The premium priced ASUS GTX 980 Ti STRIX DCIII OC certainly does perform

12 Comments

1. 

   Master Chen on September 9, 2015 at 1:22 am

   Nope. Not even close.
   Zotac’s
   Nope. Not even close.
   Zotac’s AMP Extreme is the best 980 Ti currently (if you don’t count in Colorful’s monstrosities).

   Reply
2. 

   Philbert579 on September 9, 2015 at 1:33 am

   While I suspect a typo, I
   While I suspect a typo, I would be very excited for 8GB of VRAM from software tweaks!

   Reply
   • 

     Jeremy Hellstrom on September 9, 2015 at 10:54 pm

     dang it, fixed.  But maybe

     dang it, fixed.  But maybe Sebastian could teach you how to add that extra 2GB

     Reply
     • 

       Master Chen on September 9, 2015 at 11:12 pm

       Or just buy a Radeon. %)
       Or just buy a Radeon. %)

       Reply
3. 

   Anonymous on September 9, 2015 at 4:30 am

   If you are only going to use
   If you are only going to use the GPU for games then Nvidia may make some little sense at the Moment, but if you want to do more with the future Vulkan/DX12 based games then AMD’s asynchronous compute appears to be the way to go, if you want a GPU for more than gaming, and now for gaming too. Nvidia’s attempt at product segmentation and the striping out more of the asynchronous compute ability from its SKUs has backfired now that the games makers and the graphics APIs are able to make more use of this hardware asynchronous compute ability.

   I’m sure that Nvidia’s short sighted marketing department took great advantage of Nvidia’s power usage advantage at the expense of compute, but now the entire gaming software stack is starting to move towards Utilizing all of AMD’s available hardware compute/graphics abilities included in the hardware in the ACE’s and their hardware based asynchronous compute ability.

   So now not just games but other software applications as well will benefit from AMDs GPUs. Nvidia’s marketing made its bed on hawking its products to only gaming interested consumers, and now that hardware based asynchronous compute has shown to also be of an advantage for gaming, Nvidia is left to lie in that bed. Better get some new tapeouts on the way Nvidia, because the gaming software and graphics APIs are not going back to the pre-hardware based asynchronous compute days.

   It’s not going to ever again be just a matter of doing it all on the CPU compute wise for gaming, a whole lot more of that non graphics gaming compute is also going to be done on the newer GPUs with that hardware asynchronous compute ability running more than just graphics type operations. Gaming physics and lighting require a lot of parallel types of computation that can be done on the GPU more efficiently in addition to just pure graphics computation.

   Reply
   • 

     Jmac on September 9, 2015 at 2:39 pm

     This is a GAMING card. Your
     This is a GAMING card. Your observations may or may not be correct, but it has nothing to do with this piece of hardware.

     Reply
     • 

       Anonymous on September 9, 2015 at 3:30 pm

       And gaming engines will be
       And gaming engines will be using all of the asynchronous compute ability that is available, with much more done on the GPU, and much less done on the CPU. Say goodby to the latency inducing motherboard CPU for discrete gaming as the gaming engine/graphics API software stack takes advantage of all that GPU number crunching ability. Who needs those 8 CPU cores when there will be thousands of GPU cores available for game engines to utilize. Let the motherboard CPU do its janitor’s job of keeping the OS/bloatware in order while the GPU runs the games, and discrete GPUs will be getting some more CPU like abilities going forward, and may even get a few on die CPU cores to reduce latency to a minimum.

       There will be some exceptions for the motherboard gaming APUs that may be constructed of CPUs/GPUs/HBM on an interposer that will have more beefy discrete like GPUs with plenty of ACE units and the CPU cores wired up in a wider more direct manner to the GPU by thousands of traces wide interconnects etched out on the interposer’s silicon substrate. Expect the CPUs on these APUs on an interposer to be further integrated with the GPU with even more HSA compute ability offloaded to those thousands of GPU cores. GPU Hardware based asynchronous compute ability is the future of gaming.

       So all this NON Hardware based “asynchronous compute” ability kit will be quickly placed at a disadvantage, and it only runs using the outdated API feature set’s limited hardware support. Enjoy waiting for the draw call to complete to be able to do any “asynchronous” work, and that’s is either graphics or compute, but not both at the same time. Vulkan and DX12 will be doing things asynchronously from now on. Better get your hardware in order Nvidia!

       Reply
   • 

     Anonymous on September 10, 2015 at 6:15 pm

     Short sighted?
     And yet, AMD
     Short sighted?
     And yet, AMD had a CPU plan that was based on games using multiple, less efficient cores. How did that go?

     Future proofing existing technology while gimping current plans isn’t necessarily the best way to go either.

     NVidia is profitable. AMD is not. Perhaps NVidia actually has a good plan?

     There will always be NEW CARD designs as well.

     *Finally, it’s not like DX12 is going to suddenly make DX11 pointless. Not only do people continue to play and buy older games where NVidia is MUCH better at DX11 but also DX12 games aren’t going to be the majority of games sold for a long time.

     Reply
4. 

   terp02andrew on September 9, 2015 at 4:53 am

   It’s a bit premature to be
   It’s a bit premature to be drawing those conclusions.
   I’d like to see what unfolds and would like to see updated results following Kollock’s latest update (which came on the Friday before Labor day).

   If and only if nVidia’s proposed solutions continue to display inefficacy will there be a concern on long-term longevity.

   Considering the same Kepler whites papers are referenced for Maxwell, then whatever solution nVidia formulates will apply across the product line – better or for worse :p

   I’d like to see that before making any judgements.

   Reply
5. 

   Anonymous on September 9, 2015 at 5:08 am

   But how does it handle ASYNC
   But how does it handle ASYNC shaders?

   Reply
   • 

     Yorgos on September 9, 2015 at 10:57 am

     It doesn’t matter, by the
     It doesn’t matter, by the time the next gen is out, nvidia will make you give them money via driver emulation.

     Reply
   • 

     Master Chen on September 9, 2015 at 12:25 pm

     “Asynchronous Shaders? What
     “Asynchronous Shaders? What is that?” (c) Huang
     “Async Compute? Nope, never heard of it.” (c) Maxwell

     Reply

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