A tale of two possible mini-ITX builds

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

vailr

Diamond Member
Oct 9, 1999
5,365
54
91
As far as "best bang for the electrical power buck" in tiny-sized hardware, shouldn't the Intel NUC (or Gigabyte's equivalent) be considered? Maybe even throw in a MAC mini for comparison, as well. More cost initially for the hardware, but you (maybe eventually) make up for it in lower power consumption.
 

Insert_Nickname

Diamond Member
May 6, 2012
4,971
1,692
136
As far as "best bang for the electrical power buck" in tiny-sized hardware, shouldn't the Intel NUC (or Gigabyte's equivalent) be considered? Maybe even throw in a MAC mini for comparison, as well. More cost initially for the hardware, but you (maybe eventually) make up for it in lower power consumption.

Unless there is a NUC or equivalent you can plug-in a PCIe graphics card, no. But if I where after ultimate performance-per-watt, sure, I'd consider one of those.

I'm just running this as an intellectual exercise on how much performance you can squeeze out of every Watt/Joule spent. Nothing more and nothing less. I don't mind stuff using a lot of power, but it has to justify the additional energy spent for the additional performance...

Dealing in absolute performance possible has more-or-less come down to "buy a 4790K/5960X and OC the heck out of it". I don't find that particularly intellectually stimulating. Except the OC part of course... :D
 

mfenn

Elite Member
Jan 17, 2010
22,400
5
71
www.mfenn.com
I think we're somehow talking past each other. Lets worst-case and simplify it. If you have a CPU that uses 35W full-load for an hour, you'll have used 0.035KWh. If you have an 84W TDP you'll have used 0.084KWh. A difference of 0.049KWh. Since you're billed per KWh, the 35W TDP will cost you ~7"øre" (1/100 of a Danish Krone) per hour to run, the 84W will cost you ~18"øre" per hour to run. Over twice the cost for a best-case extremely minimal increase in performance.

We have a saying in Danish, that freely translated reads "A lot of little creeks make a very big river..."

It is a gross over-simplification, but am I making myself clear...? :)

I understand what you're saying. What I'm saying is that given a fixed workload (either trivially or due to waiting on other components), the 84W TDP chip is not going to run at 84W. So your worst case isn't going to happen unless you're tilting the odds by making the processors do different amounts of work.
 

Insert_Nickname

Diamond Member
May 6, 2012
4,971
1,692
136
I understand what you're saying. What I'm saying is that given a fixed workload (either trivially or due to waiting on other components), the 84W TDP chip is not going to run at 84W. So your worst case isn't going to happen unless you're tilting the odds by making the processors do different amounts of work.

For a static workload/fixed task, I agree 100%. The faster CPU will finish first, thereby getting idle faster. Its what happens with a dynamic workload (like gaming) I'm interested in. I haven't been able to find information on that anywhere, except an old article from Xbitlabs testing the S/T Sandy Bridges. But again that wasn't really what I'm looking for. The closest I've been able to find is data on power use from servers, which doesn't really translate well to a desktop environment.

I do appreciate your input... :)
 

VirtualLarry

No Lifer
Aug 25, 2001
56,542
10,167
126
I'd expect the 1037U to "feel" snappier just surfing the web and running Windows. But that said, your inability to surf the web on the E1-2500 is probably for a software problem. If you're not using NoScript and FlashBlock, you're going to have hangs and stuff even on a full-blown i7 desktop. Netflix on lower-end hardware comes down to quality settings and HTML5/Silverlight. And so on. You know the drill.

But if the cost of the H61/G2030 is close enough (and it looks like you can get those for around $90) then I'd go with that. It'll pwn.

Interesting observation. I noticed that on my G630 rig, scrolling got really bogged down on the current page, when I had about 50 tabs open. Which is about how the E1-2500 was. Well, at best, the E1-2500 didn't have as smooth scrolling as the G630 with NV GT430. But still, I was surprised that it bogged down just as bad. CPU usage wasn't 100% on the G630, it was about 50-60%. Whereas on the E1-2500, it was near 100% at times.

I think Waterfox 31.0 is just a bit more bloated than 30.0, 29.0, or 28.0.

Edit: I was all set to get the ECS H61 mITX, but they raised the price today back to $59.99. And they pulled the G3258 w/MSI Z87 PC Mate combo for $104.99.
 
Last edited:
Feb 25, 2011
16,890
1,539
126
I just use plain old Firefox, but 50 tabs open - if they're the right tabs - will lag on any system I use on a regular basis. (The i5 in my sig, a similar-specced i5 hackintosh, my 15" rMBP and the i7 workstation at work.)

It has more to do with the content, scripts enabled, etc. Bad javascript will ruin your day.
 

mfenn

Elite Member
Jan 17, 2010
22,400
5
71
www.mfenn.com
I just use plain old Firefox, but 50 tabs open - if they're the right tabs - will lag on any system I use on a regular basis. (The i5 in my sig, a similar-specced i5 hackintosh, my 15" rMBP and the i7 workstation at work.)

It has more to do with the content, scripts enabled, etc. Bad javascript will ruin your day.

Yep, Firefox has a real issue with intensive scripts on background tabs affecting performance on the foreground tab. It's really annoying, and gets worse the more tabs you have open (and as the Internet in general gets more javascript-y).

Chrome is better about this, but suffers from ridiculous memory requirements as the tab count increases. It's the class time vs. space trade-off.