I’m actually scared by the fact that even 24GB of RAM isn’t enough for a high-end workstation these days. And I’m talking about projects up to a maximum of full HD resolution! I don’t even consider 2K or, hell, 4K projects at the moment as I primarily work on TV commercials etc…

This is very interesting as I remember every time I upgraded the amount of RAM of my workstation, I was very pleased with the performance and the amounts of available memory every time I worked on something. But then, not long after, I started to reach the limits of the system’s resources.

The thing is, when you reach the cap of your CPU, or GPU for that matter, your renders just take a bit longer to finish. But as soon as you reach the RAM limit, your apps start crashing and the entire system becomes unstable! So, it is a bit of a problem.

I started reaching the limits not long ago when I had tons of apps open at the same time. Especially Nuke, Max, VMWare Workstation and Photoshop. But on recent projects I started reaching the cap with Nuke and Photoshop alone! It’s terrifying as it suggests that no matter how much RAM I put in the system, it will never be enough!

So, it seems that my next workstation upgrade will, indeed, be a RAM boost. Not sure I’ll be able to buy the full 96GB of ECC RAM at once, but, I’m certain I’ll get there very soon.

I don’t know what caused the problem, but when I returned to the studio, all my machines were restarted, running, but restarted. The Internet Router I used for external access was also down. I had to manually start up some services and buy a new router. Then, when Windows Update popped up on the file server and I hit “Install Updates”, since I figured when it was all down anyways, I may as well update the system. That was the biggest mistake! The server crashed into a BSOD and I had to restart it manually which resulted in the failed RAID array and the complete loss of the OS.

The other issue was with the MD1000 DAS with all the company and projects data on. Since I configured it as a software RAID I lost the complete array with the OS. The data is still on the drives, but I can’t access them, since the OS is down.

So, right now, the entire studio is offline and I’m working on putting it all together before we start rendering the current projects. However, I’ll also make the best out of this disaster and config the entire IT infrastructure a bit differently. I’ll use my license server (a SuperMicro SuperServer I bought about half a year ago) and make it a primary file server as well as the domain controller. I just need to upgrade its RAM and possibly add another CPU. Once all done, I’ll need to install everything from scratch, config the domain, all the software etc… Then the studio will not only be a bit faster, but also a bit leaner as I’ll be able to lose the old PowerEdge 2600 server that has been the core of the studio’s pipeline.

So, the bad news is, we’re currently offline. The good news is, I’ll be able to build the IT from ground up on a more powerful and less memory hungry hardware. I’ll also upgrade the PowerVault to sport all 15 500GB drives (currently there are only 12 500GB drives) and I’ll make everything more streamlined and a bit faster than before.

Hopefully we’ll be up and ready within a week (waiting for the HW parts for the SuperServer).

I must say, I am very pleased with the new Sandybridge i7-2600 in my render farm. It’s super powerful, it consumes less power and is a lot cooler in the tiny 1U rack configuration!

It’s been under a 100% load (3ds Max rendering) for over two hours and the CPU temperature hasn’t even reached 70°C! Now that is impressive, considering the i7-860s I have in the rack are a bit slower, consume more power, but mainly, in their tiny 1U configs, easily reach 90°C-95°C under full load after a few minutes.

This has been a huge improvement and I can confirm all the rumors and speculations on the forums that Sandybridge really is a great CPU for your general processing tasks. As if it was specifically designed for low-cost render farms.

When I read the news about the new generation of CPUs from Intel, Sandybridge, I was trhilled! And when these bad boys were finally available at the local hardware stores, I immediately bought one.

I gave my new Core i7 2600 a quick spin and after initial OS install immediately tested its performance using Frybench.

I didn’t use an actual 1U case as I’m still waiting for it to arrive (it’s getting more and more difficult to get these Chieftec 1U rack cases these days), but I tested it quickly laying on my table.

What I found is that the machine is a bit faster than the previous i7 models and produces a bit less heat while being a bit less expensive! It’s a WIN-WIN-WIN to me.

Anways, here’s the results to compare to my other machines:

Core i7-860, 8GB of RAM, 74GB 10K rpm VelociRaptor HDD, 1U Dynatron cooler, 300W powersupply, Windows 7 Professional x64:
Frybench rendertime: 00h:07m:12s
Maximum temperature reached: ~92°C

Core i7-2600 (Sandybridge), 8GB of RAM, 74GB 10K rpm VelociRaptor HDD, 1U Dynatron cooler, 300W powersupply, Windows 7 Professional x64:
Frybench rendertime: 00h:05m:32s
Maximum temperature reached: ~70°C

And just for comparison, my workstation:
Dual Xeon E5520, 24GB of RAM, 50GB RevoDrive SSD for the OS, 512GB Crucial C300 SSDs in RAID-1 for the work files, GeForce GTX580, 1KW powersupply, Windows 7 Professional x64:
Frybench rendertime: 00h:04m:24s
Maximum temperature reached: ~58°C

As you can see from the numbers, the new Sandybridge chips are faster, cooler and cost approximately as much as the older i7 generation. Plus you’re getting an integrated graphics chip (only supported on the H67 chipset), which is great for render nodes.

I’ll be definitely ordering more of these when I have the need for more render power. These machines absolutely make perfect sense for this kind of a job. Unlike the Phenom-IIs…

If you’ve read my last post about my first ever AMD machine I bought just to test out the fantastic power consumption and mainly heat dissipation attributes, I burnt myself, almost litterarly, with it.

I might have had just a bad luck, but the cpu, AMD Phenom II X6 1055T, is a fucking fusion reactor! Man, I was planning on putting that thing in a 1U rack case config with a special 1U Dynatron cooler, but as it turned out, it’s impossible! The CUP’s TJ Max is about 70°C, yet under no load, in BIOS, the temperature climbed to 45°C with the standard, included, cooler. As soon as I mounted the 1U Dynatron cooler on that thing and checked the BIOS readings, the CPU slowly climbed to about 65°C and kept on rising! That’s when I turned the machine of, just to not catch fire, or whatever!

I am so dissappointed in this CPU and AMD in general that I will never, ever, buy another AMD CPU in my life. You can call it irrational or spontaneous, but believe me, I’ve been building my computers since I was 13 or 12 and I’ve always stuck with Intels (even at their worse times), yet I’ve never been so dissappointed by their CPUs.

I’ll reclaim the CPU and see if it was just a bad unit. If not, I’ll sell the stove to someone with a huge-ass cooler for a few bucks and be done with it.

It feels kinda strange as I’ve toyed with PCs ever since I was about 7 years old (on grandpa’s 286) and since my 13th or 14th birthday, I’ve had dozens of PCs (also one Macintosh, which I’d rather forget about). All Intel. Not a single AMD. Don’t know why, I just didn’t feel the need for a change. Same goes for the GPUs. I’ve always been with nVidia (or 3dfx, if you remember, before nVidia bought them), not a single ATI.

So, today is a little special day. I bought my first AMD machine in my life! It’s another render node, to be specific. The reason is simple, or rather, the reasons are simple. A friend of mine bought himself a nice AMD Phenom-II rig not long ago and praised its speed as well as power consumption and mainly the heat dissipation, which is what I’m currently most interested in. Another factor was the cost. I was able to put toghether a, presumably, about 10% less powerful rig than my current i7s, but for about half as much!

So, all in all, it boils down to one thing: if this machine performs +/- the same as the i7s, but produces much less heat (as I heard) and costs half the money. I’m sold!

I’ll perform some real-world usage tests (hopefully including power consumption) in my rack setup on actual productions and if all goes as well as I hope, I’ll sell all the i7s and buy Phenom-IIs instead. Primarily because I have some serious issues with cooling the 1U machines in the tiny server room of mine.

Oh, I love this! I always feel like a child when new toys arrive!

This time, I ordered a high-end workstation, dual Xeon E5620, 24GB or RAM, 2GB GeForce etc… some UPSs, a SuperMicro SuperServer 1U blade, which I intend to use for license distribution, database administration and generally to offload some “administrative” tasks from the main file server at the studio. It’s a single Xeon, 6GB of RAM and about 200GB of HDD space.

I also decided to finally give SSDs a test drive. However, I’m not really impressed by the regular, SATA, SSD performance/price ratio so I decided to see if those PCIe drives perform any better. And they should, based on the specs. This one is one of the slowest (540MB/s, 450MB/s read, write respectively) and cheapest, but I only intend to use it as a test drive for speed comparisons. It’s only 50GB, OCZ RevoDrive SSD, so, if it doesn’t perform as I expect, I won’t regret the purchase. But if it does, I’ll be ordering a 512GB Z-Drive R2 P82, also from OCZ, as my primary work drive. That one should be almost two times as fast as the RevoDrive here, but it also comes at a steep 10 times higher price tag. So, that’s why I’m testing on the cheapest (which still cost about $350 USD) drive. If I buy the larger one, I’ll use this one for the OS instead. Hopefully the 50GB will be enough for the system to comfortably run on…

Oh, I also invested in some more software licenses as you can see.

Well, I’ll report on the system’s performance and setup when I have the time. Gotta finish what’s on the table right now. Stay tuned!

I had to deal with two, moderately serious, server issues at the studio.

The first one was rather strange as I was adding another render node to the loop and I wanted to check whether everything is fine with the machine, so I physically logged in to the server as an Administrator and checked the Remote Desktop connection to the newly added render node. All was fine so I logged off and that’s when the entire operating system got stuck. I only saw the “Logging Off” message for hours without change. I couldn’t log back in, I couldn’t log in remotely, I couldn’t restart the machine. So I had to hard-reset the server, which I wasn’t really happy about as I was afraid the RAID drives might have failed or anything else could have gone wrong. Luckily, nothing serious happened, I only had to manually jump-start some of the services I need to run for my studio infrastructure, but that was it.

Now the second one was a joke. Since the Windows Update offered me about gazillion updates I decided to update the server since I didn’t have any jobs in the queue at the time. However, as soon as the updates reached about 98% the server stopped responding. My entire local network was inaccessible from the outside and I was, basically, out of luck (remotly). So the next day I went to check on the status of the server and all I saw was a very unpleasant BSOD. Not even remotely as nice or appealing as the one you can see above this post.

So, generally speaking, I have no idea what went wrong and how to prevent it, aside from avoiding to locally log-in to the OS or performing a Windows Update.

Sure enough, the studio is slowly growing, both in its technical background as well as with the projects that are coming in. So, we need more horse-power!

Again, same setup, same procedure. Today I’ve installed and prepped the slave and tomorrow morning I’ll place it in the rack and setup additional programs and stuff so it’s ready for the heavy-load.

One nice change, though, since the components got a bit cheaper, I installed a bit faster RAMs and twice as big a harddrive, which is nice, although, not what I’d normally do, since I’m basically deviating from my regular HW setup and thus opening the rendering environment to possible flaws. However, only time will tell if there are any issues with the machines. So far, they’ve been performing exceptionally well!

That’s right! Althought I missed the 100 days anniversary, I am still very happy to see that my server has gone through over 115 days of labor without a single hiccup, error or restart.

So, hopefully, at the very least another 115 days will come by without any issues what so ever.