Why AMD is making weak products?

OK. Take the processor. The FX 8 processors although has 8 cores is weaker than the Intel i5. The AMD R9 is weaker than most Nvidia GPU. The processors are more affordable but not the GPUs. So why? I hate Intel is being a monopoly in the market. Let's bring the compeitition.

Update:

hmmm I'm thinking Wintel is kinda true. That theory that Windows only works well with Intel processors.... maybe that's why. Some Windows expert can shed a light here?

Update 2:

@anwar: I think you are confused with AMD APU and CPU. The AMD FX series are not APU!!! They are CPU without any integrated video. The APU well starts with A letter. A8, A10..... Those APUs are far weaker than the FX series.... I have and own one of APU and one of the CPU. The best AMD APU scores only 312 and the best CPU scores about 670 (Cinebench R15). If you think these synthetic benchmark is not accurate. Well most PC Game magazine uses them to compare between CPU and APU.

6 Answers

Relevance
  • Anonymous
    5 years ago
    Favorite Answer

    They got onto the wrong track with the introduction of their Bulldozer architecture, they bet on the wrong horse. They took the opinion that by this point multi-core products will be ruling, and single-core products would be going by the wayside. Well it was true, that single-core has virtually disappeared from existence, and pretty much everything is multicore now, but they didn't expect that despite multicore being prevalent, single core performance is still pretty important, even on multicore products. They were expecting that most programs would nowadays be fully embracing multithreading and programs will be happily spreading their workloads around multiple cores, but most programs are still basically single-threaded. The only use mult-cores have nowadays is that they allow many simultaneous single-threaded programs to be efficiently executed at the same time. That's why most people still find quad-core to be more than adequate, if all you're really using multicores for is to execute multiple single-threaded apps, then you really don't need six or eight cores for that.

    Another mistake they made with this architecture was they opted for a deep pipeline. A pipeline is a set of decoding stages for instructions, the deeper the pipeline, the higher the GHz that they can hit with the architecture, and they can hit some unbelievably impressive frequencies, 5.0 GHz for the FX-9590 for example. But the downside of deep pipelines is that if they make a mistake with a guess for which direction an instruction is going to go within the pipeline, they have to erase the entire pipeline and start from scratch with decoding those instructions again. This was in fact, the exact same mistake that Intel made with its Pentium 4 Netburst architecture, which could hit high frequencies 10 years ago, but most of it was wasted spinning its wheels when the instruction was trashed by a branch misprediction. Intel learned from that mistake by abandoning the Netburst architecture, and going back to the slightly older-fashioned design of the Core 2 architecture. Before that AMD was known for trashing Intel processors with greater efficiency processors, both in terms of power and performance. But unlike Intel, they couldn't abandon their large pipeline architectures, they had already invested too much into it, and they had to make do with whatever they had. They've finally announced that they're abandoning it for something they call Zen, which will go back to the older-fashioned shallower pipelines.

    As for their GPU architecture, that's also an entirely self-inflicted wound. But this time it was software-related rather than hardware like with their CPU's. Nvidia invests far more on their driver development than AMD does. Due to inefficiencies in the DX11 API, you have the ludicrous situation where the device driver is almost as important as the game code itself when it comes to getting good FPS in the games. The device drivers determine which game it is that they're playing, and then they determine the most efficient path through the DX11 API to best play this game, because the default path is almost never the best path to take. You see both the GPU giants doing this all of the time, you'll see a new anticipated game come out, and AMD and Nvidia will both announce new drivers to support this game. This is unique to the video driver world, you'll never see new SATA drivers if a new hard drive comes out, or new Bluetooth drivers if a new Bluetooth headset comes out. But in video drivers, you need to accommodate every new game individually.

    This happens because in DX11, you have no control over where the draw calls are made from. In fact, there is only one core ever assigned to making draw calls, that's why Intel's high-performance single-cores did so well. Now, AMD would make optimizations for games, but they stuck to the basic DX11 design code, so they only ever used one thread for draw calls. However, Nvidia went a little beyond the basic DX11 design and they were able to implement a rudimentary multithreaded model right under DX11, it's actually only a dual-threaded model, but that makes all of the difference in performance in the world between AMD and Nvidia. This also made Nvidia GPU's ironically much more suitable for AMD CPU's than AMD's own GPU's, because AMD provided all of these cores, and Nvidia actually made use of more than just one core.

    This all changes in DX12, here game developers themselves will have direct access to all of the cores in a processor, and they can use each and every single one of them to feed draw calls into the GPU's. This will mean that AMD will no longer have to waste resources in driver development for every game, the games themselves will do their own optimizations. But this will also mean that Nvidia will lose one of its biggest advantages over AMD, it will no longer control the optimization paths through DX12 like it did in DX11. It will put everybody in equal footing. This is probably AMD's desire anyway, because DX12 is heavily based on AMD's own Mantle API, which AMD probably wanted Microsoft to copy and standardize on.

    We're actually seeing that in DX12, AMD cards have a slight but definite advantage over Nvidia. AMD's GCN architecture can do asynchronous computing, which apparently Nvidia's architecture is only able to do in limited circumstances. Thus we're actually seeing Nvidia drop a few frames in DX12 vs. DX11, not a huge amount, but maybe 5% or less. So the cards to get will be AMD when it comes to Windows 10 and DX12, but Nvidia when it comes to Windows 7/8 & DX11.

  • TWB
    Lv 7
    5 years ago

    You see them as weak products because you do not understand their position in the market. They do not try to make the greatest and most powerful CPU. They make good solid middle to upper line CPU that offer good performance for the buck. If you analyze their performance compared to their price you see that they do far better them Intel. While Intel holds the king of the hill title for gamers at the top end those who do not want to spend a $1000 on CPU do just fine with AMD FX series. When you look at AMD APU A10 series they are far better then the Intel CPU with integrated GPU.

    As far as who makes the best discrete GPU is still and will always be up for debate. One year AMD holds the titles and the next year Nividia holds the title. Keep in mind when bench testing the top tier discrete graphics chips the winner is normally winning by just a few frames per second. It also depends on the game being used. Some games AMD is better and some games Nividia is better. It does get confusing trying to lay claim to the best and most powerful GPU. AMD just released their latest one and is now going to make a run at being king of the hill. As always it wont be long before Nividia makes a release. It is a never ending cycle.

    But I love watching them compete.

    Source(s): TWB 35+ years of experience in the service industry. You name it, I have probably fixed it. I did not design it, I did not build it, I did not break it, but I am the one who can fix it. What that means is that I have spent a life time taking these things apart and seeing how they are made. The difficult we do right away, the impossible just takes a little longer. I hear voices, so please be quite so I can listen to them.
  • Thomas
    Lv 7
    5 years ago

    Because AMD don't have as big an R&D budget. They've proven competitive in graphics cards at all but the highest price brackets. They've struggled more in CPUs though - AMD's "Bulldozer" architecture took a different approach to Intel's chips and one that hasn't worked out as well.

  • 5 years ago

    It all depends on how it is optimized and created. Intel CPUs are well, CPUs. AMD CPU are not really CPUs, they are APUs(Accelerated Processing Units). AMD focuses on graphics more than Intel did on their CPUs. That's why AMD CPUs/APUs run games better than Intel CPUs excluding the GPU. Playstation 3 and Playstation 4 uses AMD APUs in their system and are able to run games at 720p30fps and above. Intel CPUs can't do this.

    Nvidia GPUs have better drivers and usually lower specs than AMD GPUs. This is because of how the systems worked. Nvidia hardware allows faster rendering, and tessellation compared to AMD. However, with the recent DirectX 12, AMD GPUs improved massively compared to Nvidia.

  • How do you think about the answers? You can sign in to vote the answer.
  • Anonymous
    5 years ago

    well tobe fair the fx series is like 3 years old now I think and they are apparently coming out with their new zen CPUs soon.

  • Flexon
    Lv 6
    5 years ago

    Well why don't you join AMD and show them how it's done.

    I agree that it's not good for the competition though,

Still have questions? Get your answers by asking now.