Microprocessor Report: Intel, Fix the PC!
Gwennap stated that persistent low growth rates are evidence that the PC processor is in a crisis, which may be partially homemade. He criticized the fact that performance of CPUs isn't growing fast enough to find the interest of consumers.
"Even counting a modest boost for the new Sandy Bridge CPU, performance is increasing at just 10 percent per year for desktops and 16 percent for laptops, a far cry from the good old days of 60 percent annual performance increases. As a result, PC users have little incentive to upgrade their systems," Gwennap wrote.
The key for Intel to remain successful would not be so much in mobile application processors that could barely make up for 5 percent of Intel's processor revenue, if Intel were to hold all the available market outside of Apple and Samsung, but to "stay the course" and make PCs more competitive again.
"It can’t do the job alone, but it knows how to work with Microsoft, PC vendors, and other ecosystem players," he wrote. "Having redefined the PC form factor, the next step is to rework the PC’s functions to include performance-hungry capabilities such as voice/gesture recognition and intelligent agents in ways that benefit end users."
Of course, making the processor special again could help as well. In 2005, Intel changed its marketing strategy to turn microprocessors into a commodity product that was sold with relatively cryptic sequence numbers. The idea was to not market the clock speed, but integrate a CPU into a PC almost invisibly as the engine that enables a certain form factor and feature set. Seven years later, the CPU and its sequence number has become largely meaningless to mainstream customers. Perhaps it would be a good idea to highlight the strength and performance of x86 once again.
That, and the comfort of sitting at a desk and doing your work with a keyboard/mouse.
Also, one reason why the CPU performance jumps aren't increasing all that much is because there really isn't a demand for it, why? Because the Top of the line CPU's right now are handling every game on the market just fine.
Shit I have an Intel I5 2500K and its still handling all games to date at max graphics just fine. Why get a new CPU with a boost of 60%?
Game developers need to start making games that'll truly challenge the PC through graphics (not bad programming) but unfortunately they want to make them for the Consoles as well, (actually other way around). So now when they port it to PC the graphics are all compressed and crap.
The days of PC specific games, that are truly made for keyboard/mouse, seem to be gone.
But anyways, PC won't disappear any time soon. Analysts are f--king idiots if they think "PC IS DEAD DUR".
They don't know what animating, gaming, programming, etc requires. Obviously. Ever try playing far cry 3 on a tablet? Animate a video on a tablet? Process larges amounts of data on a tablet? Create CGI on a tablet for a movie?
Take almost any PC sold within the past 4-5 years.
Try and do anything the average consumer uses a PC for.
If it works fine for them, why would the masses upgrade?
Its not that people are waiting for bigger performance gains before upgrading, that is only the small enthusiast market. Its that people no longer have a need to upgrade, processors have long reached the speed of "good enough" for most consumers
Analyst are concerned about market growth, revenue, margins and broader consumer buyers not the small percentage of users.
If all the lay folks stop buying desktops/laptops for their use cases Intel will be crushed. Not saying this is what's going to happen, but the analyst is saying they need major improvements in PC CPU to compel people to want to buy PCs to avoid the chance that tablets, or some hybrid tablet with keyboard docks, etc will takeover
Only a smart guy like me for whom a non-removable battery and non-accessible memory slots (that is, if the RAM is not soldered to the board) are a deal breaker, but even I feel myself giving in to the temptation.
60% annually? what a load of bullcrap. Some of us actually have a working memory in our brains... yes, 386->486 was 100% increase in arhitecture alone. 486-> P5 was almost 100%, awful Pentium 4 (netburst) to amazing core arhitecture was a jump forward... But thats like once every 10 years, while every year we were lucky to see 10% increase in frequency or sth..
Not to say some of us need more powerful opencl hardware.
Look at their 'Next Unit of Computing' systems.
Look at Haswell, with it's very modest performance increases, but huge efficiency jump.
Look at Broadwell, which they have already dropped the PGA package and may be dropping the LGA package depending on Haswell's sales.
Look at Clover Trail, a half baked chipset that they rushed to market ASAP because they saw where the table market is going.
Enthusiasts and power users are going to have to get used to the fact that there are billions of users who just want to get on the internet and do some word processing. Xeon will always be there for the power user, but Core is going mobile.
So in the end it is really money+ laziness = our current situation (PC gaming sector)
Gamers make up a small % percentage of the global desktop and laptop consumers
The ATX mainboard form is sadly out of date, and there seems to be no mainboard manufacturer willing to support someone who might want to build a system with a decent RAID card, 3 graphics cards, and a decent sound card. If you want any one of those you are okay, but all 3, you are SOL.
Could we get like at least a PCIe x1 slot above the first PCIe 3 x16 slot? Could we get some PCIe slots that aren't covered up by PCI x16 video cards? Hasn't anyone figured out a modern high end graphics card takes up at least 2 expansion slots worth of room?
I have a 2560x1440 LCD that can overclock to 120Hz, and I'd like to run 3 of them in surround, but I'd also like to slap in a decent sound card, too. I don't really need a RAID card (well of course it's all one giant toy anyway, so I don't "need" any of it, heh), but I don't see why I can't build a high-end rig that's really high-end all the way. I could use the power today - and what will the future bring? What happens when we start getting real high-density monitors, capable of running decent refresh rates to support those high frame rates that give us nice smoothness when we pan our cameras in games?
I'm looking forward to Haswell, but at the same time, I wish the whole industry would get out of yesteryear thinking and bring us into what we all know is possible.
Will never happen.
X86, and X86 compatibility is here to stay.
Why?
Because the business world demands it... Too much time, money, and training invested in it.
Besides, the arch is fine, its the software...
- Star-trek like voice recognition (ability to detect free speech of anyone without any training)
- Star-trek like speech generation (natural voice, indistinguishable from a normal human voice)
- A true 3D immersive user interface, fun to use (in which I could "walk" inside my computer's content and place and arrange items spatially)
- Image content recognition (categorize my thousand photos by what's inside them, without forcing me to "tag" or "put in folders")
- AI capability (let me talk with my computer and ask him his opinion on various stuff that he's aware of)
- VR headsets for games and interactive movies (ability to watch a movie's action from the perspective I want, like in a ST holodeck)
(and many more stuff we've seen in movies)
We're still stuck with mouse/keyboard and clicking links and buttons on webpages. (Everything has just become bigger, to adapt to the higher resolution screens of today). Can't really call that progress...
And now a tablet, small laptop or other internet-enabled device can do that, why on earth would a common person need many-core CPU's which eat 100-200Watts of energy in a big case, just to point and click on a website?
Yes, give us software stuff that a tablet/small laptop CAN'T do, and we'll definitely buy new powerful desktop PC's !
I want the good old days of 50%+ every 6-12 months back!