The AMD Mantle Thread

[krumme]

AMD bundles Battlefield 4 with A10 "Kaveri"


http://translate.google.com/transla...8007-amd-bundlar-battlefield-4-med-a10-kaveri

Bold move, they must be confident.

Perhaps they even "think" it can play multiplayer

 

[DiogoDX]

Perhaps they even "think" it can play multiplayer

The internet will go full rage if mantle patch is more stable than actual DX11. But I tink will be more buggy in the start.

 

[psoomah]

Bold move, they must be confident.

Would be in line with AMD's Product Manager Devon Nekechuck sayng in an impromptu post event Q&A:

“with Battlefield 4 running with Mantle, the card (290x) will be able to ridicule the Titan in terms of performance.” - that's a very strong descriptive term from a senior exec.

A senior exec that has a very good idea of what Mantle can do answering off the cuff = 'ridicule' was likely either the word that popped into his head when giving thought to the hard numbers he had seen or was picked up from others talking about it.

Also of note is that while that statement was far more 'out there' than the very measured statements about Mantle from AMD to date, there was no retraction or walking back of that term either.

 

[krumme]

The internet will go full rage if mantle patch is more stable than actual DX11. But I tink will be more buggy in the start.

Thats to be expected for a new api. Add if Mantle gives huge performance benefits oc gfx will get more prone to crash. The same we see with eg mining or furmark tests.

I think bf4 is far better than just 2 weaks ago. Lots of bugs still in the game but very stable the few hours i have tried it.

But listening to the devs at apu13 i think one of the important aspect of mantle vs dx is its far easier to debug. I think its factors like that, that actually drives mantle adoption.

 

[Stuka87]

Thats to be expected for a new api. Add if Mantle gives huge performance benefits oc gfx will get more prone to crash. The same we see with eg mining or furmark tests.

I think bf4 is far better than just 2 weaks ago. Lots of bugs still in the game but very stable the few hours i have tried it.

But listening to the devs at apu13 i think one of the important aspect of mantle vs dx is its far easier to debug. I think its factors like that, that actually drives mantle adoption.

Yesterdays patch fixed a pretty common crash. Worked well for me last night.

I am actually really looking forward to Mantle in hopes CPU usage drops as much as it should. Since I am very CPU bound.

 

[DiogoDX]

AMD bundles Battlefield 4 with A10 "Kaveri"


http://translate.google.com/transla...8007-amd-bundlar-battlefield-4-med-a10-kaveri

Bold move, they must be confident.

The claimed 8% better than 4670K in PCMark8 is huge compared with richland?

 

[psoomah]

Yesterdays patch fixed a pretty common crash. Worked well for me last night.

I am actually really looking forward to Mantle in hopes CPU usage drops as much as it should. Since I am very CPU bound.

Considering how crucial BF4 Mantle performance is to both AMD and DICE/EA, an all-hands-on-deck scramble to get it running smoothly is understandable.

Good news for BF4 players as they will see a rate of post release improvement far beyond the usual.

 

[blastingcap]

Would be in line with AMD's Product Manager Devon Nekechuck sayng in an impromptu post event Q&A:

“with Battlefield 4 running with Mantle, the card (290x) will be able to ridicule the Titan in terms of performance.” - that's a very strong descriptive term from a senior exec.

A senior exec that has a very good idea of what Mantle can do answering off the cuff = 'ridicule' was likely either the word that popped into his head when giving thought to the hard numbers he had seen or was picked up from others talking about it.

Also of note is that while that statement was far more 'out there' than the very measured statements about Mantle from AMD to date, there was no retraction or walking back of that term either.

Quoting execs is not dispositive. Didn't that AMD rep JF-AMD or whatever say that Bulldozer would have higher IPC and it didn't? Also TITAN is old news now, GTX 780 Ti is the top NV card and rumor has it that NV is coming out with more 28nm cards in Jan-Mar. 2014.

 

[Erenhardt]

The claimed 8% better than 4670K in PCMark8 is huge compared with richland?

4670k is aboout 25% faster than 6800k, but that doesn't say a thing about mantle.

 

[BallaTheFeared]

See this is the start of what we've been saying coming to fruition. With this move AMD shows the world that their APU's are good enough for gaming. EA will be giving them extra keys for cheap and hoping they make more money on DLC - money they wouldn't have been making otherwise.

This is exactly what we said would happen with Mantle opening up cheap, high-quality gaming to the mass market, and why the devs are falling over themselves to make it happen.

40% Uplift over the i5-4670k says otherwise, I can already hit higher with an OC on the i5. Haswells successor will be out in a few months sporting a much improved iGPU across the board :|

That said even the i5 OC is pretty much scrape the barrel gaming performance for desktops.

Still nothing in the way of GPU side performance figures from Mantle... I really hope they aren't pushing Mantle to make up for their under-performing cpus, I couldn't careless since I don't have one of them.

 

[Yuriman]

4670k is aboout 25% faster than 6800k, but that doesn't say a thing about mantle.

Was curious so I checked out some benches. BF4 is More like 50% faster on the 4670K, with the advantage being generally 50-80% overall (stock vs stock) in CPU-limited situations.

 

[SiliconWars]

Still nothing in the way of GPU side performance figures from Mantle... I really hope they aren't pushing Mantle to make up for their under-performing cpus, I couldn't careless since I don't have one of them.

I guess you aren't the market. If Kaveri (with Mantle) outperforms a 4670K and 650 GT at less than half the price + free BF4 then it's a massive win for AMD, the OEM's and the average consumer though.

 

[Phynaz]

That's a pretty big "if".

 

[frozentundra123456]

That's a pretty big "if".

The bigger "if" is if it outperforms a cheaper cpu like the atlon x4 750k and a HD 7750 GDDR5 to 7790 range card.

The desktop GT650 is pretty much a lousy card in every respect anyway.

 

[NTMBK]

The bigger "if" is if it outperforms a cheaper cpu like the atlon x4 750k and a HD 7750 GDDR5 to 7790 range card.

The desktop GT650 is pretty much a lousy card in every respect anyway.

We shall see, we shall see. I'm personally hopeful that they will be able to port some console unified memory tricks back to the desktop APUs (especially in Mantle titles), but no idea how much that will really happen. As always, we shall have to wait for the (independent) benches! :thumbsup:

 

[RaulF]

I guess you aren't the market. If Kaveri (with Mantle) outperforms a 4670K and 650 GT at less than half the price + free BF4 then it's a massive win for AMD, the OEM's and the average consumer though.

That's a pretty big "if".

And lets not forget, just because it performs good in a handful of games it does not mean it will perform better for everything else in the windows environment.

 

[Enigmoid]

Here is a demo showing running 1080p at medium (a few settings low) without mantle. Without mantle. And mantle is what makes this really interesting.
http://www.youtube.com/watch?v=O07YOk4nLyo

I think its perfectly fine to asume this can be done at 720/768p at high on the same machine. With mantle this could be doable in a mobile form factor as the cpu is reliewed and memory bandwith is the limiting factor anyway. Remember kaveri tops out with less tdp on the desktop size compared to richland.

I think we will see this apu eradiate all the mobile dgpu 640/650 class hardware if not 660 class already 2014. Remember an competitive intel solution needs to be quad core to do the same on the new games using mantle. Kaveri will have a huge cost advantage for the oem.

Comming from richland to kaveri + mantle in a game like bf4 is a gamechanger as amd is then no longer cpu limited. Kaveri will clearly then be heavily gpu limited.

Ofcource the lack of ddr4 probably changed the kaveri arch to have fewer shaders and probably half gpu power. But the upside to this is its a better fit to notebooks on the current 28nm gh hpp process because of tdp limitations. And its cheaper too, than what could have been twice as fast on the gpu side. Actually i think amd is more lucky than unlucke here than they think. The market is for the low end. They need 20nm for 1024 shaders anyway.

Kaveri will not be anywhere close to that

Kaveri mobile with 512 shaders is nothing special if cards like the 7730m (512 shaders at ~600 mhz + 128 bit DDR3, basically the best mobile bin of kaveri) are anything to go by. The 7730m @ notebookcheck gets around 1500 3dmark 11 points, the 660m is around 2400 (with boost). 60% faster and not held back by bandwidth. The 8730m (384 core gcn chip at 700 mhz + 128 bit 900 mhz DDR3) gets around 1400 points with newer drivers. And cheap laptops will be extremely lucky to get dual 1600 mhz sticks, possibly 1866 mhz but nothing faster.

I have a laptop with the 660m and the 7730m. The two are in completely different classes. Its the difference between medium/high 1080p on one and low-medium 720p on the other (in most games, tested Tomb raider(highish on 660m, medium on 7730m) and metro LL (normal and low settings but much better framerates on the 660m), skyrim (ultra + 2xAA on one vs medium high on the other) at similar frame rates.

Kaveri also has to hit that level of CPU + GPU performance in a shared 35W envelope.

(And BTW 640/650/660 hardware has already been eradicated..... by the 730-750m series. Which are quite hobbled by nvidia's decision to use a 64 bit, often DDR3 bus).

 

[Enigmoid]

AMD bundles Battlefield 4 with A10 "Kaveri"


http://translate.google.com/transla...8007-amd-bundlar-battlefield-4-med-a10-kaveri

Bold move, they must be confident.

+ 40% on firestrike over i5.

Would put it around 1300 points. A 25% gain.

Not really impressive. Assuming those numbers are correct (why is the a10-4600m doing just as good as the a10-5800k in firestrike extreme?). We are also comparing the i5 which has a lower GPU clock vs. the tested i7.

Also note that if you are comparing the chips with standard 1600 mhz RAM (trinity is 2133 mhz and haswell 2400mhz) the relative performance disparity decreases even more. Haswell gains something like 8% from 1600 mhz while trinity is around 20%. Richland is an insignificant 2-4% increase over trinity.

 

[Erenhardt]

Edit: both of the double!
^you'r talking kaveri, mantle, or mobile? There are dedicated threads for kaveri.
No one knows how mantle can improve bandwidth bottleneck. There is unified access, which takes some unnecessary memory operation out. There is the possibility of preparing "to be used" resources, spreading memory bandwidth utilization in time.
All needs to be coded and optimized for. AMD wants their limited hardware to be used as efficient as possible, and that is what users should look forward to. We are paying for each FLOPS we get, make sure none goes to waste.

 

[krumme]

Kaveri will not be anywhere close to that

Kaveri mobile with 512 shaders is nothing special if cards like the 7730m (512 shaders at ~600 mhz + 128 bit DDR3, basically the best mobile bin of kaveri) are anything to go by. The 7730m @ notebookcheck gets around 1500 3dmark 11 points, the 660m is around 2400 (with boost). 60% faster and not held back by bandwidth. The 8730m (384 core gcn chip at 700 mhz + 128 bit 900 mhz DDR3) gets around 1400 points with newer drivers. And cheap laptops will be extremely lucky to get dual 1600 mhz sticks, possibly 1866 mhz but nothing faster.

I have a laptop with the 660m and the 7730m. The two are in completely different classes. Its the difference between medium/high 1080p on one and low-medium 720p on the other (in most games, tested Tomb raider(highish on 660m, medium on 7730m) and metro LL (normal and low settings but much better framerates on the 660m), skyrim (ultra + 2xAA on one vs medium high on the other) at similar frame rates.

Kaveri also has to hit that level of CPU + GPU performance in a shared 35W envelope.

(And BTW 640/650/660 hardware has already been eradicated..... by the 730-750m series. Which are quite hobbled by nvidia's decision to use a 64 bit, often DDR3 bus).

Not close. Well it just did ???

Well this thread is about mantle. How is your notebook playing bf4? How is multiplayer with 64 on the maps?

Unless you have quad core i7 processor in your notebook, but only 2 core i5, kaveri will just kill it in bf4 with mantle.

I can see what the consoles can do with jaguar cores in bf4. Its evident cpu performance with mantle will be a huge gain.

 

[krumme]

And lets not forget, just because it performs good in a handful of games it does not mean it will perform better for everything else in the windows environment.

Well the devs seems to think otherwise. Mantle isnt even in in beta stage and already its in the most important engine cowering 30% of the performance games comming. And we even have games like sc only intended for pc on mantle.

You can go back in thread and lisnt to their arguments. Q&A for apu13 is the best place to start imho.

 

[krumme]

I guess you aren't the market. If Kaveri (with Mantle) outperforms a 4670K and 650 GT at less than half the price + free BF4 then it's a massive win for AMD, the OEM's and the average consumer though.

4670 plus 650 is an very expensive solution especially from oem perspective. Kaveri operates in another price bracket and market. And its different solutions. Its not compettitive products.

What is intesting is that mantle could elevate kaveri performance in new frostbite games like bf4 to near the same level. What that means is that it will be superior to the i3 640 class solutions - that is perhaps in the same price braclet - in those games in particular.

Richland is often cpu limited on notebooks. If kaveri adds 15-20% ipc / gains at same tdp. Then mantle uses the extra cores and taxes them less we could easily see total 50% uplift in multiplayer scenarios. I think its a conservative guess considering how slow jaguar will be compared to sr kaveri cores and still do the job in the consoles.

 

[krumme]

I remember all the critisism of the jaguar cores for the consoles. The consoles was let down by weak cores.

The hard data is here today. Ps4 bf4 looks great and runs the 60 fps it should. And sells like hotcakes beating all prior products from sony. Its a ressounding succes technically and financially up till now.

Code similar if not compatible with mantle is what makes this possible. We will get that technology.

What is not to like ?

 

[Gloomy]

This is going to be a little off-topic, but I don't want to necro and Google isn't turning up anything specific. And this is the AMD kool-aid thread anyway, so...

Do we have a list of TrueAudio-supporting games anywhere?

 

[chimaxi83]

This is going to be a little off-topic, but I don't want to necro and Google isn't turning up anything specific. And this is the AMD kool-aid thread anyway, so...

Do we have a list of TrueAudio-supporting games anywhere?

I don't think so, I've tried looking too. The TrueAudio part of my cards is collecting dust.