The RX 6900 XT, Look Who’s Stocking?

The RX 6900 XT, Look Who’s Stocking?

As Showing You Two RX 6800 XT’s In Multi-GPU Mode Is Just Too Mean

The RX 6900 XT came and went in the blink of an eye, as has become tradition for this generation of graphics cards.  The MSRP of the card is $999, significantly lower than the $1500 price tag on the RTX 3090 so it is of interest how those two cards compare; not that you can do much but dream of owning one.  TweakTown did get their hands on a reference AMD RX 6900 XT to test, so if you are willing to torture yourself with the knowledge of how powerful these cards are, read on.

The clocks of this card match the RX 6800 XT, the extra cost and performance comes from eight more Compute Units and 20 more Ray Accelerators.  When you add it all up, the RX 6900 XT sports 5120 stream processors compared to 4608 on the RX 6800 XT.  It also has very good overclocking potential, which ensures it pushes past NVIDIA’s top card.  The problem is that so can the XFX RX 6800 XT Speedster MERC 319, which Josh will be posting a quick review of soon, and it costs a fair amount less at $750.

It does outperform any 6800 XT in Ray Tracing, but it can’t keep up with the RTX series when it comes to that particular feature.  The RX 6900 XT is hands down the greatest GPU AMD has produced and it is a far better value than the RTX 3090, but are getting those last few frames at 4K worth the price for you?

AMD's answer to NVIDIA's flagship GeForce RTX 3090 is here, with the Radeon RX 6900 XT dropping for $999 -- but does it crush it?

Video News

About The Author

Jeremy Hellstrom

Call it K7M.com, AMDMB.com, or PC Perspective, Jeremy has been hanging out and then working with the gang here for years. Apart from the front page you might find him on the BOINC Forums or possibly the Fraggin' Frogs if he has the time.

3 Comments

  1. knewman

    Any indication that AMD is still tinkering with their HBM technology? Will they release a product down the road with HBM3 or are they done with that tech?

    Reply
    • collie

      You’ve got me thinking with this question. I’m wondering if all the “new concepts” released in the past handful of years, HMB, interposer chip stacking, Mantle, AM1, ect., were just to stay relevant while they got their shit together getting on an even plane with Intel and Nvidia all while cashing in on the consoles.

      Just a theory but if true the timeline works.

      Reply
  2. razor512

    I wonder, could AMD take a page from Nvidia’s GTX 970 playbook but instead of a crappy move where you have 3.5GB at 256GB/s and 512MB at 30GB/s, they could do something like 12GB of GDDR6 and 4GB of HBM.

    If managed right, it could allow for the most latency sensitive data to be stored in HBM and the rest of the bulk assets are in the GDDR6.

    Or if cost is a huge issue still, then do a Ti model, where they keep the 16GB of GDDR6 and add probably 1GB of HBM, and that can be used for the most throughput intensive tasks, e.g., ray tracing and upscaling, and anything else that is latency sensitive, but does not need a ton of bulk storage.

    Reply

Leave a reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Latest Podcasts

Archive & Timeline

Previous 12 months
Explore: All The Years!