Actually you're just being moronic since the moment you've entered this thread, which can be seen on every single post you've made. Be more literate when posting before you act condescending towards others. Clearly you go by the idiocy motto that trolls and other sad excuses for human beings go for on the web, namely "normal person plus anonimity plus audience equals humanity's dump".
But surely minimum frame rates are far more important? I mean which is more likely to affect your enjoyment of a game: that it doesn't quite reach the same peak frame rate or that in certain places the game slows down so much that you really notice?Originally Posted by anselhelm
Those certain places are of course those areas where the DirectX makes the CPU or the max draw calls the bottleneck anyone looking at what Mantle does would be able to predict that Mantle will do better there.
In fact, the main problem with Mantle seems to be that AMD don't seem to know how to market it. Selling higher FPSs is the totally wrong way to market Mantle but better marketing better minimums, consistency and frametimes is far harder to market and it seems AMD haven't really tried.
Well that and the usual people who hunt forums and love seem to think that Intel or Nvidia cannot do anything wrong, while AMD cannot do anything right.
CAT-THE-FIFTH (29-10-2014)
Don't think that is quite right.
The 480 was a power hungry monster, because Nvidia messed up the *implementation* not the architecture. That gave them silicon full of defects that didn't hit the targets, forcing them to up the voltage to get the thing to run at workable clocks and disable compute units that were broken. That isn't the same as the architecture needing tweaking. OTOH, my 460 is still going quite nicely, I think that part shows that even early Fermi was quite capable.
The 480 and 580 were supposedly very similar designs, just the 580 was fixed. Basically, what the engineers would have wanted to put out as the 480 is there wasn't the usual commercial pressures of releasing new product all the time so you don't go out of business.
http://www.anandtech.com/bench/product/1135?vs=1350
So Nvidia fixed a lemon, because they are Nvidia who occasionally come out with lemons as well as the golden parts like the 8800GT. AMD don't tend to have that implementation problem, so without the step change of lemon fixing they have more work to do.
OFC we are coming up to time for 20nm parts to come out, and this is traditionally where AMD shine and Nvidia stuff their foot in their mouth like they did with the 480.
Edit to add: Just thought, the R9 280 to R9 285 refresh was pretty impressive. Better performance from the same number of compute elements, with a third of the memory width gone and 50W less power. I think that shows they are serious about updating gcn.
Last edited by DanceswithUnix; 28-10-2014 at 09:13 PM.
No mention of "Why do your Linux drivers still suck " in the questions.
Yeah, it's not great. But, it is a UE4 game on the PC that you can buy, so.....
DX11.3 mostly solves that problem. You can dev for DX12, which is the CPU optimised version of DX11.3. If you can't run DX12....well, you can mostly, but slower! (it's linked earlier on in the thread)
I'm not sure why you think they've "seen the future"? They've publicly stated they do not plan to add it.
UE4 targets by default: PC, Max, Linux, iOS, Android, Xbox one, PS4 and HTML5 out of the box. Go and compare Mantle support on those platforms.... It doesn't make that much sense right now to add a very low level bit of software like Mantle in. It's immature, only has backing from AMD, and a competitor is in development that would require less change (and will ultimately have a larger user base).
Engines are written for decades of use. They are not games - it's not always a simple thing.
I'd be very surprised if they added it given the targeted platforms - it would need a lot more market penetration for it to be a base addition. Games engines are seriously complex, it's more likely they're going to leave it to end developers given that it's pretty bespoke.
That and it's not on their roadmap.
edit:
If Nvidia threw money at Epic to stop Mantle, provide evidence. That's a pretty serious accusation.
Also: http://www.extremetech.com/computing...vidia-and-epic
Gameworks is not part of UE4. It was a bit of a misunderstanding of words.
Last edited by Agent; 28-10-2014 at 10:16 PM.
http://blogs.nvidia.com/blog/2014/03/19/epic-games/
^^ from nv themselves.
Together with Epic, we’ve incorporated support for NVIDIA GameWorks directly into Unreal Engine 4 making it easier for UE4 licensees to take advantage of our technology. NVIDIA Gameworks libraries are designed to help developers create a wide variety of special effects, such as more realistic clothing or destruction, and now these effects are available to every developer with a UE4 license - See more at: http://blogs.nvidia.com/blog/2014/03....xYZUD6u7.dpuf
Last edited by HalloweenJack; 28-10-2014 at 09:58 PM.
dbl post
UE3 is pretty much dead development wise, unless you have an existing licence / support with Epic. The push to UE4 started ages ago during a beta called 'Rocket' which was closed unless invited.
This is the key point though - Mantle needs market penetration. If it hit consoles then you're much more likely to get it integrated into an engine like UE4. That and it needs to be a proven, supported product by AMD.
Epic are pretty serious about UE4 stability and they will want proof of this before integration.
http://www.dualshockers.com/2014/03/...-both-further/
If Sony / MS take it up, we'll see, but I suspect they'll do their own versions if needed. The improvement ability on a console though is very debatable though, as the existing code is already running very close to the metal.
Also keep in mind that developers that can code at this level are *expensive*. Dropping in a new API is often met with resistance. People like to know the way things work, and change can be difficult - no matter how much AMD (or anyone else) would like to say its easy. These codebases will often have hundreds of thousands, maybe millions of lines of code. Tried and tested is important.
Mantle on Linux / SteamOS actually makes a lot of sense. Mantle really does help take out the CPU bottleneck in some situations. This means you can start pitching quite efficient, optimised SteamOS boxes at a low price.
The issue then though is getting the games to support it - which ultimately will still be built on an engine if they are of any complexity.
The key word is compatible. A game can be 'DX11' from the ground up, but have a DX9 compatible code path. DX11 features either simply get dropped, or emulated where appropriate.
I've got to stress though that DX9 was a bloody good release. DX10 and 11 were nice, but no where near as game changing as DX9 was at the time.
I think you're thinking about this the wrong way. Having a 32bit code path so you can run old applications on a Windows 64bit based system is not a bad thing. Nor is having a DX9 codepath in a game / engine.
But why does it matter?
If they need DX11 for the title screen - so be it. If they didn't then fine.
The DX revision is just the API the developers are using. It doesn't make a better game. This mindset of having a 'pure' DX11 game is wrong. You don't build most of the assets for the game any differently.
There is little point in locking a game to a higher DX version when you can run it on lower ones, turn things off, and potentially get more sales.
Why is X game Y engine, at all?
The developers probably knew UE and used it. UE4 is very fast to develop in.
Because it locks you to a specific version when installed? What about if it uses no DX11 features and runs a DX9 codepath with a DX11 requirement set?
Again - it's the wrong mindset you're in.
Calc.exe on 64bit Windows is a 32bit application, therefore Windows can't be called a 64bit OS.
HJ....please understand this....a renderpath is just that. It's the route the game takes to output things to your screen (in simple terms). This entire "cannot be called a DX11 only game" mindset because it also has a DX9 path makes absolutely no sense.
If there is something about games engines you wan't clarifying (I have over a decade of UE experience in a commercial capacity) I'm happy to explain where I can - but ask, as some of the things you're saying don't make that much sense right now.
Check the dates. Read my link.
If you're in any doubt, get the UE4 source code and look yourself. Gameworks is not in UE4. Period.
Here is another link, from one of the guys I know who works for Epic: https://forums.unrealengine.com/show...ll=1#post60203
Even if "Gameworks", which is just marketing for several bits of tech, was 'core' to UE4 - it doesn't matter. It would be cross platform and not locked down like existing stuff in UE4. You get the UE4 source code with a licence, so it's not like it would even be an issue...I just wanted to clear up a few things!
UE4 does _not_ have Gameworks "built into its core". We do use PhysX at the core of UE4, but that is a cross-platform, CPU-focused, rigid-body and collision engine, and we work closely with NVIDIA to ensure it runs well on all platforms we port to. Currently in UE4 NO PhysX or APEX feature uses the GPU.
I'm very excited about some of the simulations NVIDIA have shown on a GPU, and I'd love to make them standard features in the engine. We have a great relationship with NVIDIA going back many years, and we are talking to them about a way to bring that tech to UE4 in a cross-platform way. We have no plans to implement them until we come up with a solution to that problem though.
If there is something you'd like clarifying, then please just ask...
Last edited by Agent; 28-10-2014 at 10:21 PM.
Where were you complaining when Hexus was posting all the stuff about Nvidia tech which was basically PR bumpf?? No where.
Where all the criticism about the Nvidia effects?? Turf effects,etc?? Nvidia PR saying it makes the games XYZ better and so on??
Also,do you really want to start arguing with me about Nvidia PR and their sideswipes at the competition??
You want me to start listing them year by year for you??
How about these childish cartoons by Nvidia PR in 2009 taking swipes at Intel??
http://www.tomshardware.com/news/nvi...ides,9015.html
Nvidia PR is renowned for is agressiveness - just like Apple.
AMD has as much right to put out as much PR bumpf crap as Nvidia has done for years.
Oh wait! Only Nvidia is allowed to say anything they want.
Edit!!
Look at Nvidia criticising AMD/ATI:
http://news.softpedia.com/news/Nvidi...ds-78793.shtml
Nvidia NEVER smack talks competitors products ever!! In fact they never made a dual GPU card,right?Jen-Hsun Huang, the president and chief executive officer of Nvidia told financial analysts that AMD's Radeon HD 3870 X2 cannot be the highest-performing offering on the graphics market. Moreover, Huang criticized the very idea of joining two graphics processing cores on a single card. Nvidia seems to believe that the classical, single-chip approach pays off best when it comes to high-end graphics products.
Oh wait a second....!
So honestly,if you don't care about PR ignore it and don't even attempt to comment on just one company doing it,unless you feel the need to do so for ALL companies in a particular sector.
Every single company is doing it - bigging up their products and making the competition seem worse.
Honestly at this point,I really shouldn't bother arguing with you as we will just go in circles just like the times before.
Last edited by CAT-THE-FIFTH; 29-10-2014 at 01:26 PM.
His point is moot though - you need to do some more research about people using Mantle. There are people with older CPUs getting decent framerates in games like BF4 online,and those people have posted their own experiences on forums like OcUK. People with older Core2 quads,Phenom II CPUs and so on,who have managed to make do with a GPU upgrade only. Even look at Thief:
http://media.bestofmicro.com/L/L/427...Mantle-Mid.png
http://oi59.tinypic.com/2i6gsxk.jpg
http://oi61.tinypic.com/ngq7io.jpg
The cards tested are all under £150 - look at the scores for the Core2 quad and the sub £70 AMD quad cores.
You should realise by now,not everyone has top end CPUs,so Mantle does have some advantages and they are more of use to the large percentage of realworld gamers.
OFC,not for hardware enthusiasts with their £150+ latest CPU and £400 cards.
Plus it is here now and in games now and its going to be in games which have been announced NOW.
DX12 games are still incoming and there is no solid date apart from the middle of 2015,with no actual game announced at all last time I checked.
But OFC,like with the Fermi delays,lets just say DX12 is coming and coming and that instantly makes Mantle pointless. Didn't we see that with Rollo??
For an API which is "not of any importance" it does appear those who are the most vocal critics of it are people who will never really use it.
Remember he says this for example to me:
One of those sources was PCPER,who developed FCAT with Nvdia,but he cannot still accept that AMD might have worked enough on XFire to make it pretty decent by now. But I bet he also did not pick up on Nvidia quoting PCPER when FCAT was first released showing "how superior" they were.Citing the one internet source that reckons CF is better then SLI.SLI bridges "archaic".....erm, does anyone care if the bridge is there or not? Is that really going to make anyone with half a brain cell change their purchasing decision?
Statements like this:
Nvidia makes similar claims about their tech too.Mantle, TrueAudio and Eyefinity give the best gaming experience? Oh really?
Yes,because Nvidia would NEVER do the same would they.....!?
PR quoting best case scenarios.
Shock and Horror!
What if you do though?? If there is a game which you want to play and it has Mantle support then I don't see why you should ignore it??
No different than Nvidia users saying they want to buy a Nvidia card for PhysX effects or CUDA.
NVAPI didn't run on AMD cards or Intel IGPs too,so what??
Yet since Nvidia was using it thats fine and gave them a nice performance bump for the GTX580 and GTX680 over the HD6000 and HD7000 series cards. So if Mantle does the same,pfft???
Plus since the said Mantle games have DX11 paths currently(and probably DX12 ones in the future),I don't see why there is this negativity.
Don't like it?? Don't use it.
Simples.
BTW,since I have used NOTHING but SFF PCs since late 2005,I really am just a "tad" clued up about SFF computers!
Again,maybe you should check the power consumption characteristics of the GTX970 and GTX980 cards. They are closer to GXT670/GTX680/GTX770 level and according to the dynamic power measurements from TH can exceed TDP and power consumption of such cards too.
What you fail to realise repeatedly,is that Kepler was based on the "power hungry" Fermi. Maxwell was based on Kepler with a sharing of texture units and greater power,TDP and voltage control granuality.
So are you honestly saying that GCN cannot be evolved in the same way??
Really??
So again,I would advice you to look at some of the tech Nvidia implemented and look back at the slide above.
People say that all the time. See my next post.Not doomed, but they're not in a good position. In the CPU market, AMD fell dramatically from way ahead to far behind, and the market as a whole has been worse for it with Intel uncontested in absolute performance and in performance/watt. To have the same happen in the graphics market would be bad for everyone.
The second generation Fermi parts had better power containment features and the TSMC 40NM process was better yielding by then.
People forget that the GTX580 was not massively better off than the GTX480 in regards to power consumption,but yields were better by then so they could enable the whole GPU which meant better performance,and with the better TDP confinement mechanisms,meant performance/watt improved.
But,I find it also quite funny that people don't realise that even a "power hungry" R9 290X or GTX780TI is consuming LESS power than a GTX570 or GTX580 in most cases. It just shows you that 28NM has seen a decent decrease in card power consumption overall.
Last edited by CAT-THE-FIFTH; 29-10-2014 at 04:34 AM.
Agreed.
Looking at the last few years of GPU releases.
2002-2003 ATI is doomed due to the cancelation of the 8500XT and Nvidia having the TI4200 and TI4600
2003-2004 Nvidia is doomed since the R300 is first to DX9 and faster than the FX series.
2004-2005 ATI is doomed since the 6000 series Nvidia cards support DX9C and they don't. Nvidia is lower power consumption.
2005-2006 Shifts between ATI is doomed and Nvidia is doomed.
2006-2007 ATI is doomed as the G80 and G92 have better performance and lower power consumption than the HD2000 and HD3000 series
2008-2009 ATI is first doomed since they having nothing to compete with the GT200. Then AMD launches the R700,and now Nvidia is doomed as Nvidia has to price cut huge chips against AMD ones which are smaller
2009 - 2010 Nvidia is doomed since ATI/AMD is first to DX11,has lower power consumption and smaller dies than the GTX400 series. Although soon with the GTX460 series,AMD is doomed.
2010 -2011 AMD is partially doomed since they don't have the fastest card anymore,and Nvidia has better tessellation
Late 2011 to early 2012. AMD is doomed due to the HD7970 and even more doomed with the GTX680.
Late 2012 to middle 2013. AMD is doomed as they have nothing to compete with once the Geforce Titan and GTX780 are released
Late 2013. AMD is still doomed with the R9 290 and R9 290X releases,since Nvidia quickly launched the GTX780TI and the AMD cards have black screens,throttle and run too hot and explode all the time.
Late 2012 to 2013. Nvidia is doomed due to AMD winning console contracts.
Early 2014. AMD is doomed due to the GTX750TI
Late 2014. AMD is doomed due to the GM204.
Potential next doom point - Nvidia releases 20NM GM200/GM210 in small quantities at £1000 and even if AMD has the fastest card in the R9 390X at £500 before then is still doomed.
And so on.
When it comes to GPU releases people seem to have very short memories indeed. At every launch the standard stuff happens.Hilarious.
Its so predictable,so much so I predicted the response to this launch months before it came out. I believe I mentioned it here or on OcUK or some other forum.
Funnily enough I got quite good at predicting the response to Apple and Android launches for a few years too.
Last edited by CAT-THE-FIFTH; 29-10-2014 at 03:49 AM.
I wouldn't say unsurpassed. While AMD isn't completely beat, i'd say nVidida has the edge on most things. Still would use AMD in a pure budget build for someone because that new Pentium line from Intel just won't cut it for me when i can get a APU from AMD that will do much better.
Right, let's actually look at that TH article, shall we?BTW,since I have used NOTHING but SFF PCs since late 2005,I really am just a "tad" clued up about SFF computers!
Again,maybe you should check the power consumption characteristics of the GTX970 and GTX980 cards. They are closer to GXT670/GTX680/GTX770 level and according to the dynamic power measurements from TH can exceed TDP and power consumption of such cards too.
The gaming-based power consumption numbers show just how much efficiency can be increased if the graphics card matches how much power is drawn to the actual load needed.And on the next page (after a disclaimer that previous results from a modified non-reference card have been removed due to being non-representative):The values above have potential consequences for the everyday operation of these graphics cards, as they represent what can be expected when running performance-hungry compute-oriented applications optimized for CUDA and OpenCL.
The tl;dr being: Compute workloads bring power consumption up to max TDP. Gaming loads do not. Overclocked cards use a lot of power (duh), reference cards do not.When it comes down to it, it's possible for our most taxing workloads to take Maxwell back to Kepler-class consumption levels. In fact, Gigabyte's factory overclocked GeForce GTX 980 actually draws more power than the GeForce GTX Titan Black without offering a substantial performance gain in return. As you can see below, the reference GeForce GTX 980 draws substantially less power, though.
Unless you are aiming for a compute workstation or have an overclocked non-reference card (with a higher rated power consumption), power draw will be significantly lower when gaming.
Now, AMD certainly have the opportunity to implement more rigorous power gating for a Tonga successor. But that just brings them down closer to Maxwell territory. In order to significantly surpass that, they need a new architecture, or a new process, or both. And we all know that changing architecture AND process is a recipe for low yields and lower than expected performance (e.g. early Fermi).
Looking at TH's own GPU efficiency chart from that same article, you can see the challenge AMD has to produce a dramatic increase in efficiency just to remain competitive with current parts.
There's no one insurmountable problem AMD need to overcome, but the combination of moving to a new process, a long time without an architecture refresh (or even longer if they roll out a whole line based on Tonga) and the need for at a minimum performing significant power modifications to Tonga? Something has to give. Doing everything at once will be a big gamble, and TSMC is continuing to have trouble getting 20nm big-chip yeilds up to an acceptable level. AMD deal in volume, producing a few golden-samples at massive cost is dangerous with their current finances.
Remember also, TDP isn't only relevant in mobile and volume-constrained scenarios; for multiple generations now, top-end GPUs have been constrained by the 300w PCI-E max power consumption limit. Over-limit cards have been produced, but those sell only to end users. Pre-build OEMs are very reluctant to install out-of-spec parts due to the increased support liability (you not only need to pick one out-of-spec part, but need to demand assurances from manufacturers of all interacting parts that operating alongside an out-of-spec part will not void their warranties too, which costs a lot more), which cuts out a large market segment.
It should be obvious but AMD is not 'better' than Nvidia, and Nvidia is not 'better' than AMD. A healthy competition between broadly equivalent parts is needed to keep prices for consumers down and drive for development up. But sticking your head in the sand and ignoring the realities of semiconductor development helps no-one.
Pfffffffffffffffft. In other news, AMD learns trolling.
Lets look at the article again.Firstly flat out the TDP contain mechanism pushes TDP to a much higher level which means under compute workloads big Maxwell will not be as efficient.
What is even funnier,is that you fail to realise on all levels that both the GK110 and Hawaii have MASSIVELY higher DP compute performance and those large memory buses and loads of RAM chips are there for certain GPGPU operations which are memory bandwidth limited. So you are comparing a stripped down gaming part with parts which have additional functionality plus Hawaii has audio DSPs consuming more power too. People like you did the same with Tahiti vs the GK104 too.
Lets look at the gaming power consumption figures in more detail though.
The same was noted with the GTX750TI(GK107) since TH uses very high speed measuring equipment - which again does not negate what I said before,or this slide you on purpose keep ignoring:
http://images.anandtech.com/doci/797...8-AM_575px.jpg
Maxwell is very peaky in power consumption but any body who had bothered to keep up to date would have known this for ages.
I keep many CPU and GPU rumours and info threads up to date on here and OcUK,so actually bother to read a lot of the bumpf AMD,Nvidia and Intel release.
What you fail to realise like with AMD,Nvidia has been developing a whole load of similar tech which is going into their SOCs,just like AMD. You are stuck in a microcosm without seeing the bigger picture at all.
AMD is at a massive process node disadvantage to Intel,so it needs to develop similar tech for its APUs too.
Read that slide again and read what Maxwell has.
Plus trying to sound cool by saying AMD needs a new uarch all the time is funny.
The last MAJOR uarch change from Nvidia was the GT200 to Fermi jump - the last for AMD was the jump from VLIW4/VLIW5 to GCN.
Or do you want me to explain each of the last few generations to you??
Agreed.
Looking at the last few years of GPU releases.
2002-2003 ATI is doomed due to the cancelation of the 8500XT and Nvidia having the TI4200 and TI4600
2003-2004 Nvidia is doomed since the R300 is first to DX9 and faster than the FX series.
2004-2005 ATI is doomed since the 6000 series Nvidia cards support DX9C and they don't. Nvidia is lower power consumption.
2005-2006 Shifts between ATI is doomed and Nvidia is doomed.
2006-2007 ATI is doomed as the G80 and G92 have better performance and lower power consumption than the HD2000 and HD3000 series
2008-2009 ATI is first doomed since they having nothing to compete with the GT200. Then AMD launches the R700,and now Nvidia is doomed as Nvidia has to price cut huge chips against AMD ones which are smaller
2009 - 2010 Nvidia is doomed since ATI/AMD is first to DX11,has lower power consumption and smaller dies than the GTX400 series. Although soon with the GTX460 series,AMD is doomed.
2010 -2011 AMD is partially doomed since they don't have the fastest card anymore,and Nvidia has better tessellation
Late 2011 to early 2012. AMD is doomed due to the HD7970 and even more doomed with the GTX680.
Late 2012 to middle 2013. AMD is doomed as they have nothing to compete with once the Geforce Titan and GTX780 are released
Late 2013. AMD is still doomed with the R9 290 and R9 290X releases,since Nvidia quickly launched the GTX780TI and the AMD cards have black screens,throttle and run too hot and explode all the time.
Late 2012 to 2013. Nvidia is doomed due to AMD winning console contracts.
Early 2014. AMD is doomed due to the GTX750TI
Late 2014. AMD is doomed due to the GM204.
Potential next doom point - Nvidia releases 20NM GM200/GM210 in small quantities at £1000 and even if AMD has the fastest card in the R9 390X at £500 before then is still doomed.
And so on.
When it comes to GPU releases people seem to have very short memories indeed. At every launch the standard stuff happens.Hilarious.
Its so predictable,so much so I predicted the response to this launch months before it came out. I believe I mentioned it here or on OcUK or some other forum.
Funnily enough I got quite good at predicting the response to Apple and Android launches for a few years too.
I will quote my last post again for you.
The hilarious doom and gloom cycles on the internet. Makes me laugh.
At every new GPU launch you can see people emerging from the woodwork making the same grandiose claims.
Its something I expect every year.
I bet you thought Nvidia was doomed when Fermi was first released?? ROFL.
So you can keep peddling your doom mongering repeatedly,but don't expect me to agree with you since history says another thing it appears.
So we can agree to disagree and keep it at that. I am getting a bit fed up having to repeat the same stuff again and again now.
Last edited by CAT-THE-FIFTH; 29-10-2014 at 12:31 PM.
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
LinkBack URL
About LinkBacks
Reply With Quote
