Regarding the RX 480 AOTS benchmark at Computex

[deviant88]

please everyone just stop and wait for the cards, nvidia has lied a ton about the gtx 1080 running cool, being faster than 2x 980 etc (maybe in cherry picked game or something) so lets wait for AMD cards, i personally think 2x 480's are not faster than 1x 1080 in every single game with 50% headroom left to spare that would make 1x 480 equal to 1x 1080 which makes no sense, the AOS game is an AMD partenership title i do believe its also cherry picked but somewhat accurate lets just wait to see the reality

[Prysin]

22 minutes ago, deviant88 said:

please everyone just stop and wait for the cards, nvidia has lied a ton about the gtx 1080 running cool, being faster than 2x 980 etc (maybe in cherry picked game or something) so lets wait for AMD cards, i personally think 2x 480's are not faster than 1x 1080 in every single game with 50% headroom left to spare that would make 1x 480 equal to 1x 1080 which makes no sense, the AOS game is an AMD partenership title i do believe its also cherry picked but somewhat accurate lets just wait to see the reality

it is only 50% headroom in scenes with extreme amounts of CPU workloads.

[Trixanity]

33 minutes ago, xAcid9 said:

Normal batches mean less draw calls mean less work for the CPU to feed the GPUs so i don't understand why you think Normal batches become "CPU" bottleneck instead of GPU bottleneck.

 

 

Anyway this thread go derailed pretty fast since it's clearly the 1080 is at fault here. 

Source: AMD. They say the normal batch is CPU-bound. I suppose it's because the GPU utilization is at 51% hence the GPU isn't doing much work. An increase in draw calls usually means lessening of CPU-bound scenarios. So that would indicate that an increase in draw calls reduces the CPU bottleneck. I guess this also depends on what one perceives as a bottleneck. Is the GPU being choked by a lack of draw calls and does that mean the GPU is the bottleneck? Or is it the CPU which is not providing the performance (or draw calls) necessary to keep the GPU occupied? I would personally consider the CPU a bottleneck in this scenario. Or perhaps I'm wrong and the GPU is unable to distribute the work properly to achieve maximum utilization when it doesn't receive enough draw calls. If that is so, perhaps a driver update can fix that.

 

In either case: you'll have to take this one up with AMD. It's their statement.

[xAcid9]

47 minutes ago, Trixanity said:

In either case: you'll have to take this one up with AMD. It's their statement.

Even if that doesn't really make sense?  

 

I read @Thracks post in Reddit, yeah less GPU utilization mean it could be bottleneck by the CPU but why the CPU got bottleneck in less demanding scenario in the first place?

[Prysin]

32 minutes ago, xAcid9 said:

Even if that doesn't really make sense?  

 

I read @Thracks post in Reddit, yeah less GPU utilization mean it could be bottleneck by the CPU but why the CPU got bottleneck in less demanding scenario in the first place?

driver overhead

[Trixanity]

15 minutes ago, xAcid9 said:

Even of that doesn't really make sense?  

 

I read @Thracks post in Reddit, yeah less GPU utilization mean it could be bottleneck by the CPU but why the CPU got bottleneck in less demanding scenario in the first place?

I'll assume the experts know what they're saying although blind trust is never a good thing

 

Well, hard to say. All I can do is speculate since I'm not exactly a software engineer or anything of the sort. It is a conundrum. So I'll try to come up with something: 

 

The game is still doing lots of CPU tasks (physics, AI, etc) despite the graphical fidelity going down. So the GPU is doing less despite the CPU still being able to feed the GPU adequately because the GPU tasks are smaller. However, even if we assume that the CPU tasks are still reduced (as in normal batch being less intensive on both CPU and GPU versus heavy batch), the CPU is still constantly busy, so we're in a scenario where increased CPU power (eg overclocking, more cores) could increase the FPS to reduce the bottleneck (potentially). Well, depends on whether we consider that a bottleneck. Let's just say the performance/FPS in this scenario is limited by how fast the CPU does things, not the GPU because the GPU is doing its job without breaking a sweat and increasing GPU power would not increase the FPS because the system is not lacking GPU power, but CPU power in order to improve performance.

 

Since AotS is an RTS: many units, complex AI, dynamic weather (don't know if AotS has these things but I'm guessing it's probable) etc. is very demanding even with for example less units on screen (or off screen) in a less intensive benchmark would still be CPU limited with modern CPUs. However with modern APIs and modern programming techniques the bottleneck might be lessened or removed by proper CPU scaling (clock speed and cores).

 

It could be that or just driver overhead as suggested

 

Hopefully my post is somewhat coherent and not entirely wrong.

[bVinyls]

I'm still gonna put a 480 in my system, It's cheap, and good from what I've seen. 

[Entropy]

On 6/2/2016 at 9:58 AM, AmbarChakrabarti said:

 

$400 GPU setup beats a $600 GPU! Holy Balls!

And you only have to worry about crossfire issues. I went for two 970s instead of a 980. Sli never again. One 1080 is still going to be objectively better than two of any cheaper card in sli for that reason alone. 

 

It's impressive that they're selling it at such a low price,  but one card for me. 

 

 

Disclaimer: I am aware of the difference between sli and crossfire. Both have undeniably annoying issues in many titles. 

[rattacko123]

On 02/06/2016 at 11:56 PM, Fulgrim said:

GTX 1080 is incorrectly executing the terrain shaders

I liked the incorrectly executed terrain shaders tho 

The image on 480s looked kinda bland

[The Flying Sloth]

On 6/3/2016 at 2:51 PM, App4that said:

http://www.fudzilla.com/news/graphics/40808-amd-radeon-rx-480-crossfire-rendering-differently-vs-gtx-1080

https://www.reddit.com/r/Amd/comments/4m692q/concerning_the_aots_image_quality_controversy/

[Xsilent(X)]

On 3-6-2016 at 11:00 PM, Entropy said:

And you only have to worry about crossfire issues. I went for two 970s instead of a 980. Sli never again. One 1080 is still going to be objectively better than two of any cheaper card in sli for that reason alone. 

 

It's impressive that they're selling it at such a low price,  but one card for me. 

 

 

Disclaimer: I am aware of the difference between sli and crossfire. Both have undeniably annoying issues in many titles. 

 

Maybe to you but i never had experienced isseus with SLI or Crossfire on. and i have a lot of games

[Jahramika]

On 6/2/2016 at 10:11 AM, No Nrg said:

I still think it's a bit ridiculous that people continue to use this game no one even plays as a benchmark in performance and call it definitive proof of one being better than the other.

 

Show me benchmarks from a whole list of games with one beating the other consistently across a majority of the list and maybe I won't call bullshit.

Well AOTS is the only ground up Dx12 title to date which gives it street cred. What other Dx12 title can pair a Nvidia and Amd gpu and actually benchmark it.

[Cryptonite]

On 6/3/2016 at 2:29 PM, TidaLWaveZ said:

 

I'm sure it is.  I'm not out to bash the RX 480, I don't see any thing wrong with it.  I honestly prefer a single card after a few lackluster experiences with SLi/Crossfire.  For my situation I think the extra $140 for the GTX 1080 is justifiable, and I also honestly think that the two RX 480 solution will not end up being comparable to the 1080(though I could be wrong).

Yes, I'm sorry I didn't make it clear that I meant that the 480 is a sweetspot single GPU solution. if it performs around the same as a 980. I didn't mention or mean to mention CF. I meant to just include that a single 480 might be good enough for a lot of people and that people wouldn't need to even consider buying anything more expensive unless they planned on a 4k build, unless superbly more demanding games came out which I doubt since my 980 is still more than capable to max every single game I play at 1080p at ~ 60+ fps. I even DSR to 4k on a lot of my games just for fun. (the difference is minimal at 1080 since MSAA etc are adequate.) 

 

So I'm super excited for how amazing the performance might be for the $200 - $300 range and would expect a lot of people that were even planning on buying a high end solution for 1080p to fall back to this option if it delivers on it's promised performance.

[Notional]

Officially confirmed that NVidia fracked up their shitty drivers yet again:

 

 

[shdowhunt60]

1 hour ago, Notional said:

Officially confirmed that NVidia fracked up their shitty drivers yet again:

 

 

This seems to be a recurring theme this year: nVidia rushing out drivers of dubious quality.

[Notional]

3 minutes ago, shdowhunt60 said:

This seems to be a recurring theme this year: nVidia rushing out drivers of dubious quality.

And people blaming AMD for the result 

[suicidalfranco]

21 minutes ago, Notional said:

And people blaming AMD for the result 

Just LTT drama things 

[Valentyn]

2 minutes ago, suicidalfranco said:

Just LTT drama things 

Sadly it was all over on many prominent tech forums and reddit; so many people claiming AMD were "obviously" running at lower graphical settings and lying.

[suicidalfranco]

2 minutes ago, Valentyn said:

Sadly it was all over on many prominent tech forums and reddit; so many people claiming AMD were "obviously" running at lower graphical settings and lying.

LTT is the only forum i visit, don't even know what's reddit ¯\_(ツ)_/¯

[Fulgrim]

13 minutes ago, suicidalfranco said:

LTT is the only forum i visit, don't even know what's reddit ¯\_(ツ)_/¯

Pure cancer, please avoid it.

[GlassBomb]

1 minute ago, Fulgrim said:

Pure cancer, please avoid it.

Some would say the same about LTT. It all depends on who you are asking.

[Notional]

@Fulgrim Mind updating your post with the twitter update?

 

3 hours ago, Notional said:

Officially confirmed that NVidia fracked up their shitty drivers yet again:

 

 

 

[Fulgrim]

2 minutes ago, Notional said:

@Fulgrim Mind updating your post with the twitter update?

 

 

will do fam

[zMeul]

1 hour ago, shdowhunt60 said:

This seems to be a recurring theme this year: nVidia rushing out drivers of dubious quality.

what are you talking about?

 

AMD used pre-release drivers instead of the publicly available 368.25 WHQL, since May 26th

[vanished]

6 minutes ago, zMeul said:

what are you talking about?

 

AMD used pre-release drivers instead of the publicly available 368.25 WHQL, since May 26th

well there was the whole 364 fiasco... that's what I'm thinking