AMD Ryzen - Games in 1080p can improve more than 35% !

[zMeul]

1 minute ago, Megah3rtz said:

Basicly it's about how to optimize your build.

right, OCing is "optimizing" your build 

[Megah3rtz]

Just now, zMeul said:

right, OCing is "optimizing" your build 

k

[zMeul]

1 minute ago, Megah3rtz said:

k

if that's the only thing you can reply with .. we're done here

[Megah3rtz]

Just now, zMeul said:

if that's the only thing you can reply with .. we're done here

 You don't even realize what I'm pointing out, I guess so ! 

[Organized]

1 minute ago, zMeul said:

if that's the only thing you can reply with .. we're done here

Ryzen threads are done when you start posting in them.

[leadeater]

35 minutes ago, WereCat said:

I think that many reviewers keep their games on dedicated SSD and just switch it between PC.

Imagine how more time it would take to benchmark everything if they had to do a clean install of every single game on every single system.

And I agree with this game being in minority of games that actually benefit from this.

Yea I'd do the same, but in the case of Ryzen new everything would be the way I would do it.

[Morgan MLGman]

40 minutes ago, zMeul said:

if that's the only thing you can reply with .. we're done here

I'm not sure if you're doing it on purpose or you really didn't get the point but he meant that switching HPET to OFF in the "Balanced" power plan gave almost a 20 FPS boost which translated into 18.5% of performance bump, the rest of the settings were unchanged. Don't you think this indicates an issue with the power plans in W10 and the HPET when paired with Ryzen CPUs? Just answer honestly to that question.

 

[LAwLz]

What the test actually shows:

Do a clean install when you change platform, or else your performance in some games might be lower than they should.

By the way, overclocking and/or faster RAM gives higher benchmark scores.

 

What people will read:

OMG WITH THIS ONE TRICK RYZEN'S FPS IN 1080P BENCHMARKS INCREASES BY 35%!

INTEL IS FINISHED! AMD WON!

 

 

I can't wait for Reddit to find this and then parrot it over and over and over. It will be a fun couple of weeks where the game developers gets all the blame for Ryzen's (comparatively) poor gaming performance...

Because clearly people can't just accept the fact that Ryzen 7 isn't a CPU meant for gaming.

 

Interesting tests though. Seems like a really bad idea to handle things the way F1 does, but it makes me wonder if more games does it that way.

[Morgan MLGman]

51 minutes ago, LAwLz said:

What the test actually shows:

Do a clean install when you change platform, or else your performance in some games might be lower than they should.

By the way, overclocking and/or faster RAM gives higher benchmark scores.

 

What people will read:

OMG WITH THIS ONE TRICK RYZEN'S FPS IN 1080P BENCHMARKS INCREASES BY 35%!

INTEL IS FINISHED! AMD WON!

I said exactly the same thing about what Nvidia did with their DX12-improving driver update, it's just hilarious This misleading bullshit is common practice, sadly.

 

They presented those number as a relative comparison to launch drivers performance, launch game version, ran at 4K with a GTX 1080, this heavily suggests to people who don't know as much as we do about this stuff to just go like this:

"Oh, I heard Nvidia increased DX12 performance in my GTX 970 by 33%! So great!" When he gets 1-2FPS in reality... Except Hitman, performance in this game really improved.

 

Funniest that they called it a "Breakthrough DX12 driver performance"

GeForce GTX 1080 @ WQHD	GeForce 378.66	GeForce 378.78	Difference %
Deus Ex : Mankind Divided	69	70	1,43%
Hitman (2016)	92	104	11,54%
Rise of the Tomb Raider	93	94	1,06%
Gears of War 4	95	94	-1,06%
Ashes of the Singularity	86	88	2,27%
Sniper Elite 4	89	89	0,00%
Battlefield 1	95	96	1,04%
The Division	72	75	4,00%

[SpaceGhostC2C]

3 hours ago, kladzen said:

 Now manually editing the game file to correctly look at a 8C/16T CPU the game gains even more performance..

 

a wooping 35.53% more performance in 1080P gaming!  From 98.5 fps to 133.5 FPS 

 

 

According to the table, correctly presenting the CPU as 8c/16T only meant a 2.89% increase. The bulk of the gains come from clean-installing everything and using the best power management options (+26.8% combined). Then it shows that faster RAM is relevant. Finally it shows OCing the CPU is kind of irrelevant for this game, despite what some trolls may say (mere 2.36% increase).

 

So, the bad news is that 16c vs 8c/16T is practically irrelevant. The good news is that it was game-specific anyway, while there are substantial potential gains immediately available to users: you don't need to wait for any optimizations or patches to do clean installs and set HPET and/or the power plan that works best with your CPU.

It also serves as a sanity check: any reviewer should be able to replicate the 126.75 result by using the same settigns (since all of them were using as fast RAM as their buggy boards allowed them to), otherwise either AMD numbers are wrong or they are being lazy and re-using installs, skipping Windows configuration, etc.

[Okjoek]

At this point the whole world can improve up to 35% I think.

[DocSwag]

1 hour ago, LAwLz said:

What the test actually shows:

Do a clean install when you change platform, or else your performance in some games might be lower than they should.

By the way, overclocking and/or faster RAM gives higher benchmark scores.

 

What people will read:

OMG WITH THIS ONE TRICK RYZEN'S FPS IN 1080P BENCHMARKS INCREASES BY 35%!

INTEL IS FINISHED! AMD WON!

 

 

I can't wait for Reddit to find this and then parrot it over and over and over. It will be a fun couple of weeks where the game developers gets all the blame for Ryzen's (comparatively) poor gaming performance...

Because clearly people can't just accept the fact that Ryzen 7 isn't a CPU meant for gaming.

 

Interesting tests though. Seems like a really bad idea to handle things the way F1 does, but it makes me wonder if more games does it that way.

It's not even a true 35% over what reviewers are getting anyways. Most reviewers have tested it with HPET on and many did it with a fresh install. Most reviewers had memory OCed around 3000 MHz or so amd showed both OCed and not OCed results (not to mention most people can only get 4 ghz, not 4.1 ghz). Changing it from 16 cores to 8 cores only gave a 2% or so boost in fps, and I'm willing to bet this is a very isolated case, or amd would have shown more games that you can do this on. So really, this shows that in one game out of the multitude of games tested, Ryzen can get a 2% boost in performance. Wow.

 

People need to just face it, Ryzen is not a CPU for pure gamers. But it is for content creators. Or someone who games and does content creation. 

[HalGameGuru]

I'm confused by the reusing old OS trope. Is SysPrep no longer a thing?

 

I'd like to see them take that initial setup and ONLY change the F1 ini file and see what the change in performance is.

[rcarlos243]

[imreloadin]

What I found the most interesting was this:

 

You get an 18.5% increase by turning off HPET with everything else being the same.

By just changing the power plan to High Performance and leaving HPET turned ON you get an increase of 19.8% from the baseline.

 

Now I know this kind of removes the dead horse that zMeul keeps liking to beat by saying "But AMD helped create HPET!" everywhere but now users can just leave it on and switch their power plan to High Performance and they get more performance. This is with the same RAM speeds and NO overclock. By literally just going into the control panel and switching the power plan you can get an additional 19.8%

[Misanthrope]

9 hours ago, leadeater said:

Also how many reviewers used existing OS install versus clean install, I have no idea

Neither do I. Yet you assume that something that should be common practice among serious reviewers is something they all collectively missed in other to skew everything against AMD?

 

This suggestion is pretty worthless: Instead of assuming this is a methodology problem common among all reviewers somehow why don't we take a closer look at the OP and realize that AMD rushed a product without any thought put into optimizing or figuring out how to make it work with existing software?

 

That seems like a far more plausible explanation: AMD slipped into gaming CPU irrelevance for the past 5 years so obviously most games available today didn't even think about optimizing for them at all, this is a problem that AMD will need to fight against but I don't see this ridiculous implications about reviewer methodology affecting things.

[leadeater]

17 minutes ago, Misanthrope said:

Neither do I. Yet you assume that something that should be common practice among serious reviewers is something they all collectively missed in other to skew everything against AMD?

Um no? I assumed all reviewers did it properly and all results shown were correct.....

 

9 hours ago, leadeater said:

Going with the clean install assumption all the reviews should have accurate performance results so nothing is going to change there, games that implement similar hardware profiling should be retested with clean installs and new settings.

Read all content of posts, don't just stop mid way through otherwise your reply might not reflect what was actually written.

[Misanthrope]

20 minutes ago, leadeater said:

Um no? I assumed all reviewers did it properly and all results shown were correct.....

 

Read all content of posts, don't just stop mid way through otherwise your reply might not reflect what was actually written.

If I have only one point of contention it is not a misquote. If it suits this new "rule" that wants to turn LTT Forums into endless, pointless walls of text and scrolling through unreadable replies I'll limit myself to bolding the parts I am replying to, you can assume I concede or have no comment to the parts I do not reply to then.

[SpaceGhostC2C]

48 minutes ago, imreloadin said:

What I found the most interesting was this:

 

(..)

You get an 18.5% increase by turning off HPET with everything else being the same.

By just changing the power plan to High Performance and leaving HPET turned ON you get an increase of 19.8% from the baseline.

Don't forget it's not keeping everything else equal, though: they moved form reused installs to clean installs, so you are seeing the combined effect.

[PCGuy_5960]

9 hours ago, Megah3rtz said:

When you use high speed DDR4, you are overclocking the infinity fabric. So @zMeul was right

[Hunter259]

1 hour ago, imreloadin said:

What I found the most interesting was this:

 

You get an 18.5% increase by turning off HPET with everything else being the same.

By just changing the power plan to High Performance and leaving HPET turned ON you get an increase of 19.8% from the baseline.

 

Now I know this kind of removes the dead horse that zMeul keeps liking to beat by saying "But AMD helped create HPET!" everywhere but now users can just leave it on and switch their power plan to High Performance and they get more performance. This is with the same RAM speeds and NO overclock. By literally just going into the control panel and switching the power plan you can get an additional 19.8%

That's not what it is saying. They turned HPET off (for some stupid reason) AND reinstalled windows and F1 2016. I mean this is some fucking terrible testing methodology, changing multiple things in one run. 

[PCGuy_5960]

1 minute ago, Hunter259 said:

That's not what it is saying. They turned HPET off (for some stupid reason) AND reinstalled windows and F1 2016. I mean this is some fucking terrible testing methodology, changing multiple things in one run. 

And a perfect way of hiding results 

[imreloadin]

28 minutes ago, SpaceGhostC2C said:

Don't forget it's not keeping everything else equal, though: they moved form reused installs to clean installs, so you are seeing the combined effect.

 

5 minutes ago, Hunter259 said:

That's not what it is saying. They turned HPET off (for some stupid reason) AND reinstalled windows and F1 2016. I mean this is some fucking terrible testing methodology, changing multiple things in one run. 

Has anyone ever tested to see what kind of FPS improvement you get by going from an old install of the game & windows to clean installs of them? I'm really curious to see what the results are.

[SpaceGhostC2C]

7 minutes ago, Hunter259 said:

That's not what it is saying. They turned HPET off (for some stupid reason) 

Apparently you haven't been following the topic: they've been recommending this, and explaining the reason, from day release day (essentially, software-based clock management interferes with the CPU's own clock management). That's why using "High performance" removes the need to switch HPET off (or at least in this test).

[Hunter259]

19 minutes ago, SpaceGhostC2C said:

Apparently you haven't been following the topic: they've been recommending this, and explaining the reason, from day release day (essentially, software-based clock management interferes with the CPU's own clock management). That's why using "High performance" removes the need to switch HPET off (or at least in this test).

It's turned off due to a sleep bug more or less. The bug causes Windows timing to slow down which artificially inflates benchmark scores. HPET also can cause a very small hit to performance in certain cases. AMD just said to do that due to the bug and trying to gain every last drop of performance they can.