Inspiring

Question

Samsung 960 Pro Speed Freaks

Forum|Forum|9 years ago
January 7, 2017
5 replies
6887 views

How about this from two Samsung 960 Pro M.2 in RAID 0:

But I will have to say that exporting from Premiere with the PPBM disk intensive Disk I/O timeline cannot take advantage of this sequential write speed. Our Premiere Pro benchmark stops improving at ~2000 MB/second so this configuration would not be practical usage.

I just could not resist showing these astronomical transfer rates

This topic has been closed for replies.

D

DanHidalz

Participating Frequently

1. Ok, I just thought that it was a decision that had something to do with performance. I have no problem using a PCIe drive with my actual SSD but I'll sure have it finding a laptop with two SSD and one of them PCIe. So provided the fact that laptop processors are not that powerful yet, going with just one PCIe SSD seems to be more than enough.

2. I edited it after writing it, I wanted to say "disk intensive", not "CPU intensive". I mean, if it is in any way, similar to editing, rendering and exporting ProRes 422, RED 4K or similar content. But ok, I get the overall idea of the CPU intensive test; I've been reading the project's website.

3 & 4. All ok with those points, thanks!

5. A wrote you a fifth point above. Take a look whenever you have time. Thanks a lot for your help Bill Gehrke!! You're a valuable member of this community

danielh19294473 escribió:
I take advantage and ask this little additional question to Bill and all of you apart from what I posted before.
5. I know that this kind of benchmarks are made to be as similar as real life performance as possible (being made inside PPro and all that stuff) but how much representative it actually is of real world performance?
I would't be asking this stupid question if I wouldnt have found this: https://www.pugetsystems.com/labs/articles/Adobe-Premiere-Pro-CC-2015-4-Storage-Optimizati on-854
As they analyze it seems to be little to no difference even when working with 4K. Is anything wrong with their procedures or is it just that we're benchmarking nearing a synthetic approach and its not as representative?
Thank you a lot for your help guys!

Bill GehrkeAuthor

Inspiring

I spent about an hour composing your reply and some how lost it. Here goes again

The PPBM benchmark was not designed to be as "near real life as possible" That would require constant changes as codecs are added and features change . PPBM was designed to be done in Premiere but to test your hardware, specifically your CPU your GPU and your disk drive(s). The identical benchmark works in CS6 up and through the current CC 2017. I do have different project versions and scoring scripts but the timelines are all the identical.

Puget undertook a massive project and did it very well but it is not a "portable" benchmark that you can run. They being a for profit organization could pay their employees to do that massive job and it is of great value. On the other hand I have not received any compensation for PPBM. Harm (when he was still with us) did ask for contributions to defray the cost of the software he used in presenting the data on ppbm7.com. He being a business man designed a very complex web site which I cannot easily update, He did create a document for me on how to add a new submission. It is 24 screen grabs with annotation on each so I have not been able to even keep the result submissions up to date let alone make changes to it.

On my "experimental" desktop I have one document that currently has over 400 benchmark runs recorded as I changed the configuration. If someone is not happy with their editing experience and they run PPBM, I sometimes can point out a weakness in their system. I am amazed how many times we have people complaining their GPU is not working--mainly because they are not using GPU accelerated effects or features. Running PPBM demonstrates usually 100% GPU load for them.

I have another web site that I am going to be publishing more test data from PPBM testing.

D

DanHidalz

Participating Frequently

A million thanks for taking all that time to explain all that clearly for all of us.

It's all clear now about it!

I recently took the benchmark and I was happy to see that it had been updated to work with CC 2017.

Hope you find the way, or someone who can help you, to develop and continue with the project. It's been of great help since I met it.

All the things I know today about Premiere and video editing related hardwareI owe it to Harm, your posts and the PPBM project.

Again, thanks a lot for all your time.

jasonvp

Inspiring

Frustration: I've had one of the 960 Pro 1TB drives on order since October of last year. It finally shipped from Amazon on the second of January and arrived, safe and sound. Or so I thought. When I opened the Amazon box, I was greeted by an already-cut-open Samsung box. The drive had been removed before the Samsung box was packaged up and sent to me.

Basically I paid Amazon for a $630 cardboard box. And this was Amazon's doing, too, not UPS'. Naturally, they no longer have any drives to replace it and won't until, perhaps, Spring.

Dammit!

Bill GehrkeAuthor

Inspiring

Yesterday Newegg sent me a notice that they just received a shipment of those 960 1 TB SSD's

M

Mike 1 Irish

Participating Frequently

I have one on order since the middle of december last. The 1Tb Evo is available but the 1Tb Pro isn't. I am in Europe and I have one on order from AMAZON UK and AMAZON in Germany.

Mike

D

DanHidalz

Participating Frequently

I take advantage and ask this little additional question to Bill and all of you apart from what I posted before.

5. I know that this kind of benchmarks are made to be as similar as real life performance as possible (being made inside PPro and all that stuff) but how much representative it actually is of real world performance?

I would't be asking this stupid question if I wouldnt have found this: https://www.pugetsystems.com/labs/articles/Adobe-Premiere-Pro-CC-2015-4-Storage-Optimization-854

As they analyze it seems to be little to no difference even when working with 4K. Is anything wrong with their procedures or is it just that we're benchmarking nearing a synthetic approach and its not as representative?

Thank you a lot for your help guys!

MyerPj

Community Expert

Hi Bill,

I've been following your posts, very informative, thanks!

I'll be doing a new build in Feb or March, and was planning on doing an Samsung 512gb 850 (pro or evo) boot, programs and then a 512gb 960 pro m2 drive for media. Was just wondering though, since I have to buy both new drives... can / should I go with two 960's using one for boot and the other for media, etc? The cost would only be the difference from an 850 to a similar 960.

(I'm planning on using an Asus x99 Deluxe 2 with a 6850 processor and Asus 1060 video)

Any tips would be much appreciated...

mp

Bill GehrkeAuthor

Inspiring

Well you forced me to look at the 950 Pro 512 GB versus the 960 Pro 512 GB and there is only $2.00 difference between these two on Newegg today. Of course for the boot drive you really do not need 512 GB, 256 GB is more than enough. Here there is a $50 difference for the M.2 plus for the second M.2 device you need a PCIe adapter card.

Since I have two 960 Pro's I guess it makes sense for me to go to my desktop and run my PPBM with two of those. I will not be able to do it tonight but will try for tomorrow and see if there is any difference over my running off a SATA III boot device. My gut feeling right now is there no significant difference. I will have to break that RAID configuration that I now have and reload a boot configuration on one drive.

M

Mike 1 Irish

Participating Frequently

danielh19294473 wrote:
1. How do you think is it best to set up the drives for Premiere in terms of storage with this specific RAID?
2. Just a raid and all media cache, previews, media, projects, backups and exports all together or would it be better to split it and arrange it in a different way?
3. Why do you think it is not getting benefits from over 2TB/s? Could it be a bottleneck anywhere or could it be related with the benchmark itself?
4. Apart from that, can I ask you how much that system scored in ppbm?

I will not use this RAID as it shows me no advantage over a single device. My editing configuration (as opposed to my experimenting configurations) probably will be a single SATA III 6Gbit/s SSD and projects on the 1 TB m.2 Samsung 1TB SSD.
No to RAID. Unless somehow I get into more complex media (I do have a Sony FDR-AX100 that produces 4K 100 Mbits/second XAVC-S video that works perfectly) I will not consider anything else. I have the default location for the "Media Cache" and "Media Cache Files" on the boot SSD and all else on the M.2 drive.
It is called the "Law of Diminishing Returns" The bottleneck is now fast the CPU can hand the data to the storage device and may also somewhat Premiere limited. We designed the disk intensive benchmark to test the hardware from within Premiere by using the must native to Premiere the AVI format.
Here is a work in progress that will show you more PPBM result than you want

[Link Fixed by Mod]

Hi Bill. The link seem to be broken!

mikeklar

Inspiring

Interesting, thanks for posting.

Cheers

Michael

Sign up

To post, reply, or follow discussions, please sign in with your Adobe ID.

Sign in to Adobe Community

To post, reply, or follow discussions, please sign in with your Adobe ID.

Scanning file for viruses.

This file cannot be downloaded