Downgrading from first-gen i5 to ULV?

A very brief search on Intel's site shows only BGA packages for first-gen core-i ULV processors, but it's worth asking anyway.

Does anybody know if it's possible to downgrade an i5-450M to a UM or even LM processor? As my Envy 14 ages and my desktop becomes my primary computer, I'm more concerned with the portability of the Envy 14 than its power. I'm not really certain how big the difference would be switching from a 35W processor to a 17W or 25W processor since all three would be mostly idle most of the time while I'm unplugged (light web browsing, standard-res video watching, very light non-3D games), but if the swap could give me 25% more battery life I'd happily sacrifice the excess processing power.

So does anybody know of first-gen Core-i ULV processors that could be swapped into my system?

 

Theoretically yes, alot of eBay sellers have PGA versions of ULV Arrandale CPU's, they just cost an arm and a leg over the standard voltage ones.

Commander Wolf bought a SU9300 PGA and swapped it into his E6400, it was just a BGA CPU soldered onto a PGA.

 

Dang, I was definitely hoping this would be a cheap swap since the laptop itself is probably only worth $400-500 these days, and half of that is the Radiance screen.

I'm seeing seller for BGA转PGA (BGA converted to PGA - probably the same thing as Commander Wolf got) processors here in China on TaoBao - often called Chinese eBay though it's not quite the same. I checked a bunch of i7 UM and LM parts really quickly, prices range from about $28 for an i7-620UM to about $95 for an i7-680UM or $110 for an i7-660LM.

I'm definitely considering making the change, but I have a couple questions:

-Will a UM or LM processor make a notable difference in my battery life or will it be a waste of time and money to downgrade? If so, will the difference between UM and LM be significant as well?

-Will I need to modify or replace the cooling system to accommodate the new processor?

-Knowing that every bios is different, would a ULM/LM i7 typically be a plug-and-play drop-in replacement for the i5-450M or would I run the risk of it not working at all for BIOS, chipset, or socket reasons?

I'm just trying to do my homework in advance here. The Envy 14's a real bear to open up and the last time I tried I got down to the fan assembly and found a screw that was in so tight it actually stripped the blade on my cheap phillips screwdriver, so I ended up wasting a couple hours disassembling and reassembling the machine for zero benefit. It's been opened up by a shop since then when they replaced the fans, so that shouldn't be a problem this time (and I got a better screwdriver). The theoretical "buy the cheapest one and try it out" path is open, but far from ideal since it not only increases the eventual cost if I buy two processors, but also means I have to do two CPU replacements to get to where I want it.

 

Well, since you asked...

In the case of the SU9300 there was no practical change in battery life. The SU9300 basically had the same power profile as the P8x00 I had previously; it's just capped at 10W at 1.2GHz. Obviously if you are always running full CPU load on battery you will see a substantial difference using SU9300, but then you will also be doing less work, so again, kind of useless.

In the case of the SU9300, the entire thing (from bottom of package - not the pins - to the top of the die) was slightly thicker than a regular CPU, so if you have other chips on your heatsink (ie a dGPU with a thermal pad), you may have complications with contact.

Depends. Dell is pretty good at letting this stuff slide, but I Lenovo or HP may be more of a about it.

 

I don't think HP whitelists CPU's, Lenovo is pretty picky about non-factory/supported CPU's.

 

I'll have to look into the power profiles for the Arrandale UM/LM chips. I kind've suspected it might not make much of a difference in battery life, but isn't battery life about 50% of the point of ULV CPUs (with the other half being low heat emission)? They're a generation apart so it's possible, though maybe not likely, that low-voltage Arrandale chips might idle at lower power states than the regular-voltage chips. More homework for me.

Being thicker is another legitimate concern. The Envy 14's cooling system is connected to the processor and the GPU, so I wouldn't want to compromise the system like that.

The more I think about this (it's an idea I've only been toying with for a couple hours), the more I think it's not going to work out to any real benefit, and, illogically, the more I want to try it. I might just look for a regular voltage dual-core i7 (or would they run significantly hotter?) to get my jollies instead, though searching TaoBao for "i7-660M" or "i7-680M" just ends up giving me a bunch of gaming laptops with Kepler GPUs.

 

before you pulled a trigger I would highly recommend you tweak processor for power saving like disabling 1 core or further disabling hyperthreading. I bet simple core disabling will not put cpu much lower then performance of ULV.
To disable 1 core write msconfig in Run and press enter. go to advanced options of boot and set number of processors to 2 instead of 4. If you decide that you want to also disable HT because you still get enough cpu power then set it to 1. To revert it to default just uncheck box.
Also buy a SSd instead of HDD and buy low volt RAM 1.35V. I do not know if samsung 1600 ram will downclock to 1066 but worth to try IMO.
About BIOS... if it is AMI then it is peace of cake to check existence of proper CPU microcde while AWARD is little bit longer.

 

I mean that might be the marketing - and it will give you a computationally costly increase in battery life if you're constantly fully-loading the CPU on battery, but as far as I've seen with Penryn there hasn't been a significant difference in idle and low-load power consumption between the ULV, LV, and 25W parts. I don't actually know if such is the case with the Penryn 35W parts or Arrandale/SB/IB stuff, so you'll have to do your own research on that.

That being said, I believe the primary purpose of the ULV/LV stuff is to not melt the dinky cooling systems in Ultraportables and Ultrabooks, which is more in line with what I was doing.

 

That would be my guess too, every recent Intel CPU has rather low power consumption under light load. The minute you need more CPU power though, they're bound to use more power though, but if all you're doing is Internet browsing, I doubt it will make much of a difference.

 

I might try disabling a core and/or HT when I need battery life. Does this really make much of a difference, though? I'm not going to bother swapping in low voltage RAM - if the low voltage were a 50% difference and not a 10% difference it might be worth it, but not for such a minor difference. I'm already running an SSD, as well. It's not really that I'm all that dissatisfied with my battery life (though it would be nice if I could get more without the slice battery, making it more portable for certain situations), more that I've got the upgrade bug and want to change something for the heck of it, but I'm not going to spend money to do so unless it's going to do me some good.

Yeah, it seems more and more likely that it wouldn't really do anything for me. The fastest Arrandale dual-core is .4 GHz faster than mine and has one more MB of cache, and I don't want to lose switchable graphics with a quad-core, so I guess this idea is probably gonna be sidelined. Saves me some money, but I lose out on having a little fun.

 

Do some tests on your own and tell us time difference with disabled core for NBR forum.

 

I would download and install throttlestop and use that to control the cpu speed and turbo modes, you can also lock the cpu to certain speeds while on battery to inprove battery life, I don't think you will see any noticable increase in battery life swapping cpu's, just an increase in price, don't think it's worth the investment, try throttlestop first

 

I might do that. My initial thought of swapping was partially the itch to tinker and partially the theory that ULV CPUs would have lower power usage in low usage scenarios. Looks like I'm probably wrong.