Jump to content

TheAttorney

Members
  • Content Count

    108
  • Joined

  • Last visited

1 Follower

About TheAttorney

  • Rank
    Sophomore Member

Personal Information

  • Location
    UK

Recent Profile Visitors

997 profile views
  1. From memory, I've never got VBUS=off to work, either before or after my IR, with or without HMS inserted before my DAVE DAC. So I've never had the opportunity to see how VBUS isolation affects SQ. But even with USPCB not set to its best potential, the fact that I only slightly prefered the massively more expensive TQ shows that the USPCB can punch well above its weight - or that the TQ was underperforming in this particular setup.
  2. No, I'm using a directly connected USB stick (thumb drive). I don't see any fundemental difference, wrt ramroot usability, between NAS vs direct connected filestore.
  3. There is now a knowledge base article about ramroot, which gives much more information than any of us could provide. You can get to the knowledge database from the Support link in the Euphony app, or directly from the link below (which will require a login if you're not already logged in to their support site): https://euphony-audio.com/hesk/admin/manage_knowledgebase.php?a=edit_article&id=14
  4. Yes this will work, but it's simpler and faster to keep ramroot always enabled and just hit the "Save root fs to disc" after an upgrade or after any config changes. This will ensure that RAM and disc are always in sync. This statement is rather misleading depending on the context. My take is that.... Enabling Ramroot copies all the root filestore to RAM, such that no disc access occurs in the running of the Operating System, resulting in improved SQ due to (it's suggested) reduced latency and noise. With just Ramroot enabled, you still get full access to your music library and music files. If you check the "Copy app data to RAM" option just above the Ramroot Enable button, you will additionally copy /data to RAM. This completely stops all disc access, which further reduces latency and noise. And now you can remove your external drive which possibly may yet further reduce noise. However, that additional option is a moot point for me because usability takes a severe nosedive in exchange for a dubiously small improvement in SQ - at least when caching to Optane and 100% buffer enabled. So this additional option I feel is more for those who want to test what is ultimatly possible, rather than for those who want to enjoy their music collection with minimal fuss. The biggest boost to SQ is to just get the Operating System loaded into RAM - and for that there is no loss in music library function.
  5. If it's like my SR7, then it's the rated voltage minus 10V. I.e. a rated voltage of 12V means you have a range 2-12V. On top of that, there is some leeway beyond the rated voltage: I used this to advantage when I had a 12V rail to drive the 15V Chord HMS. Some have reported that HMS sounds better at the top of its voltage range, and I was able to adjust the 12V SR7 rail to get to just above 14V (but I didn't do an A/B comparison to check if this voltage change had any impact on HMS SQ).
  6. One observation with Ramroot: After upgrading from '10 to '12 release, and having shutdown and restarted a couple of times since then, I noticed that I was back to the '10 release. I hadn't bothered to do the "save root fs to disk" before (on assumption that upgrade would sort that out automatically), so this time I did save fs and this time it stayed at '12 after reboot. Maybe it was obvious to everybody that you need to do that save function after every upgrade if using Ramroot, but I thought I'd mention it just in case.
  7. Only 2 days later, another "new feature" release 20190912 has appeared: I haven't spotted any new features , but SQ seems to have further improved in a particular way. The previous release did indeed have more air, better defined space, greater clarity and focus in Ramroot. But further listening to my brighter recordings showed that it was not shy in showing up the brighter aspects of the recording. This new release seems to have added a more full bodied, natural presentation (what Euphony Stylus was originally praised for), without any apparent loss of that ramroot clarity. My reference point has changed several times in the last few weeks, so I'm no longer sure whether the latest changes are real or imaginary. Maybe I'm just getting used to the new sound? Or maybe something else has changed in the background? But whatever, I'm really, really enjoying this new release.
  8. I gave ramroot a try.... and I like it... a lot 🙂 No issues at all with installation and activation, and works well with my Optane still being used to cache the directly connected USB stick filestore. I've been sticking to my simple settings on my NUC7i7DN: 1.0Ghz, HW=off, core isolation, caching and buffering as previously posted. And now with ramroot, I'm even more delighted with the SQ - considering the low stress being put upon my CPU (typical core temp 37C, and Porky's cooling fins don't even rise to "barely lukewarm"). So much sound from so little energy (around 8W from earlier measurements)! The exact CPU speed seems to matter in this configuration: I slightly prefer 1.0Gz to, say, 0.8, 0.9, 1.2 etc, but this could just be imagination because the differences are much more subtle than, say, ramroot. Also, this release improves the upgrade process for those totally reliant on the WIFI function to access Euphony/Stylus - WIFI on previous releases became inactive after the upgrade process. Euphony Support said they'd fix it at next release - and they did 🙂. Great support.
  9. With hyperthreading disabled in BIOS the last couple of days, I'm really enjoying the sound. One bonus is that the max frequency doesn't seem to matter as much as before: It sounds good whether it's set to 0.9Ghz, 1.9Ghz or around 4Ghz. That's not to say they sound exactly the same, just that I can as easily engage with the music with any of them. Even Core Isolation doesn't seem to matter as much - maybe I'm just getting comparison fatigue? So I'm sticking with max frequency of 0.9Ghz, which results in a huge CPU temperature drop to just under 40C (with correspondingly reduced power consumption). A further bonus is that these lower speeds do not suffer the temperature fluctuations of turbo mode. Strangely, at the low speeds, "Everything Else" is back to showing more CPU activity than Stylus, so I've come full circle on that point. So, with HW off, Max Freq= 0.9Ghz, Core Isolation = 0-1 stylus 2 gstp 3, I get a great sounding, cool running, low energy, no fuss solution that I'm delighted with 🙂. Even if a max'd out turbo'd and hyperthreaded version sounds a bit better (and I'm not sure it does), I've stopped trying. No more changes from me until the ramroot function gets fully released.
  10. No, I'm not using ramroot. I tried again last night with the same results as before. It could be a CPU/motheboard-related issue: If my 7i7DN board doesn't fully respond to CPU frequency settings, maybe it also doesn't fully handle core isolation as well? RickyV may be able to answer that question when he tries my tests - as he also has a 7i7DN-based server. But more interesting for me is that today I've found a better sounding Core Isolation setting 🙂. I disabled Turbo and HW (hyperthreading) in BIOS to see if that changed the Core Isolation reboot issue (it didn't). I then reinstated Turbo because that is required to get CPU max frequency above its base frequency. But left HW off (Fidelizer's designer believes HW is detrimental to SQ in his experience, so I wanted to test that out). So, with only 4 real cores available (and Turbo on), I found that 0 stylus 1-2 gstp 3 gave me even greater clarity and dynamics - the best yet. Strangely, with HW off, stylus now causes the highest CPU load, and Everything Else the least - the opposite of what I saw before - goodness knows why! That's why I allocated 2 cores to stylus to stop it going beyond 70C. Overall, CPU temperatures are a few degrees C lower than with HW, so a good result in every way. This uplift in SQ is not fully conclusive at this stage because I also recently made a tweak to my headphones - I think I have seperated the two changes, but more time is needed. In the meantime I do recommend others try the disabling HW - of course this could be CPU dependent. Just hit the Core Isolation Apply button. This will display all the relevent processes. As your default position is an empty Core Isolation field, hitting the Apply button will display current values without changing them.
  11. Firstly, the Core 0 temperature increases after applying my particular Isolation settings (as previously posted). After reboot, the temperature reduces to the default values. If I hit the Apply button at this point, the temperature starts increasing again. Secondly, I believe that hitting the Apply button with an empty field will show the actual core values without changing them. If I do this after applying Isolation, the display shows the correct values. If I do this after reboot, the displays shows the default values (0-7 for everything). Tip: Remember to Ctrl/C your isolation settings before clearing the field, so you can quickly re-instate then later. Thirdly, by listening. If I re-hit the Isolation Apply button after reboot, I get a boost in SQ. Any subjective test is not totally reliable, but that's what I hear. It's harder to notice this on reboot because of the larger time lapse and because even default sounds pretty good anyway.
  12. There appears to be an anomaly with the Core Isolation feature: It reverts to the default of No Isolation after reboot, so I need to hit the Apply button again after each reboot - worth knowing for those who don't run their servers 24/7.
  13. My default position is Euphony Stylus booted from 32 Optane M.2 card, and redbook music files on a 256GB USB stick directly plugged into one of my NUC's USB slots. With 100% buffer, caching, turbo, HW and recent core isolation. And I'm very happy with the sound. BTW, caching doesn't always appear to be immediate when adding new files to the playlist, and Stylus gets tempremental when I try to add several albums in very quick succession. But it gets there eventually I again removed the USB stick whilst music was playing from cache (there's about 20GB available for cache on a 32 GB Optane, which holds several redbook music files). And I again couldn't reliably tell if there was a SQ change when removing the USB stick, so even if there is a slight SQ impact, I'm prepared to accept that in exchange for an incredibly simple and low cost filestore solution for my NUC server. I also tried ramroot from the trial Euphony on a different USB stick, but didn't get very far. Firstly, on a brief initial comparison, ramroot did seem a bit better than boot from USB stick, but it wasn't sounding obviously better than my memory of the boot-from-Optane default. At this stage, it was too short a test and too many variables for anything to be conclusive, but then I ran into a problem: Having activated ramroot, I then hit the disable button and rebooted. After ages, it still came up in ramroot, even though the info in the ramroot field and button seemed to acknowledge that I had indeed disabled it. So I retried and the same thing happened. As I don't have a monitor to connect to my NUC server, I can't follow the boot process for any clues, so I soon got bored with this and gave up. It doesn't help that my trial ramroot was created onto the only spare USB stick I had, which is several years old and probably contributing to the painfully slow boot up time. Maybe the Euphony team could consider why I wasn't able to disable ramroot? In the meantime I'll just wait for this function to be fully released, so that it can go straight to my Optane drive.
  14. Yes, I did try this (WYR WY) at the time, and it didn't sound as good as WY WYR, but by that stage I was getting testers fatigue and it was all starting to get muddled. So don't let that stop anyone else from trying it.
  15. I did try that, plus a few variations, on my 7i7DN standalone Stylus server when you posted this, but I couldn't decide if the subtle differences I was hearing were real or imaginary. So I stopped trying. But yesterday, when I was looking at the temp/CPU graphs, it seemed to me that, when just playing music (with no DSP), Stylus was putting least load onto the CPU, gstp was more load, but "everything else" was the biggest load. So I allocated more cores to "everything else": 0-3 stylus 4-5 gstp 6-7 And this gave me a (literally) clear improvement in SQ (in short, a reduction of smearing), and I'm now a believer. Of course, there may be further variants, but nothing obviously improved on this and I soon got bored with trying. This was with turbo and HW on. One unwanted side effect of these changes is that any change from the default 0-7 seemed to increase core temperature. E.g. with default, Core 0 typically fluctuated between 58-65C. With above new setting, Core 0 jumped to 68-75C, whereas Cores 3 and 4 stayed lower. Now, the strange thing is that allocating even more cores to "everything else" did not reduce that highest temperature and, in general, isolating cores did not always correlate to what I was expecting in terms of CPU load and temperature. My amateur conclusion to the above is that core isolation, especially with turbo/HW, is more complicated than one might think. I'm guessing that the CPU is at least partly overriding the isolation request in order to stop itself from overheating or to run more efficiently. But irrespective of whether that is true or not, I feel the above new setting is very worthwhile for me.
×
×
  • Create New...