Any word on when Mantle + TrueAudio will launch for Thief?

[Ryan Smith]

Why in the world do the review sites go along with AMD's instructions for benchmarking these things?

290x on low settings @ 1080p? Lol.

To be clear here, while I do read the reviewer's guides that AMD and NVIDIA send out, it's mostly for technical details not covered elsewhere and information on any bugs/errata I need to know about.

But I don't use these guides for testing. Any tests I run are tests of my own choice/design. In the case of Thief I wanted to see what a CPU limited scenario would look like, and it's easier to get CPU limited on a 290X at Low settings than it is any other setup.

In reality I expect 290X owners to be running at Very High settings, and that's why the bulk of our conclusions are drawn from that setting. However I also think it's likely that 260X owners will be running at Low quality with a CPU similar to the configuration I used, which does make the Mantle performance gains there interesting since we're seeing a tangible difference in that scenario.

 

[blackened23]

@blackened23

What happens when 20nm GPUs hit and even high end CPUs are possibly always the bottleneck? An api like mantle is going to be more and more important moving forward because CPUs currently do not scale performance with each generation like GPUs do.

Maybe, but as far as I can tell, AAA games have generally become more GPU limited over time - not vice versa. Some select sub-genres of non AAA titles can be more CPU limited, such as MOBAs, MMOs, and stuff along those lines. It just depends on the game. Let's throw out some examples here. Metro::LL - gpu limited. Crysis 3 - gpu limited. So on and so forth. AAA SP games are GPU limited mostly and this has increased over the past 5 years. Regardless, I can appreciate what Mantle is doing in terms of CPU limitations. That's fine with me. And MS is doing the same with DX12, but I don't know when that will come to fruition.

On top of this, I should mention that Mantle is benefitting the high end just fine. 5-10%? What's not to like about that? It's a free value add for your whatever AMD card you own. And that's cool. Stuff like that drives competition. Just like nvidia has their plethora of value adds for their users. I just don't see why anyone would go out of their way to do unrealistic tests such as 2000$ setup with low graphic settings. The benefit with Mantle. It is what it is. I don't get why you would try to make it something it isn't. And i'm not pinpointing AT specifically, this has happened at other websites. But let's get back to the low settings + 2000$ hardware here:

So anyway per the topic. My only thing is keeping the testing real world. When I see ivy bridge with a 290X tested at low settings, I scratch my head. That isn't real world. If you want to test low or mid range, hey that's great. Keep it real world though. As an example, FX 6300 with a 260X. That's a very real world setup. 4770k with a 290X, 4960X with a 290X, 7850K with a R7-260. These are the types of systems people go out and buy, and test them accordingly. And in a single player AAA title, you would generally want high settings on the high end setup. And on the lower end? FX 6300 with a 260X? Test low settings. Low settings on the 2000$ high end rig does not make sense for this game. But low settings DO make sense on a 7850k + 260, or FX 6300 with a 260X. Makes COMPLETE SENSE there. Makes ZERO sense on a 4960X.

When you see a test with a 4960X at low settings, 1080p, with a 290X? Yeah. Scratching head. Not just here either. Other websites have done this, and it is equally as confusing. Now I am not certain that AMD is suggesting this sort of stupidity in their reviewer guide, and apparently they don't make an explicit requirement of it, yet more than 1 review website has done this. Testing low settings on extremely high end hardware that costs 1500-2000$. It doesn't make sense to me. Just keep it real world, you know. If you want to test mid range, test mid range. If you want to test low end, test the LOW END. Don't test low settings on a 2000$ setup. Doing that doesn't completely mirror what a true low end setup and is just a confusing test. If the type of testing of high end $2000 hardware with the lowest graphical settings possible makes sense to others, that's fine. It just doesn't to me, put these low setting tests in the proper context with low end hardware. Feel free to disagree. Not trying to make a big fuss on this, don't really care to argue it anymore. Just my .02, that's all.

 

[VulgarDisplay]

As others have said. If you play the game to win you use low settings to get higher fps. If you play to camp and admire scenery by all means use high settings.

 

[blackened23]

Are you using lightboost now? I thought you were talking about how you wanted nothing but IPS panels back in that g-sync thread? That TN sucked and all that sort of thing? What changed your mind? Yeah I like lightboost as well. But IPS is great in others ways. Multi monitor is great huh. Take your poison for gaming (lightboost) and use IPS for everything else. Hey I use both IPS + lightboost too. Both good in different ways. Which I guess is what you do apparently. Anyway. Well you're clearly talking about BF4. And what you're saying is completely 100% debatable. I know I never lowball my graphical settings in BF3 or BF4. But hey whatever. You do what you like. I know what I like, and I don't like to completely lowball my graphics generally speaking, so long as my framerate is good.

Anyway, I'm talking about Thief which is what was tested. 2000$ setup with low graphical settings. Please. Re-read what I posted without having thoughts of BF4 in your head. And AAA titles have become more GPU limited over time, not vice versa. It does depend on the type of game and game genre. And like I said, I can appreciate what Mantle does in that respect (cpu limitations), and i'm sure DX12 will do something similar when it gets here. I am not AGAINST what mantle is doing for CPU limitations. I think that's great. I think it's great that MS is going to do similar with DX12 allegedly.

I am only saying. Keep the testing real world. Don't do unrealistic testing with a single player AAA game, by testing 2000$ hardware with low settings. And i'm not finger pointing at AT. Other websites have done this. Which makes me suspect the reviewers guide HIGHLY suggesting this. But I don't know. Repeating for emphasis: real world testing, real world setups. Not 2000$ setup with 1080p low settings in a single player game. That is all. I am not against Mantle doing it's thing for CPUs. I am for it. I want DX12 to do the same. What I am against? I am against unrealistic testing. Don't read what i'm stating as some sort of pot shot at Mantle. That isn't the intent.

 

[Rezist]

Not to mention Mantle is much smoother and more consistent.

Probably the biggest thing most realworld users have seen BF4 is the minimum boosts, Mantle is a better tech to counter G-Sync then FreeSync will be because it's easier to implement. If you can get a decent 20-30% boost on the minimums with tighter frame latency the overall effect is much smoother.

 

[Headfoot]

This seems to be the way at the moment in AMD for some reason, they have "released" these technologies but only to a select group of companies. Everyone else is shut out from using them. Doesn't really make a lot of sense. I for instance am quite keen to see just how much performance these DSPs have, what sort of things can be run on them, if they have access to GPU data and such. But its all private right now, can't access any of it despite the cards being out for quite a while.

I don't understand the secrecy, but it is not the way to build an ecosystem around an API or piece of hardware.

I couldn't agree more. I don't see what they have to gain by closing off the API to people who want to hack potentially the next killer app for them (for free might I add!!).

Maybe they hacked the API together real quick and are trying to beautify it before public launch?

 

[Headfoot]

To be clear here, while I do read the reviewer's guides that AMD and NVIDIA send out, it's mostly for technical details not covered elsewhere and information on any bugs/errata I need to know about.

But I don't use these guides for testing. Any tests I run are tests of my own choice/design. In the case of Thief I wanted to see what a CPU limited scenario would look like, and it's easier to get CPU limited on a 290X at Low settings than it is any other setup.

In reality I expect 290X owners to be running at Very High settings, and that's why the bulk of our conclusions are drawn from that setting. However I also think it's likely that 260X owners will be running at Low quality with a CPU similar to the configuration I used, which does make the Mantle performance gains there interesting since we're seeing a tangible difference in that scenario.

Ryan, thanks for your fast update on TrueAudio and Mantle in Thief!

For at least myself, I would be very interested in seeing what the Mantle benefit is for 4c/4t CPUs like the ever popular 2500k/3570k/4670k as I imagine a fairly substantial portion of enthusiasts are running these CPUs. Although I would also be interested in seeing how 2,3,4 core Core2 and Phenom II CPUs fair in both BF4 and Thief, as my "guest" computer can just barely run BF4 on a dual core phenom II.

Anyways, thanks for your work

 

[Headfoot]

Yeaa. The good old tigersharc. Man if the software was open to this the sound could even be perfectly adapted to the phones and the ear. Imagine calibrating sound in your game eg bf4 to your own phones, ears and cognitive system. And recalibrating after you have played a lot in game. Then select your 3d sound profile for the day eg some you named "Today i am angry on Metro and dont want to camp" lol.

That would be awesome. I am so in for anything that improves game audio. I still remember how great Doom 3 was on my dinky little 5.1 in a box computer setup back in the day

 

[OCGuy]

As others have said. If you play the game to win you use low settings to get higher fps. If you play to camp and admire scenery by all means use high settings.

You realize you are the only person propagating this that I have seen? Thief isnt BF4 with 64 players, or BF4 TWL.

AMD isnt investing resources in Mantle for the 6 people who purchase $500 cards to put them on low with mainstream monitors.

@RyanSmith- I understand what you are trying to show ( wasnt picking on you specifically, every review does it) but more like why you are trying to show it. Also maybe "instructions" gives the wrong impression.

I bet most people would run a 40yrd dash faster than they normally could if they were given PCP and being chased by a lion, but that wont get you on the track team.

AMD has gone out of their way to move the goal-posts in their recent product reviews. From last second "driver" changes to "increase speed" (was really a BIOS update allowing higher fan max), to removing base clocks, to benchmarking in non-repeateable pigeon-holed multiplayer scenarios, to hardware combinations that are laughable.

I have no problem with AMD pressing this stuff. They are a business. I am disappointed that so many respected reviewers are going along with it.

Surprisingly [H] of all places went sour on Mantle. I have been harsh on them, but I have to give credit when it is due.

 

[hungtran]

Since GPU speeds are growing faster than CPU speeds, the difference between Mantle and non-Mantle will just become more pronounced on next gen AMD cards. Game developers are already planning years ahead. By that time, CPU bottlenecks will be more apparent even at higher settings as GPU bottlenecks are removed.

 

[SPBHM]

gains are quite good for slow CPUs + fast GPUs
mantle really have a huge effect on CPU performance dependency, it makes easier to understand how those underpowered jaguar cores are enough for the consoles... no need to waste time with DX+drivers

if more (most) games adopted mantle I would feel tempted to keep my old i3 and upgrade to a fast GCN card.

 

[hungtran]

You're part of a market a game maker didn't have before Mantle. The hurdle to becoming a potential customer of the game goes from buying a new cpu+motherboard+gpu+maybe ram to just gpu.

 

[VulgarDisplay]

You realize you are the only person propagating this that I have seen? Thief isnt BF4 with 64 players, or BF4 TWL.

AMD isnt investing resources in Mantle for the 6 people who purchase $500 cards to put them on low with mainstream monitors.

@RyanSmith- I understand what you are trying to show ( wasnt picking on you specifically, every review does it) but more like why you are trying to show it. Also maybe "instructions" gives the wrong impression.

I bet most people would run a 40yrd dash faster than they normally could if they were given PCP and being chased by a lion, but that wont get you on the track team.

AMD has gone out of their way to move the goal-posts in their recent product reviews. From last second "driver" changes to "increase speed" (was really a BIOS update allowing higher fan max), to removing base clocks, to benchmarking in non-repeateable pigeon-holed multiplayer scenarios, to hardware combinations that are laughable.

I have no problem with AMD pressing this stuff. They are a business. I am disappointed that so many respected reviewers are going along with it.

Surprisingly [H] of all places went sour on Mantle. I have been harsh on them, but I have to give credit when it is due.

There are thousands if not millions of competitive gamers who do exactly what I am saying in EVERY multiplayer game, not just BF4.

I miss mantle so much already that I'm sending back my gtx780 just to get it. The difference in gameplay in BF4 is night and day.

 

[frozentundra123456]

You're part of a market a game maker didn't have before Mantle. The hurdle to becoming a potential customer of the game goes from buying a new cpu+motherboard+gpu+maybe ram to just gpu.

Sure, if there are only two games that you want to play. And what, maybe five or ten in the next couple of years?

 

[3DVagabond]

Sure, if there are only two games that you want to play. And what, maybe five or ten in the next couple of years?

It's new. It's still in beta. It's not going to happen overnight.

If you are on of the devs putting out the games that support it you have increased your market base for your game on systems that support it. You don't really care if one other dev supports it or 100.

 

[Keysplayr]

As others have said. If you play the game to win you use low settings to get higher fps. If you play to camp and admire scenery by all means use high settings.

To add, if you want to play the game to enjoy it, crank everything up as far as your graphics card will take you.

 

[blackened23]

gains are quite good for slow CPUs + fast GPUs
mantle really have a huge effect on CPU performance dependency, it makes easier to understand how those underpowered jaguar cores are enough for the consoles... no need to waste time with DX+drivers

if more (most) games adopted mantle I would feel tempted to keep my old i3 and upgrade to a fast GCN card.

You're part of a market a game maker didn't have before Mantle. The hurdle to becoming a potential customer of the game goes from buying a new cpu+motherboard+gpu+maybe ram to just gpu.

We know mantle has benefits. But this argument is pretty silly. Mantle has 2 confirmed games for 2014, with the others being beta (ie star citizen) or have no confirmed release date. To compound things, AMD seems to have suggested during their 290X marketing that all FB3 games would get Mantle, and it turns out that wasn't true. Far from it. It is up to the developer of these games. NFS Rivals isn't getting Mantle, Dragon Age inquisition is a big unknown and the devs haven't commented - and it doesn't appear on AMD's Mantle portal. Go figure. And then, Plants vs Zombies 2 doesn't seem to have a PC release date and it is also missing from AMD's mantle portal. Go to AMD's mantle portal if you believe otherwise. 2 confirmed 2014 games.

The point i'm making here is not a potshot at Mantle. My point is this. You don't "upgrade" a CPU for a good experience in 1-2 games while still having a completely sub-par experience in all other games. That's what happens when you upgrade to a 290X with a core 2 duo. Now the situation is different if you have something like an i7-970. Or a 2600k. I am not talking about this situation, because these CPUs are still VERY strong. Those CPUs aren't significantly different than the current generation, and are reasonably close (40% or so) when overclocked. That's different. The argument I see is stupidity , such as pairing a 290X with a low CPU such as a core 2 or A10-6800K. 40$ CPU with a 700$ GPU? Come on. Argue reality, not fantasy. You don't upgrade for 2 games by keeping the crap CPU. You upgrade for all games. So you'd be in a situation where you maybe get a good experience in 1 Mantle game. While all other games still run like crap. Or would run substantially better on a 4670k, which isn't expensive, BTW.

So, bottom line here. You upgrade for all games. So back to Mantle. I think it's great what Mantle is doing, it's a good value add for AMD users. I'm good with that. It drives competition, and Mantle is definitely benefiting the high end with 5-10% increases. Maybe more depending on the game type (eg cpu limited or gpu limited). I am also all for the CPU efficiency stuff that Mantle brings, which DX12 will also bring. So, once again, this is not a Mantle potshot. This is about reality vs fantasy. So let's get back to reality here. You have gotta be kidding me If you're throwing north of 6 benjamins for a GPU (290X), that you would just upgrade for 1-2 games by keeping some cheese core 2 duo that cost you 40$ on ebay. You upgrade for all games. So if you're in a situation where you have a junk Core 2 Duo or AMD 750k, then you Mantle might help you in 1-2 games. What about the other, I dunno, 200000000000000 games? No such luck. Still slow. Still need a better CPU. Mantle is doing good things, but it is still only covering less than .0000000001% of all PC games. So I am continually confused by this argument to say the least.

 

[el etro]

I am only saying. Keep the testing real world. Don't do unrealistic testing with a single player AAA game, by testing 2000$ hardware with low settings.

Hardware costs $2000, but it still haves a universal indicator of CPU/GPU performance(Example Cinebench and 3Dmark) that will reflect on proportionally weaker configs.
We have to consider too that i7 49xx is top-tier processor with top class clocks and cache sizes and this will reflect on the CPU performance in the game.


Not everyone haves enough hardware to play AAA games on maximum settings. Once we take off the MSAA of the games, we will get a much more CPU-bound situation(especially at 1080p), where Mantle gives great benefit.
What if we do a survey with battlelog players with enthusiast cards(R9 265 and above) about how much FPS they get at 1080p ultra/high/medium details? By i remember, nor in Singleplayer my GTX 670(i5 3570) could play BF4 maxed out with great FPS all the time.


And of course the combinations(hardware + settings) used with a video card may vary for each user. core i7 users that have a 290x may not have a hell of a benefit with mantle, but what about core i5/i3/FX 63xx/FX 83xx users that haves the card(the majority of enthusiasts)?