Intel Launches Next-Gen Atom N450 Processor 165
MojoKid writes "Intel has unveiled its next-generation Atom N450 processor, and a review of the new Asus Eee PC 1005PE netbook that houses it shows decent gains in performance and lower power consumption. The Atom N450 has been re-architected similar to Intel's other notebook processors in that it now has an integrated memory controller and graphics core on the CPU itself. In addition, Intel's serial DMI (Direct Media Interface) now replaces the system bus to the Southbridge IO controller. From a performance standpoint, the Atom N450 single core chip offers a nice performance gain versus previous generation Atom CPUs and it appears Intel has dual-core variants of the chip on the horizon as well."
Re:Still chokes on flash? (Score:4, Informative)
If you think Flash sucks on Windows then obviously you've never seen it run on Mac OS X. Adobe is a complete disgrace on that OS.
RTFA, please (Score:2, Informative)
Re:Still chokes on flash? (Score:5, Informative)
If you think Flash sucks on Mac OS X then obviously you've never seen it run in Linux. Adobe is a complete disgrace on that OS.
Re:So... (Score:3, Informative)
It is, supposedly, X3150, so basically the same part that's in G31. 3100/X3100? Anyway, seems it's "proper" Intel GMA, with good Linux support.
Re:RTFA, please (Score:4, Informative)
Makes no sense: the 945G and variants had a GMA 950.
Re:RTFA, please (Score:3, Informative)
RTFA: "The graphics core is a basic DX9 instantiation that is a kin to Intel's GMA500 graphics core in the previous generation Intel 945G chipset"
I have a 945GM system whose graphics part is called GMA950. It uses the common opensource Intel drivers. On the contrary, GMA500 aka Poulsbo is the problematic one with closed drivers.
http://en.wikipedia.org/wiki/Intel_GMA [wikipedia.org]
Re:Intel and Linux (Score:3, Informative)
I also have a small tangental question. I always hear about huge performance gains that can come from properly writing code to take advantage of SSE2,3,4,etc instruction sets. I also hear that almost no one does write code to take advantage of these instruction sets. If Intel really wants to push their hardware, why not write such optimizations for the Linux kernel?
The kernel doesn't do much CPU-bound processing. It is math and media libraries where these vector instructions would be actually useful. You can already have some of their advantages using a decent compiler. Basically, that means different binaries for processors with different capabilities, so your average binary distro is not going to have any fancy instructions. I suggest trying Gentoo if you actually want to use your modern CPU.
Re:Intel and Linux (Score:4, Informative)
Your post completely missed the original poster's point - the Intel GMA500 is a major outlier in terms of Linux support.
The GMA950 series is well supported by Linux (with the exception of the re-architecture issues that hurt Ubuntu 9.04 so badly).
The GMA500 is simply minimally supported in Linux and all indications state that it will stay this way. The GMA500 graphics core was outsourced to another company, as was driver development.
As to SSE2/3/4 - They only benefit for certain operation types. Most kernel ops won't benefit, and also, using SSE usually means hand-coding in assembler - compilers that generate good vector SIMD code are rare. The kernel developers tend to prefer to avoid hand-coded ASM whenever possible.
However, I do recall that RAID checksumming code and memcpy() were once implemented using MMX to improve them, so these sections might benefit from SSE (and might already do so.)
Re:Re-Architecting English (Score:2, Informative)
Yes, the architecture changed: No more FSB, which also means no more alternative chipsets. The only chipset available for the new Atoms is Intel's one-chip NM-10. Other changes are not really architectural changes but would not have been possible without the abandonment of the FSB architecture: The analog video output is limited to 1440x1050 and the LVDS port for the LCD only drives up to 1366x768. Intel would not have dared crippling the chip so seriously if manufacturers could circumvent it by using a different chipset and a separate graphics core.
Re:Midnight Blue? (Score:5, Informative)
Once you have removed the stickers, you are often left with difficult to remove adhesive gunk on the laptop. An easy way of removing the gunk without damaging or scratching the surface is to spray a little silicone based lubricant in the area and wipe with a paper towel. It quickly wipes off and the silicone lubricant won't damage plastic like petroleum based lubricants (like WD-40) sometimes do.
Re:Still chokes on flash? (Score:3, Informative)
On the other hand Flash Player for linux is the only x64 flash player out there.
Re:Intel and Linux (Score:3, Informative)
The various SSE instruction sets provide SIMD instructions, which is an acronym for "single instruction, multiple data". As the name suggests, they allow you to perform operations on multiple pieces of data with a single instruction. SIMD is great for media applications, where you often have to do the same mathematical operations over and over again to lots of data at once, however pretty much all of the stuff that happens in a kernel is logic-heavy tasks that only deal with single pieces of data at a time, and thus can't really take advantage of these instructions in any way.
Re:RTFA, please (Score:3, Informative)
That doesn't make any sense. The 945 chipset uses the GMA950, the GMA500 is actually a totally-outsourced PowerVR chip. The 'native' Intel chips (i810 through G45) are all tatally supported by Intel's open-source drivers, the GMA500 is almost impossible to get working in Linux.
The new built-in N450, D410, and D510 graphics chips are based on the GMA3100, if I recall, they're even called 'GMA3150'. That means they're supported by open-source drivers (and possibly by Mac OS X!), but the performance is bad enough that even Google Earth will make you want to cry.
They really should have used the G45 series of graphics for these things, instead of the G33. They're -worlds- apart in functionality and performance.
Re:meanwhile, where are the ARMs? (Score:4, Informative)
http://www.arm.com/markets/embedded_solutions/armpp/25333.html [arm.com]
Doesn't take much looking.
Re:Power use? (Score:4, Informative)
Even the original Atoms used less power than the most power-efficient single-core AMD platform.
Platform TDP for the Yukon platform (RS690E northbridge, SB600 southbridge) ranges from 19 watts with a 1 GHz Sempron, to 26 for a 1.6 GHz Athlon. (29 for a dual-core 1.6 GHz Turion.) The most efficient Athlon-based Yukon is 1.2 GHz, and platform power consumption is 24 watts.
Platform TDP for the typical N270+945GSE+ICH7M is 11.8 watts, N450+NM10 is 7 watts. Granted, the Yukon stuff doesn't really compete with the Atom, it competes with Intel CULV.
CULV has a 14.5 watt chipset (GS45, ICH9M) TDP, add 5.5 watts for single-core, 10 watts for dual-core CPUs.
Oh, and I'll toss the VIA Nano in, it fits somewhere between the Atom and the CULV and Yukon platforms in performance.
The fastest current Nanos for netbooks are the U2225 and U2250, both at 1.3 GHz (the U2250 is at "1.3+ GHz") and 8 W TDP. (IIRC, though, the Nano is significantly faster than Atom.) The matching VX800U chipset has a 3.5 W TDP, so 11.5 W total platform TDP - less than the old Atom platform.
The upcoming U3200 is at 1.4 GHz (and even faster than the clockspeed implies, apparently,) possibly 5 W TDP, and 2.3 W for the VX855, so 7.3 W platform TDP.
Re:Midnight Blue? (Score:3, Informative)
I prefer orange oil based cleaners. They are often marketed as label or gum removers.
Not only do they smell good, they also don't damage plastics. Oh and they're also a great insecticide and will keep ants away because all insects hate the smell - after all the oil is the oranges' natural defense.