Radeon RX 470 Benchmarks

Well that's obviously a ****ty demo, dx12 runs worse in both cases. I would expect to see at least 5-10fps boost, not 5-10fps lower avg fps.


I mean dx12 is famous for lots of drawcalls and them woods/grass should take advantage of that, but it looks like something is choking it.

 

Nope, common misconception, the fact is there is nothing in DX12 that will alter how fast GPUs can process compared to DX11, the main limitation being lifted is as you say the drawcall limitation due to proper multithreading in the API.

Just because there are many trees does not mean many drawcalls ! You can have one drawcall for a set of 100 trees, or you can have a drawcall for each tree. You can have a drawcall for each leaf on the tree + drawcalls for all the branches + drawcalls for the main trunk.

If there's a flock of birds in a game, they could all be rendered through a single drawcall, or you can batch the work into several drawcalls each covering a single bird.

 

Ok I was thinking about dynamic/diverse trees or more objects on screen at the same or even faster fps. As its suppose to be.


Also how can't it alter? now it in dx12 it has more threads, not just 1 main..

dx11 8ms, dx12 4ms
http://i.imur.com/8hW1mKd.jpg

amd slide
http://cdn.arstechnica.net/wp-content/uploads/sites/3/2015/08/anddx12.jpg


I still think its a poor dx12 implementation in that Cryengine tech demo. It should have saved few fps, not lowered them further.. Or enhance visual fidelity with more interactivity at same fps.

 

No it really shouldn't have, there's absolutely no advantage for DX12 over DX11, as far as I know, if the DX11 path isn't bottlenecked by the API, there will be absolutely no advantage whatsoever none, zilch.

 

heh I don't get your logic in this one..



Did you even bother to check those links?

Also if its zlich none nada, then why does dx12 run 5-10fps slower? It should run exactly the same if there is no "difference".

 

Don't have DX12 numbers to compare, but, The Talos Principal allows us to compare DX9 vs DX11 vs OpenGL vs Vulkan . (one of the few to allow this btw).

http://steamcommunity.com/app/257510/discussions/0/412447331651720139/?ctp=18

Post #266 (missing OpenGL though)

When optimised, DX9 still fastest on AMD (results are different on Nvidia though). Still better overall average lowest and high fps.

Those expecting magical performance on AMD with Vulkan...not yet. Still needs to be optimised just like DX12.

 

With Dota 2, I can say that DX9ex vs DX11 vs OpenGL vs Vulkan, that Vulkan is king for me.

 

those numbers look real good, but until HH confirms it I'm not giving this any hope

 

Got some numbers to share? Would be interesting to see how far apart the api's are in a less demanding game like Dota 2.

 

http://www.phoronix.com/scan.php?page=article&item=dota2-vulkan-redux&num=1

 

Exactly the article I just found myself lol.

4K results are interesting.

 

No, I can share experience.
DX9ex Game starts well, decent fps, smooth. But GPU utilization is very low and GPU downclocks (Clockblocker or not, same effect). Then CPU utilization gets very low too and slowly in course of 1~2 minutes game crumbles to ~15 fps.
On DX11 same thing happens, it just takes more time, maybe 4 minutes and it hovers around 25fps.
With OpenGL being around DX11.
And Vulkan simply keeps ticking at high fps, GPU still has full clock.

I do not know why Dota 2 chokes for me, but it does not do it under Vulkan.

I would do comparison on tablet, but intel still does not have driver ready for Cherry Trail.

But this is nice comparison. CPU utilization looks pretty much same. But game is smoother with Vulkan.
And GPU monitoring shows that GPU utilization is higher and so is GPU temperature (indication of workload).
But neither did extract 100% GPU potential. Maybe fps limiter on 60 is enabled. I remember that I had to do something to dota 2 to get over limiter.

 

Dota 2 Reborn R9 270 performance looks interesting. Kinda confirms The Talos Principal results (no Vulkan results though);

http://www.gamingpcbuilder.com/dota-2-reborn-hardware-performance-benchmark/3/

 

haha omg... look at that ancient 9600GT go

 

Again AMD mostly using AMD sponsored/optimized games to show off there gpu's, with results that don't reflect the true performance you will get in 99.9% of other games!

Overwatch can be run on a toaster, and 3Dmark is just a benchmark that is getting old now, and does not represent the performance in the latest games very well.

Also funny how they are using a high end Intel cpu, i wonder why they don't use there own AMD cpu's:eyebrows:

As always wait for the facts with a guru3d review testing a broad range of games, and not mainly AMD sponsored games.

 

Overwatch is a Blizzard game, and they have been favoring NVIDIA GPUs since WarCraft 3. It's actually an impressive showing just because of that. It's like the 1060 beating the 480 in Hitman kind of impressive. And yes, they would use the fastest possible CPUs to test their GPUs, that makes sense.

He's testing 15 titles. 6 are NVIDIA sponsored, 6 are AMD sponsored, 3 are independent. Please at least learn to count up to twenty before accusing people like that.

 

You misunderstood, i said wait for the guru3d reviews meaning guru3d test a wide range of games, unlike AMD who mainly test AMD sponsored games!

 

I am sorry then. I didn't get it right.

(Are these posts for ants?)

 

I think it won't be long before Guru3D has propper a review for this little card , then we can see the facts as how this VGA performs.

 

This should be somewhere between 290 and 280x.