|
[10/03/08]
So today, my biggest gripe with the E6400 goes away for good!
My biggest concern with computing these days, and probably for a good year or two now, has been with power consumption and noise. Back in the days of my CpT C and 500m, it really wasn't a big deal, cuz the processors didn't dump much heat and the fan(s) rarely, rarely came on. Hard drives were kind of annoying, but 4200rpm drives are pretty quiet on the whole (unless you get some old clunker with a dying bearing or something silly).
I really started noticing all the noise computers made when I put the old Dell tower in my room. I don't quite remember the rationale for that (except that it was for the lulz), but soon the Dell tower wasn't good enough and I got Colette. In the original configuration, Colette was pretty noise, with a cheapo case and a bunch of 80mm (and smaller fans) running at full tilt. Over the next year or two, I spent considerable time and effort quieting her down, and I've detailed that in the past (too lazy to go reference older posts at the moment).
That being said, it gradually came to be that the most important thing in a computer was pretty much it's power, and subsequently, noise footprint. Power's not the whole story, as many computers these days seem to have really aggressive fan profiles, but I'm not going to get a laptop with a hot processor, a noise fan, and then active noise cancelling speakers in order to make it inaudible. That's clearly inelegant, and that kind of thing matters to me.
The opposite side of the coin is passive cooling. Whenever someone asks me about netbooks or something, I always volunteer the (few) that have passive cooling. Considering the low power consumption of the Atom, it's the perfect candidate for such a thing, but so many manufacturers insist on putting a shitty, whiny fan into their chassis and keeping the whole thing at 32C, whereas I'm pretty sure filling up that fan space with a bigger heatsink would solve the cooling problem just as well.
Of course, that's ultimately speculation, but if the mini 9 can passively cool a 9" chassis, there's no way, it's ridiculous that some of the bigger 10" and 12" netbooks can't.
But I digress. The HP TC4200 really opened my eyes to what kind of noise profile was possible even in a small notebook with a mainstream processor. While my D830 ran it's fan most of the time, the TC4200, undervolted and all, would only crank it up after long periods on my bed or under full load. The fan consistently came on at 60C and shut down at 50C, and thus never ran long.
If every notebook had such a nice and simple fan control, I wouldn't be making this post. Undervolted, the P8400 in my E6400 and the Pentium M in my HP both put out what seem to be similar power figures under light to mid load. The P8400 tops out at a higher wattage, but I don't really mind running a fan if I'm playing SC2 or something. The problem has always been the shitty fan control on the E6400.
I don't want to go into the details, but it pretty much sums up what I hate about "modern" fan control. It turns on at a low, low temperature, and runs forever. It isn't even controlled by a temperature sensor that's picked up by HWMonitor, so it's very, very hard to work around. I can't just lower the temperature and get less fan noise and that is the crux of the issue.
So after the owning the machine for about a year, it began to dawn on me that what I wanted to do was get rid of the damn fan altogether.
The first thing I did was an experiment. I unplugged the fan, booted the machine, and used RMClock to lock my P8400 to its lowest multiplier (for a speed of something like 1600MHz). Then I loaded up the CPU and let it run. I wanted to see if Penryn-based processor running at a lower clock speed would generate enough heat to saturate the passive cooling capabilities of the E6400's stock thermal solution.
It almost did. At 1600MHz and 0.8750v, the temperatuers topped out a bit above 60C, which isn't dangerously hot, but fairly hot. Given that was a full-load speed, it's not that bad, but still a bit hotter than what I'd like. Furthermore, a software solution like this is a little dangerous. If the machine freezes for some reason or another while I'm not at it, the software cap will go away, and the thing will try to run the P8400 at full speed, stock voltage, without a fan. That's a fairly dangerous situation I'd like to avoid.
So what I needed was a processor based on the same Penryn core, but hardware-capped to somewhere around 1.5GHz. Intel doesn't actually make a standard socketed processor like that. The slowest Penryn is a 2.0GHz T4200, which is definitely not going to work here. Older Socket P processors are capped at lower clock speeds, but they are also based on older cores that dump more heat per clock.
But there is a niche market apparently just for this. Intel's ultra-low voltage processors are based on the same cores as their full-power counterparts, but capped at lower clock speeds and rated at lower TDPs. They are also typically soldered to motherboards, and you subsequently can't drop them in a typical mainstraight notebook.
I don't remember why I first looked into this, but for the longest time, there have been a small number of PGA-modded ULV processors on eBay. This means that the ULV processor has been soldered to a PCB with pins on it, such that you CAN put it in said mainstream motherboard. It's a little bit thicker than a standard Socket P part, but thankfully in the case of the E6400, it doesn't matter. I got an SU9300 for just over $100.
Yesterday, it arrived. Today, I put it in.
I ran the load test again, and they were very good. At stock speeds, it stayed under 60C on my desk, and I went so far as to put it on my bed, where it gained another 5C, but 65C isn't too dangerous. Undervolted, I dropped 5C off both of those figures, and the best part is, I can still play Starcraft II! What an victory for silence.
Alright, that's probably it for today.
|
