- Date:
- Tuesday , July 29, 2008
- Author:
- Kyle Bennett
- Google +1

Intel Atom vs.VIA Nano
A battle of the most powerful processors on the face of the Earth? Nope. Who sucks the least? Not really, but closer. Marketing teams will have you believe one thing, but we are here today to show HardOCP readers what they likely care about. Which one just works better, Atom or Nano?
Synthetic Benchmarks
We know how much you guys love these synthetic benchmarks when it comes to calling a spade a spade, but we all also know that many times these benchmarks don’t mean dick either.
Still they are always fun to see and certainly they go more to showing you areas that might be an issue or product bottleneck rather than really showing you what’s “right.” Below are four of the most popular synthetic benchmarks that are still used for CPU testing. Real world applications to follow.
SiSoft Sandra CPU Operation Benchmark
Dhrystone has been around for years and is a number crunching benchmark as are most of these synthetics we have here today. VIA’s Nano handily serves up a big plate of synthetic defeat to the Atom. At least we are seeing where the extra 10 watts the Nano system was pulling is going to. (Score: Atom-1 / Nano-2)
SiSoft Sandra Memory Bandwidth Benchmark
Sandra’s memory bandwidth numbers are STREAM based and measure sustained bandwidth, not spikes. Our Atom has a slower 533MHz memory bus, but was running with tight timings of 4-4-4. Our Nano was running at 666MHz bus with 5-5-5 timings. We ran these tests over and over and finally took the highest scores. While very close, the Nano edges out another win, but just barely. (Score: Atom-1 / Nano-3)
Super PI Mod v1.5 Benchmark
Mmm, love me some pie, and Super Pi ain’t bad either. Super Pi is a tried and true tool of tweakers and overclockers. This synthetic simply counts pie to a certain number of integers. In our scenario, 1 million digits. We simply time how many seconds it takes to do this. Lower is better and you can see that the Nano handles the Atom again with ease. Using another mod (HiperPI) that allows us to run 2 instances of Super Pi, we set it to run for 2 instances since our Intel Atom shows “two” cores. Remember that the Atom is Hyperthreading enabled. And while the 154 second scores looks to be a bad defeat, it is actually quite handsome. The Atom managed to squeeze in two workloads that took 95 seconds each into 154 seconds. A feat in itself! Still Nano gets the win, again. (Score: Atom-1 / Nano-4)
wPrime
If you are thinking wPrime is a prime number counting synthetic benchmark, you would be totally wrong. (Don’t blame me, I did not name the thing.) wPrime actually is a square root calculator. It was built to be well threaded, so unlike Super Pi that creates two workloads and sends one to each core, wPrime is threaded so that the one workload is distributed across however many cores designated. Technically the Atom only has one core (currently) but it does Hyperthreading. And as you can see above, it pays off when it comes to wPrime. Atom’s Hyperthreading edges out a needed Atom victory. (Score: Atom-2 / Nano-4)
