Thread: Survey: Startup times and methods

I'm curious if other people are as dissatisfied with their startup times as I am. Their doesn't seem to be much I can do to reduce startup times other than removing some RAM (grrr...) or switching over to STR (which many people seem to advocate against) Hopefully this thread with enlighten me or make me realize I'm in a boat with many other people. ;-)

How does more memory slow your bootup time? Did you disable the memory test in the bios?

Mike

oh boy...here we go again...

don't have all the final info right now, but my new system uses the Lippert Thunderbird-MM Mini-ITX P-M board. Using a heavily slimmed nLite XP setup will all my drivers installed and using an 80GB laptop 4200RPM HDD bootvis calculates a 13 second cold boot time. This board is quick as **** booting up. There is hardly any BIOS delay, it goes so fast I can't even see the diagnostics screen listing all my hardware. Anyways, once I get my system finished up I'll post according to your format...

check out carshell in my sig, it should reduce cold boot times, not sure about hibernation results

Personally I am running the hardware in my sig with unmodified windows (other then running carshell as my shell) and have a 30 second boot time from opus power on to music.

A few comments on the replies so far...

1. Additional RAM doesn't appreciably effect cold start times, but resume time from hibernation (suspend to disk) is roughly proportional to the amount of RAM in the machine. Based on my off-the-cuff numbers, it looks like the Epia board might be using PIO (rather than U/DMA) to restore RAM contents from disk.
   
2. The Epia boards seem to have a particularly large BIOS startup time, and I am well aware that other boards go through BIOS startup much more quickly. This is one of the reasons I have asked the total boot time to be broken down into the various sections.
   
3. I did make the assumption that resuming from hibernation would be faster than a cold boot, but I realize now that this might not always be the case -- A fast, well-tuned, system could potentially cold-boot in less time than it would take to recover from hibernation. If anyone has actually seen this in practice, please be sure to post about it.
   
4. In a correlary to the previous point, I personally have not tweaked my system for optimal cold boot speed because the particular front-end I am running (which I cannot talk about at the present time) takes approximately 30 seconds to launch, so I know that, de-facto, resuming from hibernation will be faster than a cold boot.
   
5. I am particularly interested in the number of people using STR (S3 suspend) vs. everyone else. I'm well aware of the power drain issues there, and I am curious how many people have worked to overcome them for the 2-second start-up time regardless of CPU speed, BIOS type, or system startup configs.
   
6. Perhaps I should add one more section to my survey:
   
   
   1. Did you select your hardware platform (BIOS and/or CPU) specifically for boot speeds?
   2. Did you spend substantially more for this board than you would have for another just to obtain reduced startup times?
   3. Do you regret purchasing the mobo you did as a result of slow startup times?
   
   

-p.

1. Hardware Configuration
* Mac Mini 1.42ghz 256mb RAM
* Carnetix P1900

2. Software Configuration
* Mac OS 10.3
* iTunes for music
* Powermate for control

3. Startup Strategy
* Sleep

4. Startup time, broken down by component of possible
* startup controller delay - 3 seconds
* BIOS delay (n/a for STR) - N/A
* boot time or restore from disk/ram time - 3 seconds
* application startup time (cold boot only) - N/A
* total time from startup initiation (ignition or remote entry) until music comes from speakers - 6 seconds
* If using STD/STR, you do not need to provide the times to cold boot the system, but if you don't have a 100% recovery rate, provide an estimate of failure rate either in terms of "every X days" or "every X startups". Haven't had a failure yet, 6 weeks in car.

5. Subjective information
* On a scale of 1 (not important) to 10 (very important) how important is a quick startup time? 8
* On a scale of 1 (not satisfied) to 10 (very satisfied) how do you feel about the startup time of your system? 10
* On a scale of 1 (not important) to 10 (very important) rate how interested you are in improving your startup time. 1
6. Any other configs, notes or comments that don't fall into any of the categories above.

Gotta love sleep mode for those fast recovery times! Have you put a meter on your rig to measure current draw in sleep mode?

Given the general attitudes I saw towards sleep mode I didn't even bother to measure sleep-mode current on my rig -- a huge mistake as that would have been simple when it was on my bench. *sigh* Perhaps I'll do that this weekend.

-p.

I was going to measure it but never got around to it. Doesn't matter as the P1900 cuts it off if the battery voltage gets too low. I'm sure of that as I let it sit for 10 days while on vacation.

I have measured battery voltage drops overnight. It was doing about .2-.4 over a 12 hour period.

Originally Posted by gnomad

A fast, well-tuned, system could potentially cold-boot in less time than it would take to recover from hibernation.

My cold boot time and hibernation are about even at 11 and 12 seconds to desktop (with internet disabled) since I dropped down to 256mb ram. I don't have the delays for different components recorded though. I've been working on a post for the thread in the FAQ Emporium, but haven't finished it up yet. But anyone can run that fast even with slow hardware like the m10000 (well, maybe not with a 4200rpm drive). It just takes a bit of work. Spend some time in the software optimization forum here and the Nlite forum on MSFN.

Regading Epia BIOS comment, disable everything not needed and run optimized settings. Do some tweaking and see how changes affect your time. Like SFiorito, the hardware list is so fast I can barely see it, and I often have to reboot to hit delete in time when I want to change a setting.

Anyway, I'm very pleased with my startup time. I wish I could break 10 seconds, but even a test install of an XP build so stripped as to be nearly useless except for music didn't get me there. But still fast enough that I don't care about surviving crank. Although the ITPS does in fact survive it about one in four times.

If you are really seeing a 0.2 - 0.4 voltage drop overnight (that is, your initial measurment does not indicate a surface charge) then you have a serious problem as a 0.4 volt drop represents 50% of your battery life! (reference: http://www.uuhome.de/william.darden/...4.htm#vrla_soc)

The more serious problem is that starting batteries are not meant to be discharged below 80% of overall capacity, and draining them below 50% on a regular basis will reduce their lifespan dramatically. (The reasons for this are explained in the above-referenced FAQ.)

So, let's hope you don't have a valid measurement there! The whole issue of surface charge and non-linearity makes voltage drops a tricky number. Direct measurement of the sleep current is the important value...

-p.