[Guru3d] Hitman (2016) DirectX 12 updated benchmarks review

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

seitur

Senior member
Jul 12, 2013
383
1
81
Its a trend? I must have missed it.

DX12 have everything against it. And its plain simple due to economics and time. DX12 looks good on paper, but its designed for an utopian PC world. Or a console world.
AAA gaming generally needs graphical increase over and over to keep the model afloat and at the same time it does need big sales.

With changes to semiconductor industry, increasing cost and depletion of finances of average consumer this is becoming an growing issue.

In last two decades solution was to simply throw more and more hardware at the problem. Consumers had more money (+ fast increasing debt) and shrinkage of process was bringing cheaper transitors over, so it all worked.


It does not work anymore.

That is why one of possible solution is software. DX 12 or/and Vulkan can potentially allow for graphical fidelity to keep increasing in times of slower hardware advancement and consumer adoption.

Sure the downside is increased cost of software developement, but there is not like there is lot of choice in the matter.


Unless either:

a) economic situation will drastically change, rebound and general population will start to have more money

b) semiconductor industry will shrink cost by alot and thus hardware will get cheaper. Will it happen though? Will EUV come and it's effect will be strong enough?


Of course there is scenario, in which current model for AAA gaming industry collapses which is certainly possible, but I would not bet on it as much as some do.
 

Adored

Senior member
Mar 24, 2016
256
1
16
So you guess on JPRs categorization. While you refuse Steam numbers?

If there was a big variance you would see it month to month. So again, nothing wrong with the Steam numbers :)

How many Steam accounts are you basing the numbers on? 125 million?

I have 3 different Steam accounts and only one even gets used. I haven't been offered the survey in years either even with multiple OS changes, formats and a bunch of new mobos.

There are 4 PC's in this room I'm currently in, 3 of which have Steam installed and yet none has ever taken the survey.

I think it's quite likely that an awful lot less than 125 million people have taken the Steam hardware survey, and a lot more likely that the numbers are simply based on the people who have taken the survey, not the total number of people with Steam accounts.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
AAA gaming generally needs graphical increase over and over to keep the model afloat and at the same time it does need big sales.

With changes to semiconductor industry, increasing cost and depletion of finances of average consumer this is becoming an growing issue.

In last two decades solution was to simply throw more and more hardware at the problem. Consumers had more money (+ fast increasing debt) and shrinkage of process was bringing cheaper transitors over, so it all worked.


It does not work anymore.

That is why one of possible solution is software. DX 12 or/and Vulkan can potentially allow for graphical fidelity to keep increasing in times of slower hardware advancement and consumer adoption.

Sure the downside is increased cost of software developement, but there is not like there is lot of choice in the matter.


Unless either:

a) economic situation will drastically change, rebound and general population will start to have more money

b) semiconductor industry will shrink cost by alot and thus hardware will get cheaper. Will it happen though? Will EUV come and it's effect will be strong enough?


Of course there is scenario, in which current model for AAA gaming industry collapses which is certainly possible, but I would not bet on it as much as some do.

The only way you are going to get more graphics performance via the API is to go something like FP16 instead of FP32. Something that is coming soon. :)

DX12 is mainly for the CPU part.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
How many Steam accounts are you basing the numbers on? 125 million?

I have 3 different Steam accounts and only one even gets used. I haven't been offered the survey in years either even with multiple OS changes, formats and a bunch of new mobos.

There are 4 PC's in this room I'm currently in, 3 of which have Steam installed and yet none has ever taken the survey.

I think it's quite likely that an awful lot less than 125 million people have taken the Steam hardware survey, and a lot more likely that the numbers are simply based on the people who have taken the survey, not the total number of people with Steam accounts.

You could have a point back when Steam was a million or 5 users. Not today, basic statistics.
 

kraatus77

Senior member
Aug 26, 2015
266
59
101
The only way you are going to get more graphics performance via the API is to go something like FP16 instead of FP32. Something that is coming soon. :)

DX12 is mainly for the CPU part.
FP16 is already in gcn. so as with CUs nvidia is catching up there too. :thumbsup:
 

airfathaaaaa

Senior member
Feb 12, 2016
692
12
81
You could have a point back when Steam was a million or 5 users. Not today, basic statistics.
still you miss the point the survey cannot be taken seriously when it doesnt have the competitors cards on the survey
unless you are suggesting that no one bought any 3xx card and the 7900 series is ruling the place
http://store.steampowered.com/hwsurvey
do you even hear your self before you post something or you just go and bable around just to create white noise?
 

kraatus77

Senior member
Aug 26, 2015
266
59
101
According to steam, nvidia doesn't have 80% dgpu market share, but i guess that data doesn't count.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
According to steam, nvidia doesn't have 80% dgpu market share, but i guess that data doesn't count.

Installed base vs shipping. AMD is losing installed base on Steam.

They went from 29.33% to 25.5% the last 18 months.
 

Erenhardt

Diamond Member
Dec 1, 2012
3,251
105
101
This is just circumstantial evidence, not proof, in fact we can use it to prove exactly the opposite of the point you are making.

Async compute helps improving performance when part of the GPU is not fully utilized by scheduling independent work, so a very efficient GPU might see little to no improvement with async compute, simply because it doesn't have lots of unused resources in the first place. I have no idea which case applies to Maxwell (perhaps both in different situations) but it doesn't prove what many on this forum think it proves.

No.

You just showed that you have not idea what async compute is and how dx11 rendering pipeline works.

Async compute does not magically wake up sleeping cores, like those lazy peons.

It enables things that worked in a serial manner, despite processing in dedicated units, to be executed in paralel.

So from dx11: Hey X go do A thing, when you are finished let Y know to do B, then Z will do C,
async give the ablitly to go: X does A thing while Y does B and helps X do A thing, and while they are doing their own thing z does C in the background.
 
Last edited:

Adored

Senior member
Mar 24, 2016
256
1
16
I think we can safely say that Roy's Numbers were right and that the 970 is the start of the "Enthusiast" bracket for JPR, because he's using their numbers.

SJ8ucWp.jpg


More than half of those will be 970's I'd wager.
 

tential

Diamond Member
May 13, 2008
7,355
642
121
I think we can safely say that Roy's Numbers were right and that the 970 is the start of the "Enthusiast" bracket for JPR, because he's using their numbers.

SJ8ucWp.jpg

So Roy has a JPR report.
Shintai has Steam numbers that he swears are out there, but refuses to post....

So you guess on JPRs categorization. While you refuse Steam numbers?

If there was a big variance you would see it month to month. So again, nothing wrong with the Steam numbers :)


Just trust him, the numbers are there he will discredit Roy's numbers, but give you no numbers in return.
They're somewhere in Steam land.....
 

Piroko

Senior member
Jan 10, 2013
905
79
91
Ah, the beloved Steam Hardware Survey. Also known as:

An opt-in survey among 100 train enthusiasts and 100 common people has found out that 90% of participants love trains. Participation level was 50%.

Opt-in statistics in which participants know the topic before opting in are skewed towards enthusiasts. They can show trends, but not absolute numbers.
 

renderstate

Senior member
Apr 23, 2016
237
0
0
No.

You just showed that you have not idea what async compute is and how dx11 rendering pipeline works.

Async compute does not magically wake up sleeping cores, like those lazy peons.

It enables things that worked in a serial manner, despite processing in dedicated units, to be executed in paralel.

Who talked about sleeping cores? Try reading again what I wrote since we are saying the same thing: to be executed in parallel == scheduling *independent* work for (hopefully concurrent, if resources are available) execution.

The point I made is fully valid. Observing little to no perf improvement with async compute *does not* prove async compute is not working. You need to know how resources are being used to make such statement.