*** AMD "Zen" thread (inc AM4/APU discussion) ***

[N19h7m4r3]

OH MY! 295X2 is beating the Titan X Pascal in DX12 Rise of the Tomb Raider as well. Where as Fury X gets smashed to a pulp by GTX 1070.

7 mins 47 seconds!

 

[akarypid]

Saddest part of this is the conclusion: RoTR is an outlier! (translation: thanks for the heads up guys, we'll make sure to STOP testing that when Vega is out, if NVidia hasn't issued an update yet).

Meanwhile, sheep, get this into your heads: OUTLIER OUTLIER OUTLIER! BAD BAD BAD OUTLIER!

Bring on those Overwatch scores!

EDIT: to be fair to this reviewer, at least he didn't use this game in his original review either.

 

[RavenXXX2]

Yep there's 'a bug' as he says:

AdoredTV is right. Nvidia needs to sort that out before Vega. Their driver can't feed their card fast enough, whereas the AMD driver can..

EDIT: I wonder if this is due to the GPU architecture, meaning that async compute stuff really provides for that huge an impact. If that's the case and we're seeing the true limit of the Pascal architecture (i.e. it's a hardware limitation and that's the best their software can do), then it would be really bad for NVidia... Having said that, the difference seems to huge for that to be true so maybe it's just the NVidia software (which CAN be fixed).

TXP at 50-60% utilization on one level in one game at 1080p. I'm sure Nvidia are concerned.

 

[akarypid]

TXP at 50-60% utilization on one level in one game at 1080p. I'm sure Nvidia are concerned.

Well, one game in one specific area is too little to consider this a widespread problem. The only time this will be settled is when Vega is out and we can have a view of a single high-end GPU (i.e. removing GPU bottleneck on AMD side with a single card) used for benchmarking CPUs (low settings) alongside the NVidia one (1080ti/TXP).

Until then... it's all speculation.

 

[N19h7m4r3]

TXP at 50-60% utilization on one level in one game at 1080p. I'm sure Nvidia are concerned.

One of the few DX12 games were NVIDIA were ahead on DX12 performance for comparable parts usually.

Remember this isn't about NV vs AMD GPU; it's about how those GPus are used on AMD vs Intel CPU.
AdoredTV showed that the CPU performance gap significantly closes compared to Intel, when used with an AMD GPU vs an Intel one in Rise of the Tomb Raider.

Then there's also this.

 

[CAT-THE-FIFTH]

Yep there's 'a bug' as he says:

AdoredTV is right. Nvidia needs to sort that out before Vega. Their driver can't feed their card fast enough, whereas the AMD driver can..

EDIT: I wonder if this is due to the GPU architecture, meaning that async compute stuff really provides for that huge an impact. If that's the case and we're seeing the true limit of the Pascal architecture (i.e. it's a hardware limitation and that's the best their software can do), then it would be really bad for NVidia... Having said that, the difference seems to huge for that to be true so maybe it's just the NVidia software (which CAN be fixed).

It might explain all the weirdness I was seeing under DX12 too. But OTH,it makes me wonder if some of the reason Vega is not out yet,is AMD RTG fine-tuning their drivers for Ryzen. Vega+ Ryzen might be a monster!

 

[opethdisciple]

If there is an issue Nvidia do need to do something pretty quick. Either wise the majority of Ryzen buyers will be buying an AMD GPU if their in the market for a new card.

Just coming out and saying we are aware of some issues an we are working on it would go very very far.

Preferably I'd rather stay with an Nvidia GPU but if nothing has changed I can see my self going AMD once Vega arrives. Simple for the fact I now own a Ryzen. (was this part of AMD's plan all along?)

 

[akarypid]

I on the other hand am not ready to declare this changes anything. I'm pretty sure that with Vega we'll see better performance out of Ryzen, as AMD will definitely tune their drivers to their platform, but it remains to be seen how far this will go. Sure, in RoTR the difference may be significant but I'd like to see performance across a range of games repeating this before I go 'wow'.

The reason I find all this fascinating is because it proves that the dynamics of performance are very complex. In that sense, I really appreciate the spirit of the AdoredTV's videos and the fact that he boldly goes ahead to make claims / projections in spite of all the criticism he gets when he errs. It's much more interesting than your run-of-the-mill 'here's the bars stacked on top of each other, make of them what you will' which are boring. We all know a TXP is better than a 1080 is better than a 1070 is better than a 1060 and so on...

 

[old gamer]

If there is an issue Nvidia do need to do something pretty quick. Either wise the majority of Ryzen buyers will be buying an AMD GPU if their in the market for a new card.

Just coming out and saying we are aware of some issues an we are working on it would go very very far.

Preferably I'd rather stay with an Nvidia GPU but if nothing has changed I can see my self going AMD once Vega arrives. Simple for the fact I now own a Ryzen. (was this part of AMD's plan all along?)

the amd master plan, these videos are a year old, laughed at the time but don't seem to be so funny lately

and

 

[opethdisciple]

the amd master plan, these videos are a year old, laughed at the time but don't seem to be so funny lately

and

Interesting stuff. Watching it now.

It seems with out saying it games that have both DX11 and DX12 options what their really saying is use DX11 if you have Nvidia and DX12 if you have AMD card.

 

[akarypid]

Another review showing the same: https://thetechaltar.com/amd-ryzen-1800x-performance/5/

There seems to be some concern over whether or not NVIDIA cards play well with Ryzen CPUs. So we tested using a GTX 1080 and turning threaded-optimization on and off to see what kind of a difference it makes. As it turns out, Ryzen is affected in a few games, though only significantly in Rise of the Tomb Raider. The other question concerns NVIDIA’s DX12 support and how that affects performance on this platform. In some games, notably Rise of the Tomb Raider, the GTX 1080 and Pascal Titan X tend to get better performance using DX11, though that’s something that is seemingly processor agnostic. That is, DX11 gives better results on both Intel and AMD processors in certain games with NVIDIA GPUs.​

Seems like RoTR is a one-off case that NVidia need to look at, as it doesn't they can't seem to multi-thread properly there.

The second observation has nothing to do with CPUs and simply states what we all know, that NVidia does not do as well in DX12.

 

[akarypid]

Interesting stuff. Watching it now.

It seems with out saying it games that have both DX11 and DX12 options what their really saying is use DX11 if you have Nvidia and DX12 if you have AMD card.

This one's very good as well, though a bit more technical:

It also explains why DX11 was so good for NVidia (which ironically boils down to multi-threading on the CPU side since they lack a hardware scheduler).

 

[opethdisciple]

Yep. During my time with my 970 and obviously following the progress of the 1000 series from Nvidia, I've been feeling like there was an issue with DX12 for Nvidia.

I guess the affordability of these high thread CPU's from AMD have just highlighted the issue.

No doubt Nvidia will fix this but when..... I imagine they will sort out the shortcomings when they launch the true successor to the current Pascal cards.

They will be DX12 from the ground up with rewritten DX12 drivers I bet.

But I am in the market for a new GPU now, so looks like it will be Vega when they launch.

 

[CAT-THE-FIFTH]

[

This one's very good as well, though a bit more technical:

It also explains why DX11 was so good for NVidia (which ironically boils down to multi-threading on the CPU side since they lack a hardware scheduler).

Its interesting as a few of us were wondering what the impact of Nvidia moving schedular functionality to software with Kepler would do in the future,as AMD moved the other way with GCN.

 

[akarypid]

[


Its interesting as a few of us were wondering what the impact of Nvidia moving schedular functionality to software with Kepler would do in the future,as AMD moved the other way with GCN.

Well the gist of it is here:

As long as Nvidia dictate PC port optimisation, they will stay ahead, because they have the $$$ to do it.

What is changing though is that other forces are now in play. Like I like to say: it used to be NVidia vs AMD and NVidia was the 500 pound gorilla. Suddenly, Microsoft came into play and (for reasons of its own) sided with the approach that favours AMD (as in DX12). Suddenly, there's a 1000 pound gorilla in play and all AMD has to do is wait for all the pieces to fall in place...

EDIT: ...while the rest of us hope studios opt for Vulkan instead of DX12!

 

[opethdisciple]

Well the gist of it is here:

As long as Nvidia dictate PC port optimisation, they will stay ahead, because they have the $$$ to do it.

What is changing though is that other forces are now in play. Like I like to say: it used to be NVidia vs AMD and NVidia was the 500 pound gorilla. Suddenly, Microsoft came into play and (for reasons of its own) sided with the approach that favours AMD (as in DX12). Suddenly, there's a 1000 pound gorilla in play and all AMD has to do is wait for all the pieces to fall in place...

Yea there is definitely a disparity with what is happening on the consoles and what is going on on the PC.

Consoles are all AMD, whilst PC games are being dominated by Nvidia.

---

It wasn't just DX12 that has been rubbish on my 970... I found Vulkan in Doom was also crap compared to OpenGL.

DX12 in BF1 = rubbish as well.

 

[CAT-THE-FIFTH]

EDIT: ...while the rest of us hope studios opt for Vulkan instead of DX12!

Its because many people are tied to Windows due to gaming and MS is increasingly looking like they might end using a subscription model looking at what they have been doing with Office.

 

[akarypid]

Its because many people are tied to Windows due to gaming and MS is increasingly looking like they might end using a subscription model looking at what they have been doing with Office.

I only hope it becomes the standard for everyhing non-MS. I hate lock-ins. That's why I never bought into Apple's model and their store. If I had an iPhone, I'd have bought hundreds of apps by now making it super-costly to switch to another brand. At least with Android I've used HTC, then Samsung, then Nexus and now a Sony.

I'm buying as many Vulkan games as I can get my hands on. Spread the word!

 

[CAT-THE-FIFTH]

I only hope it becomes the standard for everyhing non-MS. I hate lock-ins. That's why I never bought into Apple's model and their store. If I had an iPhone, I'd have bought hundreds of apps by now making it super-costly to switch to another brand. At least with Android I've used HTC, then Samsung, then Nexus and now a Sony.

I'm buying as many Vulkan games as I can get my hands on. Spread the word!

I am secretly hoping Elder Scrolls 6 has Vulkan.

 

[Panos]

I am secretly hoping Elder Scrolls 6 has Vulkan.

There no future single player ES games for years to come and as long as TESO is supported.

So we need to hope for Vulcan implementation on the TESO engine, instead of the planned DX12. That would help Cyrodiil also. Because currently even if we drop an AMD Naples with it's 32c/64t to it, it will be utilised fully on siege battles. Even my 6800K is getting crippled on those situations, and I bet even if I had a 6950X @4.5 it would be crippled also. And is not bad programming. Is just few hundred players per side trying to defend/attack at the same time, using siege engines on top.