• Announcements

    • colonel_mortis

      PayPal Problem - Some payments not going through   20 May 2017

      PayPal appears to be experiencing problems at the moment, resulting in people being billed but the transaction showing as failed on our end. If you have signed up for a new subscription, and the transaction is showing as pending on the forum, so you haven't gained floatplane access yet, and you have waited more than 10 minutes after purchasing it in PayPal, please go to https://linustechtips.com/main/failed_transaction_fix.php and follow the instructions. If you have any issues, please send me a PM (hover over my picture or username and click "message"), including the transaction ID, and I will try and sort it out for you.

lots of unexplainable lag

  • Content count

  • Joined

  • Last visited

Everything posted by lots of unexplainable lag

  1. Watch this, then make your choice off of it:
  2. Reference coolers are designed to be the bare minimum required at stock clocks as to not take away sales from aftermarket solutions.
  3. With a new monitor? If yes, get the RX580 with a FreeSync monitor, even if the GPU is more expensive. You'll save that cost by not spending a fortune on a G-Sync monitor.
  4. Yeah, I think it's save to say AMD wants to step away form traditional VRAM and use the word VCache for it or something since it's not really used as traditional VRAM. AFAIK this first gen of I.F. can already scale up to like 512GB/sec or something so raw data throughput won't be an issue.
  5. Note this isn't your typical multi-GPU design we're talking about here. This is about lowering cost of a GPU die by a bazillion % by having multiple smaller dies connected via Infinity Fabric rather than one massive die size and then connecting two massive dies together. Take the Fury X. 4096 Stream Processors. You can either make one massive die of 4096 stream processors with all of the other logic attached (display stuff, memory controller etc) or make 4 1024-SP dies connected via Infinity Fabric to the separately manufactured display stuff, memory controller etc. You end up with way higher yields on your silicon wafers by having more dies per wafer and thus having a higher succes rate per wafer which drives die costs into the ground. Connecting them via Infinity Fabric with the other logic is the final step in the process, and now you have a high-performance GPU for like 2/3rd the cost. Like, 1080Ti performance for 1070 cost. Or AMD can go absolutely bonkers and make like 4 2048 (so RX570-sized minus display/memory logic etc) dies and connect them for a 8192-SP monster of a GPU that eats everything in its path alive at 1080Ti costs. It's entirely possible to do that. Not to mention that to software it's seen as one GPU rather than multiple in CrossFire eliminating the need for game devs to implement CF/SLI scaling. tldr This is AMD going "Hi, we're AMD, and welcome to Jackass".
  6. Honestly, with all of the nonsense going on there, if they want to basically have a pair of cameras placed just above your eyes they should just disconnect from the world and float off into the Atlantic.
  7. Forget that, how about 4 1024-SP dies on a single interposer for a 4096-SP part at the price of like an RX580.
  8. Paper launch during Computex, products launching in June. That still qualifies as H1 2017 so he's keeping it for at least 6 more weeks.
  9. Strange. Why would AMD make a render of a product that seems to need more power? Or the card Raja was holding was an engineering sample and the render is the real thing.
  10. No, RX Vega was never planned for this talk with investors. RX Vega is oriented to consumers, and we have Computex for that. AMD holds a press conference on May 31st giving them the entire month of June to still release the thing. Fiji was announced during Computex 2015, RX480 during Computex 2016 and they all launched in June following the events.
  11. Seeing how the Ryzen 8-cores are priced, I expect some really insane prices for these things.
  12. That lit-up R-logo though... That sexy blue aluminium casing... That's cool. Also, hello double 8-pin power plugs.
  13. NVMe is a protocol designed for solid state storage. The M.2 connector can be utilised in two ways: either connected via the good ol' SATA bus using the same protocol used by hard drives AHCI (still giving some better performance compared to SATA 3 SSDs because of the higher bandwidth supplied by the M.2 port) or via the new protocol NVMe which was designed for SSDs from the very start thus having waaaaaaay higher read/write speeds. Read on it here: https://en.wikipedia.org/wiki/NVM_Express
  14. M.2 is a small version of a standard PCIe 3.0 x4 connector. The connector can be used on both a SATA bus or the PCIe bus, but in this case it's most definitely the PCIe bus.
  15. Pretty much yeah. It uses the 32GB of what I assume is HBM2 or something, then it has slower but still direct access to 1TB of speedy NVMe storage.
  16. Upgrade basically everything. Locked i5 and a 1080Ti, there's bound to be some bottlenecking somewhere. At this point, you'd be best off getting something like an Ryzen 5-1600 with a B350 board (or X370 if you want some higher-end motherboards that can do SLI). And new RAM, since what you have is DDR3 and new Ryzen/Intel boards/CPUs want DDR4.
  17. nvidia

    Why specifically that old-af GPU? Get an RX560. Faster, 4GB of VRAM, all the latest display standards (including FreeSync) and, well, a new card that won't be bottlenecked in like a year.
  18. Yup. It completely bypasses having to call the CPU for it to access a large storage pool of ROM meaning it can access huge amounts of data way faster. It's not 1TB of VRAM, It's 32GB of VRAM and then a 1TB NVMe SSD.
  19. Literally the 4th or so post on this. Also, please follow the guidelines for posting in this forum.
  20. Nope, GDDR6 is like GDDR5(X), it's an on-PCB tech, not something you put on an interposer. It's waaay too big for that.
  21. Don't see 4GB HBM2 happening, even if it's more than enough with the caching tech in Vega. It just won't sell at all because dumb consumers seeing 4GB on an AMD card and 8GB on an nVidia card and thus going for the nVidia card even though the AMD card is better.
  22. 1525 mHz is already a confirmed minimum (25 TFLOPs FP16 with 4096 SPs), so this isn't entirely unlikely. The 16GB HBM2 though, maybe. Maybe only for high-end WS boards or Titan XPp-class GPUs, but for gaming-oriented GPUs it's just gonna be too expensive.
  23. graphic cards

    The GTS is kinda between the RX480 RS and RX480 GTR (way better than the RX480 RX with if I have to believe Buildzoid 50A power stages meaning its VRM can push 300A to the GPU which it will never ever go to) and the RX580 GTR-S basically re-uses the RX480 GTR PCB meaning 360A or something. In the RX480 lineup, the GTR pretty much was the go-to card for the high-end RX480 and people got 1400+ mHz on them sustained. Same applies to the GTS and GTR-S. They're pretty much the best boards for an RX580.
  24. graphic cards

    XFX RX580 GTS or GTR-S too.