Samsung 960 Pro Speed Freaks

Report · Jan 06, 2017

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

Report · Jan 16, 2017

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!

Report · Jan 17, 2017

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!

Report · Jan 17, 2017

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

Report · Jan 17, 2017

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

Report · Jan 17, 2017

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

Yep; they're already out of stock.

Report · Jan 20, 2017

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

Wednesday afternoon I got email from FedEx saying that something was incoming tomorrow (Thurs) that needed my signature. It was coming from "Ingram Micro" in PA, which is one of the big electronics distribution centers here in the North Eastern corner of the US. And it weighed a pound.

Hm.

I verified my order with Samsung, and they claimed it was still "Processing." So it couldn't be the drive, could it? A few hours later, they emailed me and verified: yep, drive has been shipped! And sure enough, I finally have it. Woo hoo!

Now to reinstall Windows 10. Again.

Report · Jan 20, 2017

Congrats! That is exactly how I got my first one that vwas ordered directly from Samsung.

Report · Jan 17, 2017

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!

Report · Jan 18, 2017

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.

Report · Jan 18, 2017

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.