brock2621 & squidman,
There's stuff for both of you here... so pick and choose.
And squidman, you think that YOU are longwinded... About Cases
I had trouble finding an EEB case myself. I didn't understand the differences between EEB and the same-sized EATX form factors.
So I looked at the pricey EEB Cooler Master Cosmos II. It was HUGE enough for an easy build; but it was too big to fit under my desk. And there is a lot of "aerodynamic laundry hanging out to dry" on the outside (making it look kinda like a piece of exercise equipment). I could have gotten used to that but going to a more mid-range case I knew I could save enough money to buy another SSD. I almost bought the EEB compatible NZXT Switch 810; but it was not quite as sturdy as the EATX compatible Cooler Master HalfX. I settled on the HalfX and I'm very happy with it. It's got great airflow potential, it's well built, has two 3.5 hotswap bays for BU rotations off-site, plus it's roomy enough for my RAID array (five 3.5 drives), four 2.5 drives, and three optical bay devices. Again, the looks are more gimmicky than I would have liked; but it's a workhorse.
But with this choice of case came the EATX issue. I found that 2 out of the 10 pre-drilled standoff holes didn't line up with the board holes. One option I've seen people suggest is to leave them off. I wouldn't recommend that with this board, as both of them are found under vast spans in the middle where there's very little surrounding support from other standoffs. One is actually in the proximity of the RAM slots so the mobo would definitely flex a lot if you try to replace RAM after the board is installed. You could drill and tap new holes in the case; but that comes with it's obvious difficulties. My approach was a compromise. I filed the threads off of the brass standoffs which came with the case. Since I had a vice to hold them securely during the process, it was fairly easy. The boards little mounting screws still attach them to the board; but they only rest on the case surface. But support is very decent.Gaming Pedigree SSI-EEB cases
Cooler Master Cosmos II (~$330 US)
NZXT Switch 810 (~150 US)
Silverstone Raven RV01 (~$240 US)
Silverstone Raven RV03 (~$150 US)
Silverstone Raven RV04
Chenbro CA-10569BK (~$90 US)Gaming Pedigree EATX cases
Mountain Mods custom (~$250-$600 US)
Corsair Obsidian 900D (~$350 US, very big & simple; but latest model with tons of drive space & flexibility)
Corsair Obsidian 800D (~$280 US)
Corsair Obsidian 650D (~$190 US, note: 3-screw, no mod install of Z9PE-D8 - http://forum.corsair.com/v2/showthread.php?p=578030
Cooler Master HALFX (~$190 US)
Rosewill Blackhawk-Ultra (~$180 US)Server Pedigree SSI-EEB cases
(most with PFC PSUs):
iStarUSA D-490S rack-mount series (~$240 US)
Intel SC5600LX (~$900 US)
Intel SC5650WS (~$400 US)
Intel SC5600BRPNA (~$730 US)
Intel SC5400LXINA (~$680 US)
Intel Server Chassis P4308XXMFEN (~$400 US)Media-Center/HTPC Pedigree SSI-EEB cases
Silverstone Series SST-GD07B (~$140 US)
My experience with Compositng software on this mobo.
I haven't a lot of experience with Nuke yet; but I've done some work with the PE version and I've done the 2-week temporary tryout of NukeX. In NukeX I did some 3D tracking on 480 frames worth of 1920 x 1280 HD footage. I haven't detected any performance issues. It's been stable, seems snappy enough, and rendering is quick. But I don't have much to judge it against.
I have had really good results with AE (CS5 & CS6). When building my machine, I paid attention to the Adobe CS5 performance white papers. So I got the largest number of cores I could afford (E5-2630s) and chose 64GB of RAM. I also came up with a multiple-drive strategy as well: SSD boot drive, SSD application cache drive, and a RAID 10 HDD array as storage/rendering target drive... (plus BU HDDs). With the 12 physical cores and 64GB of RAM, I am able to go into preferences and assign to AE a total of 19 virtual cores and the maximum allowable RAM for each core (3GB) for fastest rendering performance. That still reserves 5 out of my 24 virtual cores plus 7GB of RAM for the OS and other non Adobe apps. Adobe says that there's a possibility of over assigning cores; but I haven't seen proof of that yet. And they don't explain what sort of composite will exhibit this behavior. Recently I did a series of tests in CS6 with a fairly complicated composite. But one experiment like this couln't possibly prove or disprove their point. Anyway, it did serve to show me the benefit of more cores. It contained several varieties media: two different codecs of video, sequences of large PNG files with alpha, multi-layered Illustrator and Photoshop files, effects like Blur, several effects from the Trapcode Suite, text with animated effects, and more. I started with 4 cores & the maximum RAM per core (3GB). I worked my way up to my max of 19 virtual cores with 3GB/core. Performance improved in linear fashion with the addition of each core. What took 12 minutes with that max setting, took over an hour with "Render Multiple Frames Simultaneously" turned off. And when I set it to render with the new Ray-Traced 3D Renderer (using my GTX670 GPU), it took 2 1/2 hours! But that's because I didn't have any 3D in the composite. I have yet to test the pure GPU Ray Traced 3D Render with that type of composite. But I really love how Adobe and Octane are focusing on exploiting the procesing power of Nvidia CUDA technology. That helps digital artists on a budget because they can use gaming cheaper/faster gaming cards. Neither uses SLI; but that doesn't matter. If you have multiple cards they will use all of the CUDA cores that you have installed. But AE does limit the total RAM used for GPU processing based on the card that has the least amount. So if you install a 4GB card and a 2GB card, it will set the maximum for each to 2GB, for a total of 4GB.My experience with 3D modeling software on this mobo
I am not that well versed at C4D yet; but I have played with it and it is quick and stable. Autodesk Maya seems stable on my machine as well. I do have issues occasionally with 3DsMax tho. But who doesn't? 3DsMax is a buggy patchwork of code. On my machine I have seen strange geometries fly out from my quad's verticies on screen during rotation of complicated geometries. Also it crashes when I try to do Camera Per Pixel Mapping
. But I have seen this behavior with this program on several other different HP workstations (didn't check the GPUs). The screen issues make me wonder if the Quadro/Tesla/Maximus approach that Autodesk recommends is more than just words. Like many peeps these days I am using a gaming GPU because price/performance is superior to workstation GPUs... And Nvidia GeForce particularly for Adobe applications. My card is an EVGA GeForce GTX670 Superclocked w/ 4GB. The NVIDIA gaming cards don't get the benefit of worstation level drivers tho, so the precision is reduced. Plus there are other processing technologies within the Quadros that are vital to 3DsMax that are disabled in the GeForce cards. I know that there are some ways of hacking various GTX models to turning them into faster versions of their Quadro counterparts; but where one's livelihood is at issue, the stability & reliability found in stock components is more important.About the SRX as an alternative mobo
The EVGA SRX wasn't available when I got the Z9PE-D8 over a year ago; but even so, it looked pretty cool and I was drooling over it. Since then I haven't seen too many reviews about the SRX. I just happened upon one the other day by Tom Logan of Overcloc3D, so I watched it eagerly. He is as usual long winded, and he didn't have much good to say about it. He whines a lot about color mismatches (which is probably silly to anyone building a WS); but he's more into the OC/gaming culture where that is part of the game. He does make some good points tho about the SRX's practical/technological inadequacies. He thinks the ASUS board is much superior.
EVGA SRX mobo review:
ASUS z9PE-D8 mobo review:
GPU/Display For Octane
I don't believe it would be money well spent getting an additional GPU just to drive the display, if that's what you are saying. The GeForce cards would support your monitor(s) using a bit of RAM; but as with Adobe Rendering apps the more CUDA cores you can throw at a render the faster it will go. So if you double up, go with two identical Kepler Nvidia cards. It's easy to go overboard with complex variations and it easily causes unseen issues. If two of the same don't work, send them back and try two others.
Here's a thought... If the board isn't too touchy with 3GB GTX 660s or a Titan (I don't know the latest on compatibility and the BIOS situation), two 660s would supply the same number of CUDA cores & RAM to Octane, AE & Premiere as a Titan GPU costing double that. That's 2688 CUDA cores and 6GB of RAM. There may be single cards or dual setups that will hit your value/performance sweet spot better.Smart Caching
I haven’t taken advantage of the smart caching feature of this board. I simply used an SSD as a boot drive. But i had issues that caching might have circumvented. Because my boot drive is only 120GB, I had to manually optimize my setup to make efficient use of the limited space available. First, I had to disable sleep/hibernation feature because Windows 7 reserves C: drive space for state storage. Whatever your RAM Capacity is, that's how much space Windows reserves on the C: drive. Because I have 64GB RAM, over half my drive would be taken up for hibernation purposes. There would not be enough room for all of my applications. I also needed to move all of my personal folders from the C: drive to my storage drive (which I'd probably do anyway). I disabled search indexing., disabled the pagefile on my c: drive, and enabled write caching. Using board driven caching would conceivably give you similar boot performance; but without the setup backflips I just described.Water Cooling
These Xeons don't get very hot at all. My wimpy Dynatron R17s in my very well-ventilated case, plus dry 78F/25C ambient temps keeps CPU1 averaging around 38C and CPU2 at around 42C. I'm sure that the back one is hotter because it sits in the warmer air stream of the first. Using the H70 water cooling like you suggested would conceivably even out their temps a little bit and also drop temps down a few more notches; but it isn't necessary. And a custom rad that they both shared would definitely even them out. I didn't care for the stiff hoses on that cooler and didn't like the idea of potential leaks.Dual Xeons vs OC'd i7
Not much of a contest if your board works and you can afford it. The Xeons with their superior number of cores will be all around workhorses that will outperform the single CPU in the widest range of 3D and rendering situations. It is possible that a fast, single multi-core CPU like an OC'd i7 could edge out a pair of slower E5-2620 Xeons in rendering if you're using Octane and a couple of fast CUDA GPUs. It's hard to say tho without testing. And I'm sure that in general Windows performance, the OC'd single CPU would feel snappier with the fast i7 than with a pair of Xeons... no matter what they were. But you'd need to weigh those 2 situations in your workflow against what the better general grunt of more cores would buy you.XFX ProSeries PSU
I don't know how good this PSU is but at first glance it does look decent. It employs active PFC which is good. Since it does, you will need to get a PFC compatible UPS if you are choosing to add one to your collection. And I'd recommend that you do if you live in storm country or have flaky power. "Active PFC compatible" UPS units are typically of the "Pure Sinewave" type and unfortunately more expensive. But without a compatible one, and in the event of a power failure, your PSU will very likely shutoff during the short transition to battery backup power. So what's the use in going to the trouble and expense of setting up a UPS if that happens? I have the CyberPower 1500PFCLED. It has saved my butt several times while I was working to stringent deadlines.
This is all my opinion of course but it's backed by blood, sweat, and tears. Your mileage may vary..
BTEdited by BlackenedTush - 7/15/13 at 7:16am