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

leadeater

Moderator
  • Content Count

    10,346
  • Joined

  • Last visited

Awards


This user doesn't have any awards

About leadeater

  • Title
    Fanatic
  • Birthday 1987-09-23

Profile Information

  • Gender
    Male
  • Location
    New Zealand
  • Occupation
    Systems Engineer | IT

System

  • CPU
    Intel i7 4930K
  • Motherboard
    Asus Rampage IV Black Edition
  • RAM
    16GB G.Skill TridentX F3-2400C10-4GTX
  • GPU
    Dual Asus R9-290X
  • Case
    LD PC-V8
  • Storage
    4 512GB Samsung 850 Pro & 2 512GB Samsung 840 Pro & 1 256GB Samsung 840 Pro
  • PSU
    EVGA Supernova NEX 1500 Classified
  • Display(s)
    Dell U3014 30"
  • Cooling
    Custom EKWB, 3x 480 RAD everything cooled inc ram (why not?)
  • Keyboard
    Razor Black Window Ultimate BF4
  • Mouse
    Mad Catz R.A.T. 5
  • Sound
    Custom build speakers, home theater sound
  • Operating System
    Windows 10

Recent Profile Visitors

11,131 profile views
  1. leadeater

    Ryzen 3700U w/ Vega 10 spotted

    Guess you haven't watched many recent GN live streams, the answer is 7980XE
  2. leadeater

    Ryzen 3700U w/ Vega 10 spotted

    What CPU?
  3. leadeater

    Amazon Picks HQ Locations

    My house has gone up 20% in the 19 months I've owned it and I haven't been displaced by the raise, if I were to sell I could buy in exactly the same area I live right now. Renters get displaced but home owners do not. Exactly no tech companies or large corporate of anything have opened up anything of any kind near me recently either, explaining why such a big raise would have to dive in to about 3 different core areas and it's not really worth covering them.
  4. leadeater

    64 core AMD Rome CPU

    More of a server thing, faster PCIe standards allow the very high end network cards (100Gb/400Gb) to work in smaller slots, 400Gb barely works in a PCIe 5.0 x16 slot and that doesn't even exist yet. PCIe 4.0 allows 100Gb card to be used in x8 slots rather than 3.0 x16 slots. This means you can have more devices in the server as each one uses less lanes. Lane usage adds up rather quickly; 6 NVMe 24 lanes, HBA/RAID card 4 lanes, dual 10Gb/25Gb/40Gb cards 8 or 16 lanes, 4 GPUs 64 lanes which is 100/108 lanes and you might need more than that if you need Infiniband or some other kind of higher end dual stack networking.
  5. leadeater

    Ryzen 3700U w/ Vega 10 spotted

    Why not call it Ryzen R7 2750U? Or anything starting with a 2, this seems dumb. AMD you have 4 digits to use, how are you running out and resorting to numbering things as another generation when it's not.
  6. leadeater

    AMD RX 590 benchmarks

    Previous 90 class cards were the Fury X's/Vega 64's, without sucking for what they actually are. All AMD did was create a line of GPUs pushed even higher in power target, the general increase in CUs and SPs was standard generation by generation affair. Significant amount of the TDP increase was hidden by the HBM. One of the reasons Vega 56 looks so good is it's more in line with what the 90 class cards are/were, with a lower than usual TDP to boot. Far as I'm concerned Vega 56 (at 250W TDP) is the successor to the 290X/390X and all the Fury X/Vega 64 nonsense was a manifestation of having literally nothing to offer so resorted to 'extreme hot-wiring' to make something out of nothing.
  7. leadeater

    (Official?) Battlefield V with RTX benchmark

    SLI does work with DX12, DX12 has many multi GPU modes which in my view is kind of not the best thing as it gets confusing quickly as to what is being talked about.
  8. Something something, sacrifice chickens, system is secure now. My work here is done.
  9. leadeater

    (Official?) Battlefield V with RTX benchmark

    It's not actually a bandwidth problem, actual NVLink can't be used for games (as it stands). SLI is used across the NVLink, the technique is unchange even though the connector and fabric has. SLI always has a master and slave GPU and the master does a significant amount more and certain tasks have to run before work on the 2nd GPU can be done, there's a lot of loss because of that. If you move multi GPU up to API/Game Engine you can just treat GPUs as work units and throw work at them how you like, no master slave and no dependencies beyond what you have in your own code.
  10. leadeater

    (Official?) Battlefield V with RTX benchmark

    Better than throwing 66% of it out the Window . Multi GPU really needs to be handled at the game engine and graphics API level to make proper use of more than 1 GPU.
  11. leadeater

    (Official?) Battlefield V with RTX benchmark

    NVLink is the improvement from the even more shitty HB SLI bridge,
  12. Russian scientists have already been arrested for doing that (mining but same diff)
  13. They are still treated as nodes, workload managers allocate jobs to nodes based on a lot of rules. CERN for example can only run up 300k of their 500K CPUs due to power and cooling (there is multiple locations where nodes are located, all under OpenStack). When you submit a job you give it workload tags so the system knows where to run it i.e. GPU node or memory heavy nodes. The workload manager will make sure jobs don't get allocated to nodes in racks that are at max power capability, stuff like that. Have a look in to SLURM, that's the most popular workload manager. Most of the top 500 use it. https://slurm.schedmd.com/ https://en.wikipedia.org/wiki/Slurm_Workload_Manager Intel and GPU based systems don't combine in to single logical entities, IBM on the other hand does have the capability for that type of thing. You don't actually want to make massive blocks of logical compute nodes across chassis and cabinets though, you start to hit barriers like bandwidth and latency. It's more efficient, depending on task, to move that logic up to your code and make it aware of the hardware boundaries because you can get much better control that way.
  14. leadeater

    Amazon Picks HQ Locations

    Wasn't the original point related to how the new HQ would be a detriment, should get that answered first before accepting rail roading comments about the business as a whole. If the best example is a business going under because of bad forecasting of business prospects then that has nothing to do with the HQ at all.
×