Slashdot Mirror
VIA Nano CPU Benchmarked, Beats Intel Atom
Vigile writes "Back in May, when the Isaiah architecture was first disclosed, VIA declared a performance victory over Intel's upcoming Silverthorne technology. Since then, Isaiah has become the VIA Nano processor, and Silverthorne changed to the Intel Atom — and now we can finally see tests comparing the two technologies. The Nano's out-of-order super-scalar design is definitely an architectural leap over the Atom's in-order single-issue design, but with Intel including HyperThreading technology in their CPU the competition is closer than expected. The Nano does win the performance tests by a considerable margin, but what might be more impressive is seeing the Atom use only 4 watts of power under full load!" As reader Mierdaan points out, that's 4 watts more than at idle, for about 60 watts total.
If you're not in a rush to get one of the Atom/Nano based computers, wait for the next generation. Although both CPU:s are excellent in performance, the next iterations will bring two cores and far better efficiency.
The first generation of any product line is usually fairly rushed and experimental. That does not mean the product itself is bad, but we should expect a big jump from the next generation.
Full Tilt
I took my graduate level architecture class from Dean Tullsen at UCSD, who invented 'hyperthreading' although it was called Symmetrical MultiThreading (SMT) back then. As I recall the entire greatness of the architecture was recognizing that all the fancy hardware introduced to allow out-of-order speculative execution could actually be leveraged to allow the processor to drive multiple independent threads at the same time, without much additional overhead. So if intel's atom (haven't been following it) uses an in-order core and hyperthreading that just don't make much sense. Anyone care to provide an explanation?
Imagine that the HD uses 20W during the test. If the Atom takes 120s to complete it, that's 240 joules of the total that come from the HD; if the Nano takes 60s, that's 120 joules from the HD. My point is that, if you use a low power device, both lines would go down by the same amount, so the integral for the Atom would go down by more than the one for the Nano. This effect could be big enough to make the Atom more attractive.
Unfortunately, they decided to use a power hog to test the energy use of low power systems, making their test very flawed and, therefore, unreliable. I'll wait for a better test before making any conclusion.
I was liking both processors up until I found this gem in the article:
Here we see Intel, up to its obnoxious "You'll use our technology only as we prescribe" games. This is the same philosophy that leisurely milked the market for 33Mhz CPU bumps every 6 months, while they sat on years worth of better technology, until AMD lit a fire under their ass.
Don't be fooled again.
Reading through that article, I found this:
My my. Swap CentaurHauls for AuthenticAMD, and Nano's performance magically jumps about 10 percent. Swap for GenuineIntel, and memory performance goes up no less than 47.4 percent. This is not a test error or random occurance; I benchmarked each CPUID multiple times across multiple reboots on completely clean Windows XP installations. The gains themselves are not confined to a small group of tests within the memory subsystem evaluation, but stretch across the entire series of read/write tests. Only the memory latency results remain unchanged between the two CPUIDs.
Whoops! I wonder what they'll have to say about that...
The driver model was wretched in Win98. Plus it was still DOS under the hood. No, going NT-based with XP was a good move on Microsoft's part. Now if only they had gone ahead with Neptune instead of Windows ME, they would have saved themselves a major embarassment.
The Via Nano is currently built on a 65nm process and the Intel Atom on 45nm so Via has some room if they move to the 45nm process.
Makes me wonder if building the Atom on 45nm is costing them production of high price multi-core chips in order to squeeze themselves into this UMP market before the sector solidifies more. And seeing things like the TI OMAP 35xx chips it makes me wonder when the UMP market moves off x86. There is already pressure to bump up the price of the UMP as to not take away from the lowend laptop market.
LoB
"Anyone who stands out in the middle of a road looks like roadkill to me." --Linus
You know why they did that don't you?
The atom has next to no features by itself. It really is a "dumb-terminal" amongst CPUs.
The Nano is capable of much more, even with its chipset. Namely, last I checked it's capable of playing h.264 at 1080p, and can play Crysis.