release
dev builds
extras
themes manual
wiki
device status forums
mailing lists
IRC bugs
patches
dev guide

Rockbox mail archive

Subject: Re: DRAM 2MB -> DRAM 8MB

Re: DRAM 2MB -> DRAM 8MB

From: Warp Dag <warpdag_at_hotmail.com>
Date: Tue, 17 Sep 2002 07:01:46 +0200

Actually I don't agree at all!!!

Just think 5 minutes ;) Filling a 2MB or a 8MB buffer while the hard drive
is spinning takes about the same time!!! I mean, for a song, spinning up the
drive, getting 2MB of data and put it into sleep mode takes exactly the same
time than the same process with 8MB of data (the time to fill the buffer is
ridiculous compared to the time the hard drive takes to spin up). Thanks to
god the drive can get more than 1MB/sec!!!

Also, the DRAM power difference between a 2MB and 8 MB chip is meaningless.
There are a lot of 8MB DRAM chips out there that burn less power than the
original 2MB chip we currently have.

Finally, about your average power consuption calculation is flawed. Here is
the correct formula (I assume the drives will never take more than 4s to
fill the 8MB buffer):
2 MB : 50mA + 300mA * 2/100 + 500 * 3/100 = 71mA average
8 MB : 50mA + 300mA * 4/450 + 500 * 3/450 = 56mA average

But you know, you should not think in "average". Because the idea is that it
will reduce the spinning stuff, and you must know that each time you take a
heavy load on the cells you loose a lot of energy, more than what you think
in fact. For instance, if you pump 100mA continuously, your battery will
last longer than if you pump randomly at 100mA average (batteries don't like

Bye, David

>To: rockbox_at_cool.haxx.se
>Subject: Re: DRAM 2MB -> DRAM 8MB
>Date: Mon, 16 Sep 2002 17:02:03 +0200
>
>Stuart Tedford wrote:
>>Excellent work.
>>
>>This mod should save a lot of battery power. Could you post the results
>>when you get round to doing a test?
>
>More RAM space for cool apps - YES
>
>Doubled battery life - NO!
>
>As we have a larger buffer to fill, the HDD has also to spin longer, so
>nothing gained here. What we would gain is less time spent with drive
>spinup if the buffer is 4x large.
>
>Assuming we use:
>
>50mA when the HDD is off
>300mA when the HDD is running (reading 1MB/second)
>500mA to spin up for 3secs
>
>So with 2MB = 100secs of MP3 we have:
>50mA + 300mA * 2/100 + 500 * 3/100 = 71mA average
>
>and with 8BM = 450 secs of MP3 we have:
>50 mA + 300mA * 2/100 + 500 * 3/450 = 59mA average
>(not counting the extra current for the larger DRAM)
>
>The average current is reduced only by about 20% by making the buffer
>4x larger .....
>
>These values are just rough estimates, but I just wanted to show that
>a double buffer size does not make double play time.
>
>For this you would have to make the current consumption of the rest of the
>circuit much smaller.
>
>Regards,
>