As a scientist, I assure you that science isn't just about getting paid to prove a theory. There are certainly instances where that is true but for the most part science should be blind to opinion. However, science should be open minded, history is filled with examples of scientists who rejected ideas without testing them (and eventual become accepted). Over skepticism leads to over sights. Thus, an open skepticism is best for science to move ahead. Belief in established protocols is another problem with science, in reality not science, but the fact that scientists are human. Humans do human things even those that practice science. The fact is we all have bias.
Now as far as benchmarks are concerned. While my degree is not in electronics, nor physics, but I love both topics. I have learned a lot from reading anandtech and various tech sites. What I have come to understand is that benchmarks are far from being science but generally thought to be good enough to test current technology. Ryzen is new technology and will have some similar strengths and weaknesses that current technology has but in some cases, it may do some operations differently or even do things that hasn't been done before. That does not mean on its own merits that it is slower or faster. Since benchmarks test different things according to its designer they can give wildly different results with the same processor. In other words, benchmarks are useful but not particularly scientific. The same can be said about experiences, they too are useful. Why do you think people that are experienced at something are far more efficient than someone who isn't? New people coming into the mix may stir things up and bring new ways, but in general they are less efficient. So I would argue that a persons experience also has relevance. Benchmarks are not necessarily a good predictor of what you might experience. If we ever develop a truly scientific benchmark that treats all players the same then it might be able to predict your perception.
Yep, I saw this after my first post. Maybe this Ryzen/Nivdia phenomenon is only in a few games. Something that should be more easily patched. I'm sure there will be more investigations since the "can has been opened" but it may not be a far reaching thing. Time will most likely tell.
It seems pretty obvious that Rise of the Tomb Raider is a special snowflake of a game with plenty of problems. And trying to prove a point by looking really hard for a single part of a game that might correlate your opinion is just not very interesting. No secret that AMD does better at DX12, is it? Next generation Nvidia will probably have proper async compute, and then there will be more and better DX12 games around to use it. Hopefully the Radeon team over at AMD has managed to make a GPU that competes in the high end by then.
And data is data. If a reviewers benchmarking methodology is sound, then the data is probably sound too. Subjective opinions are not very interesting are they, as everyone has a ton of them. I wish Jim over at AdoredTV would do some frame time analysis ass it tells more of the story how games perform om different hardware.
Wow, that explains a lot why the CPU bound section of Rise of the Tomb Raider that AdoredTV tested with Nvidia doesn't scale as well as it should in DX12 compared to AMD on Ryzen, and explains the discrepancies in CPU loads. Thanks for the showing that.
@wendell I know you're doing your own little test of the Nvidia vs AMD on Ryzen platform so the video @MichaelLindman posted might interest you.
Anyone got time to crunch some numbers? I can share my raw dataset with a select group of guys.. we don't have a lot of people so I would like to have a bnuch of people look it over and let me know. @gtbtk -- the other thread has a quickie version of my own RotTR testing and even 7700k vs 1800x @ 4.1 7700k still had the edge slightly. I can repeat some of the tests though if you think that would help.
I have both DX11 and DX12 results for 1080(ti) and Asus Strix Fury on both platforms.
No one has had time to analyze the results yet -- just a cursory glance.
There is no guarantee that nvidia will support async properly at all.
As long as they can stop companies from using by not implementing it in games for WHATEVER reason, then amd lose out as the majority of developers wont bother with it... or go with nvidia's hobbled solution which has a limited queue depth to a fraction of the load of what most amd ACE engines are capable off.
Lets be clear what async compute does in real world terms.. it mostly cuts down on the need for game launch driver optimization as the ACE engines are now doing the job that the 'Game Ready' driver developer used to do at RUNTIME and would close the gap between 'amd vs nvidia' gaming benchmarks on newly launched titles by a fair degree.
This has been nvidias edge for a lot of titles as it can afford to micro manage day one drivers for game launches, amd can't. It's almost like the rich staying rich because they can 'pay to win'.
All Nvidia have to do to sandbag asynchronous compute is not use it, give me some hope and tell me I am wrong but sadly I don't think I am.
I don't think they have a choice really. The consoles do it and have done it for quite some time now. I don't think Nvidia can really skimp on this and not get sidelined. Besides, I think they can make more money supporting it properly.
You make great points here. It's perfectly fine to use other's benchmarks as long as they are in the margins with what the majority have discovered. In fact, it's the aggregate scores that put the spotlight on those who are slanted with bias.
