• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Lets talk socket 754...

A

artificio

n00b
Joined
Sep 4, 2004
Messages
16
After reading this article, I was wondering if I could get my 64 bit Sempron 3000+ down to that level of power consumption w/ a m-atx motherboard. Right now, I've gotten it down from 120W to 85W at full load via cpufreq, and with a custom built kernel should be able to drop it a bit more via voltage/fan speed/etc...

That being said, I doubt I can cut more than about 5W at full load this way, so, are there any m-atx boards that would cut out at least 20-25W compared to my ECS 755-A2? Any one have any experience with the power draw of 754 m-atx boards?
 
Are you looking for a motherboard that will undervolt the CPU? I'm not sure you're going to get a 20-25W decrease from just swapping out the motherboard even if you disable all of the non-essentials.
 
I'm trying to get consumption on this thing down into the 40-50W range at full load based on what some other members said but after some research it looks like their claims were just hot air. So far, at 1.1V with a single SATA drive and a pci VC peak consumption is around 80W, and with .9V it's around 70W.

The claim I was referring to was that an undervolted AMD system could use the same power a C7-D system uses, but thus far my system (Sempron) still uses 68W at full load while my C7-D uses 37W with equivalent equipment at full load. If, like ya said, there's very little difference between mobo power consumption, a m-atx board won't help, and the Sempron requires significantly more power than the C7-D even undervolted.
 
i just sold my socket 754 and a motherboard for 90 bucks. hahahaha i guess some guy needed a 754 set up bad.

i got so many damn heat sinks laying around everywhere so i gave the guy a brand new big typhoon heat sink
 
Yeah, it just sounds like the sempron isn't going to get the job done, however, for the added power consumption, the sempron should be much more powerful anyway.

You check out some of the low power notebook s754 CPU's. That might be your best bet. Not sure if you're going to get as low as the Via chips, but it might get you another 10W or so.
 
A mobile processor may help... The TDP for my SDA3000AIO2BX is 59W, and going from 1.8ghz/1.4V to 1ghz/.9V drops that to 32W with system use at 68W (compared to 95W at full load/voltage). A mobile proc could only drop that to 61W, maybe lower, but there isn't as much room to go down in terms of voltage. I still have some parts to swap/test in order to determine what draws what, but it seems like the VIA system is the low power winner with a TDP of 15W and max load of 44W w/ an extra PCI sound card... The Sempron system with a mobile proc would need a max load TDP of 8W at some voltage in order to match the Via's power consumption. If that's even possible. I haven't switched around the hard disks or power supplies yet, so that may bring the two closer together in an extreme case, but that doesn't seem very likely.
 
Swapped power supplies and at 1ghz/.9V and full load the Sempron only sucked down 5W more than the C7. :eek: Ah well, I suppose I can use the C7 for something else or just ebay it and recoup 75% of my purchase price.
 
Is this just a hobby or are you trying to do something with such low power consumption?

How does the Sempron at 1Ghz compare to a C7 in terms of performance?
 
I cant imagine getting power consumption going to much lower from just changing a motherboard. The only way to do it might be the mobile chip. Number of RAM sticks and voltage might add a few watts as well.

Wasnt socket 754 like 110nm or 130nm? And is the Mobile chip in that article the same size? Your chip might just not be capable of hitting the lower wattage due to being older technology.

It might be a better route to just go AM2. Pick up a budget board ($20-$40), 3600x2 ($40), and 1GB Or 2GB DDR2 ($20 - $40). Save yourself the headache of trying other parts for half the cost and not being satisfied. I have a few 3600x2s that can do stock speed @1.0v or less.
 
I'm just putting together a setup that'll suck down a little power as possible for data logging/convenience. I dunno how much lower a AMD2 setup can go, but I'm pretty sure it would take a decade or so to pay off the ~$50-100 difference in price via a 10W savings by undervolting. According to AMD's data sheet, at 1ghz/1.1V the SDA3000AIO2BX only uses 21.7W, so I imagine that at 1ghz/.9V it uses around 15W. Being 64bit capable is nice since it really speeds some stuff up, and since it's a 90nm it's not to bad for lower voltage/stock speed. IIRC dropping from 1.4V to 1.1V won't hurt the 1.8ghz speed, although at 1V and stock speed it freezes. I think I'll try some benchmarks between the 1.5ghz C7 and 1ghz Sempron later on to see what kinda differences there are.
 
Alrighty, the 1.5ghz C7 was noticeably slower than the Sempron@1ghz/.9V in all of hardinfo's benchmarks. Surprisingly, this underclocked/undervolted Sempron even outdid the 1.5ghz mobile Intel proc used by the program's author for comparison on a couple of the benchmarks. So... I figure that an extra 10-15W is worth using something that's about 30-100% faster underclocked/undervolted, especially since it would go unused for a 4-6 hours each day, and I can just cat waketime in order to have it start streaming stuff to the household in the morning.
 
You must log in or register to reply here.
Back
Top