AMD Live Webcast From Computex 2016

This is probably a good way to gain market share. If 2 $199 cards can even get remotely close to a 1080 (wouldn't have to beat it really), there will be some people buying them them 2 at a time. Problem is that real dx12 titles need to come out (Not dx11 titles with dx12 patches).

Of course, we need some real world benchmarks 1st.

I like the marketing tactic though. Nvida and AMD have the same goal but going in the opposite path to get there. There could potentially be a new performance leader going back and forth every few months. Nvidia launched high end now with low end coming in the fall. Amd has low end now with (hopefully) something coming before 2017. so if they kept on this tract, they would just rotate performance leads between them both.
 
This is probably a good way to gain market share. If 2 $199 cards can even get remotely close to a 1080 (wouldn't have to beat it really), there will be some people buying them them 2 at a time. Problem is that real dx12 titles need to come out (Not dx11 titles with dx12 patches).
If we go back through history I think you'll find two ~$200 cards in CF/SLi pretty much always beat the single X80 ($500-$600) chip.
Anybody who gets excited at that slide is either brand new to PC's or enjoys flashy marketing. It's literally status quo.
 
AMD needs to really work on improving their product reveals. Their presentations are so boring.
 
TRG & Microsoft really seem on good terms, so DX12 looks in good shape for them. Vulkan is based on AMD tech, so that's fine also.

Looking good overall for TRG.
 

I was also questioning this. Why did the left one look so much better than the right. It was really noticeable. Then again, I'm a game designer...
 
I wonder what the 300 dollar version of Polaris will be?

Lisa mentioned there will be polaris based gpus from 100-300 dollars.

A 480x?

More to come I suspect. Also, I wish they showed the performance with a single 480 and threw the 1070 into the mix, I get they were trying to make a very specific comparison about the cost/performance you get with dual gpus vs a single 1080, but that utilization thing made me wonder how utilized would a single 480 be in ashes of the singularity in dx12?


If two 480s being utilized at 51% can slightly overtake a 1080 with 98% utilization, could a single 480 at a similar 98% utilization match the 1080 all by its lonesome? That would have been an infinitely more impressive display, but I suspect it might be harder to max out GCN for a developer, and I would LOVE for someone in the know to explore why that is?


The power advantage is basically nil with these cards vs nvidia, but that is partly expected. It seems that while pascal has better preemption, is STILL does not have the same sorts of additional hardware schedulers (that use additional power) that allow for concurrent async compute like amds gcn cards, so while nvidia takes fewer hits and performance degradations, they still don't get the same boosts as AMD cards.


We need that Maighan guy to chime in, he seems to know secrets and additional details.

I especially want some exploration about if and why amd cards have a harder time being utilized to greater effect. And perhaps I am making bad assumptions, is it true that if amd cards were utilized at around 98% on the 480 it would be similar performance in dx12 in ashes? Or does it not scale linearly?

We need more tests, and more dives, I want answers!
 
Glad to know I wasn't the only one that caught that. Graphic settings weren't equal.
 
I was also questioning this. Why did the left one look so much better than the right. It was really noticeable. Then again, I'm a game designer...
Wait, you thought the left looked better? o_O
 
People on Reddit say that the demo is dynamic and the Nvidia driver doesn't draw all the snow sometimes. I dunno. /shrug.
 
Ohhhh dear.

I understand that there's no such thing as bad press but man, that's veering a bit too far.
 
I was also questioning this. Why did the left one look so much better than the right. It was really noticeable. Then again, I'm a game designer...


Left better than the right? I thought the one on the right looked better than the left shot. More terrain detail and more units on screen, but I need to rewind the tape to see if that was just a single time because I don't remember if the entire simulation had the same result. It does look like that shot has the nvidia card on the right with visuals turned up though compared to the other one. Again, we need more tests, independent tests.
 
Left better than the right? I thought the one on the right looked better than the left shot. More terrain detail and more units on screen, but I need to rewind the tape to see if that was just a single time because I don't remember if the entire simulation had the same result. It does look like that shot has the nvidia card on the right with visuals turned up though compared to the other one. Again, we need more tests, independent tests.

Sorry, I literally mixed up my left and right. Yes, the Right one looks WAY better. the units cast ambient shadows and the terrain has more features, as well as more units on-screen...
 
Oh wow, will we see frontpage posts about "misleading marketing tactics" by AMD? Let's see how Raja vs Huang criticism plays out.
popcorn.gif
 
the left is supposed to be the 480s. didn't see anybody post this but here is the captured stream:
AMD Live Webcast From Computex 2016
aots demo is at 36:30ish

edit: listen to his VERY careful wording at 37:15 he says "kinda sorta similar performance" and "maybe fewer" THEN HE STOPS and corrects himself(shit don't say lower settings, don't say lower settings!!) and continues "maybe people with really crazy eyes..."

god dammit AMD! WTF are you guys doing?!
 
Last edited:
Sooooo... nvidia still having a hard time with AoTS, ok I guess.
 
Oh wow, will we see frontpage posts about "misleading marketing tactics" by AMD? Let's see how Raja vs Huang criticism plays out.
popcorn.gif

So apparently Nvidia cards will render more snow versus AMD cards, but the other things are a bit eyebrow-raising.
 
While the AoTS discussion is eyebrow raising, the Zen reveal might be the most cringe-worthy thing I've seen.
I'm just not sure it's something I'll forget for a long time.

Hopefully the actual hardware is still good haha.
 
While the AoTS discussion is eyebrow raising, the Zen reveal might be the most cringe-worthy thing I've seen.
I'm just not sure it's something I'll forget for a long time.

Hopefully the actual hardware is still good haha.

What Zen reveal?

oh...
 
Well, at $229 the 8GB version of Polaris makes mGPU look good! I'll take an extra 4GB VRam for $30! :LOL:
 


So I guess there is more to come? He may have said more than he is suppose to. So looks like we might see a gtx 1070 competitor for may be 300? Straight out said 8gb will be competing with gtx 1070 but we don't have the price on that yet. So it seems like amd gave out more info on tech day about polaris than we saw here. Guess he couldn't hold on lol
 
Yes, it seems that there will be reveals all the way to the June 29th hard launch.
 
So I guess there is more to come? He may have said more than he is suppose to. So looks like we might see a gtx 1070 competitor for may be 300? Straight out said 8gb will be competing with gtx 1070 but we don't have the price on that yet. So it seems like amd gave out more info on tech day about polaris than we saw here. Guess he couldn't hold on lol

Watch this be the little Polaris chip, and AMD actually has a big Polaris chip waiting to be announced.

I think there is an icecube's chance in Hell of that happening, but hey, I've been surprised before.
 
This is precisely the reason I said earlier that I'm skeptical until we see benchmarks outside of an AMD-controlled environment. AMD have become notorious for doing this sort of thing.
 
Nice to see the card can also go single slot under water. And with the tiny PCB, that can make for some nice small builds at a much more affordable price than the Nano.
 
Back
Top