Discussion AMD Ryzen MegaThread! FAQ and Resources

Page 50 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Please refrain from linking Salazar here in the future, or post that you are linking his channel specifically if you do.

Salazar is not objective, he is a clickbait YTer with no clue, and I prefer not to generate revenue for sods like that.

Jayz2cents has said that Ryzen is great for gaming, and he cannot understand the fuss about it. GamersNexus...well...they do not understand much of what they talk about. Like the time they tried to say GPU encoding was best for adobe premier to insult Ryzen, when adobe premier does not support GPU encoding.
 
So, this guy tested the 1700X with varius RAM speeds, and his gaming benchmark results show an interesting story. Ryzen is very sensitive to RAM speed, as discussed here, and there is less loss from inter-CCX communication. Then, performance goes up the faster RAM you use. Not much news, but I liked that there are numbers and tests.

https://youtu.be/RZS2XHcQdqA?t=3m21s

154hkee.png
 


Everything that he said however was true in that video. Just like when Juan said the same thing pages ago. Jayztwocents says its great in gaming but he is also very vocal about how its not what a gamer should get if that is all they do. As for gamernexus well i personally find him to be very objective to everything unlike lets say Joker who is an obvious fanboy who doesn't know how to run a benchmark properly.
 


It doesn't really matter if it's played or not. See DOTA2, which is one of the most played games in the world and it got dissed as fast as you're dissing AotS.

Toms also has it's own view and the numbers are weird. I wonder what the differences in configurations are.

Cheers!

EDIT: Missed words.
 


Actually based on Steam numbers AOTS Escalation and PA Titans are the two most played current RTS titles besides Starcarft II.... So yes it's played. RTS actually has a reasonably solid fan base online, it doesn't attract the numbers that your COD or Battlefiled games do but quite a few smaller studios revolve around strategy titles one way or the other. It's also worth noting that *online player counts* aren't the best gauge of RTS popularity- most of the best RTS titles feature a large single player aspect and given how steep the learning curves are for RTS online many players tend to stick to the single player side of the game (case in point the recent excellent 'Homeworld Deserts of Kharak'- which by all accounts sold well but doesn't attract many players in multi player).

Edit: Also the performance uplift is quite impressive for a small patch. It goes to show it is possible to improve game engines for Ryzen and that Dota 2 wasn't a 'one off'.
 


Indeed, it was seriously broken on RyZen... That is the reason why a patch has improved a lot of this game.

EDIT: Most sites are reporting the gains measured by the internal bench tool included in the game. Pcgameshardware has used an external tool to measure the real performance gain and got a very different picture

https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fwww.pcgameshardware.de%2FRyzen-7-1800X-CPU-265804%2FSpecials%2FAMD-AotS-Patch-Test-Benchmark-1224503%2F&edit-text=&act=url
 


I have watched enough of Salazar's videos to know that he mostly does not talk about fact, and when he does, he spins the facts to suit his agenda.

On a scale of 1 to 10, I rate his channel negative ten.

Jayz2cents main argument was that if you are a gamer only, you should not buy the 1800X. If you are a gamer that wants 144 fps, then Ryzen would not cap frame rate in some games. Which is valid.

If you are a 1080p/60fps gamer like 75-80% of people out there, then save the money, buy the i5 priced 6c and call it a day. No way that you will not hit 60 FPS in games.

GN and a few others took shots at Joker for his review, but he did a video talking about his methodology on it, and showed his settings were slightly different from the others. He also showed how multiple runs of the same benchmarks showed different numbers at different points on the same processor, and intel processors. He may want AMD to come back, but he showed his settings on video, and discussed his build specifically. Other than frame time captures, every other YT reviewer does nothing more than that.

Aside from that, Wendell at Level 1 Techs (IMO, the best reviewer out there, because he is a long time software developer for Linux) has done some videos discussing the strengths. His points are mainly that Ryzen plays games dramatically smoother than any other processor he has played. He said it was almost impossible to quantify with raw data, but he said that there are no hitches when you play on Ryzen. Multiple others have said the same things, and commented that the 7700K sees stutter fairly frequently. Some have postulated this is because the gap between maximum and minimum FPS is so significant that it brings the stutter into the realm of noticeable. I tend to agree with that hypothesis, but there is not enough raw frame time data out there to be able to legitimately point to a mountain of proof that this is the case, because the feel of the game as it plays is just not something raw data can legitimately describe other than frame time consistency, and even that does not necessarily tell the entire story.
 


In regards to the bolded section, yes there are ways to quantify it. Maybe he just does not have the tools to do it.

Now, I am *really* interested in that, since I haven't seen any widespread sentiment around that. If it was the case, then why I haven't seen that being said in reviews? Maybe I haven't read them hard enough? lol.

Have some quotes and links?

Cheers!
 
Just watched Wendell's Ryzen review there, very good I have to say.. he say's that the infinity fabric runs at half the memory speed.. this is the first time I've ever heard this statement.
We all knew it had an affect but half the ram speed thats very specific, I never heard that anyway..
So when we are overclocking the ram we are also speeding up communication the cpu makes with everything else.. ccx to ccx..everything it talks to, it talks to faster. So overclocking your ram can increase the speed your cpu communicates with your gpu...Everything goes through the fabric right.
Overclocking memory on an intel machine really pales in comparison.

Can't help wondering what the limit is on the speed of the fabric...?
Well we certainly haven't hit it yet... as guy's are getting ddr4 3600mhz to run stable...

So it looks like there's even more juice in this baby yet.. do i hear ddr4 4000mhz anyone...?

Few more BIOS updates an I guess we will find out :)

Jay

Edit: just thinkin looks like there's more juice in naples as well as they only ran the bench's with 2400mhz ram..
 


Time to do some YT diving 😉

https://www.youtube.com/watch?v=88IOcFE4yho&feature=youtu.be

https://www.youtube.com/watch?v=8c1MSLy63Lc&feature=youtu.be&t=51m32s (<- Joker, Wendell, and the guy from Tech Deals all discussing this issue specifically...)

I confess, I have not watched the first video completely...so not sure what the total content is there...but the second video is entirely worth watching, and there is a fair discussion about differences in the "feel" of the game. The guy from Tech Deals talks about frame time consistency a great deal, but Wendell and Joker discuss just the raw feel of the games. Wendell even discusses the fact that minimums were higher on the 7700K but that it would "hitch" or "hiccup" (read: stutter) about every 10-15 frames.

EDIT: Here is Eurogamer commenting:

That said, AMD told us that games could run faster than Core i7 at 1440p even if beaten at full HD, and we have been able to replicate this. Take a look at the Crysis 3 test - there's a telling shot below. Ryzen 7 1800X is very slightly off pace at 1080p, but takes a five per cent lead at 1440p. How is this possible? Well, as we mentioned, non-complex scenes see the 7700K shoot ahead of Ryzen at 1080p, while the AMD chip commands a valuable advantage at the lower end in much more taxing rendering scenarios. By running at 1440p, the GPU overhead cuts off those sky-high leaps the Intel chip enjoys in simpler areas of the test scene, and those complex areas where Ryzen takes point weigh more heavily into the overall average.

The Crysis 3 result is fascinating because the GPU cutting off the i7 7700K's advantage won't just be noticeable at 1440p - it should also manifest with a less capable graphics card at 1080p too. This is a fascinating route forward for further testing with Ryzen, and we've approached AMD for its views on other games that may exhibit this behaviour. But we should be clear that it's not a silver bullet that makes Ryzen 7 the better chip - only Crysis 3 validated AMD's point out of our test titles and we would expect the Core i7 7700K to maintain its leadership in most games.
Click to expand...

http://www.eurogamer.net/articles/digitalfoundry-2017-amd-ryzen-7-1800x-review
 


Precisely, the best sort of review in my mind. Talks high level overhead, and then dumps the raw data for you to digest with some commentary regarding key points.

I almost wish some YTers would just data dump and leave the data to interpretation.
 


I will note using an external program will affect thread scheduling to some degree, no matter how little load said program is taking up.

Fact is, once we start benchmarking use cases where all cores are being utilized, using external programs to measure performance WILL have a non-zero impact on the final result.
 


I just caught that too. VERY interesting and it appears that there is some unconcious media bias showing here. There is no doubt that Nvidia's cards are faster, which was the choice of AMD to focus on mainstream and getting their 'big GPU' ready. But when everyone is using the same hardware, and getting slower results, it's odd that nobody thought to test AMD hardware and drivers. It DOES stand to reason that AMD's DX12 drivers would be optimized for Ryzen.

And it really makes me want to see what happens when Vega drops. That should be quite a revealing set of benchmarks.
 


The interesting aspect of that, is the short term implications for reviewers. They no longer can assume that Intel is the defacto-platform for testing GPUs. Most of the painful CPU bottlenecks that AMD had with PD are basically gone for the most part, so going forward, there are 3 main options: test Intel only, test AMD only or test both configs (effort x4).

From a pure effort perspective, the last one is the one I'd personally would love to see, but reality and money behind sponsors would keep reviewers tied to Intel. Yes, the money for the Hardware, unless it's an independent reviewer, comes from the sponsors most of the time. I bet there are a few cases where the parent company gives the greens, but that is hardly the case. I would love a challenge back on this last point though :)

In any case, long term view on this is also something I don't like. You'd be tied to AMD vertically if their GPUs perform better with their hardware only. I would hope to see the driver only favors the faster CPU (by other metrics) and not specific ones. Also, I would hope nVidia says something about that. I bet they are working on driver optimization, but some confirmation would come a long way, I guess.

Cheers!

PS: Thanks 8350 for the links!
 


And that's the kick in the nuts right there. The AMD DX12 drivers were faster for both Intel AND AMD CPUs than Nvidia's drivers for their cards. I think it's pretty clear that a 1070 OC'd like he tested is definitely faster than a single RX480, but two of them are equivalent to a 1070.

Here's the proof.

http://www.gamersnexus.net/guides/2507-crossfire-rx-480-benchmark-vs-1070-and-1080-power-temperatures-fps/page-3

At 1080p, the DX11 for the CF 480s is pathetic. No faster than a single 480. DX12 CF 480's in AOTS hits 72fps. DX12 1070's are around 70fps.

At 4k though, the DX12 CF 480s are faster than some of the 1080's and jumps several FPS ahead of the 1070s.

FPS gains from switching between DX11 to DX12 is much higher on AMD cards than it is on Nvidia cards. That means (from what I can see in the results):

AMD has really crappy DX11 drivers. Nvidia has really good DX11 drivers.
AMD has really good DX12 drivers. Nvidia has really crappy DX12 drivers.

This really highlights the issues with the particular API's underlying the choice in video cards used for reviews. I'm sure if every game was DX12 capable, and using AMD cards instead of Nvidia cards, you'd see a much closer CPU situation than you do now.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom