Sign in with
Sign up | Sign in
Your question

Cache size have impact on CPU temperature?

Last response: in CPUs
Share
June 3, 2005 8:27:20 PM

I'm building a HTPC and want it to be as cool and quiet as possible. I am considering a 3500+ Venice or 3700+ San Diego. However, based on a PSU calculator I consulted, the Venice consumes 67 watts while the San Diego consumes 89 watts. Assuming all things equal (clock speed, HSF) will the greater wattage directly translate into higher temps? BTW, is this difference in power consumption due to the larger cache on the San Diego?

Thanks!
June 3, 2005 9:28:22 PM

Quote:
will the greater wattage directly translate into higher temps?

Yes

Nothing is as easy as it looks
June 4, 2005 2:33:24 AM

Boy, you sure cant argue with that. In this case watts = heat. So yah 89 is more than 67.
Why then do I think the San Diego is a better core for you?
Those are max values. With the Amd chips, it is really hard to max out the cache usage. Even if you did, the extra cache makes the core almost 50% larger. That means the watts per square cm is lower on the San Diego core. Since the core has more contact area with the heat spreader, it is also better able to disapate heat.
All in all, this means that the San Diego core is better able to shed excess heat, and can tolerate more heat, but at the same time, is less likely to see it's max heat output.
Related resources
a b à CPUs
June 4, 2005 6:38:05 AM

Go with the Venice. There are lots of nice Micro ATX cases that resemble an audio component and fit nicely into a 17" wide rack, don't get one of those stupid cube-shaped things. MSI has a good Micro ATX board for that CPU.

<font color=blue>Only a place as big as the internet could be home to a hero as big as Crashman!</font color=blue>
<font color=red>Only a place as big as the internet could be home to an ego as large as Crashman's!</font color=red>
June 4, 2005 4:16:41 PM

I actually got an Ahanix D-Vine 4 case. I got it because it accepted a full-sized ATX board and PSU. I also decided to go with a Venice 3200+ for cost reasons. I know that wasn't part of the original choice I presented, but the discussion still helped with my decision.

Which brings me to a follow-up question. Just what is the difference (besides clock speed, obviously) between, say, a Venice 3200+ and 3500+? They both have the same cache size and power consumption. What makes one run at 2.0GHz and the other at 2.2GHz? Is 3200+ just a defective 3500+? Is there a difference in transistor count? Does the 3500+ have higher quality parts that allow it to run at a higher clock speed? I just read an article where they were able to overclock a 3200+ to a max 2.7GHz stable, while they were only able to get a 3500+ to that same max 2.7GHz. Leads me to believe it's the same 'stuff' inside with the clock speed fixed at a lower rate. But then again, what do I know...that's why I'm asking the question!

Thanks for all your help.

Barley
a b à CPUs
June 4, 2005 10:03:17 PM

Sometimes they test processors at various clock speeds to determine what the highest suitable clock speed is at a certain voltage.

I've heard they'll bin processors depending on what part of the die they come from, with those from the center being tested for the highest clock speeds, and those near the edge being tested for lower clock speeds.

Processors have often been down-binned when the market demands more lower speed (cheaper) processors than the company is producing. In such instances one that might have previously been considered for 3500+ speeds might be dropped to, say, 3200+ speed to fill the market need.

<font color=blue>Only a place as big as the internet could be home to a hero as big as Crashman!</font color=blue>
<font color=red>Only a place as big as the internet could be home to an ego as large as Crashman's!</font color=red>
June 5, 2005 12:57:01 AM

>Assuming all things equal (clock speed, HSF) will the greater
> wattage directly translate into higher temps?

No. Thermal density matters too, when it comes to cpu temperature. Venice is smaller than San Diego, therefore has to dissapate its heat on a smaller surface, which should result in higher core temps (for equal power). However, the majority of this effect is negated by the heatspreader and the heatsink (assuming both being equal), but since no heatsink will ever achieve a infinite thermal conductivity, a 80mm² 80W chip will be (ever so slightly) hotter than a 100mm² 80W chip.

But this is an academic discussion, rather than a practical one. Because there is not only the heatsink effect, there is also the fact that cache consumes considerably less power than the cpu core per mm². San Diego and Venice have identical sized cores, and I estimate at least 90% of the power is consumed there, and not in the caches, so really, any difference would be nearly impossible to measure, but in theory the smaller chip will be hotter than the bigger one, all other things (especially power consumption and cooling) being equal.

Finally, the die temperature may not be your biggest concern, but the heat created inside the box, that needs to be moved out might, and there 80W is 80W regardless of cpu size or core temperature.

I hope I didn't confuse you too much, but if I did: just get a Venice core :D 

= The views stated herein are my personal views, and not necessarily the views of my wife. =
!