Hi,
I'm considering buying a netbook based on an Intel Atom processor, and wonder what engine performance to expect on such platform. Benchmarks I read so far are not always relevant for this kind of activity.
So has anybody got the number of nodes per second achieved for free engines like Fruit, Toga, Crafty, Glaurung, etc ? And for Rybka ?
Pascal Georges
Intel Atom processor and nodes per second
Moderator: Ras
-
Carey
- Posts: 313
- Joined: Wed Mar 08, 2006 8:18 pm
Re: Intel Atom processor and nodes per second
The reviews I've read say the Atom is very under powered for a netbook or low end notebook.
It's only suitable for the more limited mobile kinds of devices. Supposedly, that is what Intel is actually aiming the device for but manufacturers are doing it in sort of a grey area low end netbooks because they'd rather deal with Intel than other vendors.
Intel isn't minding the extra sales, but I don't think they are actually marketing the Atom for those devices.
As you point out, chess is a little different from other things, but it's still basic raw computation, so any benchmark for just plain computation (superpi, etc.) would probably be okay. I'd stay away from video playback benchmarks though, since video decoding can take advantage of SSE etc.
Also, doesn't the Atom have a deep pipeline? If so, that's going to hurt, just like it does on the P4.
So personally, I'd just call it a low end, stripped down P4 and assume performance would be pretty bad.
That's not to say that netbooks based on other low power processors are fast either. But devices designed for netbooks & low end laptops are likely to be faster. Just maybe not quite as power friendly.
It's only suitable for the more limited mobile kinds of devices. Supposedly, that is what Intel is actually aiming the device for but manufacturers are doing it in sort of a grey area low end netbooks because they'd rather deal with Intel than other vendors.
Intel isn't minding the extra sales, but I don't think they are actually marketing the Atom for those devices.
As you point out, chess is a little different from other things, but it's still basic raw computation, so any benchmark for just plain computation (superpi, etc.) would probably be okay. I'd stay away from video playback benchmarks though, since video decoding can take advantage of SSE etc.
Also, doesn't the Atom have a deep pipeline? If so, that's going to hurt, just like it does on the P4.
So personally, I'd just call it a low end, stripped down P4 and assume performance would be pretty bad.
That's not to say that netbooks based on other low power processors are fast either. But devices designed for netbooks & low end laptops are likely to be faster. Just maybe not quite as power friendly.
-
Karmazen & Oliver
- Posts: 374
- Joined: Sat Mar 10, 2007 12:34 am
Re: Intel Atom processor and nodes per second
HERE more info:pgeorges wrote:Hi,
I'm considering buying a netbook based on an Intel Atom processor, and wonder what engine performance to expect on such platform. Benchmarks I read so far are not always relevant for this kind of activity.
So has anybody got the number of nodes per second achieved for free engines like Fruit, Toga, Crafty, Glaurung, etc ? And for Rybka ?
Pascal Georges
http://www.tomshardware.com/reviews/Int ... 81-16.html

the objetive is more energy... I don´t know how many Kn/s calcualte hiarcs on a pocket pda, (cpu intel ARM), near 50-70. Kn / s...
Atom have a lot of kn /s for only 4 W of power...
bye from spain.
-
Carey
- Posts: 313
- Joined: Wed Mar 08, 2006 8:18 pm
Re: Intel Atom processor and nodes per second
Thanks for the link. Interesting.Karmazen & Oliver wrote:HERE more info:pgeorges wrote:Hi,
I'm considering buying a netbook based on an Intel Atom processor, and wonder what engine performance to expect on such platform. Benchmarks I read so far are not always relevant for this kind of activity.
So has anybody got the number of nodes per second achieved for free engines like Fruit, Toga, Crafty, Glaurung, etc ? And for Rybka ?
Pascal Georges
http://www.tomshardware.com/reviews/Int ... 81-16.html
the objetive is more energy... I don´t know how many Kn/s calcualte hiarcs on a pocket pda, (cpu intel ARM), near 50-70. Kn / s...
Atom have a lot of kn /s for only 4 W of power...
bye from spain.
I also find it interesting they didn't compare Atom's results to other CPUs in the same category. Instead they compared to notebook & low end desktop CPU's.
Most of the reviews I've seen have have VIA's nano processor easily outperforming the Atom with only a slight increase in total power consumption.
But I haven't seen any Fritz results for a direct comparison to the Atom for chess.
Netbooks are an interesting category of computers... What balance point do you actually go dor power vs. performance?
My mother is trying to talk me into getting her one for christmas. (She got her first computer just a year ago. Her first iPod 6 months ago.)
For me, I'm still waiting for one with a fold-out screen. Unpack a paperback book sized thing to a full keyboard and a 24" screen....
Of course, all this talk about performance makes me think about Bob Hyatt..... Do you realize that it wasn't too many years ago that he would have given about anything to have a system with the performance of an Atom or Nano...?
Heck, these things are easily outperforming Cray-1's & Cray-2's that we (I) used to drool over.
Can you imagine him showing up a world computer chess tournament and him pulling out a ePC or something?
-
pgeorges
Re: Intel Atom processor and nodes per second
Thank you very much for Deep Fritz figures. This shows Atom processors on eee PC netbooks are good enough for playing and analysis.
I personaly I prefer PDAs over PCs because of :
Pascal Georges
I personaly I prefer PDAs over PCs because of :
- - mobility
- noise
- boot time
Pascal Georges