Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

New Workstation Build - i7 X99, Titan X, Overclocking...

Would anyone using an Areca 1883 and Corsair C70 Case share what Mini SAS - Drive Cables they have used? I bought the cable below first, but the power cables won't allow the case door to be installed:

https://www.amazon.com/gp/product/B01GPD5KFK/ref=oh_aui_detailpage_o01_s00?ie=UTF8&psc=1

I'm wondering if this one is low profile enough:

https://www.amazon.com/Cable-Matter...id=1470239730&sr=1-18&keywords=mini+sas+cable

Or if there is another better solution?

I didn't actually plan to put spinning disks in this build, I have the X99 Deluxe II Board with a Thunderbolt 3 to Thunderbolt 2 that is working very well with my Pegasus drives. This has the space, so I might as well throw an array in...
 
Can you use separate data and power cables? That's what I usually do, allows you to choose 90-degree or low profile SATA/SAS power. The one you posted should work tho.
 
Thanks David. In the end I just cut the molex connectors off of the power and drive cables and soldered everything together into a more tidy wiring loom, solving the space issues.
 
Hey everyone. New here but I'm having a problem running 2x GPUs on the x99-E WS with Thunderbolt. I have been speaking to ASUS tech support for weeks and they're clueless, they often tell me to do conflicting things, like moving to slots which would be impossible to reach, and moving them back again, etc.

My setup is this:

X99-E WS
Slot 1: Wifi/Bluetooth combo card
Slot 2: ThunderboltEX II AIC
Slot 3: GTX 980 #1
Slot 4: Blocked
Slot 5: GTX 980 #2
Slot 6: Blocked
Slot 7: Decklink 4k Extreme

For troubleshooting I'm taking out everything except the two 980's and the TB card but it still won't work. I'm running a Hackintosh but it won't even work in Windows. Following Eric's advice I got this thing to BOOT but the TB driver is unloaded in Windows. The INT19 option is missing from all BIOS revisions 3+. I revertd to an OLD BIOS from last year and it still didn't work.

Also, doing "Enable Above 4G Decoding" will cause OSX not to boot at all, as far as I understand, so that's not an option. But even if I turn it on, this still doesn't work in Windows currently.

Any help would be appreciated. This board should have more than enough lanes to get this to work.

P.S. Has anyone tried ASUS' 12g raid card exclusive for their server boards? It's $200 so JUST A LITTLE BIT cheaper than the Areca one...
 
Anyone here running a hackintosh with a similar build? Or another board like the Designare with TB 3 and two GPUs? I'm curious.
 
update: I got Thunderbolt to work with 2x GPUs in WINDOWS by enabling "Above 4G Decoding" AND forcing slot 2 to 4x. It would not work on auto bandwidth. However, enabling "Above 4g Decoding" makes OSX not able to boot. Anyone know a way around this?

Apparently you can boot by turning off USB 3.0 completely. What about if I mess with the TB I/O resource settings?

This article has me thinking: http://appleinsider.com/articles/08...d_64_bits_santa_rosa_and_the_great_pc_swindle
 
Last edited:
I put together a system based on this blueprint and it's been insanely good. I'm so grateful for this community. Thank you Jeff for sharing all your insights.

I had my first hard drive failure yesterday, which does not surprise me given the amount of use I've been given this machine. What kind of failure rate are you all seeing from your comparably built systems?

My raid is formatted raid 6, and I'm using an array of seven Seagate 6TB SAS drives. I've been doing a lot of editing straight off 6K raw files, which is now my standard workflow and I love it. But I wondering if all the editing I'm doing is putting extra wear and tear on the drives? Coloring seems less demanding on the raid.
 
My original build that kicked off this thread will play 4K from RED ONE/EPIC MX no problem, doesn't even break a sweat, layered streams in Premiere, etc... When we get into Dragon and more demanding REDCODE versions, the system has to work harder. But still 4K clips from Dragon are typically handled with ease. That said, most of us don't shoot 4K all that often on Dragon, we shoot 5.5K or 6K variants. These are a lot more demanding. But the system still handles 6K from Dragon reasonbly well for editing/ coloring. But don't go thinking you're going to edit with lots of clips in your timeline and the debater on high and at full resolution. Not going to happen. For simple stuff I can usually operate at 1/4 res playback, but often find myself at 1/8th for general work so everything stays smooth in Premiere. I can play back full quality, full res 6K, scaled to a 4K monitor just fine if it's a single stream and it usually does have to buffer a bit just to keep things smooth.

The reality of this is that as we move to larger formats the requirements go up exponentially. 6K is about double the pixels as 4K and on top of that, the newer cameras are recording more data per pixel. In most current workflows, it's rarely necessary to work in full quality and full resolution scaled down to 4K. There are reasons to work in HD and UHD/4K resolutions and there are methods or solutions to make this happen.

I see you're asking in various threads about playing 4K in real time. But What is it you're trying to do?
 
Michael Blieden said:
I put together a system based on this blueprint and it's been insanely good. I'm so grateful for this community. Thank you Jeff for sharing all your insights.

I had my first hard drive failure yesterday, which does not surprise me given the amount of use I've been given this machine. What kind of failure rate are you all seeing from your comparably built systems?
Click to expand...

I haven't seen a failure on the SAS drives I've put in this build... yet. I would say I have about a 3% to 5% failure rate on HDDs annually. It's easier to see the failure percentages when running lots of drives. For users with fewer systems/RAIDs to compare, it often seems like a terrible thing when a drive dies and some can go for a few years without a single failure and not understand why others seem to have more trouble. The best thing to do is just expect an eventual failure and be prepared for that contingency. And know that sometimes having a backup isn't enough. Nothing worse than having a RAID die in the middle of a production or running up against a deadline and while it's nice you have a backup to restore from, you find yourself in a bind when it takes 12 or 24 hours to fully restore that backup.

My raid is formatted raid 6, and I'm using an array of seven Seagate 6TB SAS drives. I've been doing a lot of editing straight off 6K raw files, which is now my standard workflow and I love it. But I wondering if all the editing I'm doing is putting extra wear and tear on the drives? Coloring seems less demanding on the raid.
Click to expand...

When editing you're often pulling from more files at the same time and are working the drives harder. This may or may not actually put more wear on them, usually you can tell if the drives are working harder by the amount of noise from the heads. But like any tool, they're going to eventually wear out from use. HDDs are complex mechanical devices with fast moving parts and not only are they fast, but they have to maintain a fine degree of accuracy. It's never a question of if a drive will fail, it's only a question of when. All things considered, HDDs are pretty reliable. 5% or less in terms of failure is pretty good. SSDs have their own share of problems too and seem to have about the same rate of failure now for the better units. Cheaper consumer units tend to fail more often, especially when used in RAID environments.
 
Jeff Kilgroe said:
It's never a question of if a drive will fail, it's only a question of when.
Click to expand...

So true dude.

UPDATE: The drive that failed was running way too hot. Here is a not great photo of an earlier configuration (pictured with 4 drives instead of 7) that's similar to what I was running when the drive failed. You can see the hard drives in two stacks of 2 on the lower right corner of the enclosure. When I increased the array to 7 drives it was a stack of 4 in the front and 3 in the middle. The drive on the bottom of the middle stack was running the hottest and that's the one that failed. Sandwiched between the front stack of drives and the power center, and sitting at the bottom of the pile, this drive was getting almost no air circulation.

20150927_005710-e1475392266201.jpg



Here's the configuration I'm trying now, with all 7 drives stacked vertically in front of the fans. So far drive temps are much more manageable.

20161001_234530-e1475392138742.jpg


Side note....I currently do not have the Asus AI Suite installed because the fan control software was somehow disabling the GTX 980ti after waking from sleep. But now I'm considering reinstalling AI Suite so I can have that awesome fan control back. Anyone else have a bad or good experience with this?
 
Last edited:
A new thread on a mini blade based workstation farm?

A new thread on a mini blade based workstation farm?

Jeff, maybe we should have a thread on building a cheap blade based extendable processing farm, considering the Nvidia CUDA units and Intel XI Phi future, like I suggested for the Mac mini-Blade in the MacPro thread. To also act as a sort of sticky that canbe updated with freshly found parts and designs.

In this design (originally vertical blades added together clamped by an extendable clamp or bookends, for desktop use, rather than a rack) you keep adding new blades, and removing old ones depending on how much processing power you need. It's ultimate usefullness really depends on his much software can make use of the extra blades as an extendable resource processing blades, but also allows a number of Nvidia Cuda cards. Otherwise you can treat it as a more flexible conventional workstation, with a blade forna conventional motherboard and extra blades for more drives and periphials. I wonder if such a schmem could use a conventional tower in place of a primary blade, lined up together?

I have see a site with all sorts of blade form factors cheap (was that here?) Including small slot ones where you have 6-8 slotted mini CPU units in one u1-u2 rack unit.

I say this, as older workstation CPU's, and the MacPro seem to be a bit stalled.
 
You must log in or register to reply here.
Back
Top