The AMD Mantle Thread

[psoomah]

I have less than no confidence that microsoft is smart enough to dig themselves out of any of the holes they have put themselves in recently.

Yep.

Looks to me like Microsoft is (was) planning to split DX, with their new and improved streamlined Xbox version reserved for their closed store ecosystem apps and games and the older version left pretty much as is for those WIndows 7 and 8 freeloaders too cheap to pay for proper apps from their store. Like providing no SP2 for Window 7 to drive people to Windows 8 so they acclimate to the 'metro' interface and going to Microsoft's store to buy apps. I just did a Windows 7 re-install with SP1 and the initial update was still 135 items. Yeah, frack u Microsoft.

Welcome to the bad timing awards. You're making your slow ponderous continuous face planting drive to switch to an Apple style closed ecosystem just as all heck breaks loose in the fast and agile viable alternative OS/platform universe.

Now you're too late. DICE has said flat out it's concentrating on Mantle and the PS4 for it's future Frostbite R&D and optimizations. Good luck keeping up with the PS4 on EA games. To DICE/EA your proprietary DX programming requirement represents a drag to progress. In 5 years, when EA is doing write once, fast and easy close to the metal Mantle ports to every other platform in sight, they'll still have to muck around with your 'DX only' game programming requirement. What Dice and EA already know, the rest of the developers will soon realize. Mantle and PS4 optimizations are the future of much higher quality gaming and a substantially better profit picture at the same time.

And there in your rearview mirror is a menacing black van with a Mantle wearing skull and bones motif driven by a crazily grinning Gabe Newell.

Sony doesn't like it either, but there was a price to pay for going with such a stock AMD hardware and middleware solution - being in no position to impede Mantle in any way than a statement of 'no support'. But hey, all the future optimizations will benefit the PS4 over the Xbox One so there is that.

 

[bystander36]

Sounds like Tinfoil hat time.

Seriously, Xbox has a streamlined DX, because Xbox has a single architecture.

 

[psoomah]

There's also the sheer difference in scale here - your example (kavari + R290) isn't exactly balanced - the R290 is an absolute monster in comparison.

Oxide demonstrated an FX 8350 running at half speed having no problem feeding a 290x, a 4 core Kaveri at full speed should have no problem feeding a 290x.

After all, one of the major advantages of Mantle is removing the CPU bottleneck.

 

[bystander36]

Oxide demonstrated an FX 8350 running at half speed having no problem feeding a 290x, a 4 core Kaveri at full speed should have no problem feeding a 290x.

After all, one of the major advantages of Mantle is removing the CPU bottleneck.

It was a demo. Seriously, wait until we have a real game to test it on. Demo's miss a ton of stuff required in games. It is not a real comparison. In a game, the CPU resources will be used to deal with a lot of added features, such as physics and AI.

It's also utilizing multiple threads to feed the GPU, using new tech to achieve it. This is great, but seriously, it is new tech, let's wait to see it in the real world before going crazy.

 

[psoomah]

Sounds like Tinfoil hat time.

Seriously, Xbox has a streamlined DX, because Xbox has a single architecture.

It's still DX addressing an x86 architecture.

It's a simple equation. Microsoft is requiring programmers use DX to support it's closed ecosystem strategy and therefore needs to optimize the heck out of DX to stand any chance of maintaining parity with the PS4, However that larger closed ecosystem strategy still needs to accommodate multiple hardware vendors so the DX optimizations for Xbox will be done with that in mind. It's a balancing act. Basically Microsoft is doing now what the developers have been wanting for years, but Microsoft want the higher quality DX games for it's store exclusively. It essentially wants what Mantle is going to provide, a reason to buy from one vendor over another. But in this case Mantle is going to put the electrodes to Microsoft's exclusivity crinkly brown hazels.

 

[DownTheSky]

It's still DX addressing an x86 architecture.

It's a simple equation. Microsoft is requiring programmers use DX to support it's closed ecosystem strategy and therefore needs to optimize the heck out of DX to stand any chance of maintaining parity with the PS4, However that larger closed ecosystem strategy still needs to accommodate multiple hardware vendors so the DX optimizations for Xbox will be done with that in mind. It's a balancing act. Basically Microsoft is doing now what the developers have been wanting for years, but Microsoft want the higher quality DX games for it's store exclusively. It essentially wants what Mantle is going to provide, a reason to buy from one vendor over another. But in this case Mantle is going to put the electrodes to Microsoft's exclusivity crinkly brown hazels.

M8 I get what you're saying and appreciate your enthusiasm but you sound more and more like a fanatic.

 

[NTMBK]

AMde secret sauce of TOTAL DOMINATION!!!!!

Do you even realise that you sound like a parody of a rabid fanboy?

 

[psoomah]

M8 I get what you're saying and appreciate your enthusiasm but you sound more and more like a fanatic.

Heh, I find the ever evolving graphics/APU/CPU scene much more fascinating than any other area of technology, so that is what I discuss, however I am also cursed with a capability of discerning relative significances of the pieces of the puzzle and how they fit together.

What you discern as 'more like a fanatic' are my ever evolving attempts to effectively convey the significances of the pieces I see clicking together to others.

I used to write/blog/post about politics, religion, cars, food, peak oil, Fukushima and so on, i'd post on a wide variety of forums. I got bored with it all after a while. The dynamics of AMD/Intel/Nvidia/Mantle/Microsoft/Sony in the gaming/graphics space still intrigues me though. It's fun to work it all out. So I guess I am a 'fanatic' in that it's all I still write about.

 

[psoomah]

Do you even realise that you sound like a parody of a rabid fanboy?

Yeah.

My friends would laugh at that question.

One day in Hawaii a girl known as Ocean walked up, looked at me for a bit, then said "My god, how many dimensions do you exist in?"

Kinda like that, only in a forum space.

And being a retired diagnosed schizophrenic recluse Vietnam vet living with a broken neck and fracturey upper spine that had a stroke 10 years ago.

A big part of doing this is keeping my brain cells clicking together figuring stuff out and have some fun doing it.

Figure I'm doing pretty good everything considered.

 

[selni]

Oxide demonstrated an FX 8350 running at half speed having no problem feeding a 290x, a 4 core Kaveri at full speed should have no problem feeding a 290x.

After all, one of the major advantages of Mantle is removing the CPU bottleneck.

Yeah except that's totally irrelevant. If you'd read the rest of the post you chose to pull that single sentence out of you'd see you're quoting a discussion on using mantle to do APU + GPU crossfire, you know that right?

 

[3DVagabond]

It'll come down to how easy it is to implement the APU/GPU functions. If the code is basically portable/reusable with little work fine. If it takes a lot of coding that has to be written on a per game basis, I don't think we'll see it used much.

 

[psoomah]

It'll come down to how easy it is to implement the APU/GPU functions. If the code is basically portable/reusable with little work fine. If it takes a lot of coding that has to be written on a per game basis, I don't think we'll see it used much.

Once the APU/HSA rendering paths are part of the Mantle API and incorporated into the game engine why would it be hard to include those paths for any game using that engine?

 

[3DVagabond]

Once the APU/HSA rendering paths are part of the Mantle API and incorporated into the game engine why would it be hard to include those paths for any game using that engine?

I don't know. If it's not, then we should see it used. If it is then we'd likely need AMD to implement it. At least until/if AMD APU's become more mainstream for gaming. Which, I only think it's a matter of time before that happens, anyway.

I'm sure not going to sit here and say it's going to be SOP in games when I personally have no idea. There's way too much of that on both sides of the argument regarding Mantle.

 

[psoomah]

There's also the sheer difference in scale here - your example (kavari + R290) isn't exactly balanced - the R290 is an absolute monster in comparison. AFR is totally out of the picture, so what do you offload to the APU? You can't do anything too bandwidth heavy because you've got a unified address space, not unified memory and everything has to go over PCIe. There's certainly performance to be gained here (perhaps just using the APU for general compute, but why do you need mantle for that?), but it's tricky.

Because Mantle sweeps the overhead complications out of the way and allows the developer direct memory control. A unified address space ain't peanuts, it's huge, it massively streamlines the pipeline and make it far easier to account for the latency differentials involved. As the Mantle rendering path also largely bypasses AMD's drivers, the developer can directly decide exactly what tasks it wants the APU to handle and what tasks it wants the GPU to handle based on the needs of the game. That's my understanding anyways.

 

[monstercameron]

Because Mantle sweeps the overhead complications out of the way and allows the developer direct memory control. A unified address space ain't peanuts, it's huge, it massively streamlines the pipeline and make it far easier to account for the latency differentials involved. As the Mantle rendering path also largely bypasses AMD's drivers, the developer can directly decide exactly what tasks it wants the APU to handle and what tasks it wants the GPU to handle based on the needs of the game. That's my understanding anyways.

then again the complexity of mantle, HSA and memory management between main memory[hsa memory] and dgpu memory could definitely balloon dev time...

 

[psoomah]

I don't know. If it's not, then we should see it used. If it is then we'd likely need AMD to implement it. At least until/if AMD APU's become more mainstream for gaming. Which, I only think it's a matter of time before that happens, anyway.

I'm sure not going to sit here and say it's going to be SOP in games when I personally have no idea. There's way too much of that on both sides of the argument regarding Mantle.

My apologies to 3D Vagabond. The offending paragraph was poorly worded and I sincerely did not mean that paragraph as a personal attack on you. It's been a long day of posting and my brain was getting a bit scrambled.

The almost over the top enthusiasm of Johan Andersson shouldn't be taken lightly. He's a true luminary not only in engine development, but in game development. His depth and breadth of knowledge and experience is really on another level and he has a reputation as a straight shooter, and nobody outside of AMD has a deeper understanding of Mantle.

As far as I am concerned the bullet points on his presentation slides at GPU13 and APU13 are totally valid and mean exactly what they say unless proven otherwise. I don't expect that will happen.

Everything I've read about Johan Andersson and DICE informs me there is no hidden agenda and what they communicate is valid and on the up and up.

If I were listening to a JHH presentation, everything I've learned about him would inform me otherwise.

Same goes for any information source. Just like one's friends and family. I know I assign each a 'validity' factor on a scale of 'gospel truth' to '99% bullspit"?

Anywaaaaaaaays, I just wanted to get that out there.

Infraction issued for callout and personal attack.
-- stahlhart

 

[krumme]

M8 I get what you're saying and appreciate your enthusiasm but you sound more and more like a fanatic.

No its just crazy. So what. Its part of the quality There is good and very creative points. There is no beating to death the same point more than 5 times. Its well written and funny language.

 

[krumme]

Because Mantle sweeps the overhead complications out of the way and allows the developer direct memory control. A unified address space ain't peanuts, it's huge, it massively streamlines the pipeline and make it far easier to account for the latency differentials involved. As the Mantle rendering path also largely bypasses AMD's drivers, the developer can directly decide exactly what tasks it wants the APU to handle and what tasks it wants the GPU to handle based on the needs of the game. That's my understanding anyways.

But what do you need the kaveri apu for when you have a 290 that have 700% the speed???

Unless its really easy to implement in the engine i dont see it happening. But who knows with the crazy low "2 man months" mantle implementation anything could be possible.

 

[psoomah]

then again the complexity of mantle, HSA and memory management between main memory[hsa memory] and dgpu memory could definitely balloon dev time...

It will definitely increase dev time. Ballooning is a different matter. I suppose it depends on the game, the targeted market and the expected short and long term benefits.

An example would be a slide from DICE's APU13 presentation that looked like this:
.....................................................................................................
Frostbite Plants vs Zombies: Garden Warfare

◢ Very different rendering compared to BF4  
◢ Frostbite Mantle renderer will work out of the box 
◢ Focus on APU performance
......................................................................................................

That's specifically targeting APU only desktops and laptops whose owners play Plants vs. Zombies on Apple or Android devices.

An obvious 'ballooning' of developer time to create and incorporate a different render path into Frostbite than already developed for BF4. As obviously worth it as it can then be used on subsequent games targeting that same audience and that audience is H-U-G-E compared to the hard core gaming audience.

A really really clever move by EA. Draw them in with a familiar favorite and get them hooked on really high quality graphics and gameplay, then they move up the gamer food chain ... and the AMD hardware food chain.

So it goes, each game development decision tree existing within it's own unique parameters.

 

[psoomah]

But what do you need the kaveri apu for when you have a 290 that have 700% the speed???

Unless its really easy to implement in the engine i dont see it happening. But who knows with the crazy low "2 man months" mantle implementation anything could be possible.

700% of the performance. But it's not that much, the 290X has 2,816 stream processors and the top Kaveri will have 512 stream processors, that's 550% based on that spec. Kaveri also has the full HSA/hUMA advantage, so the performance delta will be somewhat less than 550%.

Then there's the power savings when not high end gaming. There will be extensive power gating throughout, so the 290x won't spin up until it is needed. Playing Plants vs. Zombies: Garden Warfare for instance, which will be optimized for APUs, might not even spin up the 290x, or just spin up part of it. That saves a lot of heat and power usage.

It doesn't have to be 'really easy' to implement in the engine, it just needs to make a fair bit more money over time than it costs to implement.

For all developers, the bigger strategic picture provides substantial long lasting cost savings and increased profits if Mantle becomes the programming model across OSes, hardware and platforms over time, which is exactly Johan Anderssons vision.

 

[selni]

Because Mantle sweeps the overhead complications out of the way and allows the developer direct memory control. A unified address space ain't peanuts, it's huge, it massively streamlines the pipeline and make it far easier to account for the latency differentials involved. As the Mantle rendering path also largely bypasses AMD's drivers, the developer can directly decide exactly what tasks it wants the APU to handle and what tasks it wants the GPU to handle based on the needs of the game. That's my understanding anyways.

A unified address space is a programming convenience. Better to have it than not, but it's not letting you do anything you couldn't before (mantle allowing direct memory control certainly does, but that's another issue entirely). It's the physical implementation that matters - when the CPU accesses an address mapped to the GPU's memory (or vice versa), what happens? Is it an NVidia style hidden copy? Is it the same physical memory like AMD's APUs (including the consoles)? Something else? Caches etc also make this really complex.

That's the important question with regards to unified address space.

Yes, the developer can decide what tasks to run on the APU and GPU, but realistically what can you assign to the APU given the slow interconnect and huge difference in power? How much difference will that even make?

 

[Abwx]

when the CPU accesses an address mapped to the GPU's memory (or vice versa), what happens?

http://www.anandtech.com/show/7507/amd-kaveri-apu-launch-details-desktop-january-14th

 

[psoomah]

A unified address space is a programming convenience. Better to have it than not, but it's not letting you do anything you couldn't before (mantle allowing direct memory control certainly does, but that's another issue entirely). It's the physical implementation that matters - when the CPU accesses an address mapped to the GPU's memory (or vice versa), what happens? Is it an NVidia style hidden copy? Is it the same physical memory like AMD's APUs (including the consoles)? Something else? Caches etc also make this really complex.

That's the important question with regards to unified address space.

Yes, the developer can decide what tasks to run on the APU and GPU, but realistically what can you assign to the APU given the slow interconnect and huge difference in power? How much difference will that even make?

● This conversation originated from my "Basically it means Mantle can address, and flexibly distribute the workload between, an APU and a GCN dGPU at the same time." statement, so how is Mantle allowing direct memory control "another issue entirely"?
● I don't get your 'huge difference in power' thing. We're not talking a dual core Sempron here, Kaveri is an extremely powerful high performance processor in it's own right, 512 SPs aren't chopped liver. Even without Mantle in that leaked hotel room video Kaveri was smoothly running BF4 with everything set to 'high'. Programmers can assign pretty much anything they want to it.

 

[selni]

http://www.anandtech.com/show/7507/amd-kaveri-apu-launch-details-desktop-january-14th

Again that discussion is about a dGPU + APU - obviously the APU can have truly unified memory, but what happens when the dGPU tries to access APU memory and vice versa?

 

[Abwx]

Again that discussion is about a dGPU + APU - obviously the APU can have truly unified memory, but what happens when the dGPU tries to access APU memory and vice versa?

Sorry i confused the two issues, right that it s a different
matter once a dGPU is added.