Thoughs on the AMDs, DX12. and why Intel is better than AMD "discussion"

[gonf]

Please don't reply if you are just saying stuff like Intel is the way or stuff like that. I know Intel is faster than AMD for now etc etc. this post is about AMD not Intel. so no fanboy stuff please.

Let start with watching this video. AMD explaining what they are tying to do with their gpu first.
https://www.youtube.com/watch?v=v3dUhep0rBs

I remember reading something like this few years back explaining why Intel is faster "more Efficient" than AMD. so I always have this though in my head that the reason Intel is over AMD is software stuff....let go back to a bit about basic 0 and 1 computing. If I remember correctly back in the days. each time a CPU clock go from 1 to 0 our software can execute a command. that is why the faster the 1 to 0 and back to 1 the faster our computer will be. that is the hz of our cpu which is 1/4.7ghz will be our time of execution. at least that how it works for my microcontroller and my microcontroller class at my college years. that is the reason why AMD is slower to Intel with higher clock rate have been a mystery to me for many years.
I also remember reading something about AMD been trying to push their software to game developer "which fail they did" few years back. until mantle come out. the interesting is that MS jump right in saying "we will have our DX12 out very soon and it will be better than mantle" which how i see it as MS saying we will finally write code for your AMD in our DX series since if we don't you will write your own and push it out.

here comes the DX12/mantle added to the equation.
http://www.tomshardware.com/reviews/amd-mantle-performance-benchmark,3860-4.html

does this mean AMD is closing in finally or there is another layer of code AMD need to upgrade before they can finally do the job?
Let say if Dx12 is all they need. will that help their older CPU let say the FX8350? or it only about the new one that come out in 2015 or later?

 

[Supernova1138]

I don't think DirectX 12 is going to be particularly relevant for the current FX chips, simply because even if the API becomes available this year, it's still going to be about 2 years before most games will even support it. Most developers aren't going to be rushing to incorporate a new API into their code really late into development. Supposedly the Xbone might support DirectX 12, and maybe that will accelerate things a little bit, but I see DirectX 12 being adopted at a similar rate that DirectX 11 was, we'll see maybe two or three games next year that support DirectX 12, and it won't be until 2017 that we see more widespread adoption.

If DirectX 12 is going to help AMD, it's probably going to be on their rumoured Zen CPUs slated for next year, which are apparently also moving in the 'more cores' direction with a 16 core consumer chip rumoured. By the time you see a lot of games support DirectX 12, the FX chips are going to be 5 to 6 years old and probably be bordering on obsolete at that point.

 

[gonf]

it started sooner then you though.
take a look at this video
http://channel9.msdn.com/Events/CPP/C-PP-Con-2014/How-Ubisoft-Develops-Games-for-Multicore-Before-and-After-CPP11

Game Developer have been making games for the xbox1 and Ps4 which both are using AMD Multicore system. to port the game from a Xbox1/Ps4 to PC. if most of the programming is alrdy done on the first place. why wouldn't they use the new API?

 

[synphul]

In a sense, if I understand correctly dx12 is supposed to refine things in such a way as to reduce load on the cpu. Even the current top end high dollar gpu's aren't bottlenecking decent cpu's unless in heavy sli/crossfire (which has it's own headaches). The way I see it, gpu's are the ones that need to improve. Current mid to high end cpu's are already waiting on gpu's to catch up in terms of balanced performance even without the boost of dx12. Once dx12 is implemented, gpu's will be even further behind playing catch up. Hopefully dx12 brings some sort of boost to gpu's as well.

"Obviously in-game performance won’t simply double just by switching from DX11 to DX12. Performance in CPU limited scenarios will definitely see a great improvement. But in GPU limited scenarios the performance improvement will be more subtle."

Read more: http://wccftech.com/dx12-revealed-compared-dx11/#ixzz3YHEIacok

 

[gonf]

how I see it.
DX12 is what MS have to do because mantle came out. They need DX12 so everyone will stay on windows to play games so people will keep paying for windows. it like a chicken and egg thing.

here is another link about AMD with DX12
http://wccftech.com/amd-directx-12-async-shaders-multithreaded-command-buffer-dx12/

and yes CPU is faster then GPU. but isn't thats the point of Computer? CPU should be the Core of computer. everything should be in sink with the CPU clock. it like the sun in our solar system. every thing should spin around it.
but yes GPU is very important in gaming. but CPU should still be the core of the pc.

and I'm hoping DX12 will up some what of a % on graphic texture. like every dx upgrade they will up some kinda water effect or wind effect. like back in the days

 

[freeskier93]

AMD's issues have nothing to do with software and everything to do with architecture and node size.

 

[Calculatron]

AMD's issues have nothing to do with software and everything to do with architecture and node size.

I'm sorry, I can't hear you over the sound of Tomb Raider benchmarks.

 

[gonf]

AMD's issues have nothing to do with software and everything to do with architecture and node size.

Please go in deeper. of why and how. we are open mind to true reason.
here is a older video of explain why it is somewhat software and code of why AMD is slower than Intel. I know it a 2013 video. but that is the reason why AMD pull out mantle.

https://www.youtube.com/watch?v=A9YwkSLqpXw

 

[freeskier93]

AMD's issues have nothing to do with software and everything to do with architecture and node size.

Please go in deeper. of why and how. we are open mind to true reason.
here is a older video of explain why it is somewhat software and code of why AMD is slower than Intel. I know it a 2013 video. but that is the reason why AMD pull out mantle.

https://www.youtube.com/watch?v=A9YwkSLqpXw

I'm sorry, but this discussion has been rehashed so many times over the years it's not hard to do your own research. I have better things to do then write an essay on Intel vs. AMD, so I'll keep it short.

First of all, I should make sure I'm not going crazy and we are in the CPU section right? DirectX and Mantle are graphics APIs, I'm not sure why the belong in a discussion about raw CPU performance, not that any amount of software will magically make your CPU more powerful. When it comes to raw performance/efficiency metrics AMD is still lagging behind Intel. This really shouldn't be news to anyone, AMDs benefit has always been the price to performance ratio.

Maybe it will help if I mention that my desktop is built with an AMD A8-5600k. I root for AMD, I really hope they can turn things around next year with the release of their new Zen core architecture, but if Zen is a flop they aren't going to survive against Intel in the consumer market.

 

[gonf]

CPU performance and gpu performance are somewhat link to one and other.
Dx12 and mantle are the API which not only about gpu. It is also about CPU too. please read some of the link.
It not magic that software will change the performance of computing. it how computing work in basic
if you don't want to go deeper to explain anything. Than please don't post here. Most of the post over the years are just peopled saying Intel is more efficiency and perform better. But they don't go too deep to expand. They will say because Intel will do more stuff in lower Hz etc etc. that will be as deep as they get.
also if you don't want to write essay. you can post a link.

http://en.wikipedia.org/wiki/Megahertz_myth

it doesn't have to be 100% on topics any interesting link with useful information is welcome in here. because this is an "discussion"

I know Intel uses lower power but please this post is not about that.

 

[synphul]

Intel vs amd isn't about gpu's and dx12. Those are two totally different areas. Intel's edge over amd isn't even so much the power efficiency as it is the architecture. Frequency or speed in ghz between the two means nothing, the only time it matters is to compare cpu's from the same architecture, brand and family. It's exactly because intel can do more per hertz, or instructions per clock. That's at the core foundation of the cpu's abilities and translates to performance throughout the range of what a cpu does. It would take a very long time to delve into all the intricacies of cpu architecture and construction to sort out the details so the easiest way to quickly sort out performance variation are benchmarks.

At the end of the day, it's much less important 'why' one cpu performs better and more important that it 'does'. Synthetic benchmarks can give a rough idea but usually they're optimized to max out a cpu fully and may not tell the whole story. People don't run synthetic programs, they run real world programs which is why real world benchmarks are important.

For instance cpu 1 may have a higher sysmark score than cpu 2, but cpu 2 may render faster in adobe. Or it may encode movies faster or compress files into zip/rar formats faster. It can also vary from one type of program to another. The best way is to find real world benchmarks comparing cpu's for the tasks you actually do. If you need to compress a lot of files on a regular basis, it doesn't matter if one cpu scores higher on a synthetic benchmark. Pick the one that compresses faster, or whatever the task(s) is/are you need to do. Having a faster benchmark score means nothing if in real world use you're waiting longer to do your work.

 

[gonf]

Yes.
let say you need to move 1000L of water from swimming pool A to swimming pool B.
CPU A is 6 people who can run speed 10 with water bucket that hold 200ML.
CPU B is 8 people who can run speed 12 with water bucket that hold 150ml.
But the road will only let you pass 2 people at a time for CPU b but 3 people for CPU a
The road would be the limitation of code and programming.
So yes it hard to compare the 2 and it not just a 1+1=2. There are too many thing in the equation.
And that is why we use speed of completing a task as a benchmark.
But outcome can be easily differs if the programmer decide to use different kind of code.