The AMD Mantle Thread

[el etro]

The pc.lab slides strikes again.:biggrin:

http://forums.anandtech.com/showpost.php?p=36020818&postcount=161

As a side note to put some light on the pclab reviews. Some time ago (2010 or so) they didn't get their review samples from AMD. They went all out war, publicly complaining about it (originPC style). Since then not a single amd review had positive conclusion.

We call it PCLamb here for a reason.

-------------------------------

It's the only site which compared both.
Until there is a another site which compares nVidia and AMD/Mantle i take them more serious than people who are hyping mantle for whatever reason.

Agreed, but your claims of mantle tests on many sites being marketing hype is BS until the sites do these tests.

 

[sontin]

PClabs published more than these "nVidia in front" numbers.

But then i guess all numbers are fake.

 

[ShintaiDK]

http://www.anandtech.com/show/7728/battlefield-4-mantle-preview



Preview is up.



Unfortunately it's just average FPS, I want to see some graphs showing if Mantle has smoothed out the framerates too like some of the slides we saw. Next week a more indepth review though.

Looks nothing more than what an optimized driver would deliver. And even one case with DX being faster

 

[caswow]

Looks nothing more than what an optimized driver would deliver. And even one case with DX being faster

how come you dont mention they teste with Intel Core i7-4960X @ 4.2GHz- You know a cpu everybody has

 

[ShintaiDK]

how come you dont mention they teste with Intel Core i7-4960X @ 4.2GHz- You know a cpu everybody has

Its tested with 2Ghz 4 cores and 3Ghz 2C/4T as well.

 

[SPBHM]

this really needs to be tested with heavy MP64 action (I know it's pretty hard to test like this with consistency) and more average CPUs,

 

[Gregster1970]

Just checking out Anand's testing, and for a 290X coupled with a decent CPU, we are seeing gains of nearly 8%. Now you have to remember that AMD and Dice have put a lot of work into this, so 8% isn't bad and takes the averages up from 71.8 to 77.4.

I say thanks and well done. BF4 now averaging 77.4 instead of 71.8 is a welcome boost

 

[el etro]

Then again your avg joe probably doesnt own a 290x either.

But with Mantle he can start to think about the 290x(or at least the non-x version).

Correct me if I'm wrong, but disabling cores in the BIOS does not bring a 4960X down to the performance level of an i3 for example. 5x more cache for starters.

And everyone here remembers Ahtlon II X4 vs Phenoms Gaming comparison.

http://www.hardwareluxx.de/index.ph...nisse-im-kampf-mantle-vs-directx.html?start=2

Crossfire frame time results.

Thanks!

Crossfire frametimes improved too. Mantle helps CPU processing in every single way.

Remembers it:

link: http://www.neogaf.com/forum/showpost.php?p=98939900&postcount=28

 

[PPB]

Dont know if nv fanbase already scared Johan away, but here's my questions:

Do you think that Mantle with its inherent CPU bottleneck alleviation will influence in the way games are done on the FB3 engine (and their future itinerations) by the different game dev studios at EA?

For example, will this push for games with even more draw calls (if my memory servers right, in your BF3 slides you talked about the game implementing 3-4K draw calls, with some exceptions where it could as high as 7k draw calls), even tho there is still a DX code to deal with, that wont play nice with higher draw call values?

And, if that was indeed the case, dont you think Mantle will push the bottleneck even further to the server/network side of the equation, as BF4 is already pushed in that sense (low tick rate to deal with all the information needed to be processed by the server, client side hit detection to alleviate the server from doing that kind of work, etc)?

 

[Cloudfire777]

Hardwareluxx have also tested BF4 in Multiplayer with 64 players.

Intel Core i7-3960X @3.9GHz and R9 290x.

Improvements were:

1080p: +25%
1600p: +16%
4K: 7%


They also tested R9 290X in Crossfire:

1600p: +28%
4K: 25%

http://www.hardwareluxx.de/index.ph...nisse-im-kampf-mantle-vs-directx.html?start=1

 

[cytg111]

It's the only site which compared both.
Until there is a another site which compares nVidia and AMD/Mantle i take them more serious than people who are hyping mantle for whatever reason.

So, to stay scientifically correct and stay within the 0.05 boundries for statiscial correctess you put your back to _one_ data point and that _one_ data point alone.
Is that correct?
Think it is!
Stop posting.
Plz.

 

[sontin]

One data point is more than 0.

So maybe you should show your counter. Otherwise

Stop posting.
Plz.

 

[jason166]

Looks nothing more than what an optimized driver would deliver. And even one case with DX being faster

I understand the idea behind having a repeatable test and it would be nice if BF4 supported a bot mode for testing, but I don't care about single player performance and I suspect most people who play the game feel the same way.

What matters for us is multiplayer.

 

[cytg111]

Looks nothing more than what an optimized driver would deliver. And even one case with DX being faster

Please just bite on it. bite bite bite.

Are you a developer on direct3d optimiaztions? No? Then youre just making up numbers.

We could problary optimize you and you'd stop posting fud galore.
(ps. you folks be the reason why redi left... too many trolls too many agendas... u suck)

Warning issued for personal attack.
-- stahlhart

 

[DamnedLife]

So, to stay scientifically correct and stay within the 0.05 boundries for statiscial correctess you put your back to _one_ data point and that _one_ data point alone.
Is that correct?
Think it is!
Stop posting.
Plz.

Forhet that guy, he is most likely a stock holder who is bitter or getting paid. Changing goal post in the middle of one sided conversation, which it MUST bu you to prove the facts, while they do nothing but push us back some skewed benches and their opinion which don't reflect the facts.

Infraction issued for callout.
-- stahlhart

 

[sontin]

But Golem.de's numbers aren't ridiculous?
Maybe they have been paid to publish fake numbers?

BTW: There is no difference to their November 2013 tests.

It goes both ways.

 

[cytg111]

Forhet that guy, he is most likely a stock holder who is bitter or getting paid. Changing goal post in the middle of one sided conversation, which it MUST bu you to prove the facts, while they do nothing but push us back some skewed benches and their opinion which don't reflect the facts.

Yep, he likes the attention - the phrase "dont feed the trolls" seems like a good fit.

Sleep times here, looking foward to reading up on 8h history when I wake up . L8r.

Infraction issued for callout and thread crapping.
-- stahlhart

 

[sontin]

PClab is still updating their article. They even give a reason why nVidia cards so much faster in the Multiplayer:

http://pclab.pl/art55953-9.html

Computer published their first impression - with a WIndows 7 system:
http://www.computerbase.de/artikel/grafikkarten/2014/erste-eindruecke-zu-amds-mantle/

 

[Cloudfire777]

Looks like both Computerbase and Hardwareluxx found out you can gain about 15-20% on Multiplayer with a i7 CPU and 290x.

+10% in Singleplayer
+20% in Multiplayer

Hardwareluxx tested 290X in Crossfire and they got +25% with i7 and Mantle. Not +60% like DICE claimed

 

[SimianR]

Most reviews were showing the R9 290/290X outperforming the 780 @ 1080P in BF4 before mantle, and they were using windows 8 as well -that's why I'm not so sure about pclabs results.

 

[blackened23]

I understand the idea behind having a repeatable test and it would be nice if BF4 supported a bot mode for testing, but I don't care about single player performance and I suspect most people who play the game feel the same way.

What matters for us is multiplayer.

A lot of websites do benchmark multiplayer. The obvious issue with a true 64 player MP test is replication - results are much harder to replicate since every run will be different; different numbers of explosion, action, etc. So while some sites HAVE tested mantle multiplayer, others are hesitant for this very reason. It's all too obvious that repeatable and consistent results are much harder in MP. One run can have 20 explosions while another could have 50. That would cause the results to be skewed. But - again - many sites are testing MP, and many sites have tested MP with Mantle as well.

 

[f1sherman]

PClab is still updating their article. They even give a reason why nVidia cards so much faster in the Multiplayer:

http://pclab.pl/art55953-9.html

Computer published their first impression - with a WIndows 7 system:
http://www.computerbase.de/artikel/grafikkarten/2014/erste-eindruecke-zu-amds-mantle/

I used google-translate; seems they are doing live article, publishing new data as soon as available.
Also they are only ones breaking AMD Mantle Benchmarking Guide - by including NVIDIA in benchmarks

 

[blackened23]

I used google-translate; seems they are doing live article, publishing new data as soon as available.
Also they are only ones breaking AMD Mantle Benchmarking Guide - by including NVIDIA in benchmarks

Is that really happening? Is the AMD review guide prohibiting comparisons to nvidia?

How ridiculous. Not surprising though since nearly every website testing Mantle is using cheap HTPC chips and not doing applicable comparisons to nvidia. Seriously, what the heck is that crap? Can anyone confirm this? If it's true, it's bullcrap.

Can anyone get Ryan Smith on the batphone to verify this?

 

[f1sherman]

Is that really happening? Is the AMD review guide prohibiting comparisons to nvidia?

How ridiculous.

Nono... thats purely my conjecture.

But judging how reviews are done (NO NV CARD IN SIGHT) - not entirely baseless. What do you think?

 

[Haserath]

Looks good so far from ryan's preview. Everything I've heard about Mantle has been positive and it's showing good numbers.

Who would want to deny a positive thing like this?