Is my PSU enough?

  • Thread starter Thread starter zm
  • Start date Start date
Z

zm

I have:

AMD 2200+ (I want to overclock it), 1GB RAM, 160 GB HD, DVD-RW,
CD-RW.

USB ports: Scanner, ADSL Modem, micro/earphone,....

I have a 350 W PSU. Is that enough?

Thanks in advance.
 
I have:

AMD 2200+ (I want to overclock it), 1GB RAM, 160 GB HD, DVD-RW,
CD-RW.

USB ports: Scanner, ADSL Modem, micro/earphone,....

I have a 350 W PSU. Is that enough?

Thanks in advance.
Click to expand...

Video card is typically the 2nd or 3rd most demanding device
in a box... you didn't mention it.

Also no mention of specific make/model of PSU... all 350W
PSU are definitely not created equal. In general, a decent
name-brand 350W is enough but a generic may be liability
even at stock speeds, after it's aged a bit.
 
zm said:
I have:

AMD 2200+ (I want to overclock it), 1GB RAM, 160 GB HD, DVD-RW,
CD-RW.

USB ports: Scanner, ADSL Modem, micro/earphone,....

I have a 350 W PSU. Is that enough?

Thanks in advance.
Click to expand...

should be ok

I am using a 350w PSU with an AMD Athlon XP2600+ 1.5gb RAM Geforce 5200
graphics 80 gb hdd, dvd-rom/cd-rw

I am getting the odd monitor black out, i would like a 400w PSU.. and
eventually 600w after all the modding im gonna do its gonna need it

plus OC'n
 
zm said:
I have:

AMD 2200+ (I want to overclock it), 1GB RAM, 160 GB HD, DVD-RW,
CD-RW.
Click to expand...

Why do you want to overclock it? Do you actually think you will notice
any difference? I prettty much doubt it.
 
kony said:
Video card is typically the 2nd or 3rd most demanding device
in a box... you didn't mention it.
Click to expand...

Geforce FX5200
Also no mention of specific make/model of PSU... all 350W
PSU are definitely not created equal. In general, a decent
name-brand 350W is enough but a generic may be liability
even at stock speeds, after it's aged a bit.
Click to expand...

Q-TEC ADT-350

What I don't like in this PSU, is that when I'm browsing (right cursor
w/mouse) or in Word running through the pages (right cursor w/mouse),
the PSU makes a noise like zzzzzzzzzzzzzz.

Once a week or so I use a 2nd. HD to backup with Ghost my 1st. HD
 
On 10 Oct 2004 01:41:50 -0700, (e-mail address removed)
(zemariano) wrote:

Geforce FX5200

Q-TEC ADT-350

What I don't like in this PSU, is that when I'm browsing (right cursor
w/mouse) or in Word running through the pages (right cursor w/mouse),
the PSU makes a noise like zzzzzzzzzzzzzz.

Once a week or so I use a 2nd. HD to backup with Ghost my 1st. HD
Click to expand...

The video card isn't much additional load but Q-Tec PSU
aren't very good, you'd be better off with even a 300W-350W
Sparkle/Fortron, Antec Truepower, or 360W+
Thermaltake/Enlight.... just examples, there are a lot of
brands good enough for that component load but generally
generics are best avoided.


"Sometimes" the motherboard itself can be the source of
noise that changes based on load. If this is a possibility,
try touching the inductors while reprducing events that
change the noise, when firmly grasping the part it should
change or reduce the noise (but DON'T do that with parts
inside the power supply itself unless you know what you're
doing).
 
kony said:
On 10 Oct 2004 01:41:50 -0700, (e-mail address removed)
(zemariano) wrote:



The video card isn't much additional load but Q-Tec PSU
aren't very good, you'd be better off with even a 300W-350W
Sparkle/Fortron, Antec Truepower, or 360W+
Thermaltake/Enlight.... just examples, there are a lot of
brands good enough for that component load but generally
generics are best avoided.


"Sometimes" the motherboard itself can be the source of
noise that changes based on load. If this is a possibility,
try touching the inductors while reprducing events that
change the noise, when firmly grasping the part it should
change or reduce the noise (but DON'T do that with parts
inside the power supply itself unless you know what you're
doing).
Click to expand...

Thanks again (and again), Kony.
 
AMD 2200+ (I want to overclock it), 1GB RAM, 160 GB HD, DVD-RW,
CD-RW.
Click to expand...

If it works at the moment, then it will still work when you overclock it!

The only reason overclocking would require more power than running at
standard speed is if you overclock far enough to have to increase the
voltage on the CPU and that would only draw a little more power: I think the
athlon XP takes about 60 watts, so upping the voltage by 10% (that's quite a
lot) would require 10% more power. Keep an eye on your temperatures though -
you might need more cooling fans, but they would only draw a few watts each.
 
AMD 2200+ (I want to overclock it), 1GB RAM, 160 GB HD, DVD-RW,
If it works at the moment, then it will still work when you overclock it!

The only reason overclocking would require more power than running at
standard speed is if you overclock far enough to have to increase the
voltage on the CPU and that would only draw a little more power: I think
the athlon XP takes about 60 watts, so upping the voltage by 10% (that's
quite a lot) would require 10% more power. Keep an eye on your
temperatures though - you might need more cooling fans, but they would
only draw a few watts each.
Click to expand...

Incidentally, I just bought a new PSU - its a Nexus NX-3000 ATX PSU 300W. It
cost £15 on the QuietPC.co.uk site. If I turn off all the other fans in my
system, I can hear it, but its pretty quiet. My system runs just fine with
it:

Athlon 2400+
2 IDE drives
2GB SDRAM ( 3 modules)
1 DVD/CDRW combo drive (used to be fine with DVD + separate CD writer + SCSI
card).
PCI sound card
AGP Radeon 8500
PCI wireless card
PCI network card
 
Gareth Tuckwell said:
If it works at the moment, then it will still work when you overclock it!
Click to expand...

Uhm... No.

Why do you think the CPU gets hotter when overclocked? Where does all that
power come from?

Just overclock and monitor temps, voltages and run Prime95, etc. to test
stability. If the machine becomes unstable then start investigating what
brings it back to stable. There is no single solution.
 
AMD 2200+ (I want to overclock it), 1GB RAM, 160 GB HD, DVD-RW,
Uhm... No.

Why do you think the CPU gets hotter when overclocked? Where does all that
power come from?
Click to expand...

I assumed that overclocking a CPU only produces more heat because you have
to increase the voltage in order to achieve stability at the increased the
CPU speeds

If a CPU produces more heat at faster speeds even with the same voltage -
where does the extra heat come from if the power (volts x amps) intake is
the same? I have experimented with voltages and now run my CPU at its lowest
(stable) voltage - 1.5v instead of the 1.65v. This reduces the heat
generated by the chip by several degrees, but I have not changed the clock
speed. I assumed therefore that the power used by the chip was directly
responsible for its heat

So is speed a direct factor in the heat calculation?
 
I assumed that overclocking a CPU only produces more heat because you have
to increase the voltage in order to achieve stability at the increased the
CPU speeds

If a CPU produces more heat at faster speeds even with the same voltage -
where does the extra heat come from if the power (volts x amps) intake is
the same?
Click to expand...

Amps is not the same, the amperage goes up.

I have experimented with voltages and now run my CPU at its lowest
(stable) voltage - 1.5v instead of the 1.65v. This reduces the heat
generated by the chip by several degrees, but I have not changed the clock
speed. I assumed therefore that the power used by the chip was directly
responsible for its heat

So is speed a direct factor in the heat calculation?
Click to expand...

Yes, see my last post in this thread.
 
If it works at the moment, then it will still work when you overclock it!

The only reason overclocking would require more power than running at
standard speed is if you overclock far enough to have to increase the
voltage on the CPU and that would only draw a little more power: I think the
athlon XP takes about 60 watts, so upping the voltage by 10% (that's quite a
lot) would require 10% more power. Keep an eye on your temperatures though -
you might need more cooling fans, but they would only draw a few watts each.
Click to expand...


Current consumption goes up linearly with clock speed.

If a 2GHz CPU uses 40W, the same CPU running at 2.5Ghz
(without any vcore increase) will use 50W. This does ignore
other factors, like the fact that same CPU running at higher
speed will finish jobs faster (all things being equal), and
thus be idle more often for "typical" uses in a PC, but
would be true with both running at full load, providing the
other subsystems can keep up with the faster CPU's higher
processing ability.

Upping the voltage on the other hand, is NOT a linear
wattage increase. Power is proportional to the voltage
squared.

If a given CPU running at 2GHz & 1.3V, uses 40W, the same
CPU, still running at 2GHz, but with vcore raised to 1.5V,
will use 53W, but in this case there is no grace for
different performanc and job run times since the speed is
the same.

NOW consider if we take the 2GHz CPU @ 1.3V, and o'c to
2.5GHz @ 1.5V... 67W

That is only the CPU though, if you raise the FSB and memory
then the motherboard also has a linear power increase with
clock rate. The motherboard's CPU VRM circuit is also only
85-95% efficient (IIRC) so it is a further consumption too.
 
Current consumption goes up linearly with clock speed.
If a 2GHz CPU uses 40W, the same CPU running at 2.5Ghz
(without any vcore increase) will use 50W.

Upping the voltage on the other hand, is NOT a linear
wattage increase. Power is proportional to the voltage
squared.

If a given CPU running at 2GHz & 1.3V, uses 40W, the same
CPU, still running at 2GHz, but with vcore raised to 1.5V,
will use 53W, but in this case there is no grace for
different performanc and job run times since the speed is
the same.
Click to expand...

Excellent - thanks for the info - very interesting.

Is there a formula integrating all the above that we can use to calculate
how much heat a processor should generate given certain voltages + speed
settings? I am interesting in finding out how much cooler I could run my
CPU, or how much extra power I could squeeze out without much extra heat: I
have a thoroughbred Athlon XP 2400 at present. Normally runs at 1.65v and
2000MHz (133x15). I have undervolted it to 1.5v, but not changed speed
settings.
 
Excellent - thanks for the info - very interesting.

Is there a formula integrating all the above that we can use to calculate
how much heat a processor should generate given certain voltages + speed
settings? I am interesting in finding out how much cooler I could run my
CPU, or how much extra power I could squeeze out without much extra heat: I
have a thoroughbred Athlon XP 2400 at present. Normally runs at 1.65v and
2000MHz (133x15). I have undervolted it to 1.5v, but not changed speed
settings.
Click to expand...

First determine the following for your CPU model by
consulting the manufacturer's spec sheet (or any other
alternate method you want, since you'll never run CPU at
"100%" load even using a CPU stress test, last estimate I
read was it'd be closer to 85-93% of full load, but for the
purposes of comparison of _relative_ heat, 100% value will
suffice.

X = Watts / voltage ²


Then calc the new wattage (heat)

Watts = (X * voltage² ) * (New MHz Speed / Spec'd MHz Speed)



As for "how much cooler", I wouldn't worry so much about
that, temp at reduced core frequency can often be higher
than at higher frequency with CPU remaining stable. More
important might be that it's stable running at the noise
level you find acceptable (if your choice of fan speed/size
as it effects temp, is tolerable for the environment it's
in).

To get a better picture of the tradoffs being made, graph
out how much voltage it needs per each operating frequency
(or calculate each wattage and graph for wattage instead if
you like) then it should be easier to see based on the
curve, where the points of diminishing return are... You'll
probably need a certain minimal vcore to get the box to POST
at all, and a certain max (with "normal" cooling) frequency
no matter how much higher the voltage. Past a certain point
voltage increase has a negative effect, CPU becomes too hot
at full load.

I would GUESS (based on typical results with T'breds) that
at 2GHz & 1.5V, you're in a good zone already for MHz to
voltage compromise. While it's easier to cool the
larger-cored Bartons (and some Thortons with half cache
disabled also have larger core surface area), they cost more
too, so again another tradeoff.
 
Back
Top