The question posed was about desktops. ARM is already in the household, just in the form of iPads, Fire and phones.
See, this is exactly what will make ARM work. Simply use a custom Linux fork (Steam OS, Fire OS) and then make sure your apps work with that fork and the end user will never know the difference. You buy a "FirePC" from Amazon, download apps happily from their AppStore and play your simple games without ever noticing that you aren't in a Windows environment. And it doesn't necessarily have to be a desktop, could just as easily be a laptop (equivalent to Chromebook perhaps).
For 'near future' let's say 2020.
It just seems like ARM processors are catching up to x86 lightning fast.
Until they run out of the low hanging fruit. A9/A9X have already gotten to extremes. And the die size with it.
Though I agree with you, we've been hearing that since Apple A7 and they kept on surprising us.Until they run out of the low hanging fruit. A9/A9X have already gotten to extremes. And the die size with it.
Though I agree with you, we've been hearing that since Apple A7 and they kept on surprising us.
A9X IPC is similar to Core. So when Core frequency is low, A9X is competitive. But it certainly is no match when Core can use higher frequencies.The A9X is still slower than Core.
147 mm2 is the whole SoC area. The CPU itself is much smaller than that.Yet a dual core A9X 147mm2 on 16nm
Remind me how large Intel L3 is exactly?with 51.2GB/sec memory, twice the L1 cache, 3MB L2 cache and quite power hungry.
Do you really think simple brute force can lead to where their CPU is? Only someone with no CPU design experience can think so.Apple simply turned towards brute force.
"Will it play Crysis?" << actually an irrelevant question in this discussion.
People will run their software on the cheapest hardware that will run it, assuming broadly equivalent performance. The fastest ARM CPUs are performance- and cost-competitive with low-TDP mobile x86 CPUs, but the software doesn't really exist yet to make them usable.
When it does? People who don't care will buy whichever is on sale.
</prediction>
A9X IPC is similar to Core. So when Core frequency is low, A9X is competitive. But it certainly is no match when Core can use higher frequencies.
147 mm2 is the whole SoC area. The CPU itself is much smaller than that.
Remind me how large Intel L3 is exactly?
Do you really think simple brute force can lead to where their CPU is? Only someone with no CPU design experience can think so.
Let's see if they can still improve things. I would be surprised if they could, but as I wrote, they've surprised many knowledgeable people.
And epic software needs EPIC instructions so that means HL3 is Itanium only confirmed.not likely because complex applications need complex instructions.
Until they run out of the low hanging fruit. A9/A9X have already gotten to extremes.
The IPC isn't similar.
And when scaling up it goes bad for A9X. Its easy to make high IPC, its hard to make high IPC and high clocks.
A dual core GT2 is what, 89mm2? And that's with PCIe links too. The cores themselves are what, 8mm2 each? So that's pretty much equal for the 2.
The L3 cache is 4MB on the compared products. But the L2 is 256KB. I thought you knew a L3 would be slower due to latency. Memory is 29.8GB/sec contra 51.2GB/sec. The A9 due to its slower memory got a 4MB L3.
Brute force can lead many places when money is of no concern.
Now, if Microsoft brought back an ARM version of Windows and Office, I could see that happening for the general office computer.
Maybe not desktops, but I see a potential in thin client (which are replacing desktops).
And when scaling up it goes bad for A9X. Its easy to make high IPC, its hard to make high IPC and high clocks
Uh, we have a couple versions of Office on ARM. In fact Microsoft pays companies to pre-install a version on their devices:
Right. However the mobile Office offerings from Microsoft are heavily dumbed down feature wise.