Jump to content


  • Content count

  • Joined

  • Last visited

Community Reputation

1,034 Excellent

About rmpfyf

  • Rank
    1000+ Post Club

Profile Fields

  • Location
  • Country

Recent Profile Visitors

3,078 profile views
  1. rmpfyf


    Shielded pairs and the bundle is screened. I was counting on this getting you out of bed it's actually >20dB better than UTP at 50Mhz, gets up to 35dB at some freqs. These data are for common EMC tests. Connectors are plastic; there's at least a screen (no idea how joined and though I've got two I'm not keen to pull them apart). Cables are Elecom, picked them up in Akihabara a few days ago.
  2. I found that I didn't like the M.2 using motherboard power. Yet to see an adaptor that gets around this (though it probably exists, particularly if it's just done into a usual PCIe slot. Don't see the point in SATA M.2, would need to be NVMe. I've found it to be bootable, though I run Linux and I imagine the BIOS has a bit to do with this at any rate... as mentioned above if it's not a drive the BIOS can 'see' then you can't boot it. Honestly a SATA SSD on a block with it's own PSU isolated from the PC is going to sound pretty good.
  3. rmpfyf


    It's genuine. More interested in the shielding properties.
  4. rmpfyf


    In the spirit of isthereanyfriggndifference, I've managed to acquire some CAT8 Ethernet cable. This stuff is thick. Has anyone else had a play? It's supposed to have a good 20+dB on CAT6.
  5. Sorry mate out of the country presently so missed this for a few days. The speed changes are BIOS. Essentially how you configure your BIOS gives Linux a range of options to play with; if you configure it to enable changing CPU speeds, Linux will give you that functionality if support is compiled into the kernel (usually is though you can rip it out). One of my PCs is locked at 800Mhz. Another isn't, and I do it in software as required - so if I'm playing a file back with convolution and/or resampling the CPU goes to max speed to process the audio, then drops to minimum speed for playback. XXHighEnd and a few others do similarly. I haven't worked out how to disable the iGPU entirely. Which would be a powerful thing. Intel's CPU manual suggests there's a state for this though I've not sat down enough to exhaust trying... which would then have me ask someone with more clues If anyone knows how please share. For controlling USB power in Linux I use uhubctl https://github.com/mvp/uhubctl - you need to download (git cone) and compile, then run. Preferably put it into a script you run at startup. I would have only your audio plugged in to your USB card and nothing else anywhere. It may even allow you to de-power the unused port on your card (I'm assuming possibly incorrectly that it's got two). If you a** it up you just reboot and you have your regular system back. Give it a shot or write back asking for exact instructions, either is fine, we are all here to help each other I would run your OS from a SSD simply because OS latencies are going to be lower than for USB, and you are trying to reduce OS jitter at all costs. My best results have been with the SSD powered independently. I couldn't get an NVMe solution sounding better, probably because though the latencies are lower the motherboard power consumption gets a bit wild (which in turn has an indirect effect on jitter); probably the ultimate is a nice, shielded and isolated NVMe SSD with independent power. It wouldn't be hard (though the usual suspects, PPA etc, are likely to come out with a bog-stock Samsung or similar at 3x the price for a bit of ally/a few bits of wire/a rubber grommet). I'll be home in a few days and I'll pull up notes on CPU isolation and affinity management. Basically you can tell which programs to run on what cores, you can tell which devices to run their interrupts on what CPUs, and you can change this whilst music plays to hear the difference dynamically. Then it's a matter of: Having things that annoy the CPU whilst it's meant to be playing music annoy it less (or not at all - LAN connectivity is a big one here, and surprise surprise shutting down networking during playbacks sounds best... but then you have no control) Having things that need really super fast important access to your CPU do so (e.g. your USB interrupt, or the PCI interrupt your USB card sits on) Getting your kernel configured to run your CPU so that it's not doing too much work under low-load running (the sort of which happens when you're running music). I have no idea what SnakeOil does here, I have an idea what my Linux does because I built it... which is not a criticism, SnakeOil is excellent. I just don't know if you could change this, or if you'd even want/need to. I'd be betting this bit is already well-configured by the SnakeOil team.
  6. rmpfyf

    electric cars

    Sadly they'll do so until the weight of public opinion exceeds partisan interests. And we used to have real leadership.
  7. Not true. Batteries are still making money, there's just less money to be had for Frequency Regulation Ancillary Services (FCAS). People are adding batteries every day, big and small, and most FCAS markets are easy to play in with the right market interfaces... there's just less money in it now. 'The cost of maintaining power quality' isn't just FCAS. It's fair to say the amount paid for FCAS has dropped around 90%, but that's nothing very little to do with there being a large battery in the mix, and very much to do with there being a state-owned asset in the mix that's bidding for these services at a rate much cheaper than what other asset owners were trading their assets at into the same market. If the status quo was a bunch of batteries operating at silly-high rates, then a coal generator bidding very cheaply would have achieved broadly the same ends. SA - with a similar volume of deployed wind energy - was getting on just fine for accessibly-priced power quality with two coal generators up to 2016, and had the SA govt taken their owner's offer of extending plant life for a very, very miserly sum... SA energy rates and state taxation would have cost a lot less to 2018 with a small carbon penalty, at which point a battery could have been procured for a good bit less as costs fall and more vendors come to market. Neither is it quite true to suggest that no fossil fuels are involved. The battery needs to be charged, and it's not connected with a wind farm behind the meter - it's a registered generator in its own right (two in fact) and it gets its electrons from the same grid everyone else does. Which is to suggest that whilst Neoen may operate their assets 'net zero', this talk of 'no fossil fuels' is BS. Mentions of 'cost' need also be sensitive to whether or not Tesla made a decent margin on that project enough for it to be a truly equitable, repeatable thing. Or whether there were other possibilities that had better cost/benefit... though no iconic-ish CEO with a Twitter account. Would also add that McKinsey (the data source for the article you cite) might want to up their game if they want to play in this one - that data is old (and nothing new at any rate)_. FCAS prices did indeed fall... until a Basslink suffered an outage, after which they went well up for a good bit there. Seems if you want some decent resilience against this sort of thing, it'll take more than one big battery bidding low enough to aggressively undercut players in FCAS markets. I'm all for renewables, just less for greenwashing! The big battery is a net good thing, it'd be great to see a few more. 'Right mix of renewables' - nothing new! Australia's done some really good work at microgrid level for a while now, it's the same tech on the big grid.
  8. rmpfyf

    Aussie Politics

    More a legal voice, the ironic bit being that DL's got himself a LL.B and unless he slept though his classes he'd know his liabilities. As if the colour of one's exterior has anything to do with justifying that sort of behaviour, Art? DL always wanted smaller government after all, I'd suggest he's about to experience it.
  9. rmpfyf

    Aussie Politics

    Anytime. Broad churches and all that. Only if with a time machine. Regrettably without flux capacitors we've SHY in parliament. Lift.
  10. rmpfyf

    Aussie Politics

    Should we? Putting our fingers in our ears about that sort of behaviour erodes our democracy.
  11. rmpfyf

    Aussie Politics

    Sure is, and the discussion at hand concerns the quality of public discourse. Sadly diminishing, and crucial to the success of any political process. Wish we could ignore the discussion in this context or remove it altogether (which is what the PM should do). 'Till then it's unfortunately a part of 'Aussie Politics'.
  12. BIOS - more than that - did a post a few pages back though mostly it's device management, speed syncing and P/C state management. Network - yes, with isolation on the last leg. I do a little more with CPU affinity though you're limited on an i3. Isolate it at IRQ and power level on the motherboard. You'll need to install some software to power off excess ports (if your motherboard supports this much), again CPU affinity, and nix all other USB if you can (admittedly this lesser part is less of an issue with a separate USB controller for audio as per your card). Would consider going to 1 RAM stick 1R. Try removing your second and see how you sound. 1x stick lights up 200+ less traces on your motherboard and drops some latency loss. Think EMI. There's a few other things you can do to cut kernel jitter though honestly Snakeoil OS is an excellent build. You'll have more flexibility on 4 genuine cores and more cache. See about an i5 T-class.
  13. rmpfyf

    Aussie Politics

    Life isn't a meritocracy, Rob. She can get voted in or out too, much as anyone else in the chair. Quite a separate point to the discussion at hand though.