Hi,
some kind of experiment: few days ago I've connected 3 devices for SETI-at_home number crunching:
1) Sony Xperia Z2 smarpthone (ARMv7 Processor rev 1 (v7l), Snapdragon 801 Quad-core 2.3 GHz Krait 400)
2) Samsung Nexus old smartphone (ARMv7 Processor rev 10 (v7l), TI OMAP 4460, Dual-core 1.2 GHz Cortex-A9)
3) Acer TravelMate-6292 old laptop (Intel(R) Core(TM)2 Duo CPU T7300 @ 2.00GHz [Family 6 Model 15 Stepping 10], 2 cores)

Smartphones are connected through USB watt-meter between power supply and a phone and laptop is connected through the watt-meter between power supply and a power socket.

Below is a result of Total Credits, Consumed Energy (kWt*h) and the ratio: Total_Credits / Consumed_Energy:

Device               Total Credits     Consumed Energy, kWt*h    Ratio
Sony Xperia Z2       1088              0.332                     3277.11
Samsung Nexus        180               0.092                     1956.52
TravelMate 6292      605               2.78                      217.63 

Is such Ratio a valid way to measure energy efficiency of devices? Watt-meters of smartphones are connected after power-supply. Even if efficiency of power supply is 0.5 (I doubt about it), it looks like smartphones are more efficient devices than the old laptop.

You already point one uncertainty - efficiency of USB power supply.
Could you power these phones via USB plugs of very that laptop? Then measure with the same single power meter consumption of laptop alone and consumption of laptop + 1 or 2 phones.

Also carefully check that AR of tasks processed by all 3 devices is the same.
I would recommend to run benchmark with known to be exactly same tasks and compute power consumption vs processing time instead of your metric. Credits awarding too unstable to use it as serious measurement instrument.

---

SETI apps news
We're not gonna fight them. We're gonna transcend them.

Could you power these phones via USB plugs of very that laptop? Then measure with the same single power meter consumption of laptop alone and consumption of laptop + 1 or 2 phones.

That's a very good idea! I will do that.

How to check that AR of tasks is the same?

Could you power these phones via USB plugs of very that laptop? Then measure with the same single power meter consumption of laptop alone and consumption of laptop + 1 or 2 phones.

That's a very good idea! I will do that.

How to check that AR of tasks is the same?

Since credit is highly variable I try to avoid using it when determining efficiency of devices.

Instead I prefer to calculate the Wh per task of comparable AR tasks.
For estimated values I start with published TDP values.
http://setiathome.berkeley.edu/forum_thread.php?id=80803&postid=1840236#1840236
For whole system calculation I measure the average power usage for the system under the task load I want to compare.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Could you power these phones via USB plugs of very that laptop? Then measure with the same single power meter consumption of laptop alone and consumption of laptop + 1 or 2 phones.

That's a very good idea! I will do that.

How to check that AR of tasks is the same?

If you look at a completed task you will see a value like WU true angle range is : 0.411818.
An AR of 0.42-44 is a "normal" AR and good to use for comparison, but so long as you are comparing similar AR tasks then you will be ok.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

I would agree that a modern smartphone is going to be significantly more energy efficient than an older CPU.

But long term RAC is probably better guide as things even out over time. (Different tasks, slow wingmen etc) Basically your laptop should RAC higher, but due to bad luck it may have a lot of credits pending?

My own experiment is pitting an old P4 HT against a Samsung J1 phone. These are 2 machines I have running SETI pretty much 24/7 (backup server) and the phone is a spare and just plugged into a PC usb port. I discount the power supply efficiency as it's running off a PCs standby power, that is "on" anyway. If it's having to pump 2 extra watts into a USB port from the 5Vstb rail, the draw from the wall will only be 2.x watts extra.

The P4 seems to RAC around 500 (with optimised apps)
https://setiathome.berkeley.edu/show_host_detail.php?hostid=8104632

The phone is pushing close to 400 after about 6 weeks.
https://setiathome.berkeley.edu/show_host_detail.php?hostid=8183273

So at least comparable RAC, with 20 or 30 times less power draw.

What surprised me was the computing power that a cheap cellphone actually has. The phone is close to the RAC of the old PC, and uses less power then the PC when it's actually shut down....

If it's having to pump 2 extra watts into a USB port from the 5Vstb rail, the draw from the wall will only be 2.x watts extra.

And given how low the power requirements of a mobile phone are, that 2W is a extremely significant value.

So at least comparable RAC, with 20 or 30 times less power draw.

Given the randomness of Credit New, the only way to compare the 2 is to compare the run times of comparable (ie same angle range) WUs.

---

Grant
Darwin NT

Given the randomness of Credit New, the only way to compare the 2 is to compare the run times of comparable (ie same angle range) WUs.

Yes, but a long term average does a similar thing. Both machines are maintaining a fairly steady RAC with a bit of variation up and down. Hence the range of power efficiency. It's going to vary depending on the mix of work units, wingmen etc. But that's only 10-20% variation. The results are consistent, if not exact.

Not only the randomness of Credit New, but the change in the source data and the applications "conspire" to make using using such measures "risky". The BLC data causes applications to behave very differently to the old Arecibo data, and then we have the Astropulse data and applications. To get anything meaningful you need to look at all these variables, and be very careful in your analysis to ensure that the samples you are studying aren't "polluted" by such influences.

Then of course there is Credit New's propensity to tend toward zero over time, overlaid with some very interesting cyclic behaviours.

---

Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Guys
Just a small point using mobiles as hosts. Remember to take the handset out of the holder/wallet, otherwise you may get "waiting for battery to cool down".
HTH
George

---