    There's a lot of cool new options/capabilities. 

    One of the biggest being the Rig Player.  It's similar to a Kemper Profiling Amp.

    For the time being, you can't create/load your own profiles (have to get them thru Overload), but hopefully they'll open up that process.

    With a plethora of stock Amp models, the ability to load your own Cab IRs, the ability to load Rig profiles (Amp and Cab sections can also be used separately), and the ability to alter Tubes in the Pre/Power Amp sections of many models, TH-U is one deep AmpSim plugin.

    Check out the demo...

    I loaded it and had a go with some of the (few) included Rig profiles.  They sound/respond pretty good to my ears.

    The SLO and Dumble models were also sounding pretty good.

    • Like 1
  17. With Win10, there are numerous things you want to disable.

    • Automatic Updates
    • Performance throttling
    • Power Management
    • Fast Startup 
    • Privacy Settings

    The more you're pushing the machine (especially at the smallest ASIO buffer sizes), the more important the OS configuration.

    • Like 1
    • Thanks 1
  18. Hi Clint,

     

    We've seen high DPC Latency from the latest Nvidia RTX-2xxx series.  

    Nvidia will ultimately get that ironed out... but avoid those for now.

     

    No trouble at all with the GTX-1xxx series...

    I'd recommend a GTX-1060.  It'll work just fine for most gaming, AutoCAD, video editing/rendering... and isn't overly expensive.

     

    We don't sell individual parts.

    • Thanks 1
×
×
  • Create New...