Sik, on 28 May 2012 - 06:17 PM, said:
Overlord, on 28 May 2012 - 02:30 PM, said:
ARM will become viable as a main desktop platform about the same time people stop caring about x86 compatibility.
IE never =P
IE never =P
Well, it isn't like it isn't possible to dynamically recompile CISC x86 instructions to RISC ARM instructions on the fly. QEMU's emulation technology works by doing that. A more specialized form could be applied to Windows if Microsoft wanted to preserve Win32/Win64. Apple was rather famous for using such a technology for the PowerPC to x86 transition. It also already exists in Android and desktop Linux. But the truth is, Microsoft wants to drop the legacy APIs because they want to change the underpinnings of Windows yet again. The Midori project should be nearly ready for commercialization midway through Windows 8's lifecycle, which is just in time to consider switching out the internals of Windows for Windows 9. Microsoft wants to make the transition to a fully managed environment for its operating systems. Even the XBOX 720 is being prepped for a fully managed environment. Native code programming is a dying art in Microsoft land.
AerosolSP, on 26 May 2012 - 07:27 PM, said:
Hmm. Those AMD APUs look interesting. They seem to be a bit behind when it comes to gaming performance though, but it's difficult to find benchmarks for them with setups that take advantage of Fusion's benefits. Atleast, if I'm right in saying that you won't really see performance gains in Fusion system until you start crossfiring 2 Radeon's with one of these APUs?
What I'm trying to say is that most things aren't really optimized for an APU like this so it's hard to recommend one for a gaming build, right now.
EDIT: And what happens if Intel and Nvidia decide to jump into bed together?
What I'm trying to say is that most things aren't really optimized for an APU like this so it's hard to recommend one for a gaming build, right now.
EDIT: And what happens if Intel and Nvidia decide to jump into bed together?
Well, the APU itself is quite powerful. It's a Radeon HD 6000 series GPU. It's miles better than the GPU included on Sandy Bridge and Ivy Bridge. The major benefit is that if you install a dedicated Radeon HD card as well, you can CrossFire it to get far better performance. While I personally think anything more than two GPUs is overkill, I've seen people do 11 GPU CrossFireX combos (CrossFire with five Radeon HD 6870 X2 cards and an AMD APU). Supercomputer performance for under $2000 is pretty awesome, but obviously overkill.
Because of the way that AMD structures CrossFire, pretty much all games can take advantage of it, and several non-games can too. But if you don't need CrossFire, the APU's GPU is powerful enough that you can leave the PCI Express slot open for future expansion.
As for an Intel+NVIDIA tie-up? Not likely to happen. Aside from the fact that the U.S. anti-competition laws would kick in and cause the Department of Justice to jump down Intel's throat for this, NVIDIA and Intel are actually on the outs. They have been nominal partners for the last few years, since NVIDIA decided it wanted to develop its own CPUs. Briefly, they developed designs for x86 CPUs, but then they jumped headfirst into ARM with Tegra. In any case, it took some time before Intel approved NVIDIA northbridge chips for Intel motherboards after the latest socket change. The relationship between the two companies is quite strained. The likelihood of Intel buying out NVIDIA is low, and if such a thing came to be, the U.S. government would immediately block it since Intel already has a graphics platform to work with. They don't actually need NVIDIA for anything. AMD bought ATI in order to create APUs in order to solve the IGP problem and to compete with Intel on the integrated system platform front. The fact is, Intel does not care about its graphics platform, which is why it suffers so much compared to all of its competitors in the graphics field.
This post has been edited by Conan Kudo: 29 May 2012 - 12:32 AM
