Why does Ryzen 7 1800X performs so poorly in games?

juanrga said:
I hope this article clarifies my point and that of reviewers

http://www.techspot.com/news/68407-...ottlenecking-cpu-gaming-benchmarks-using.html
Click to expand...

Did you read the article or did you assume that even though championed 1080 low testing that it had a good point? I am not trying to be an ass about it and the guy makes good points and I fully understand why they are isolating the CPU in testing but the guy on there proved my point. Notice he stated in the benchmarks that he used that the 2 core Pentium matched the i7 at 1440p. That has always been my point. If your playing something with a GPU bottleneck its better to skimp on CPU and spend on GPU. Or in the case of Ryzen getting a pretty fast CPU with twice the cores/threads for the same cost. People don't play with a CPU bottleneck outside one or two rare upper end competitive environments. People buy better video cards and better monitors to get the best look out of a game. They buy a GPU upgrade to get better performance out of a new game. Whatever performance you had in the previous game is at worse the baseline in that game going forward (meaning if it was acceptable then it will be now) and if a CPU bottleneck is formed people just up the eye candy to match the performance they desired. It would be one thing if the gaming performance was indicative of full system performance, but we know that is not the case. It's not even about cores and clock speed. We know the compute prowess is there. Even if games don't get optimized for Ryzen, even if core usage doesn't grow, even if the Windows bugs don't get worked out, there isn't any evidence that this ~10%-15% at low res performance difference actually affects game play outside the aforementioned edge case.

Then you add the fact that if games were to use CPU's more efficiently and more often on other things besides the information that it hands off to the GPU. Meaning that even if performance is mostly driven by GPU but a beefy CPU is needed. Ryzen is in better shape.
 
OrangeKhrush said:
The aggregate is pretty good
Click to expand...

It games like an i5, which we knew since CanardPC tested the engineering sample.

001440.jpg


And SMT only improves things by 2--5%, which we have knew since Hardware.fr review. This is the same that happened with Bulldozer, when people was awaiting magic from a CMT-aware windows patches which only improved performance by 2--5% on average.
 
Presbytier said:
Also of note is the fact there are still GPU limited games even at 1080. The division is clearly hitting a GPU bottleneck there no other explanation as to why all the games are at the exact same frames even the Pentium is not far off.
Click to expand...

Right, with bottlenecks removed Ryzen must be between i3 and i5.

2017-03-06-image-16.jpg


2017-03-06-image-18.jpg
 
juanrga said:
It games like an i5, which we knew since CanardPC tested the engineering sample.

001440.jpg


And SMT only improves things by 2--5%, which we have knew since Hardware.fr review. This is the same that happened with Bulldozer, when people was awaiting magic from a CMT-aware windows patches which only improved performance by 2--5% on average.
Click to expand...
Wrong. And how old is that graph because its clocks don't look right with the released product (not to mention the 8370 is a 4.3 or 4.4/4.7boost). And windows updates for Bulldozer were far higher than 2-3% being the scheduler was loading modules before moving to other cores which has a huge impact. With Win10 they did it again when they started treating the FX as an HT enabled CPU. By Fixing that Issue, with CPU bound I guarantee the difference is far higher than 5% on my 8350.
 
juanrga said:
It games like an i5, which we knew since CanardPC tested the engineering sample.

001440.jpg


And SMT only improves things by 2--5%, which we have knew since Hardware.fr review. This is the same that happened with Bulldozer, when people was awaiting magic from a CMT-aware windows patches which only improved performance by 2--5% on average.
Click to expand...


Firstly windows tweaks cannot be predictive so rather don't go there given your history of getting it wring.

Correct it games in linear process like an i5 7600K which also beats an out the box 5960X, 6900K and 6950X in the same linear process. Add effectively scaled titles that work ie Doom(vulcan) and its a bloodbath for big cores
 
Lets assume Juanrga is right for once and AMD scheduler fixes things by 5%

2017-03-06-image-18.jpg


That will be 142.8 say 143 vs 154 about 8% difference in a game that cannot scale above 4 threads.

I can live with that, it is kind of how i expected it to be. Side by side my 5960X at 3ghz vs my review kit 1700 at ghz both CPUs teemed with Titan pascal GPU's are thread for thread the same
 
JustReason said:
Wrong. And how old is that graph because its clocks don't look right with the released product (not to mention the 8370 is a 4.3 or 4.4/4.7boost). And windows updates for Bulldozer were far higher than 2-3% being the scheduler was loading modules before moving to other cores which has a huge impact. With Win10 they did it again when they started treating the FX as an HT enabled CPU. By Fixing that Issue, with CPU bound I guarantee the difference is far higher than 5% on my 8350.
Click to expand...

That old graph is, as I explained, the Canard old review of an engineering sample. It put Ryzen gaming performance just behind an i5, and reviews of Ryzen like the last from TechSpot, which was linked in #676, which I was replying, show the same conclusion. The top Ryzen chip plays games like an i5

1080_All.png


We also know that Bulldozer patches for Windows improved average performance by 1--2%

AMD was pretty honest when it described the performance gains FX owners can expect to see from this update. In its own blog post on the topic AMD tells users to expect a 1 - 2% gain on average across most applications.
Click to expand...

http://www.anandtech.com/show/5448/the-bulldozer-scheduling-patch-tested

OrangeKhrush said:
Firstly windows tweaks cannot be predictive so rather don't go there given your history of getting it wring.
Click to expand...

Coming from you, that is tacit confirmation I am right also in this one. :)

OrangeKhrush said:
Correct it games in linear process like an i5 7600K which also beats an out the box 5960X, 6900K and 6950X in the same linear process. Add effectively scaled titles that work ie Doom(vulcan) and its a bloodbath for big cores
Click to expand...

People has been for years recommending a high clocked quad-core i7 for gaming. The problem with Ryzen is that it games worse than a 6900k at stock... and 6900k overclocks better than Ryzen.
 
That's an example chart to demonstrate GPU bottlenecks.

This was the actual benchmark.

Battlefield.png


It's well above the i3 and i5.

And what about BF1 MP testing where it really matters?

https://www.computerbase.de/2017-03.../#diagramm-battlefield-1-dx11-multiplayer-fps

It's not even close anymore, even faster than the 7700K.

K4GR5oD.jpg


People who say it's an i3 or i5 do so out of ignorance. No, not everyone play poorly threaded games.
 
juanrga said:
That old graph is, as I explained, the Canard old review of an engineering sample. It put Ryzen gaming performance just behind an i5, and reviews of Ryzen like the last from TechSpot, which was linked in #676, which I was replying, show the same conclusion. The top Ryzen chip plays games like an i5

1080_All.png


We also know that Bulldozer patches for Windows improved average performance by 1--2%



http://www.anandtech.com/show/5448/the-bulldozer-scheduling-patch-tested



Coming from you, that is tacit confirmation I am right also in this one. :)



People has been for years recommending a high clocked quad-core i7 for gaming. The problem with Ryzen is that it games worse than a 6900k at stock... and 6900k overclocks better than Ryzen.
Click to expand...


The only tacit confirmation is that you pick and choose whatever bench suits you irrespective of whether it is incorrect. At this point I prefer to wait for motherboard vendors to fix their products, and for windows to correct their scheduling which is completely messed up and they run benches again. Either way it games well with high end equipment, and in non gaming tasks it flat out murders any i5 or i7 right up to a 6900/6950X
 
Battlefield 1 as the case with 4 is heavily preference on higher frequency and quad cores as minimum specs, our tests on i3's and Pentiums is that the experiencing is choppy and the experience is horrible and it is the reason why a quad is minimum recommended part.

mainstream i7's play well but we notice multiple beyond 25ms spikes which leads to hiccups and skips or rubber banding, far less than i5's and the dual cores but it is still there. Moving over to Intel HEDT parts and the lag spikes noticeably fall away even though the FPS is lower than a 7700K or OC'd 6700K it is just a smoother gaming experience fully loaded with SweetFX API added.
 
All I see is the average was over 100 fps which is great for gaming. I dont play the game benchmark so I am not sure about it, but I think my new Ryzen will play Battlefield 1 just fine for me. People are always welcome to buy the 6900k they just have to spend 1000 bucks to get one. I think I will pocket the extra cash I saved on Ryzen and buy a new video card to replace the old 290x I have.
 
OrangeKhrush said:
Battlefield 1 as the case with 4 is heavily preference on higher frequency and quad cores as minimum specs, our tests on i3's and Pentiums is that the experiencing is choppy and the experience is horrible and it is the reason why a quad is minimum recommended part.
Click to expand...
Battlefield 1 heavily prefers 6c12t over 4c8t and 12c above all else.
 
Presbytier said:
Depends on the mode. In single player it seems that 4/8 CPU run much better, but in multiplayer it absolutely prefers 6/12 over 4/8.
Click to expand...
Who cares about BF1 multisingleplayer in the first place?
OrangeKhrush said:
I7 quads beating hex and octo cores and a couple i5' matching up to generational i7 quads again showing no scaling at all.
Click to expand...
Too high framerates to be multiplayer, anyways /s

EDIT: Damn...
 
Last edited:
THUMPer said:
GTX1080ti with ryzen, 7700k. 1080p, 1440p and 4k gaming
http://www.eteknix.com/nvidia-gtx-1080-ti-cpu-showdown-i7-7700k-vs-ryzen-r7-1800x-vs-i7-5820k/4/

ryzen is showing off for sure!

And to the guy that says who cares about BF1 multi player. I guess you play single player. Because MP is where BF1 shines. You can play SP for 8 hours and beat the game. But MP is where the re-playability is.
Click to expand...
I did not realize people even buy battlefield for their singleplayer campaigns
 
THUMPer said:
And to the guy that says who cares about BF1 multi player. I guess you play single player. Because MP is where BF1 shines. You can play SP for 8 hours and beat the game. But MP is where the re-playability is.
Click to expand...
Presbytier said:
I did not realize people even buy battlefield for their singleplayer campaigns
Click to expand...

Oh hell, i just realized i have meant to type single player but somehow it flew by.

OrangeKhrush said:
Which games are those
Click to expand...
StarCrafts and Dotas of the world.
 
lolfail9001 said:
Oh hell, i just realized i have meant to type single player but somehow it flew by.


StarCrafts and Dotas of the world.
Click to expand...

StarCraft is old probably get high fps on a strategy game, i think a pentium G4400 is overkill

Dota again poor multi-core optimization, dual = quad = hex. I remember playing dota on a trinity APU like a boss
 
lolfail9001 said:
I mean, the second one. On that one even 7700k can sometimes fail to keep 60 fps up, Ryzen can easily get stuck in 30s.
Click to expand...
For a second I was thinking DOTA was what they where showing being streamed but I guess it was LoL instead. Seems odd Ryzen would perform poorly in DOTA. What GPU are you using?
 
Presbytier said:
For a second I was thinking DOTA was what they where showing being streamed but I guess it was LoL instead. Seems odd Ryzen would perform poorly in DOTA.
Click to expand...
Nah, it was Dota 2 streamed, but there's a reason they only did v-synced stream of replays.
 
shad0w4life said:
1080ti has these trading blows now...minimum FPS goodness too
Ryzen-Deus-Ex-Mankind-Divided-1080p-Ultra-Preset-MSAA-Off.png

Ryzen-Rise-of-the-Tomb-Raider-1080p-Very-High-Preset.png
Click to expand...
I don't trust those. The 1800x was showing a nearly 20% deficit in Tomb Raider against the 7700k with a Titan or 1080 in nearly all reviews last week.
Ryzen-Benchmarks-Stock.012.png


Here is the ars review of the 1080Ti on a 5960k
1080ti.009.png
 
Last edited:
lolfail9001 said:
You call losing like 20% of minimums from 10% overclock "goodness"? Gosh.

P. S. http://m.pclab.pl/art73194-27.html
Click to expand...
You picked some random site with zero info on their test platform as proof? good ole pclab.pl, the #1 source for benchmarks.

But how dare I post saying intel and amd are trading blows in games now after BIOS updates on motherboards are released.

Linus tech also showed some oddities and he was running 2400Mhz, but 4k they traded blows, GTA V Ryzen had double the min FPS of 7700k

The quality of trolls has really gone down hill
 
shad0w4life said:
You picked some random site with zero info on their test platform as proof? good ole pclab.pl, the #1 source for benchmarks.

But how dare I post saying intel and amd are trading blows in games now after BIOS updates on motherboards are released.

Linus tech also showed some oddities and he was running 2400Mhz, but 4k they traded blows, GTA V Ryzen had double the min FPS of 7700k

The quality of trolls has really gone down hill
Click to expand...
They are not trading blows you picked a random site that's numbers are no where near anyone else's.
 
Don't forget NV just pushed out their new DX12 driver for the TI. Did they 'accidentally' make AMD look better?
 
Presbytier said:
Not entirely sure what you are getting at there????
Click to expand...
I think he's saying that GPU drivers can help the new CPU from AMD.

On the other hand, I hope one day Kyle tests CPU's and GPU's games using multiplayer too, this is where the CPU is stressed out the most and the GPU not so much. Single player game benchmarks are linear and can run off dual core cpu's without breaking a sweat.
 
Last edited:
You must log in or register to reply here.
Back
Top