Hopefully this doesn't double post, sorry if it does.

I'm trying to put together a 24/7 media streaming server (probably Ubuntu) that I can fool around with some VMs (likely two VMs running at any one time) on. The trouble is, I don't have any parts that I can recycle for the project and I can't seem to decide on a CPU to use.

My first thought was to use a quad core X4 640/5. After browsing around a bit though, I can't help but feel that the chip is more than a little outdated and a newer architecture would be a better choice for both power and performance per dollar. At the same time, I can't help but feel like the X4 640 is already overkill for my needs.

I realize some of these are completely overkill, but these are some of the CPUs I was mulling over:

X4 640
X4 840
X6 1045T (depending where you look, it's only $30 more than an X4)
i3 2100/2125
i5 2400 (this is far too much processor in my mind)

I've kind of dismissed APUs (I think the IGP would be wasted for my uses), SB Celerons and Pentiums (not sure they could handle streaming with multiple VMs all at once), Atoms and Zacates (tried Atoms before, too weak for me) and of course the super high end SBs and BDs.

Am I missing anything or unfairly judging the APUs and budget SB CPUs? I'm not looking for a crazy ultra low power solution but power efficiency is important. After undervolting, would the inefficiency of the AMDs be negated when compared to the Intel SBs? If there is only a few watts difference, that's not a huge deal, but if it starts hitting 10 or more.

If anyone would be kind enough to give me some suggestions on a CPU and ATX/mATX motherboard combo that would fulfill my needs above as well as allowing easy undervolting (whether through the MB or linux-PHC, etc.), I would appreciate it. My budget is ~$500 (not including hard drives), but spending less money is always welcome.

Thanks for the help.

Actually, an AMD APU would probably work OK for you... depending on just what you want to run on your VM. Their main issue is that when red-lined, they draw a lot more power than the Intel chips. See the tables on this page -- http://www.silentpcreview.com/article1211-page5.html At idle, the difference is just a few watts, though, so if you never push the CPU or video portion of the chip that hard, then, it's viable power-wise.

Note that the A8-3850 is a 100W TDP part. We'd asked for a A8-3800 sample, which is a similar quad-core clocked 10% slower with a TDP of 65W; AMD didn't supply us the part, unfortunately, but TechReport got one, and their review suggests that it is much closer in both idle and peak power to the Intel i3-2100. http://techreport.com/articles.x/21730/2

Still the pricing of the APUs is not great at ~$140. The i3-2100 is a dualcore but it is hyperthreaded and the price is actually a bit cheaper -- by $20?

I would not bother with undervolting. It was useful for many years, but Intel chips do a good enough job, your energy savings are too small really to be worthwhile.

Finally, you don't say what your audio performance requirements are but if you want the best audio results, you're best off making a PC dedicated for the purpose. I don't agree with everything this author has to say here, but I think his basic approach is sound: http://www.computeraudiophile.com/conte ... r-CAPS-v20

There's no reason to discount the Pentium or Celeron SBs.

What does "streaming" mean?
If it's just streaming a file, that's not limited by modern x86 CPUs as the throughput will be modest. Atoms would overpowered for the job. But since most Atoms don't have the VM extension, you probably wouldn't want them. Zacates would be worth considering however.
As to non-CPU bottlenecks, streaming several files simultaneously from the same spinning drive could be a problem with crappy software for instance. You'd have the same problems with a fast CPU as with a slow.

MikeC, thanks for the detailed reply. I actually looked at that exact SPCR article before posting here. I think I misread some of the graphs and thought the Llanos drew more power. I didn't realize undervolting wasn't quite worth the trouble anymore. I think what you said is making me consider the i3 2100 a little more.

HFat, by streaming I meant the usual file server duties, but also UPnP and hopefully transcoding on the fly. I think I know what I want, but for me, it's hard to actually know until I actually have the hardware in hand to play with.

Maybe I'm wrongly assuming, but I thought HDD bottlenecks could be at least somewhat negated by using RAID; at least that was going to be my solution for simultaneous streams.

A G850 looks nice, but don't SB Pentiums and Celerons lack QuickSync? Wouldn't this hamper video transcoding? Or am I making a bigger deal out of an increase in transcode time.

A Zacate build is also intriguing, but it just seems like it would be underpowered. Then again, I have no experience with Zacate.

I might be completely off here, but my personal experience with VMs (non bare-metal on an i5-760) is that they work best when you can devote one whole core to each VM you run. Am I wrong about this? I was hoping to go with a quad core to handle the VMs, but the i3 seems to be about the best compromise for what I'm looking for. Can anyone comment on how many VMs can be run smoothly on an i3 2100? And just for kicks how many VMs can be run smoothly on a SB Pentium/Celeron?

Thanks again for the help.

Could, yes: but there's no guarantee. Good software doesn't need RAID to begin with. Bad RAID software will not help. As long as you're careful not to have any junkware (like pseudo-security scans) running, even with not-so-good software a single drive should be able serve a few streams anyway. But really crappy software can ruin anyone's day.
Unless you're lucky enough to be able to rely on quality performance information (which is very scarce), you'll have got to run tests with the software you want to use to know for sure what hardware you're going to need. Most people don't bother and just throw ridiculously overpowered hardware at every problem...


QuickSync is not a panacea. But if you're sure you're going to be able to rely on it in practice, it could well be worth the cost of an i3. You'll save electricity and get lower temperatures and/or noise from it.
Myself, I'd avoid transcoding in real-time by loading my server with files in the right format.


It's probably underpowered for real-time high-res transcoding. But for file serving and other basic server tasks it's overpowered.


In principle, you can run as many VMs as you want as long as they're not crunching numbers. In practice, the number of VMs is typically limited by RAM. You should be able to run as many as you could realistically want as long as you don't have bad software and/or configuration which stresses your hardware doing nothing. Here again, bad software can ruin anyone's day...
The branding of the CPU (i3, Pentium or Celeron) has no bearing on how many VMs you can run. I'm not sure you could use QuickSync or even basic hardware-assisted decoding or from a VM by the way.

As HFAT said, maybe. RAID will help, but probably not as much as you think.

A simple example, a two drive RAID0 can do a sequential transfer at about double that of a single drive. However, it cannot transfer two or more sequential reads simultaneously at double that of a single drive.

Access times are not reduced significantly by RAID, so the more streams you run simultaneously the more you are going to be seek limited and not transfer rate limited. Intelligent RAID control/software can mitigate this somewhat, but there is no magic.


A big downside of QuickSync and any video card based transcoding is quality. That speed doesn't come for free. There is a reason x264 (generally considered the best quality H264 encoder) doesn't use anything except pure software encoding. You need to decide if the speed/quality tradeoff is worth it to you.


I've only used a Zacate as a desktop CPU, never in a server. Subjectively it's faster than an Atom, but still no where near even a Celeron SB. If transcoding is in your requirements then I wouldn't use it.


As Hfat said, you'll likely run out of RAM long before you run out of CPU unless you are doing something processor intensive on your VMs. Transcoding might count, but you could give it lower priority and it probably wouldn't interfere much.

At work we average over 20 VMs with 2-4 cores allocated each on 8 core machines. That's up to 80 virtual cores on an 8 core box. RAM is always the issue over CPU. Right now the highest CPU load on any VM host is 16%, so just over 1 core fully used.

If your VM software and your hardware supports VT-D (intel) or IOMMU (AMD) then you may be able to do this. VT-D lets you assign a hardware device to a VM directly. I've done it with storage controllers and NICs and I've read about people doing it with video cards. Theoretically you should be able to assign the "video card" part of the Intel CPU to a VM.

VT-D however needs both CPU and motherboard that supports it. Some consumer level boards do have it, ex some Z68 boards. Be very careful that you confirm VT-D support if you use a consumer board. Server boards and Xeons are very likely to support it, but check first.

I'm learning more than I thought I would in this thread. Thanks.

I didn't know QuickSync sacrificed quality. QuickSync isn't supported by Linux, and it doesn't look like it's coming anytime soon, so I guess it's a moot point for my build. I'd love to attempt something like ESXi on a server, but I think it's a bit over my head, out of my budget, and really unnecessary for my needs. I think VT-d/AMD-Vi is way outside of my needs.

I always thought core count played a larger role in the number of VMs you could run. I was planning on purchasing at least 8GB of RAM since it's dirt cheap now. I don't think I could keep track of 20 different VMs.

I think I confused myself a bit and somewhere along the way started thinking I need something more suited to an HTPC than a server. I prefer "snappy" performance over just fast enough, and I think that's how I got headed in the wrong direction.

Anyways, it looks like I can narrow my selection to either one of the SB Pentiums (G620/840) or a cheap Phenom II X4 (maybe an 840). There is a power consumption difference favoring the Intel at idle and load, but really at idle it should be negligible, and I can't foresee the server plodding away at full load that often (guessing here). If I pair with a Z68 or AM3+ mobo, I have some sort of upgrade path if whatever I end up with isn't powerful enough. The Llano's seem nice, but it looks like Trinity is going to abandon the FM1 socket, so possible upgrade options could be limited. Also, I'm guessing the IGP is going to be largely unused.

So, which would be the better choice?

Intel's CPUs really are technically better. I don't think the power savings would be insignificant. If you pick an efficient motherboard and PSU (note that there's an Intel board with DC input), you'll burn significantly less at idle.
Aside from special cases where AMDs are cost effective, you'd generally be wanting their chips for the GPUs or features Intel stripped from the low-end products. If you don't care for either, go with Intel.