Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

[Markfw]

I've already provided the benchmark, with a graph of the results posted right here in this thread. Just because you are unwilling to consider Ryzen's decade old draw call performance, doesn't magically make Ryzen perform better at draw calls.

A draw call benchmark does NOT mean a game benchmark, just your word for it. I don't take your word since all these other benchmarks say otherwise.

Please listen to what I am saying DRAW CALL BENCHMARK DOES NOT EQUAL GAME BENCHMARK

 

[gorobei]

draw call limits is a very last gen issue. it is the limit we get with dx9(and 11 for that matter) and very little is going to change it now since most devs arent going to go back and recode it to work with newer gen hardware.

it came up after the mantle and pre ps4/xbone reveals which is how long ago (~3.5 years). Dan Baker of oxide and AotS showed the limitations of dx9-11 on drawcalls with the starswarm bench.

if you watch his recent presentation from capsacin&cream he covers more core/threads engines, VR and min frame times. he argues that 16ms(60fps) target of the last gen is a flawed premise and there are better paradigms for this coming gen.
he has a couple of performance logs showing the processing timeline and bottlenecks in respects to vr 90fps. one of his points being that with core scaling (greater than 4c/8t ie up to 20 threads) and mantle/vulkan/dx12 the draw call limit is now trivial and with better raster solutions they can get down to 5-9 ms frametimes.
https://www.youtube.com/watch?v=QJOIvACRY6g

 

[bjt2]

Yet, that isn't what the real world data says. Which means your prediction model doesn't work. Isn't science great?

Why? That sorry excuse of a modern game isn't relevant... Heck, even when it was relevant, people still avoided benchmarking it, because it runs so bad.

Since we are benchmarking old games anyway, we should benchmark world of warcraft and age of empires too.

You've seemed to have flipped the scientific method on it's head. You already have your conclusion set, and you are looking for data to support it.

One can always find a bench in which his lovely chip shine and the other hated chip went bad and label this bench the definitive and only true one... In the past was cinebench. Now that AMD has catched up and beated in MT, the definitive bench is another. That is a good candidate. Even if it does not mean anything...

 

[Head1985]

A draw call benchmark does NOT mean a game benchmark, just your word for it. I don't take your word since all these other benchmarks say otherwise.

Please listen to what I am saying DRAW CALL BENCHMARK DOES NOT EQUAL GAME BENCHMARK

look at fallout4

 

[french toast]

All of the games being measured for draw calls are all based on the same dx9 engine are they not? Some have been upgraded to dx11 but the under pinnings are starting to creak.
Hopefully bethesda moves to a ground up dx12/vulcan engine.

 

[Dresdenboy]

All of the games being measured for draw calls are all based on the same dx9 engine are they not? Some have been upgraded to dx11 but the under pinnings are starting to creak.
Hopefully bethesda moves to a ground up dx12/vulcan engine.

This makes the Bethesda-AMD partnership even more interesting.

 

[Zucker2k]

So MajinCry is on to something, then?

 

[sushukka]

I have no idea why people are trying to make an octocore HEDT platform competitor CPU a gaming CPU. If the 7700k could do the things Ryzen 7 could do, there would be no need for Ryzen.... There are other things beyond playing only 1 game at a time and running nothing else.

The point is that with Ryzen it seems that Intel's HEDT/gaming CPU segmentation has been artificial/money making strategy. Ryzen shows that you can have 8core processor which performs extremely well in both segments...and with around same price than the cheaper 7700k. As I said before, Intel have to think their product lineup segmentation again to get it in competitive shape. Also the upcoming R5, R3, APUs etc. will continue this havoc in Intel's comparative i7, i5, i3 series providing more cores, more perfomance with cheaper price (eg. 1600X 249$ vs 7700k 350$).

 

[sushukka]

I've already provided the benchmark, with a graph of the results posted right here in this thread. Just because you are unwilling to consider Ryzen's decade old draw call performance, doesn't magically make Ryzen perform better at draw calls.

Quite persistent fella. Thing is, what has been said many times, who cares? I'm buying my processor for future use and all the tests so far have been more than good from Ryzen's perspective. If you have found some bumps in lower level performance metrics like draw calls it seems not to affect the real world usage especially with modern dx11/dx12 games. Therefore you shouldn't artificially try to make some global scale point with those results. And as somebody said earlier it seems that you haven quite active with this information spreading. To me it seems that you're trying to make a mountain out of a molehill. Why? Proving headstrongly your (minor) point, mock AMD, Intel fanboy (as you don't even have Ryzen anywhere to make those tests)...?

 

[KompuKare]

My problem with MajinCry's graph is that there is little control for those results. Assuming there is validity to the results, wouldn't it make far more sense to sort them by graphic card? Also, (and I haven't looked at all the people who contributed) shouldn't other hardware be controlled like memory etc.?
Since the graph is based on FSP/(GHz * 10) where does the other variation come from? The 2500K scores at 4.5GHz and 5.1GHz for instance? Is there more going on there like memory speed or messing with the FSB (unlikely)? Why is that Haswell Pentium G3258 by far the fastest CPU?

While I think the results could be meaningful all they really show is that Bethesda have once again used a very old and poor engine for FO4 (and SSE I guess too). So yes, I do plan to go back to SSE once SKSE is out but I would not base my choice of CPU on that one game.

Anyway, can't help but think that if we wanted to know FO4 or SSE performance we'd want some more gameplay based bench. Rather than just open-world draw call performance, how about a specific busy spot and use the console to spawn lots of NPC or some complex scripted event?

I mean the worry about these Bethesda games is not so much about whether a CPU can manage to render max draw distance in busy parts of the game, but rather what happens when you've got tons of busy mods installed. Sweclockers benched Fallout4 in their Ryzen review but only with 2666 memory. And they didn't try any of the turn off SMT or similar experiments.
http://www.sweclockers.com/test/23426-amd-ryzen-7-1800x-och-7-1700x/20#content
Those results aren't great but they're not that bad either. At 720P only a bit behind Skylake i5.

 

[TimCh]

So MajinCry is on to something, then?

MajinCry is onto to the fact that if the workload is single threaded, the thread is jumping between CCX modules and slow memory is used then the performance is going to be suboptimal with Ryzen.

The problem is that he does not seem to care about the fact that most modern workloads (including games) does not work that way and that if someone is playing a DX9 era games they are most likely going to be fine even with Sandy Bridge level performance.

 

[krumme]

draw call limits is a very last gen issue. it is the limit we get with dx9(and 11 for that matter) and very little is going to change it now since most devs arent going to go back and recode it to work with newer gen hardware.

it came up after the mantle and pre ps4/xbone reveals which is how long ago (~3.5 years). Dan Baker of oxide and AotS showed the limitations of dx9-11 on drawcalls with the starswarm bench.

if you watch his recent presentation from capsacin&cream he covers more core/threads engines, VR and min frame times. he argues that 16ms(60fps) target of the last gen is a flawed premise and there are better paradigms for this coming gen.
he has a couple of performance logs showing the processing timeline and bottlenecks in respects to vr 90fps. one of his points being that with core scaling (greater than 4c/8t ie up to 20 threads) and mantle/vulkan/dx12 the draw call limit is now trivial and with better raster solutions they can get down to 5-9 ms frametimes.
https://www.youtube.com/watch?v=QJOIvACRY6g

Fantastic vid.
That man is a genious.
I remember 3 years ago in a Mantle q&a with Johan A from Dice where Johan said the new game engines would still be dependant on the mainthread and Dan replied that that would be solved in the future. He surely had the model in his head back then already.
The concept of photon time and pops with the new model that scales near perfectly even with insane number of threads plus decoupled shading is darn interesting prospect.

 

[krumme]

So MajinCry is on to something, then?

Not the slightest.
We have a few games that is basically still stuck on what resembles a dx9 engine.
But its non issue going forward. On dx12/vulcan its trivial to get the needed drawcall capacity.
Its yesterdays problem.
His graph is idiotic and misleading. A new api like dx12 gives north of 1000% drawcall uplift. Typically there is a 3300% perfuplift for a dx9 vs dx12 game engine for drawcalls.

Seeing MajinCry graph as a gamer and computer enthusiast is just sad imo. At best he doesnt understand it. No matter what; its selfdestructive old reactionary discource he is forcing throug with the typical " if its in a graph its true".

 

[FalcUK]

What is even more interesting is the work AMD are doing with Bethesda to optimize their games as announced with their partnership.

This could all be hilariously blown away if the Fallout engine gets updated to utilise Zen and Vega

 

[unseenmorbidity]

Poor performance ? where are these "bad benchmark" numbers ? All you cry is about draw calls. I could care less about something that does not prove a point. What works badly ? Where is your proof ? Where are your benchmarks. This is trolling, and you keep it up, and you are going to get infracted.

R7 1700 running at stock with 2133 MHz memory, which is the worst case scenario really.
1440p nearly maxed out on a OC'd RX 480 4gb. 1750 MHz memory & 1370MHz GPU

I was getting about a consistent 50-60 FPS, which was about the same as I was getting in town. I woud not call that bad, but I wouldn't normally play like this either. I would drop some settings a tad, and so I could get higher FPS.

There was a higher AA setting, SSAA 4x+ CMAA, but my little 480 couldn't handle that. It dropped to 30 fps.

EDIT: It was not boosting to 3750 in game.

 

[Blake_86]

Isn't that what they should be doing though? What's the point of loading up Watch Dogs 2, and walking back and forth in an empty alley?! The point is to see how the CPU performs in the actual game, and maybe make some predictions about which will be better in the near future.

I edited the pic of the benchmarks that was posted. I removed the irrelevant data, and the i7, since we are only talking about Ryzen. I also put in ratios from one tier to the next to show the gains.



EDIT: I included average gains for all the games tested, except Mass effect. It's a new game, and a pretty large outlier here.

Anyways, going from 3200 to 3600 MHz DDR4 netted about 3% gains in FPS. However, these are some of the most CPU intensive games out there. If he had tested a larger suite of games, then I would imagine the difference would be lower.

So good, but so ironic that it is very difficult to reach 3200 mhz on memory and almost impossibile to go higher, non to mantion that we are talking about 16 gb kit because if you have 32 gb kit (16x2 also worst) you can trow it in the garbage.
I understand who camplain rizen in game, because a rig whit 8 core 16 tread perfect for heavy load but limited at 16 gb is so fail!

 

[guachi]

It's not very difficult to reach 3200 MHz on memory as a consumer. It's just more expensive, especially at tight timings.

For example, it's about $35 more to go from 2x8 GB RAM at CAS 16 timings to CAS 14. (G.Skill Trident Z 3200)

Or it's $63 more to go from 2400 CAS 16 to 3200 CAS 14 (G.Skill Flare X specifically for Ryzen)

If you're willing to spend $183, you can have CAS 14 2x8GB RAM right off the QVL list (for IIRC ASUS, ASRock and Gigabyte) running at 3200 for your Ryzen system. All it takes is pulling out your credit card.

 

[Blake_86]

It's not very difficult to reach 3200 MHz on memory as a consumer. It's just more expensive, especially at tight timings.

For example, it's about $35 more to go from 2x8 GB RAM at CAS 16 timings to CAS 14. (G.Skill Trident Z 3200)

Or it's $63 more to go from 2400 CAS 16 to 3200 CAS 14 (G.Skill Flare X specifically for Ryzen)

If you're willing to spend $183, you can have CAS 14 2x8GB RAM right off the QVL list (for IIRC ASUS, ASRock and Gigabyte) running at 3200 for your Ryzen system. All it takes is pulling out your credit card.

I took a 16x2 gb trident z months ago when the price was lower. Now i discovered that if i want to use my kit with rizen i have to set at 2666 in the best scenario. Amd as bring a cpu ram dependant and now i have to downgrade from 3200 to 2660 or lower?
Madness

 

[Markfw]

look at fallout4

OK, where is the fallout4 bencghmark ?

 

[krumme]

I took a 16x2 gb trident z months ago when the price was lower. Now i discovered that if i want to use my kit with rizen i have to set at 2666 in the best scenario. Amd as bring a cpu ram dependant and now i have to downgrade from 3200 to 2660 or lower?
Madness

If you want your computer to run fastest in fa4 and not bf1 by all means get a 7700k instead. Please.The 2667 vs 3200 ram doesnt change that a all. Just stop the whining.

 

[MajinCry]

OK, where is the fallout4 bencghmark ?

I've been constantly bumping it: http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=threads/ryzen-w-amd-gpu-performance-in-fallout-4.2501467/

 

[MajinCry]

My problem with MajinCry's graph is that there is little control for those results. Assuming there is validity to the results, wouldn't it make far more sense to sort them by graphic card? Also, (and I haven't looked at all the people who contributed) shouldn't other hardware be controlled like memory etc.?

Memory seems to give a big boost to Ryzen, but that's about it, far as I can tell. And no, sorting by graphics card would only make sense if we were using 6670-class hardware. But once we're past the super low end, we're at CPU bottleneck territory.

Since the graph is based on FSP/(GHz * 10) where does the other variation come from? The 2500K scores at 4.5GHz and 5.1GHz for instance? Is there more going on there like memory speed or messing with the FSB (unlikely)? Why is that Haswell Pentium G3258 by far the fastest CPU?

Draw call performance does not scale in a perfectly linear fashion, as well as some architectures having lesser gains with faster clockrates. And I've no idea why the Haswell Pentium runs so fast. Seems to be an erroneous result.

While I think the results could be meaningful all they really show is that Bethesda have once again used a very old and poor engine for FO4 (and SSE I guess too). So yes, I do plan to go back to SSE once SKSE is out but I would not base my choice of CPU on that one game.

The reason I contrasted the draw calls with Bethesda's games, is that they issue thousands of draw calls, even in Fallout 3 and Oblivion. They really stress the CPU, as there are many individual objects to render, as well as a large number of actors being processed at any given time. Whereas with, say, GTA V, actors are little more than balls of fluff that are disposed of once they are no longer rendered.

Anyway, can't help but think that if we wanted to know FO4 or SSE performance we'd want some more gameplay based bench. Rather than just open-world draw call performance, how about a specific busy spot and use the console to spawn lots of NPC or some complex scripted event?

That's why I provided the Diamond City save. Corvega stresses the architecture's draw call performance @ 11k draw calls, whereas Diamond City stresses both draw calls @ 8k per frame, as well as actor processing.

I mean the worry about these Bethesda games is not so much about whether a CPU can manage to render max draw distance in busy parts of the game, but rather what happens when you've got tons of busy mods installed. Sweclockers benched Fallout4 in their Ryzen review but only with 2666 memory. And they didn't try any of the turn off SMT or similar experiments.
http://www.sweclockers.com/test/23426-amd-ryzen-7-1800x-och-7-1700x/20#content
Those results aren't great but they're not that bad either. At 720P only a bit behind Skylake i5.

They're using medium settings, so the draw distances are wank. There's a huge jump from medium to high, with ultra being even bigger. Low -> Medium is a marginal increase, by and by.

Also, they're using an NVidia card. DCL's help a bunch with D3D 11 games that issue a bunch of draw calls. An AMD card is better used, to show the disparity due to the driver only having one thread in D3D 11.

 

[krumme]

A dx12 engine can manage up to 100.000 drawcalls.
On every mainstream/enthusiast cpu.
Who cares if you cpu handle 90k or 110k.
That fo4 issues a few thousands and it both looks like dx9 and run relatively slow at either ryzen or core is just because it hardly uses dx11 mt but is a dx9 in its core. Nothing happens in the game. There is no heavy load of the cpu. Just 100 different bottlenecks like eg drawcall.
Its therefore foremost an Api bm.
You are 100% off topic with this discussion.

 

[KompuKare]

That's why I provided the Diamond City save. Corvega stresses the architecture's draw call performance @ 11k draw calls, whereas Diamond City stresses both draw calls @ 8k per frame, as well as actor processing.

Okay, all good answers...

However, what I meant was more along this line:

Pretty much all CPUs (or at least the ones Sweclockers benched) are able to handle those draw calls so while some are faster (or even much faster) than others that's not really what would worry me if I was playing a heavily modded Fallout4 or SSE game.

Rather, my worry would be can the CPU cope with mod scripts doing other things than just actors. Like some of the automated things you can get in large settlements (I haven't played FO4 but have heard that some mods can really slow things down when doing lots of stuff and then there's this 'contraptions' DLC where you can make factories etc.).

You already showed that some of the Papyrus script can run on different threads, so as long that relieves some of the burden (i.e. the whole game doesn't just run on one render thread and one script thread), I don't think I would be worried about the draw call performance.

Or am I missing something?

Also found this video about Ryzen and FO4:

But it seems to just be stock FO4 even if in a town. It does show that most of the Ryzen cores are nearly idle though and (not sure how they are numbered) it appears that both CCX have stuff happening even if the first one still has mostly unused cores.

 

[Markfw]

I've been constantly bumping it: http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=threads/ryzen-w-amd-gpu-performance-in-fallout-4.2501467/

I wanted a REAL web site benchmark. The video just above this post does not mean a lot either, as its not not done by a recognized web site,.