15.4 vs 1018.1 "Vbs" Benchmarks, Quad and Dual Core

I apologize for the tone. I should have picked words that are less inflaming.
But that ..gentleman has been repeatedly accusing me of lying, of being a troll, a noob, and a shill.
I have always posted public proof for any of my claims.
I am actually a computer engineer with experience (mild albeit) in designing chips in verilog, and a hardcore hardware enthusiast.
Not once did he want to engage in a productive discussion and when I pm'ed him (one and only initial message) with further proof about a subject we could not agree on, he, once again, publicly shamed me and told me that he blocked me. It's very upsetting to encounter such a ..pleasant individual in an otherwise fantastic community.
I would not even bring it up, but seeing PMonster attack PrMinisterGR truly upset me.
Let's keep it civil from now on.

 

I think for now it doesn't support crossfire

 

Anyone notice any improvements with GTA V and crossfire?
If so i will try them out.

 

We've said countless of times there's improvements, the ms has improved, hence your gameplay will be smoother. You'll definitely notice in game.

 

I did. I would try them out.

 

Thanks guys.
I wasn't following i just checked the first page and didn't see any GTA graphs.
Yeah smoothness just what we need :banana:

 

New results on the first race of the touring cars

 

Alright, awesome. Install them then.

Nice gains from win10 lol Obscure

 

Yep, you are quite right, any GPU/driver combo limited in draw call performance will take a hit on complex scenes, hindering framerates and resulting in inconsistent performance.

Thanks again for taking the time to bench these. I think there's some dispute about the dual vs quad core scenario, which is pointless. Of course the performance hit on a dual-core will be a combination of driver overhead plus game engine ineffectiveness, but that doesn't make those results trivial. In fact one just has to look at the Steam Hardware Survey, to see that there's more dual- than quad-cores between Steam users (they don't publish the sample size, but I'm guessing the confidence intervals with be within +/- 1%), making this research quite important.

Finally, I've compiled all your results (except for Civ V) in the following charts (also decided to go with your suggestion for inverting the variance chart):



I'd say the results are pretty consistent to what's been discussed already.

Overall, the 1018.1 driver seems a good performer with mostly better minimum FPS values and also lower variability in quad-core. For dual-core it's more or less a draw against the 15.4s.

 

I... I love you.

 

Mate, don't try and bullshlt a bullshltter.

The only links posted on the net lead back to either youtube accounts which a couple of you have, or to Digital Foundry, all recycling the same material.
There must be at least 30 Digital Foundry links on Guru3D in various threads.
AnandTech aren't even in the picture except for an related article Ryan did over a year ago.

And as for youtube I have said something, just not to you.

Of course they run in "1.3 mode", W8.1 has a WDDM 1.3 kernel. W10 comes with WDDM 2.0.

Drivers can't make the kernel do something it wasn't coded for, so you're mistaken. If it was that simple we'd all be looking forward to DX12 on W7. You can't make something work which isn't there.

Drivers act as communication bridge between the OS and any attached hardware devices, that's all. W7 had a partial update but it reqiured an OS patch, not drivers.

I didn't mention DX12. I spoke in regards to WDDM 2.0, not DX12.

And fwiw I'm not debating drivers can improve or hamper performance.
I just don't believe you've presented anything conclusive, and even if you did, then what? Average frame rate imrproves by 3-4 fps on dual cores, making those 2 owners happy.
I've seen the comments on other forums, most people just aren't interested, even AMD owners.

You'd need to attach an SDK debugger to know for sure it was drivers anyway......which I don't see happening.

 

Any chance you might test some more titles? Like Unity or GTA etc? It would also be nice if me or (even better) Vbs can get our hands on your frametime dumps

What I find more intriguing about these drivers is that DX9.0 performance is affected too. Maybe I have to dig up my old Witcher 2 copy and give it a run. It is the best looking DX9.0 game I know. There never was a debate about the dual vs quad core scenario, just people who didn't know how to read.
Thanks for the Steam Hardware survey too, I should have thought to look into it too. My question about this driver is if they employ some kind of frame pacing on single gpu scenarios too. The worrying part is that their lower-end performance is getting worse.
I'll probably dig up the Witcher 2 next week, Need for Speed Rivals, and Ryse. If you have any more ideas, give me something.

 

I saw the youtube links. They all pointed to 1 source. Digital Foundry. First I heard of em. are they legit?

 

Idk, but the benchmarks seem pretty legit, which is what matters. Its kinda obvious that a small site would be working on this compared to ones like techspot, anandtech, and g3d.

 

The API overhead test and the comparison between 15.4 drivers and modded VBS drivers in W8.1 are available to everybody who want to test them in their own computer.

It's not a matter of "believe" or to have "faith" in Digital Foundry, PR or VBS; It doesn't even matter WHO did the benchmarks if they can be repeated by anyone.

The benchmarks results are real, the differences are noticeable AND can be reproduced and checked by anyone using a method developed "some" centuries ago by VERY WISE people.

This method is called: Scientific method...

 

DigitalFoundry is no small site whatsoever, "In-depth technical analysis of games and gaming hardware at Eurogamer.net". http://www.eurogamer.net/?topic=digital_foundry

http://www.alexa.com/siteinfo/www.eurogamer.net Gobal Rank: 4,190

http://www.alexa.com/siteinfo/www.guru3d.com Gobal Rank: 7,874

Can we please stop this kind of unnecessary debate and focus on the topic at hand?

+1!

 

I used to think so. Until I started having a closer look at the benchmarks they were publishing...... If Techspot or HardwarePal are your go-to sites for benchtesting, then you'll like DF.

 

Man, calm down, go take a break, why are you so mad with people just trying to get their nerdy thoughts and findings out to the internet? Really what is your beef?

To the OP, thanks for the efforts man, all the data up to now makes sense.

 

You prob don't know but this has been dragging on for months, since the start of the year on different forums. If you check out threads on TPU or AT you'll see the members there got tired of hearing abiout it.

All the website images posted are from a single source... If it was such a problem why has no one else picked t up. Besides drivers are released all the time, if there were bad a set, it would only last til the next update......

Slo I don't have a beef, I'm being honest, Go read the original article yourself....

I used to think so. Until I started having a closer look at the benchmarks they were publishing...... If Techspot or HardwarePal are your go-to sites for benchtesting, then you'll like DF.



Edit: @TK: I'll be honest, after reading articles publshed by Digital Foundry, including one last year which kicked off this fiasco, I woudn't trust the guy to give me the time of day.



The testing was done by DF in a way which made the results meaningless. However that didn't stop someone trying to draw a line from the results to fit in with his angle. "Come, look at these pile of random numbers over here? They were put there by shilty AMD drivers".

If look around the tech sites u'll find miore than a few comments made about DF creredibility. The guy from DF posted here a couple of days ago in the DX11 thread, I got impression he was playing down the results from first article.... M


Last but not least, here is screenshot used by DF , it was posted quite a while back. Think it took mr Fox222 and myself about 2mins to see there something not quite right.....

 

Pill you claim that you are tired of us talking about a real problem, right?
Well i'm tired of you only having 1 proof against DF and keep posting it over and over and over again about the damn cables.

Do you know that the first SS the car is going a little to the left, and the center SS the car is going straight as the cables are, so its logically less aliasing.

Jesus! Go breath a little.
Its more than proved that nvidia have advantage on draw calls that flavours them with low end CPUs on DX11 games.
What? Let me guess! You are going to say that Star Swarm and 3dmark overhead test are buyed from nvidia or don't mean nothing, right?
Or its a benchmark and doesn't mean anythingm right? -.-

If you are tired of this discussion, just go to other threads.
I join in Guru3D because of this problem, and after that day, i only see the same proof from you and talking with nosense against anyone that prooves something.

You know.. Defending the lack of things from AMD its not get you payed and its not going to make AMD better, sorry!

IT took me 2 days to benchmark a nvidia and na AMD in my system, and the gains were visible in all games that i was severely bottlenecked.
Do you even have friends?
If you have, ask for a friend to lend you a nvidia then benchmark the games that you have GPU Usage lower than 99%, and benchmark it.
After that i hope that you can finally shut up or at least give some credit of all the things that had been posting here and in other threads.

And please Pill, Deal With It! Its real