Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

comander

Member
  • Content Count

    1,101
  • Joined

  • Last visited

Awards

This user doesn't have any awards

1 Follower

About comander

  • Title
    Veteran
  • Birthday Jan 01, 1970

Profile Information

  • Location
    /dev/null
  • Interests
    Tech, data, fitness
  • Biography
    My views are my own and do not represent my employer.
  • Occupation
    Data Scientist

System

  • CPU
    3900x
  • Motherboard
    B450 matx
  • RAM
    4x16GB
  • GPU
    RTX 2080
  • Storage
    Optane P4800x 1.5TB
  • Display(s)
    2x 27" IPS + 1x 35" Ultrawide IPS
  • Cooling
    H100x
  • Keyboard
    Topre Realforce

Recent Profile Visitors

1,728 profile views
  1. This is a relatively modest certification fee increase. Worst case scenario you'll have companies putting out fewer SKUs or foregoing certification. This should have a minimal impact on the market, though I wouldn't be surprised to see some cheaper brands or lower volume runs do something like "90% or better efficiency" if they can actually hit those figures (very possible for 12VO) or bragging "beats California efficiency standards" in lieu of 80+ certification.
  2. I have very little loyalty towards any one company. There are some minor exceptions. I trust a few companies on privacy a bit more than others so that influences smart home decisions. Apple has a little bit of trust on my end and... it's not enough to make me desire any more of their products.
  3. I'm going to overgeneralize this a bit and fall back on analogies. You do certainly have some valid points. Hypersensitivity CAN be bad. VERY VERY bad. Hypersensitivity can ruin lives for no real benefit. The analog in the health sciences would be allergic reactions and autoimmune disorders. There is a VERY real argument for gradual exposure therapy. It can literally be a matter of life or death and it's generally easier than trying to make the entire world accommodate you (or limiting yourself greatly). At the end of the day, there are tradeoffs in how situations ar
  4. Forum, reddit, misc IT projects (adding new VLAN on my mother's network for a POS terminal and configuring firewall rules; dremmeling a small server in my garage to get stuff to fit better), Oh also work (configuring data pipelines, setting up a code framework for crunching data and making ML models, writing design documents, meetings with stakeholders)
  5. PCIe 5.0 is more important for the enterprise. Imagine having 128 lanes per CPU to use for ~50nvme drives and a few 100Gbps ethernet links (or one 400Gbps) There's almost no reason to use RAID0 or similar EVER. The reliability and consistency of it is just... bad. RAID5/6 or Z1/Z2/Z3 (of 60 or two Z2 Pools) will have similarish performance characteristics with a close enough price. If even further performance is needed... cache. Use RAM. Store the most common things in memory - just be aware that 8 channel DDR4 pretty much peaks at ~100GB/s, which is slower than what you'd get f
  6. Hey guys, bit of a suggestion - can you test some of the more common keyboards for latency? https://danluu.com/keyboard-latency/ Ideally some of the more commonish mechanical keyboards, something with Topre (HHKB or Realforce) and something Hall Effect, Scissor switch or Rubber Dome based.
  7. I worked for a Telco before my current role. Any reason you don't just use a streaming service? It's cheaper, the video quality is usually better and there's no equipment to rent - just slap on a $40ish Roku/Chromecast/FireStick/AppleTV.
  8. Works just fine on my work phone (Android 11 based). Dark theme carries over from desktop settings as well when you're logged in.
  9. Be precise and quantify things. Let's say the 5900x is 2800zł and you can sell your existing CPU for 1200zł (making this up) for a net cost of 1600zł (and that your eventual salve cost for your 5900x is around 1600zł). Next let's say your cost of capital is ~10%. This means you'd need to make 160zł extra per year OR to save the equivalent in time each year. How much time do you expect it to save you? How much electricity? Can you do more projects? Will your quality of life be that much higher? As an FYI, the numbers in zł are meaningless to me and I have no idea what inco
  10. If time is money, OP should probably not waste time OCing and validating the part and instead get something like a 5950x.
  11. 3000 is the spec for all core clocks (though top turbo speed is listed at 3.7GHz). It's possible it goes higher. I pretty much ran my part at 3.8GHz at a lowish voltage its entire life. I could get it to around 4GHz but I wanted to run it at lower voltage and thermals.
  12. Once upon a time AMD and Intel both made CPUs. Instead of only selling one product they took the same chip design out of the same factories and validated them at different levels. Enterprising people discovered that the slowest parts could often run much faster, sometimes even faster than the fastest officially available. In 2005 I could take a 1.8GHz Opteron 165 and push it to almost 3GHz. In 2006 I could take a 2.13GHz Core 2 duo e6400 to 3.6+GHz. These are like 80% gains. In 2012 I could take a 3.4Ghz Core i7 and push it to 4.6Ghz pretty easily. This is a ~35% gain. In 2017 I
  13. If you look at my sig you can see I kinda went off the deep end in terms of specs around a year and a half ago. My life is not meaningfully better. It was not transformed. This is coming from a hardcore enthusiast who's big thing is computer hardware and projects related to it. One question I'd have for you - is the performance with your current CPU adequate? Would something newer/better meaningfully improve your life (free up time?). Knowing your use case really matters. "I want something good" is usually a poor strategy.
  14. My 16,000GB server is named "NAS" I'll let you guess why. For what it's worth it's faster than local storage for some things... Something about the most commonly accessed data being cached in RAM (and then an Optane drive) and the less commonly accessed data essentially being on a RAID 5 array (Z1).
×