XPS Sandy Bridge Battery Life

Alternative Title: Did I really benefit from sticking with a dual-core, or was I deluding myself?

One thing that many people are interested in when deciding between dual core and quad core is whether there are any battery life differences between them. As far as I've seen, there haven't been any direct reviews comparing the processors in these terms.

So I was thinking it would be useful for people to post their model (15 or 17), processor model number, RAM amount, graphics card (and say if you overclocked it), and battery size, and then report the results of the following:

Set to Balanced power management with 50% brightness and, while on AC power, set the computer up for its power usage (e.g. start Prime95 running), then take it off AC power, continue the task for at least 5 min, then divide the estimated time remaining by the percent power remaining (e.g. 1.5 hours with 90% remaining: 1.5 / 0.9 = 1.67 hours). This will give you the total projected battery life if you were to continue that task until the battery was drained.

Some useful test conditions, I think, would be:
1) Screen on but no applications running.
2) Browser open playing a standard def YouTube video.
3) Blu-Ray Movie playback
4) Something using the discrete GPU(?)

Feel free to add your own, but the idea is to be able to directly compare numbers with the same tests.

Of course it would be better to actually run out the battery doing these things, but these quick tests are easier and should give a good indication of the differences between the processors and other components.

Because of the power-gating, we might generally expect idle battery life to be fairly similar across the board in the near-idle condition. But does a higher load differentiate them enough to make for a significant performance/battery life trade off, or is it only a few minutes difference or less?

EDIT: And record if your wireless card is active.

 

This might be a good idea. I'm going against prime95, as it is unrealistic in terms of power usage and the fact that it can take advantage of all 8 threads on a quad core vs 4 threads on a dual core, so there will be a big difference in battery life that doesn't equate to real world scenarios. I can't remember who did this, but something like playing a blue-ray movie and checking the percentage left afterwards would be entertainment and more accurate as well as realistic to how we actually use the laptops. Playing a blue-ray probably won't enable the nvidia gpu. I've also noticed through the many 3dmark06 tests I've run, that without the AC power hooked up, my scores are around 25% lower, leading me to believe that the gpu isn't enabled on battery power, or some other battery saving function is specifically hindering performance to save on battery life. This should be addressed before any cpu or gpu intensive tasks are benchmarked for the purpose of checking battery life.

 

That's a fair point. I'll edit the post to reflect that.

 

Just what I've been looking for. Don't know if I should get a dual or quad core. I probably don't need a quad core (although I guess it would be nice), so the difference in battery life between the two is one thing that I'm curious about.

 

The question is what do you need the notebook for, and for how long do you plan to use it? Will you use it a lot away from a socket?

Is the 2630QM the base CPU in Sweden, like in Germany?

 

Well, I plan to keep for quite some time, say 4-5 years. My current desktop is problably 6+ years or something. It would be my only computer, replacing my old desktop, so it will be used for all my computing needs. Most of the time the usual everyday stuff ie internet, music, wordprocessing etc but also some gaming and light photoediting. As I will use it in my home most of the time, battery life isn't really a dealbreaker. But if I'm to buy a laptop I would like to be able to use it without wires while not at a desk, and I'm not a big fan of that 9-cell battery.

The CPU options in Sweden is ranging from i3-2310m to i7-2820qm. (I got my eyes set on the i5-2410m or the i7-2630qm)

Oh well, this thread isn't about me and my dilemmas. Let's see some comparison...

 

I just got my laptop on Monday and I haven't had time to do any extensive testing. I've actually got enough that I might post a full-fledged review on my blog. Anyway, I intend to do the battery life somewhat standard testing this weekend. But on a anecdotal note, I had the laptop off AC power from 10 am to 4 pm, without letting it sleep, basically just doing web surfing and e-mail, before it ran out of juice and into hibernation mode. That's just incredible.

 

I've got the quad-core and get results consistent with what DakkonA mentioned. Doesn't look like there is a significant difference between the dual and quad cores.

 

I have the exact question as @FS1 . Is it a huge advantage that one gets by choosing i7 2630qm over i5 2410m. My usage is good enough internet browsing,downloads, media player, mild gaming and some programming (price difference is around 100$. I can use it somewhere else.) similarly I was toggling between gfx cards GT 525M and GT 540m and after reading the views here i decided to stick with GT525M as GT540 does not offer much. I did not like the 6 gb RAM combo offered and So i went of 2 X 4gb RAM. A 500 gb hdd should be enough since external hdds are available cheap these days. overall i say i save around 230$ by opting for core i5 option over the i7. i was never going for full HD+3D option. Do add your comments on this.

 

Given that our systems are almost identical otherwise, that seems likely. Perhaps I should have sprung the extra $30 for the 2720QM after all?

Since you have essentially an identical system to me other than the CPU, would you be game for doing a couple direct comparisons? I'm thinking prime95 with 2, 4, 6, and 8 threads while watching temps and what happens with the clock speed while it works (plus battery life while its doing this). Because if a 2720QM can run 2 threads at similar speeds to the 2620M over time with essentially the same power draw, then there's really no benefit at all to having this dual core over the quad core, except perhaps that the dual core places an upper limit at how much power can be consumed at any given time.

 

Sure, I'd be interested in the comparison myself. Battery life is a huge deal for me, and I originally ordered a 2620M configuration. Then I realized that power draw would likely be comparable with light use, and that when I'm doing photo/video work (the only thing I do that would fully utilize the quad) I am always plugged in. That tipped the balance in favor of the 2720QM and I couldn't cancel so I placed a new order and sent the 2620 back as soon as I received it.

I'm not that familiar with the prime95, so let me know if there are any other options to select to ensure we are using the same tests.

 

I'll figure something out and get back to you, probably by the weekend if not tonight.

 

This was my hopeful reasoning on ordering the quad core, and also helped persuade me into upgrading at this point in technology time.

Sandy Bridge appears more so than any previous generation, able to effectively shut off the cores not in use while other cores are under load.

I figured it would be nice to have the extra power available while also having the ability to put it on the self when it isn't being used. But can I really have their cake, and eat it too?

 

I haven't ran any programs for testing the battery usage but im getting 5.5 to 6.5 hours browsing the net, and a bit of music and youtube. On min brightness with the cpu set to not go over 1ghz.
I7 2630, 1080p, 4gb ram and 540m.

 

I'm running into a bit of a problem trying to do the short-term battery life test I mentioned. Perhaps to protect the health of the battery, when only a small amount has been discharged from the battery, charging will actually be disabled and the battery only appears to be fully recharged. This makes some sense because small charge/discharge cycles are detrimental to the battery, but it's interfering with my idea for quick and easy battery life tests.

It is possible to manually disable battery charging. I think I'll do that and then look at the cumulative battery drain.

 

DakkonA, I'm running on the 2630QM without 3D screen. Battery life has been good for me. I'm not waffling but a single charge would last me for around 7hours if I were to use it for Office applications and light internet surfing. Not too bad at all though...

 

Okay I've got some hard numbers now. My base test conditions were:

HWiNFO32 Sensors Only mode, logging started, msconfig Selective Startup (All services running, no startup programs), Dell power plan with display set not to dim or turn off during testing.

I removed the laptop from AC power, then began tests, changing the condition every 5 minutes. I averaged the Battery Charge numbers and then used the equation W * h = Wh to find "h" (expected battery life) from the theoretical max charge of the 9-cell battery, which is 8.4 Wh (unfortunately, mine is already maxing out at 8.1 Wh).

Wireless on, Backlight 100% = 4.6 h
Wireless on, Backlight 0% = 7.0 h
Wireless off, Backlight 0% = 7.4 h
Wireless off, Backlight 40% = 6.1 h
", ", Blu-Ray playback = 3.8 h
", ", Prime95 (4 In-Place FFT Threads) = 1.9 h
", ", Prime95 + Furmark (Burn-in) = 1.4 h
Wireless on, ", " = 1.4 h
Wireless on, Backlight 100%, " = 1.3 h

It's kind of amazing that even with nearly as much power load as the system can take, it would still last over an hour.

I also did some tests with Prime95 with 1, 2, and 4 threads running, during which I monitored battery charge rate and clock speeds, allowing Prime95 to complete 10 tests for each thread before ending the test. I averaged the power draw for each period and subtracted the baseline power draw unrelated to the processor (1.37173 W, from the Wireless off, Backlight 40% condition above). I also recorded how long it took for the 10 tests to complete.

1 Thread: 2.16 W. Both processor cores were running in TurboBoost mode at 3.2 GHz, with one core spiking now and then to 3.4 GHz. 8 minutes to complete 10 tests.
2 Threads: 3.11 W. Both cores again in TurboBoost, but fluctuating between 3.0 GHz and 3.2 GHz. 8 minutes to complete 10 tests.
4 Threads: 3.16 W. Both cores fluctuating between 2.9 GHz and 3.2 GHz. 16 minuts to complete 10 tests.

Anyone who has an identical computer to me other than the processor, I'd like to see you do this too:

Run HWiNFO32 (in the Selective Startup environment described above with wireless off and display at 40% (= 5 blocks)), start logging, and run Prime95 with 1, 2, 4, and (if quad-core) 8 threads using In-Place FFT mode. Keep an eye on the CPU speeds while it is going on. After the tests, record the time it took to complete the 10 tests (Worker Started line to the next line after Test 10), and stop logging.

If you don't want to process the logs yourself, I'd be happy to do it for you. When you're testing, just leave half a minute or more between trial runs so that the start and end is obvious in your logs.

Just so we know we're all on the same page, Prime95 v26.5 is found here: ftp://mersenne.org/gimps/p64v265.zip and the test should start with "Test 1, 9000 Lucas-Lehmer iterations of M8716289 using Core2 type-2 FFT length 448K, Pass1=448, Pass2=1K."

The one possible issue is that it mentions optimizing for the CPU architecture. Since the processors have varying amounts of L3 cache, I'm not sure how that will affect process speed or power draw.

If anyone has another idea of how to compare processors, please let me know.

EDIT: If you're following up on the not-so-battery-life-related part, go here.