The AMD Mantle Thread

[Noctifer616]

So you don't really know?

It scales just fine, when the GPU vendor can provide a multithreaded DX 11 driver:
http://forums.anandtech.com/showpost.php?p=31520674

It scales just fine?

What you are saying, AMD is not multithreaded? So it uses just 1 CPU core? Yet we see with a 4 core just double performance? That's terrible.

Mantle numbers are :
4 core 8 thread 50k batches at 30 fps
6 core 12 thread 80k batches at 30 fps.

With mantle we see a 60% performance gain when we increased the number of cores by just 50%.

How exactly can you say that multythreaded performance in DX works just fine when we don't see the gains that we would expect?

And as other have pointed out, Civ 5 is the only example where DX multithreading seams to be working.

 

[Lonbjerg]

You can talk all the day about AMD's lack of support for DX11 multithreaded rendering, but at the end of the day, there are still only a few games that support it.

Well...since AMD's dosn't support...it hampers it's spread...but it works.
And it is one of the most improtnt things of DX11:
http://www.anandtech.com/show/2716/4

AFAIK, Civ5 is the only game where it actually has been shown increasing performance, and while AC3 and AC4 apparently support it, there doesn't seem to be any tests showing the performance figures for it
Speaking of Civ5, Nvidia pointing out the command lists may also be to give AMD a PR blow because they didn't support it. There could have been more optimizations for their drivers than just the multithreaded rendering

Lot of guessing..that nice...but of no use to me.

Dice has repeatedly said DX11 multithreaded rendering isn't working for them, and Far Cry 3 even had it patched away. While it's clear that it helped performance in Civilization V, it's also clear that's it's of no use for other games.

You got sources for that...because I remember reading DICE, Capcom ect . saying they did not implement it because of no support in AMD's driver?

 

[Lonbjerg]

It scales just fine?

What you are saying, AMD is not multithreaded? So it uses just 1 CPU core? Yet we see with a 4 core just double performance? That's terrible.

Mantle numbers are :
4 core 8 thread 50k batches at 30 fps
6 core 12 thread 80k batches at 30 fps.

With mantle we see a 60% performance gain when we increased the number of cores by just 50%.

How exactly can you say that multythreaded performance in DX works just fine when we don't see the gains that we would expect?

And as other have pointed out, Civ 5 is the only example where DX multithreading seams to be working.

AMD's DX11 dosn't support "Coammand List" (aka multitreading) in DX11...this is no secret.

That is why would like to see numbers done with NVIDIA's multhreaded DX driver, before I jugde what is from the Mantle API...and what is from AMD suddenly having a multithreaed API...since they are not sskilled enough to make it work in DX 11.

 

[Noctifer616]

AMD's DX11 dosn't support "Coammand List" (aka multitreading) in DX11...this is no secret.

That is why would like to see numbers done with NVIDIA's multhreaded DX driver, before I jugde what is from the Mantle API...and what is from AMD suddenly having a multithreaed API...since they are not sskilled enough to make it work in DX 11.

If Civ 5 numbers are anything to go by, you would see 30 fps at 30k batches with a 4 core 8 thread, and about 30 fps at 40k batches with 6 core 12 threads (if it can even scale that well).

 

[Spjut]

Well...since AMD's dosn't support...it hampers it's spread...but it works.
And it is one of the most improtnt things of DX11:
http://www.anandtech.com/show/2716/4

Lot of guessing..that nice...but of no use to me.

You got sources for that...because I remember reading DICE, Capcom ect . saying they did not implement it because of no support in AMD's driver?

One of the important things of DX11, but doesn't get used much...What one plans doesn't always turn out as good as one wanted

Yep I guess, but so do you. You point to one game where it's shown to give a boost and then you go on saying that the reason it isn't supported is because of AMD. (and if you compare AMD cards to similar Nvidia cards, Civ5 doesn't have that a big advantage today)

As for sources, look at repi's twitter, he has said alot of times that it doesn't work. He may even answer if you want to ask him yourself.

 

[Stewox]

Nonsense.



http://www.eurogamer.net/articles/digitalfoundry-could-amd-mantle-revolutionise-pc-gaming

Not to mention that the classic "layers of the earth" image shows a Mantle driver:

You prove nothing.

Have you read the entirety of my post, to understand the context of the first sentence.


There is no driver of this kind, pointless for microsoft to test since the driver, since will always work for month on the sames OS with same hardware while games will be ones that would potentially cause problems individually, there is no "updates" to the driver that would fix anything games do with Mantle, there's nothing in there except basic display stuff so it works with windows desktop, there's nothing to update what applications do by them selfs.

Hence the validation layer. But the driver isn't even about directX so AMD doesn't have to tell MS anything, but maybe there will be some kind of sticker "Ready for Windows 7" or something, I don't know, I just know this won't be an issue since the driver is so simple there's no room for it to be buggy.

 

[Piroko]

This so called game doesn't look impressive whatsoever. If AMD wanted to show mantle off, there are better ways to do it.

Compared to that studios previous game engine, used in Sins of a Solar Empire, that demo looked very impressive. Their old engine showed off some very decent large scale space battles, but was massively held back by CPU performance. I;m sceptical it even handles 5.000 draw calls before running out of CPU cycles.
I'm not hyped until they actually manage to translate that into a good game, but they got my attention.

 

[brandon888]

Guys mantle won't be amd exclusive ? nvidia can use it to ? ;/

 

[Noctifer616]

Guys mantle won't be amd exclusive ? nvidia can use it to ? ;/

AMD would need NVidia's help to enable Mantle on Kepler cards I think, but who knows if they would actually do that.

 

[Lonbjerg]

If Civ 5 numbers are anything to go by, you would see 30 fps at 30k batches with a 4 core 8 thread, and about 30 fps at 40k batches with 6 core 12 threads (if it can even scale that well).

I'd like to see it for the Oxide game...since omitting DX11 multithreaded results...kinda scews the picture.

I'm just looking for hard data...one EVERY aspect.
Seems AMD isn't interested in that...and then I draw my own conclusions...based on what the PR says...and especially on what the PR omits...or tires to ignore/hide.

 

[Erenhardt]

and especially on what the PR omits...or tires to ignore/hide.

Yep, good catch. AMD is hiding that even if Mantle makes its way to nv, the benefits will be lesser than with AMD cards.
But then, why would they talk about it, if enabling mantle for nv may never happen?

 

[psoomah]

You got sources for that...because I remember reading DICE, Capcom ect . saying they did not implement it because of no support in AMD's driver?

You got sources for DICE, Capcom etc. saying they did not implement it because of AMD's driver?

 

[psoomah]

AMD would need NVidia's help to enable Mantle on Kepler cards I think, but who knows if they would actually do that.

Why would AMD enable Mantle on Kepler cards?

 

[brandon888]

Why would AMD enable Mantle on Kepler cards?

Right question ... are they idiots ? even if they enable ... Nvidia have to pay them a lot for that ....

 

[Noctifer616]

Why would AMD enable Mantle on Kepler cards?

That's what they want according to the presentations.

 

[psoomah]

Yep, good catch. AMD is hiding that even if Mantle makes its way to nv, the benefits will be lesser than with AMD cards.
But then, why would they talk about it, if enabling mantle for nv may never happen?

Nonsensical. It's not a secret Mantle is specifically leveraging the console/PC common GCN architecture to translate console code to the PC and therefore Nvidia can never achieve 100% of the gains available to PC GCN processors from the console code.

 

[psoomah]

Right question ... are they idiots ? even if they enable ... Nvidia have to pay them a lot for that ....

Don't know about payments if Mantle is given to an open standards body, but Nvidia will definitely be way behind the Mantle implementation curve by that time.

 

[psoomah]

That's what they want according to the presentations.

Eventually make mantle available to other vendors to use, not help Nvidia implement it on it's GPUs.

 

[psoomah]

This tidbit is interresting:


Double the perfomance...just from making a driver that is multithreaded...why dosn't AMD do this first?
Or would that eat into the PR promises of Mantles performance...and thus make Mantle look less appealing?

Soon to be meaningless as in one fell swoop Mantle will blow past all the incremental DX optimization advantages Nvidia has worked so hard to implement.

 

[Falafil]

Can someone answer this question:
What happens when AMD comes up with a new architecture? Will they come up with a newer version of Mantle? And will that make games using older Mantle version unplayable on the new cards?

 

[blastingcap]

Soon to be meaningless as in one fell swoop Mantle will blow past all the incremental DX optimization advantages Nvidia has worked so hard to implement.

Actually, NV is rumored to be sticking an ARM core on future graphics cards to help with batching https://forums.robertsspaceindustries.com/discussion/78571/

 

[psoomah]

AMD's DX11 dosn't support "Coammand List" (aka multitreading) in DX11...this is no secret.

That is why would like to see numbers done with NVIDIA's multhreaded DX driver, before I jugde what is from the Mantle API...and what is from AMD suddenly having a multithreaed API...since they are not sskilled enough to make it work in DX 11.

Since AMD was already talking about bypassing DX altogether when Nvidia came out with their multithreaded driver, which means they had a Mantle type solution in mind, and as AMD is always resource constrained, it's logical enough they simply decided to bypass the time and effort of implementing multi-threaded DX drivers and focus on Mantle.

 

[Noctifer616]

Can someone answer this question:
What happens when AMD comes up with a new architecture? Will they come up with a newer version of Mantle? And will that make games using older Mantle version unplayable on the new cards?

Mantle has an abstraction layer. Any architecture that has the feature set that Mantle requires can work with Mantle. That's why NVidia could support Mantle.

 

[psoomah]

Can someone answer this question:
What happens when AMD comes up with a new architecture? Will they come up with a newer version of Mantle? And will that make games using older Mantle version unplayable on the new cards?

In the AMD Mantle presentation at APU13 this was specifically addressed. Mantle has different levels of abstractions and was specifically designed to accommodate future architectures.

 

[Falafil]

Mantle has an abstraction layer. Any architecture that has the feature set that Mantle requires can work with Mantle. That's why NVidia could support Mantle.

Nvidia cards have the feature set that Mantle requires yet AMD's 6000 series cards don't?