2400 heat & speed
Dan K
macdan at comcast.net
Sat Jul 12 14:10:17 PDT 2003
First Donald McCaig <McCaig at va.tds.net> wrote:
>> Dear Gurus,
>> My lovely two year old IBM 30 G hard drive has
>> died. DT&T will
>> replace it - but . . . Apparently there's a 7200 rpm
>> IBM hd on the market.
>> Better but is better hotter? My 240 (w heat sink)
>> takes heat babying as is.
>> Will the 7200 rpm raise the heat vs the present
>> 4200?
Then Gregg Eshelman <g_alan_e at yahoo.com> wrote:
>Heck yes. Faster drive RPM = more heat. It's
>unavoidable, much of it is from the platters stirring
>up the air inside the drive. Hard drives have vents
>to draw in outside air and for hot air to get out.
>The airflow is important. If it's blocked the drive
>will either self destruct from heat buildup or the
>air will get so hot that it's no longer dense enough
>to ram between the heads and platters (the Bernoulli
>Effect) to keep them "flying" and then it's HEAD CRASH
>time.
Gregg . . . you seem to be saying hard disk drives use
_external_to_internal_to_external_ airflow as a major part of their
overall cooling mechanism. Now I'm no 'Hard-Drive Scientist' but I have
to dispute your statement as being preposterous . . . errr, well, at
least let me say it bears no resemblance to my own understanding.
While all HDs do have a venting system, it's there simply to allow the
drive's internal air pressure to equalize to its environment. If you've
ever disassembled a HD you've seen there is no capacity for significant
airflow through the drive. In fact, that's the last thing one would want
as air inevitably carries contaminants of course leading to your
aformentioned 'HEAD CRASH'. The tiny amount of air 'flow' that does occur
passes through a very-very-very fine filter.
AFAIK, HDs are cooled entirely by shedding heat through their cases'
exteriors:
a) transferring heat through airflow around the outside of the case
b) transferring heat through to its mount, as a heatsink, assuming it to
be mounted to a material capable of absorbing the heat (eg: a PowerBook's
metal chassis.)
c) ? (I can't think of other ways but I'm sure to be forgetting something
:-)
However, while I think your premise is faulty, I generally agree with
your conclusion. A great reason a PB2400 probably shouldn't use a 7200
RPM drive is because in all likelyhood it will run hotter than a
similarly sized modern 4200 rpm drive. That said, 'This Year's Drives'
run much cooler than 'Last Year's Drives' so the only way to really
_know_ if heat will be a problem is to try one and see for one's self.
My conclusion? As PBs 2400 already have little enough capacity for
shedding excess heat, I think one shouldn't further tax the poor thing
more than neccessary.
>Besides that, the IDE interface in the 2400 can't
>take advantage of the faster data transfer speeds
>of ATA 33/66/100 drives. I wouldn't go over a 5400
>RPM drive. What can improve overall performance is
>a larger RAM cache onboard the drive. Data accessed
>from there can be transferred at the maximum speed
>the IDE interface can handle.
Agreed the 2400's bus can't use the extra speed of a 7200 rpm drive, but
so also a _modern_ 4200 rpm drive will probably be faster than is fully
usable.
As a perfect example I offer my experience moving both 4200 rpm 30GB
(DJSA-230) and 5400 rpm 32GB (DJSA-232) Travelstars from 3400s (80MB/s
RAM, OS 9.1) to Pismos (640mb ram, same OS). I tested speeds using ATTO's
ExpressPro-Tools' 'Benchmark Drive' utility, in all cases with the drives
attached to the internal buses. As the 2400 and 3400 share nearly
identical architecture, I'm assuming here the results below are relevant
to the OPs question. Also, the conclusions I draw should equally apply to
'Books older than the 2400/3400 as they are slower still.
DJSA-230 30GT series - 4200 rpm
<http://www.hgst.com/hdd/travel/tr30gt.htm>
measured sustained transfers - MB/s
3400 Pismo
read write read write
---------------------------
10.5 8.8 10.7 10.4
measured peak transfers - MB/s
3400 Pismo
read write read write
---------------------------
10.6 8.8 60.7 55.5
DJSA-232 32GH series - 5400 rpm
<http://www.hgst.com/hdd/travel/tr32gh.htm>
measured sustained transfers - MB/s
3400 Pismo
read write read write
---------------------------
10.5 8.8 16.4 16.2
measured peak transfers - MB/s
3400 Pismo
read write read write
---------------------------
11 8.9 55 45
(BTW, both drives are 66.6 Ultra-DMA mode 4 mechanisms)
In the 3400 neither TStar can top 11 MB/s R or 9MB/s W sustained
throughput, but in the Pismo the 5400 rpm drive can do over 16 MB/s R+W
sustained. Peak transfers smaller than either drives' 2MB caches were
much faster in the Pismo but significantly not in the 3400, further
illustrating how severely limited is the 2400/3400's bus speed. It also
shows how well the Pismo should respond to a mechanism faster than either
of these two models.
While these figures don't fully measure a drive's responsiveness in real
life usage, they do give valuable insight into the relative capacities of
the two PBs data paths, specifically how pokey is that of the older 'Book
relative to these drives' potentials.
The limitations of the 2400/3400 IDE bus are reason enough (for me
anyway) to not bother with 5400 or 7200 drives in such applications. I
_can_ see possibly opting for a faster drive in the case where at some
point one might move the HD to a faster PB better able to use the speed.
Even then however, even a big modern 4200 drive is no slouch and entirely
adaquate in a Pismo (at least) for most any chore. Mind you, the above
quoted TStars are already at least two generations old, I reckon the
latest 4200 rpm drives to be faster.
My conclusion? Buy a new 4200 rpm drive for your 2400 and enjoy the speed.
hth,
Dan K
PS: apologies to Gregg for my initial response above! If you know
something I'm missing, I'm all eyes. I enjoy nothing better than learning
something new. :-)
.................................
http://macdan.n3.net/
carracho://dankephoto.dhs.org:9700
hotline://dankephoto.dhs.org:9500
.................................
More information about the DuoList
mailing list