Sony Vegas Pro 10 GPU ..What, where, when….No Way!

[Clarence Reed]

I have yet to double check the Sony Vegas Pro 10 supported video cards, butt-butt I will go out on a limb cuz I pretty damn sure that their is not a single GTS card on that list last time I checked. IIRC the list was GTX, then FX And Then Quadro. ….. Ive seen GTX 400’s tear through 20min renders of AVC in 10mins. I plan to test a GTX 460 & 480 and a 590 AND THEN post the results to Youtube. I got to say Iam pretty disjointed with the lack of render time stat comparisons on the web right now. You can find a list of the supported cards on the Sony Vegas 10 site & on wiki.

[Clarence Reed]

I see that this is a rather old post. And I notice that Sony released some updates that help support CUDA cards with AVC rendering. Me and a few friends have tried various GTX 400 cards out and they have similar if not same AMD CPU. We cut through 20mins worth of AVC video rendering it with more then a few effects in 10mins if not less. On a overclocked GTX 590 we got renders in the 4-5min range. I think a Quadro 5000 could do better mebbe, for the price though I doubt it. Now the Quadro 5000 will handle 3D stuff where the GTX cards will not.

[Earle Del rosario]

Would be nice if VegasPro was enabled with the latest Quadro card, to be just as good as PremierePro is. And, if the EDL was similar to Avid’s. With retail price maintained.

[Be Sc]

Hello everybody,

My friend is using Raid 5 on the ICH10R – thats the mother board chipset. He’s averaging 5MB/s on three hard drives…and noticable CPU draw…He just uses his computer for his small business database, no mega speed needed…

5MB/s isn’t right: sounds like one of the drives is faulty. Even if the write-back cache is not enabled, the speeds with today’s drives should be in the range of 50-120MB/s writes, 100-250MB/s reads.

That’s NOT correct!

Lets assume these disks are standard SATA Disks with about 10 – 15 ms access time (say 12.5 mean).

Performance calculation of raid5 with enabled wb cache is:
Reads: Number of Disks * IOPS per Disk
Writes: (Number of Disks * IOPS per Disk) / write penalty (which is 1 for raid lvls 0,1,10 / 5 for raid 5 / 6 for raid 6)

In Numbers
Reads: 3 * (1 sec / 12.5ms) = 240 IOPS
Writes: 3 * (1 sec / 12.5ms) / 5 = 48 IOPS

As you can see, this is less performance than with a standalone disk which can do 1sec / 12.5ms = 80 IOPS.

If yiu want to setup up your raid for maximum data protection and you either have no bbu or no usv, you have to disable cache writeback for the raidcontroller AND to disable disks cache. Just switching off the controller’s cache but leaving the disks cache on may result in data loss in case of power loss.
Not using the disk’s cache results in MAJOR performance decrease. 5MB is quite normal at this point. using even slower disks (ie 5200 rpm) results in even less performance.

One more word concerning different partitions:
Sure, different partitions may be useful. But be aware: If you use more than one partition on ONE raid volume, the performance will be shared between the volumes and you MAY decrease your performance by forcing the disks read heads to jump from onw physical section to another (causing access times, lowering IOPS performance).

The performance in MB/s can be calculated by Num IOPS * block size.
Just breaking down the ratio:
A single SATA standard drive does about 100 MB (SEQUENTIAL) read OR write (not at the same time). Sequential says: There are almost no access times because the head is already positioned correctly.
Sequential Read Rate should be around 300 MB/s. Write rate would be around 60 MB/s. With both, disk cache and controller cache empty, but enabled.

So, if you have a 50:50 read-write workload, you should choose raid 10. If you’ve got a lot of reads and just some writes you should choose raid 5.

Best regards

Benny