Re: Do Athlon 64's intentially run cool......

The message <43B25F06.EAC7317F@xxxxxxxxxxxxxx>
from Johannes <johs@xxxxxxxxxxxxxx> contains these words:



> GSV Three Minds in a Can wrote:
> >
> > Bitstring <acksf.48417$Cj5.7683@xxxxxxxxxxxxxxxxxxxx>, from the
> > wonderful person Oscar the Cat <noonehome@xxxxxxxxxxxxxxx> said
> > >....I only ask because my new Athlon 64 3.2 GHz never seems to peak
> > >at more
> > >than 35C with average temperature of around 31-32C.
> > >
> > >I'm used to running at much higher temps with my older Athlon XP 3.2 GHz
> >
> > 1) neither of them run at 3.2Ghz, they just have model number 3200+
> >
> > 2) the AMD64 3200+ runs at lower clock speed, and iirc lower voltage,
> > than the prior 3200+ part.
> >
> > 3) the newer parts have ability to reduce their clock speed when nothing
> > interesting is happening, which helps with power and temperature.
> >
> > 4) don't believe the reported temperatures anyway .. well, take a large
> > pinch of salt ..

> If the temp sensor is on the chip, then it's OK. The temp can indeed be low
> if the room environment is cold, but wait until a hot summer day and it's a
> different story.

IMHO, I'd rather that the cpu's power requirements were kept to a
modest level, say 30 watts or less, with minimal power saving control.
In fact, I'd rather have no power control at all and avoid the thermal
cycling that reduces the service life of such hot components.

As it stands, as well as the issue of excessive thermal cycling stress,
another variable has been added into the stability equation. The heat
(temperature) management (cooling demands), require more thorough
testing to determine the system stability under all the most likely
environmental conditions of use (room temperatures, summer and winter.).

By pure 'Dumb Luck'(tm), I'm using an XP2500+ chip that works quite
happily at a Vcore of 1.35v (as opposed to the specified 1.65v) which
has allowed me to reduce the cpu fan speed to inaudibility and give me a
margin under Prime95's stress loading that guarantees reliable operation
in room temperatures up to 40 deg C.

I was able to ascertain this by the simple expedient of using a large
insulated carboard box to create a high temperature test environment for
the system box. I actually got temperatures up to 42 deg ambient, 67deg
cpu, 49 deg system and 51 deg HDD[1].

With those test figures, I think it's safe to say that the operator
(me! :-), will be flaking out from temperature stress B4 the computer
does.

[1] The day after that test, the second hard drive failed to be detected
by the OS and I had to use PQDI to copy the 129GB NTFS partition to
another drive on a test bench rig (relatively painlessly), but it took
me almost a week to fathom out and apply a method to recover the 20GB
FAT32 partition's contents.

Post mortem temperature testing revealed that the afflicted drive must
have been running some 5 deg hotter than Dtemp had been reporting
implying that the drive had been operated at 1 or 2 deg beyond the
"optimal range" of 55 deg C, quoted by Samsung, for no more than an
hour.

It seems Dtemp was using the master SATA drive's temp for both drives.
I've since remedied the 'Hot Slave' syndrome[2] by fitting an extra
drive cage to seperate the drives and mounting a slim 60mm fan to create
air circulation to eliminate the heat build up that typically occurs in
most stock case designs.[3]

[2] & [3] Most cases are still based on designs derived in happier days
when heat output from hard disks was at a more modest level (ignoring
the hot and noisy SCSI types) and the drive bays left almost no gap
between the master and slave drive locations. This would make the upper
drive (typically a slave, if not a secondary master) run a few degrees
hotter.

When drives spun at 5,400rpm, the 2 to 3 watts per drive was not a heat
issue. The large capacity 7,200rpm 3 platter drives now typically
account for 7 to 8 watts each and the tight packing of two drives in the
"HDD Bay" of a 'Standard' Midi ATX tower case has become a serious issue
(probably accounting for the high failure rates of all those
Maxtors.:-)[4].

[4] Maxtors are rated for 60 deg C max, some 5 deg higher tha Samsung's
5 to 55 deg C 'Optimum' range (Thinks: What! It's _all_ optimum?) and
similar or lower maximum figures given by other hard drive manufaturers.
That extra 5 deg is needed, but it's not enough on account they run
noticably hotter than that 5 deg margin over the Samsung Spinpoint
range.

Anybody getting a warranty replacement Maxtor should be giving serious
thought to fitting a cooling fan to circulate air over the replacement
unit and they might also want to consider the cooling air ventillation
layout of the computer case itself whilst they're at it.

HTH

--
Regards, John.

To reply directly, please remove "buttplug" .Mail via the
"Reply Direct" button and Spam-bots will be rejected.

.

Quantcast