Which setup will perform the best?

[kamikazekyle]

So, thanks to a Newegg combo deal, I have a new budget-minded rig coming in. I'm going to be traveling between and working at two locations, so I want to have the new rig at one, and my current rig at the other. As I spend most of time at my house, I'm going to want the more powerful of the two setups there. The problem is, from my research, I can't really tell which is going to perform better.

Purposes: video encoding, virtualization, gaming

Assume all other parts are equal (ie: no SATA 6 Gb stuff on the MSI mobo).

Setup One:
Athlon II X4 640 @ 3GHz (No Overclock, but maybe if I get an aftermarket cooler later)
MSI 870A-G54
4 GB DDR3 - 1600

Setup Two:
Core2Quad QX6700 @ 3.16 GHz (overclocked)
EVGA 122-CK-NF68-A1
4 GB DDR2 - 800

As far as I can tell, at stock speeds, the Athlon outperforms the C2Q by a small margin in Passmark, though that might/might not take into account other differences. Plus, my C2Q is overclocked by a good bit.

Most of the reviews compare the Athlon to newer i-series processors rather than the older Core lineup. However, despite the fact that the C2Q Q8600 outperforms both the above setups in Passmark, the Athlon outperforms the Q8600 in video encoding and rendering, and is about wash on game benches.

I also have a GTX 460 and a GTX 285 (both 1GB) to decide where to put, but from my research the 285 outperforms the 460 by a good margin, so that'll probably accompany the most powerful system.

So, anyone have any input? I'm kinda leaning towards the Athlon due to the better x264 benches, but finding comparisons to the older Core2Quad lineup seems to be a pain.

 

[mfenn]

And this is why friends don't let friends use Passmark for decision making. Check out some real world benches instead. That link doesn't describe your exact situation, but it should be pretty close. The QX6700 is a Kentsfield instead of the Yorkfield shown in that link, but there really isn't much of a microarchitectural difference between the two.

TL;DR The QX6700 @ 3.16 is quite a bit faster.

 

[kamikazekyle]

Wouldn't this (http://www.anandtech.com/bench/Product/188?vs=53) be a better comparison? The Q6600 is the same Kentsfield line, which when compared to the Yorkfield has less L2 cache. Ballparking a bit for the overclock and speed differences, it looks like they're probably going to be about equal for my intended purposes, barring an overclock on the Athlon.

But, thanks to some weirdness with UPS, I should get the parts in a day or two early and before the case (even though the case shipped out beforehand...), so that'll give me time to run some benches myself. Still a little off since I'll be pitting a fresh build against a several month old load, but what better way than to test yourself Besides, I wanna find out if I'll like the new case better than my current one.

Thanks for the help as usual mfenn. Much oblidged.

 

[mfenn]

Wouldn't this (http://www.anandtech.com/bench/Product/188?vs=53) be a better comparison? The Q6600 is the same Kentsfield line, which when compared to the Yorkfield has less L2 cache. Ballparking a bit for the overclock and speed differences, it looks like they're probably going to be about equal for my intended purposes, barring an overclock on the Athlon.

Your Kentsfield is clocked 33% higher, so no that is not a very good comparison at all. If you scale each of the Q6600's scores by 33%, you will see that it comes very close to matching the scores of the Q9650 that I linked (i.e. soundly beating the Athlon).

But, thanks to some weirdness with UPS, I should get the parts in a day or two early and before the case (even though the case shipped out beforehand...), so that'll give me time to run some benches myself. Still a little off since I'll be pitting a fresh build against a several month old load, but what better way than to test yourself Besides, I wanna find out if I'll like the new case better than my current one.

Of course, benchmarking your exact setup is the best way to proceed.

Thanks for the help as usual mfenn. Much oblidged.

You're welcome.

 

[kamikazekyle]

I got the parts in, and stuffed the AMD + GTX 460 rig into a beater case I had for testing. I benched the AMD rig with a 1TB Spinpoint, and my main Intel rig with the 285 GTX using my SSD and RAID 0. Basically, for the AMD it was a simple build'n'boot.

Futuremark was neck and neck with the AMD setup edging out the Intel by 100 points (ie ~7&#37.

I ran a few iterations of Y-Cruncher which isolates the tests to the memory and processor subsystems only, and with 100,000 decimal places the AMD rig outperformed the Intel setup by ~10 seconds (give or take a range here, I can't remember the original results).

At this point, I tossed in a few real world examples. I ran some encodes with x264 using High Profile on a 24 minute SD clip that had already been compressed once using h264. The average FPS encodes were right next to one another: ~70 FPS for the Intel and ~71 FPS for the AMD. One key difference seemed to be in the muxing phase as my RAID setup handled the I/O a bit better than the single drive.

Next, I loaded up a game on both boxes. I didn't have any games with a benchmark config set up, so I pretty much just loaded a demanding game that wasn't too big in file size (to minimize copy time...in this case Risen). The Intel/285 setup got 30-31 FPS whereas the AMD/460 setup got 37-38 FPS. Both had the same graphic settings to the same monitor.

Now, all the above were performed on a fresh OS load for the AMD, and a 3-month old OS load for the Intel. I formatted both boxes with a fresh OS load, and re-ran Y-Cruncher (I was doing some massive file transfers on the boxes so I wanted to isolate out the disk subsystem). I tossed in a 300 MHz overclock on the AMD setup before I ran this inbetween boots (oops). This time, the Intel setup performed the same, right around 79 seconds compute time. The AMD setup, however, cranked out a 51 second time. I re-ran it just to make sure, and yea, it was 51 seconds.

Finally, while my other box was busy with file copies, I re-ran the same encode on the AMD system. This time with the 300 MHz overclock and all on an SSD. Since the configuration was WAAAAY different, it was mostly for grins rather than valid testing. This time, I netted 79.2 FPS average vice 71 FPS. Decent little boost.

So, granted my benches weren't exactly professional and most were synthetic, and I didn't get a game test after the formats (troubleshooting some DNS and share permissions between the boxes, so right now all 400 GB of my games are stuck on my Intel setup). But, from the looks of things in the areas that I primarily concern myself with, the AMD/460 setup bests my Core2Quad QX6700 @ 3.16 GHz with the GTX 285 handily enough. While it's a pretty large margin in some synthetics, real world tests were not all that *huge* of an improvement. Since the AMD's 300 MHz overclock was just on a whim and gave marked improvement, if I slap on a 3rd party cooler I could probably get a 600-800 MHz overclock and an even greater performance differential.

Everything said and done, the AMD system is noticeably faster, quieter, consumes less electricity, and is cooler. I think I'll use that as my main rig, and my old Intel setup at the remote location. Plus, it doesn't have that very high pitched whine that my Intel setup has when it's doing processor intensive tasks. Gods, that's annoying.