Alex Gerulaitis
Forum Replies Created
-
[Eric Hansen] “how would you propose building a DPX cache for the nMP? “
Here is one option:
LaCie Little Big Disk Thunderbolt 2, allegedly 1.4GB/s, consisting of dual 500GB PCIe SSDs in RAID0 in a TB2 box.
-
Makes sense – thank you.
-
[Chris Murphy] “plus the hot spare means you’re at best 5x read speeds”
You sure it’s not 7x Chris?
(I understand real-world numbers support your thesis, yet theoretically 6 has the same read performance as 0. Perhaps it’s the implementation that brings the performance down, not RAID level.)
-
[Eckhardt Stassens] “Should we have captured the footage directly from the original VHS, S-VHS, etc. rather than from the copy on DVCAM?”
Yes although in most cases the difference is negligible: DVCAM is much higher quality vs. VHS and the quality loss is insignificant if any. With S-VHS it may be more pronounced so you’re right in doing QC tests.
— Alex Gerulaitis | Systems Engineer | DV411 – Los Angeles, CA
-
Not necessarily. Having a cold spare on a shelf is more efficient if you don’t run your array 24/7 and assuming you’ll know immediately that a drive failed.
Hot spare means that (a) it’s just sitting there doing nothing until a drive fails, (b) your capacity, efficiency are down by 1/8th, something to consider given that RAID6 already takes away 2/8th of the capacity, (c) auto-rebuild that may bring performance down quite a bit, possibly for more than a day.
I.e. hot spares make more sense on larger arrays with RAID6.
My personal preference in the event a drive is marked as failed on smaller arrays (less than 16 spindles) is to check that the drive indeed failed rather than just “marked as failed” because of a timeout. Quite often the drive is actually healthy, just hiccupped, and can be marked as “online” w/o side effects. (A rebuild may still be necessary.)
I also prefer to start and monitor a rebuild manually on smaller arrays with non-24/7 duty. Auto-rebuild in the middle of a project may not be a good idea.
If it’s uptime and data protection you’re looking for: RAID6 with a cold spare or two sitting on a shelf, and backups.
— Alex Gerulaitis | Systems Engineer | DV411 – Los Angeles, CA
-
[Michael Mohr] “A video layer and 2 PS layers. I get a lot of yellow and very little green. AE is really hindered by it.”
AE? The only thing that’s GPU accelerated in AE is ray tracing. So a graphics card upgrade won’t do much unless it’s ray tracing that’s hindered.
[Michael Mohr] “I still haven’t heard anything regarding AMD Radeon HDs, only GTX cards. Are they that inferior?”
They aren’t, it’s just GPU acceleration in Premiere Pro is more optimized for the Cuda tech that’s only available with NVidia cards. AE ray tracing – NVidia only.
-
Thanks Neil,
I still think isolating network performance from storage and measuring is separately might help zero in on the bottleneck.
Let me know if you’d like me to hop over and do some iPerf tests on your setup.
-
Neil,
Thanks so much for doing the tests – very interesting.
[Neil Smith] “Any explanations on reason for erratic double Thunderbolt Bridge transfer speeds?”
Don’t know but can only guess the bridging / device driver are not optimized for such high speeds. I’ve seen similar behavior with 10GbE links (speeds way under 10Gbs ceiling) and that was mostly attributable to NIC drivers and OS TCP/IP stack not optimized for sequential transfers.
Any chance of getting a little more granular speed measurements – perhaps with iPerf/jPerf?
Would be interesting to get bandwidth-over-time graphs similar to ones in this image, along with testing various packet sizes, and measure network performance separately from storage.
There’s a good chance this will pinpoint the bottleneck.
P.S. Clicking on the images in your post results in “404 Not Found” – wanted to see the top two in full glory and couldn’t. Any chance of fixing it… in post?
-
[Michael Mohr] “a Quadro 600 which works like something that rhymes with hit.”
Can you elaborate? 🙂
Try to enable Cuda acceleration on the 600 first. It’s got 1GB RAM which should be enough, other than that the specs are rather underwhelming and it may not pull its weight in Premiere – but at least it’s free – you already have it.
What machine is it? The GTX-770 is a good suggestion – yet may be an overkill where a $250 GTX-760 will be just fine, and you’d need to be sure the power supply can handle either.
— Alex Gerulaitis | Systems Engineer | DV411 – Los Angeles, CA
-
How large and how fast does it need to be? Any chance of using the internal SSD?
Otherwise, I’d think so, striping 2-3 boxes across TB channels should do it – something someone will eventually test – until then we won’t know for sure… 🙂
nMP specifically designed for 4K is just the usual Apple BS… excuse me, marketing. Sure it supports 4K displays (yay!) – does it support the full VFX pipeline in terms of performance and bandwidth? Let’s wait and see on that.
An ATTO 12G SAS controller was benching 5.5GB/s with 16 SATA SSDs at the last NAB on a Z820 – so in the rectangular world, it seems to be easily achievable. 🙂
(That said plain vanilla 4K (uncompressed RGB 2160p24 12-bit) is still manageable at around 500MB/s although once you get to multiple streams, higher fps or higher res (8K, 14-bit, etc.) – there’s no saying.)