Amex blue-ray HTPC - me want one =)

[Paranoid Delusion]

Mike

Have a look at http://www.anandtech.com/cpuchipsets...spx?i=2397&p=2
And also any other writeup, there seems to be a 20C difference on running temps between AMD\Intel chips, My normal athlon tops out at 34C (ambient Temp 20C) and thats whilst gaming.
I would NEVER buy a Pentium processor with that inefficiency.

[ziphnor]

Originally Posted by Okay

Well, my CPU temp. is higher when CnQ is disabled (according to Speedfan).
Vcore goes from 1.1 and up to 1.4 when CnQ is enabled.

Thats because you are not manually undervolting! You need to a seperate program to do that. The A64 3000+ runs at 1.0Ghz@1.1V -> 1.8Ghz@1.4V with standard CnQ.

With a seperate program you can force it to also run 1.1V at 1.8Ghz.

[knutinh]

Originally Posted by Paranoid Delusion

Mike

Have a look at http://www.anandtech.com/cpuchipsets...spx?i=2397&p=2
And also any other writeup, there seems to be a 20C difference on running temps between AMD\Intel chips, My normal athlon tops out at 34C (ambient Temp 20C) and thats whilst gaming.
I would NEVER buy a Pentium processor with that inefficiency.

Again, I am cooling my p4 2.8 without any percievable noise, so I will go for any solution that fits my use, wallet and competence.

As a user, CPU temperature and energy use in itself is non-relevant. What is relevant to me is noise and performance. In Norway, a hypothetical 1000W computer that made no noise would make perfect sense most of the year, as the alternative is spending the same watts on electric heating =) I know that building 1000W silent computers is impossible, but my point is to isolate tech-facts and user-interest.

The perfect CPU would allow a close relationship between usage and noise. Heat is part of the equation, but not interesting from a black box perspective. CnQ (and intels recent similar feature) are cool (double meaning there), but if you search the forums over at www.silentpcreview.com it isnt always so easy to get all the parts working properly.

In theory, an integrated CPU thermistor should drive a CPU fan directly, while CPU voltage (and thereby power consumption) was driven by OS/application demand. I still believe (writing this on a CnQ AMD 3000+) that this feature needs more tweaking befori it works as good as it should.

regards
Knut

[mzemina]

Ray - knutinh - Paranoid Delusion

Thanks for the links, I appreciate it!

Mike

[ziphnor]

Originally Posted by knutinh

As a user, CPU temperature and energy use in itself is non-relevant. What is relevant to me is noise and performance. In Norway, a hypothetical 1000W computer that made no noise would make perfect sense most of the year, as the alternative is spending the same watts on electric heating =)

As i said before, you pay alot more for the 1000W of electricity as compared to getting the same effect with normal heating. There are also some enviromental issues, you shouldnt just waste electric energy like that. Using that kind of argumentation, we should all be using power inefficient consumer electronics

From a noise perspective regarding the CPU the only parameter is the CPU thermal dissipation, if its hot it requires more cooling, which tends to be noiser. If you can cool a P4 silently ( passively? ), then fine, but in that sense any CPU can be cooled silently, and then there is no need to prefer one over the other for noise reasons at *all*. You can always take any cooling solution for a hot CPU and apply it to a cool CPU with the same or less noise resulting.

I could see your point if choosing the energy efficient solution had some disadvantage, but i dont see any. Currently the AMD / Core Duo systems are just plain better than their Pentium D counterparts both when seen from a performance-per-watt, performance-per-price, and performance overall point of view.

[knutinh]

Originally Posted by ziphnor

As i said before, you pay alot more for the 1000W of electricity as compared to getting the same effect with normal heating. There are also some enviromental issues, you shouldnt just waste electric energy like that. Using that kind of argumentation, we should all be using power inefficient consumer electronics

As I said before, I do know my bills a lot better than you =) And I pay exactly the same price for electrical heating as I do for electrical computing. In fact, my power company can`t tell if the current is flowing through a heating element or a processor =)

I am aware that most countries have more problems with cooling thier houses properly than heating it, but Norway is cold most of the yeahr, and we have plenty of waterfalls ;-) Most appartements have no other source of heat than electricity.

From a noise perspective regarding the CPU the only parameter is the CPU thermal dissipation, if its hot it requires more cooling, which tends to be noiser. If you can cool a P4 silently ( passively? ), then fine, but in that sense any CPU can be cooled silently, and then there is no need to prefer one over the other for noise reasons at *all*. You can always take any cooling solution for a hot CPU and apply it to a cool CPU with the same or less noise resulting.

I could see your point if choosing the energy efficient solution had some disadvantage, but i dont see any. Currently the AMD / Core Duo systems are just plain better than their Pentium D counterparts both when seen from a performance-per-watt, performance-per-price, and performance overall point of view.

When I bought my systems, I bought the options that made most sense at that time. I am not argueing that p4 precotts are simple to cool silently. In fact, I cant remember exactly what I was argueing ;-) I think it had something to do with your bold, "eith-or" statements, when reality is always complex and faceted.

It used to be at least, that no processor could beat the p4 at running ffdshow really, really fast.

Office computer:
AMD athlon 64 "Venice" 3000+
Asus A8N-E
Zalman 7000 Cu
NorthQ 140mm PSU
2x 512MB ddr
passive cooled Geforce 6600
160MB Samsung p120
NEC 3500 DVD-RW
Chieftec BX04
Speedfan 4.28 software fancontrol

HTPC
Intel p4 "Northwood" 2.8
Asus p4p800E-Deluxe
XP120 with PAPST 4412 G/2LL
NorthQ 140mm PSU
160+250GB Samsung p-series
NEC 3500 DVD-RW
passive cooled Geforce 6600GT
Zalman passive northbridge cooler
Silverstone LC16m
2x512 MB ddr
Speedfan 4.28 software fancontrol

-k

[ziphnor]

Originally Posted by knutinh

As I said before, I do know my bills a lot better than you =) And I pay exactly the same price for electrical heating as I do for electrical computing. In fact, my power company can`t tell if the current is flowing through a heating element or a processor =)

:lol: I know that, i just meant that electric heating is expensive Usually when you want to heat a room you should do so from a heat source placed in a good spot to do so. I dont think using a computer is ideal

I am aware that most countries have more problems with cooling thier houses properly than heating it, but Norway is cold most of the yeahr, and we have plenty of waterfalls ;-) Most appartements have no other source of heat than electricity.

Im surprised hear that electric heating is the only choice. Anyway i live in Denmark, and i spend lots of money on heating my apartment, but we use hot water for that purpose

I am not argueing that p4 precotts are simple to cool silently. In fact, I cant remember exactly what I was argueing ;-) I think it had something to do with your bold, "eith-or" statements, when reality is always complex and faceted.

It used to be at least, that no processor could beat the p4 at running ffdshow really, really fast.

There are some subtle performance differences between the different CPU's but looking at the overall picture its pretty even, and therefore i would definetly choose something that is easy to cool silently. I also think youll be hard pressed to find others that wouldnt strongly recommend the same for people buying a HTPC system now.

Btw, utilizing a low power CPU makes it possible to use a small fanless PSU such as the picoPSU:
http://www.silentpcreview.com/article601-page1.html
this means that you can easily get away with only having only a single case fan in the entire HTPC. Such a PSU would have trouble just powering a Pentium D without any other components

There is also the matter of fitting HTPC parts in a small case, where its nice to be able to use a small passive CPU heatsink. Some guy on the silentpcreview forum posted the following pic of his core duo:

(he later removed the fan from the chipset heatsink btw).

Note how small the heatsink is, and yet it has no trouble cooling the Core Duo.

But i guess if one wants a little more performance in ffdshow at the cost of ~4 times the power usage, then you are right that one should choose to use a Pentium D

Seriously though, i didnt meant to that definite, but i have serious difficulties seeing why one would go for a pentium D based HTPC design.