Shuttle XS29F: Is VIA's Nano Processor Powerful Enough?

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

jeffunit

Distinguished
May 19, 2008
117
0
18,680
Why use an 850watt power supply, when it is only drawing a maximum of 50 watts? Sure it is 80+, but the efficiency graph that is referred to starts at 20% load, which is 170w dc, which is
about 188w ac. Since this system is using about 1/4 of that power, the load is roughly 5%. There are 80+ power supplies rated at 200w or so. I am certain the celeron would use significantly less power when paired with a reasonable power supply.

This was the same problem with tomshardware reviewing the intel atom, last year. They picked some monster power supply, which was inefficient at low power loads.

Looking around, I found http://blogs.zdnet.com/Ou/?p=771
a 200w 80+ power supply for $55. I am sure there are others.
 
G

Guest

Guest
How exactly would they have used a dual core nano when no such product exists? Furthermore, when you compare products you compare them on price, performance, power usage, or other factors. Whether one product has more cores is part of its design and part of the review criteria. You compare what each vendor offers based on reviewed criteria, not what's "fair". The nano is single core, and it uses more power. Deal with it.
 
What a way of spinning out numbers... 24W at load, that's the only thing that matter... not number about power versus efficiency. What kind of review is this!?

Seriously, a system with so little power could be pretty useful for simple tasks, especially if you need 50 of them.
 
Might be good as a linux box. Despite it's rather low performance it does have rather low power consumption so it's good if you need something up 24/7 to use as a server of some kind. Still the idle power consumption should have been lower. I sure wouldn't use it as a desktop machine either except perhaps with a lightweight window manager like XFCE or ICEWM. Probably give me an excuse to get into blackbox again though :D.
 
G

Guest

Guest
personally i think you should have tested it with a distribution designed for less horsepower: ie ubuntu with xfce or lxde.
 

raptor550

Distinguished
Mar 6, 2009
34
0
18,530
How about the low powered 45watt 2.7ghz AMD processor? Or like others said, a viable competitor from VIA?

I am building a NAS/home server/media server and want a good comparison of the lowest watt platforms, not just stuff you have laying around your cube.
 
G

Guest

Guest
I know Nano is not a fantastic performer, but it´s not intended to be. Ok you´re comparing two system from shuttle, but the article is favoring Intel a little too much, wich they don´t really need. If you took systems using faster Nanos with newer chipsets, the result would be much different.
The test obviously should have been done using XP or Some XFCE distro too.
Power consumption IS low and most systems like these will spend a huge amount of time idling around.

C´mon, VIA should receive some appreciation, at least it´s an option besides Intel and AMD, which is good.
 

Major7up

Distinguished
Aug 17, 2009
446
0
18,780
[citation][nom]tacoslave[/nom]amd huron platform thats all i have to say i want to see it benched see how it stakes up.[/citation]
Do you really need to see how badly it will perform? I for one am glad that they spared us those tests.
 

Crashman

Polypheme
Former Staff
[citation][nom]jeffunit[/nom]Why use an 850watt power supply, when it is only drawing a maximum of 50 watts? Sure it is 80+, but the efficiency graph that is referred to starts at 20% load, which is 170w dc, which isabout 188w ac. Since this system is using about 1/4 of that power, the load is roughly 5%. There are 80+ power supplies rated at 200w or so. I am certain the celeron would use significantly less power when paired with a reasonable power supply.This was the same problem with tomshardware reviewing the intel atom, last year. They picked some monster power supply, which was inefficient at low power loads.Looking around, I found http://blogs.zdnet.com/Ou/?p=771a 200w 80+ power supply for $55. I am sure there are others.[/citation]

The smallest power supply the lab had was 300W. With the Celeron system, the 300W power supply drew 3W more than the 850W power supply. So even at 50W output, the Corsair power unit was more efficient than a "more appropriately sized unit". Tom's didn't get a chance to run the Shuttle power supply on the Celeron board because it didn't have ATX12V or enough molex outputs to add an adapter.[citation][nom]bollux78[/nom]I know Nano is not a fantastic performer, but it´s not intended to be. Ok you´re comparing two system from shuttle, but the article is favoring Intel a little too much, wich they don´t really need. If you took systems using faster Nanos with newer chipsets, the result would be much different.The test obviously should have been done using XP or Some XFCE distro too.Power consumption IS low and most systems like these will spend a huge amount of time idling around. C´mon, VIA should receive some appreciation, at least it´s an option besides Intel and AMD, which is good.[/citation]
Little appreciation for a system that had to be reloaded and retested several times over the period of an entire work week because of compatibility issues perhaps?
 

WheelsOfConfusion

Distinguished
Aug 18, 2008
705
0
18,980
The X27D was a poor choice for comparison because there's no dual-core Nano on the market. It's especially puzzling because the X27 with an Atom 230 is also available from Shuttle, and looks like a better apples-to-apples comparison because it's not dual-core. That way the results of Nano vs. Atom wouldn't have any ambiguity due to the number of processing cores. Failing that, the X29 (non-fanless version) with the 1.6 GHz Nano would at least be a straight clock-for-clock comparison: in this one the Nano not only has a one core handicap, it also has a 600MHz speed disadvantage (even worse against the Celeron). So the X27 vs. X29 would appear to be a much more meaningful test. As it is, this article is less informative for all the different variables, like Car and Driver pitting a Jaguar XF against a BWM 750li.
Are you guys limited to the two systems Shuttle decided to send you for a review? If so, I kind of have to ask what the point is of comparing them directly.
 

Crashman

Polypheme
Former Staff


Please read the introduction, then the conclusion. Shuttle sent these to promote the efficiency of its Nano system, a system so weak that its efficiency is lower than the Atom system they sent in spite of its lower power consumption.

Seeing that the two systems Shuttle sent were so far apart, and both target the basic office-type desktop market, a three-way comparison was set up with the basic office Celeron system as the baseline. The concept: To determine if the power-savers were fast enough and responsive enough to preserve a reasonable computing experience. The Atom was, the Nano wasn't.

Of course the idea was to test typical office applications, including web apps and MS Office. The VIA chipset wouldn't run with that benchmark suite, because the graphics driver crashed. An older version was tried of both the driver and the benchmark suite, and more problems came about. A test of the Nano's 64-bit capability was also in order, but 64-bit Windows required a patch be applied to the installation DVD. That's not something a normal small office administrator would do.

So, from the office standpoint, the Nano will remain a flop at least until Windows 7 is released later this month.
 

HalfHuman

Distinguished
Feb 13, 2006
83
0
18,630
[citation][nom]Crashman[/nom]Please read the introduction, then the conclusion. Shuttle sent these to promote the efficiency of its Nano system, a system so weak that its efficiency is lower than the Atom system they sent in spite of its lower power consumption.Seeing that the two systems Shuttle sent were so far apart, and both target the basic office-type desktop market, a three-way comparison was set up with the basic office Celeron system as the baseline. The concept: To determine if the power-savers were fast enough and responsive enough to preserve a reasonable computing experience. The Atom was, the Nano wasn't.Of course the idea was to test typical office applications, including web apps and MS Office. The VIA chipset wouldn't run with that benchmark suite, because the graphics driver crashed. An older version was tried of both the driver and the benchmark suite, and more problems came about. A test of the Nano's 64-bit capability was also in order, but 64-bit Windows required a patch be applied to the installation DVD. That's not something a normal small office administrator would do.So, from the office standpoint, the Nano will remain a flop at least until Windows 7 is released later this month.[/citation]

i still think that using vista on this sort of machines is quite a bad idea and i think nobody in their right mind would run vista ultimate on a nettop. we already knew that vista is a hog and that these systems are underpowered. we need to know if they can do the job under apropriate conditions (right os, right tasks etc). i never had imagined that 1ghz nano could be faster than atom dual core running at 1,6ghz. i strongly feel that this review must be remade to have something usefull. i've a regular reader the articles on thg and apreciate the work but this article disapointed me. i'm sure you can make it right.
 

jeffunit

Distinguished
May 19, 2008
117
0
18,680
[citation][nom]Crashman[/nom]The smallest power supply the lab had was 300W. With the Celeron system, the 300W power supply drew 3W more than the 850W power supply. So even at 50W output, the Corsair power unit was more efficient than a "more appropriately sized unit". Tom's didn't get a chance to run the Shuttle power supply on the Celeron board because it didn't have ATX12V or enough molex outputs to add an adapter.Little appreciation for a system that had to be reloaded and retested several times over the period of an entire work week because of compatibility issues perhaps?[/citation]

It is sad that you reviewed a motherboard without an appropriate power supply. Even the 300w power supply is rated at 80% efficient at loads of 20% of 300, which is 60 watts DC. If you are going to test low power motherboards, and report on their power usage, you need to have a decent power supply that has reasonable efficiency at the load it is used for.

I have seen many 80+ power supplies have efficiencies of 60% or less at loads below 20%. It is really nice that the corsair power supply is so efficient at these low loads, but unless you have measured its efficiency at these loads, nobody really knows how much power the motherboard draws. And yet you show power consumption for the board. Is the corsair 80% efficient at that point, or 60% efficient, or even less? Nobody knows.

The same issue came up a year ago, see
http://www.formortals.com/tomshardware-botches-intel-atom-energy-efficiency-tests-badly/
I would have hoped that tomshardware would have learned from their mistakes.

 

americanbrian

Distinguished
I don't mind the power supply, they both used the same one so penalties should be roughly evenly distributed, maybe.

Actually it sounds stupid when I think about it more, if the PSU is more efficient when the power draw is increased, then it would actually skew the results in favour of the atom and the celeron. completely. So I take that back. I object too.

BUT, I object even more to the OS. I bet shuttle didn't ship it to toms with vista on. either of them. the conclusion is completely off base.

It will perform productive tasks with the supplied OS I am sure. And again Photoshop? what is with that.

Lets face it. these are not office machines specifically. But they could easily do the job if you aren't an idiot who is gonna bog them down with useless software.
 
G

Guest

Guest
---HWiNFO32---
Intel Pentium U2700 1,2 GHz (800MHz FSB, 2MB L2 cache)
Single-Thread:
ALU: 15 934
FPU: 10 163
----------------------------------------------------
Intel Atom N270 1.6 GHz (HT, 533MHz FSB, 512 KB L2)
Single-Thread:
ALU: 10 267
FPU: 5 718
Multi-Thread mód:
ALU: 15 339
FPU: 10 755
----------------------------------------------------
Intel Atom N280 1.66 GHz (HT,667MHz FSB, 512 KB L2))
Single-Thread:
ALU: 11 231
FPU: 6094
Multi-Thread:
ALU: 16 062
FPU: 11 187
-------------------------------------------------------
Intel Atom 330 1.6 GHz Dual-Core (HT, 533MHz FSB, 512 KB L2)
Single-Thread:
ALU: 10 798
FPU: 5 851
Multi-Thread mód:
ALU: 31 080
FPU: 21 608
-------------------------------------------------------
VIA Nano L2200 1.6GHz (800MHz VIA V4 FSB,1MB L2 cache)
Single-threaded:
ALU: 16 049
FPU: 13 785
-------------------------------------------------------
VIA Nano U2250 1.3+GHz@1.6GHz capable (800MHz VIA V4 FSB,1MB L2 cache)
Single-threaded:
ALU: 15 859
FPU: 13 650
 

Crashman

Polypheme
Former Staff
[citation][nom]americanbrian[/nom]I don't mind the power supply, they both used the same one so penalties should be roughly evenly distributed, maybe. Actually it sounds stupid when I think about it more, if the PSU is more efficient when the power draw is increased, then it would actually skew the results in favour of the atom and the celeron. completely. So I take that back. I object too.BUT, I object even more to the OS. I bet shuttle didn't ship it to toms with vista on. either of them. the conclusion is completely off base.It will perform productive tasks with the supplied OS I am sure. And again Photoshop? what is with that.Lets face it. these are not office machines specifically. But they could easily do the job if you aren't an idiot who is gonna bog them down with useless software.[/citation]

Tom's hardware knew about that article, but not every power supply is dropping to 60% efficiency at 20% load. Take a look at the charts for the tested power supply and imagine what the curve would look like if it were extended, then take the fact that it was more efficient at 30-50W than the smaller power supplies on hand.
 

papasmurf

Distinguished
Apr 14, 2002
2,280
0
19,780
WinRAR almost perfectly reflects CPU clock speed, rather than architecture, while WinZip appears optimized towards the desktop CPU.

It looks more like win rar is better optimized considering the fact that it is compressing these files in half the time.
 

jeffunit

Distinguished
May 19, 2008
117
0
18,680
[citation][nom]Crashman[/nom]Tom's hardware knew about that article, but not every power supply is dropping to 60% efficiency at 20% load. Take a look at the charts for the tested power supply and imagine what the curve would look like if it were extended, then take the fact that it was more efficient at 30-50W than the smaller power supplies on hand.[/citation]

As you know, the 80+ people start their efficiency graph at 20% load. I have not tested this corsair powersupply. However, silentpcreview.com has tested a slightly older corsair 80+ 650w
power supply. See http://www.silentpcreview.com/article813-page4.html
At 34w ac, it was 62.9% efficient. That is a load of 3.2%.
At 59w ac, it was 70.9% efficient. That is a load of 6.3%.
The unit you used was 80+ gold, however it was an 850w unit.
Since you didn't measure the dc wattage, you have no idea how efficient the unit was, but assuming 100% efficiency, it was loaded at 5.8%. The power supply I selected from silentpcreview.com
was the first corsair unit I found there. All the power supplies that I have looked at have pretty poor efficiency at low loads.

Why don't you measure the efficiency of the 850w corsair, when it is putting out 50w? I will bet you a beer it is under 70%.
In addition, the 80+ people often test the unit at 240vac, because almost all power supplies are more efficient with higher voltage. At 120vac, power supplies are typically 2-3% less efficient. Looking at the specific silentpcreview.com article, shows a drop of 3% for this specific corsair 650 unit.
 

jeffunit

Distinguished
May 19, 2008
117
0
18,680
[citation][nom]jeffunit[/nom]In addition, the 80+ people often test the unit at 240vac, because almost all power supplies are more efficient with higher voltage.[/citation]

My bad. The unit was tested at 115vac by 80plus.org
 
Status
Not open for further replies.