Pages: [1]   Go Down

Author Topic: SSD - M2/PCIE vs SATA  (Read 4519 times)

armand

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5540
    • Photos
SSD - M2/PCIE vs SATA
« on: February 19, 2018, 08:41:39 pm »

Recently my computer was extremely slow so I started to contemplate how much more can I upgrade before I move on to a new one, which I would rather avoid as I'm not really in the mood for it. It turned out the problem was the Asus motherboard software and I suspect it was also the culprit for some random restarts I had some time ago.

The question remains though. The easiest is the video card but unless I move on to software to use more the video card the current one, a Geforce 750 Ti, is serviceable.

Next would be the hard drive. I'm quite positive my motherboard with the latest firmware supports booting from a PCIE and I can easily find adapters for M2 SSD to PCI. My current one is a Samsung EVO on SATA.

Did anybody made the switch just from SATA SSD to NVMIE SSD and is the speed difference that significant in real world use? In theory between 400-500 MB/sec to 2-3K MB/sec is a large difference but I get a feeling in real world use it's much less.

degrub

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1947
Re: SSD - M2/PCIE vs SATA
« Reply #1 on: February 19, 2018, 09:15:06 pm »

Which version of windows are you running ?
Boot disk or data ?
Win7 is a bit of a pain as you have to get the drivers loaded and get the bios selections correct.
Win 10 will just install. Don't know about Win 8.1

Win 7 64 bit - faster boot. beyond that, seems slightly faster.
Logged

armand

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5540
    • Photos
Re: SSD - M2/PCIE vs SATA
« Reply #2 on: February 19, 2018, 09:28:18 pm »

Win 10, latest version.

Everything I work with is already on SSD, I use the HDDs only for backup/ redundancy.

danielc

  • Newbie
  • *
  • Offline Offline
  • Posts: 45
Re: SSD - M2/PCIE vs SATA
« Reply #3 on: February 19, 2018, 11:51:16 pm »

M.2 is significantly faster in real world usage. I've transitioned my C drive and my LR catalogue to it and it's made a big difference.
Logged

ukmdb

  • Newbie
  • *
  • Offline Offline
  • Posts: 6
  • Nottingham Photographer - UK
    • John Price Photography
Re: SSD - M2/PCIE vs SATA
« Reply #4 on: March 06, 2018, 03:32:11 am »

Samsung 960 PRO 1TB has rated speeds of UpTo 3500 MBps Read and UpTo 2100 MBps Write
I can say the real world usage on the samsung pro is near that.
I actually do get 3,300 MBs Read and just over 2k Write.
Had mine for about a year now and have tested the speeds on a few occasions and have not noticed any drops in speed.
I would recommend the move but of course all manufacturers are different, I just swear by the Samsung drives.
Logged
Nottingham wedding photographer
https://www.johnpricephotography.co.uk
I shoot Canon > Envy the Sony

Frodo

  • Full Member
  • ***
  • Offline Offline
  • Posts: 152
Re: SSD - M2/PCIE vs SATA
« Reply #5 on: June 17, 2018, 07:20:54 pm »

Puget Systems compared:
WD Red 4TB SATA 6Gb/s (150 MB/s read, 150 MB/s write)
Samsung 850 Pro 1TB SATA 6Gb/s SSD (550 MB/s read, 520 MB/s write)
Samsung 960 Pro 1TB M.2 x4 NVMe SSD (3,500 MB/s Read, 2,100 MB/s write)

https://www.pugetsystems.com/labs/articles/Adobe-Lightroom-2015-8-Storage-Performance-Analysis-875/

They concluded: "For most of the tasks we tested, there was minimal difference between having your Lightroom files on a single platter drive versus a SSD, NVMe, or even spread across multiple drives."
Logged

Brad P

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 258
Re: SSD - M2/PCIE vs SATA
« Reply #6 on: June 17, 2018, 07:27:18 pm »

Maybe for the tasks themselves while editing a picture, but that's certainly not true for day to day work on lightroom generally.  Accessing files, loading the program, exporting files to plugins, switching libraries and folders etc. - all benefit from faster storage.
Logged

Farmer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2848
Re: SSD - M2/PCIE vs SATA
« Reply #7 on: June 17, 2018, 10:21:09 pm »

Puget Systems compared:
WD Red 4TB SATA 6Gb/s (150 MB/s read, 150 MB/s write)
Samsung 850 Pro 1TB SATA 6Gb/s SSD (550 MB/s read, 520 MB/s write)
Samsung 960 Pro 1TB M.2 x4 NVMe SSD (3,500 MB/s Read, 2,100 MB/s write)

https://www.pugetsystems.com/labs/articles/Adobe-Lightroom-2015-8-Storage-Performance-Analysis-875/

They concluded: "For most of the tasks we tested, there was minimal difference between having your Lightroom files on a single platter drive versus a SSD, NVMe, or even spread across multiple drives."

They don't state how large the catalogue was that the used, which could have quite an impact.
Logged
Phil Brown

jrsforums

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1288
Re: SSD - M2/PCIE vs SATA
« Reply #8 on: June 17, 2018, 11:10:55 pm »

They don't state how large the catalogue was that the used, which could have quite an impact.

They said that having the catalog on SSDs showed some improvement.  The actual image files did not.  One must assume that the library database has good direct access to the files needed and the faster access and transfer of SSDs was minimal vs other activities.
« Last Edit: June 18, 2018, 12:52:39 am by jrsforums »
Logged
John

Farmer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2848
Re: SSD - M2/PCIE vs SATA
« Reply #9 on: June 17, 2018, 11:52:14 pm »

We still need to know the size of the files to really understand the value of the testing.  Also, were they created just for the test or real-world files that on an HDD would likely be fragmented due to the increase in size over time (where seek and access times from HDD are much slower than SSD and therefore impact performance).
Logged
Phil Brown

jrsforums

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1288
Re: SSD - M2/PCIE vs SATA
« Reply #10 on: June 18, 2018, 12:05:09 am »

We still need to know the size of the files to really understand the value of the testing.  Also, were they created just for the test or real-world files that on an HDD would likely be fragmented due to the increase in size over time (where seek and access times from HDD are much slower than SSD and therefore impact performance).

...and if you got that info, what would you do with it?  Puget Systems has been doing Lightroom tests for years.
Logged
John

Frodo

  • Full Member
  • ***
  • Offline Offline
  • Posts: 152
Re: SSD - M2/PCIE vs SATA
« Reply #11 on: June 18, 2018, 12:10:35 am »

Fair comment Phil.  I don't have a huge catalog, but its still 1.6 GB plus 20GB previews.
For me uploading is not an issue - I can go away and make a coffee if needed. Exporting is more of an issue as I tend to do this one file at a time once processing is complete.  At the moment uploading 50MP files takes a couple of seconds, but its mainly HDR and panorama merge and general responsiveness that's an issue and I thought that was linked to processor speed.
To what degree does accessing the catalog require good read/write speeds?
A 240 GB M2/PCIE is only about twice the price of a SATA SSD. But then do I go with a 500GB SATA … and where does it stop?
Logged

Farmer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2848
Re: SSD - M2/PCIE vs SATA
« Reply #12 on: June 18, 2018, 04:15:44 am »

...and if you got that info, what would you do with it?  Puget Systems has been doing Lightroom tests for years.

I would then make an assessment of the veracity and usefulness of the test.  I've been building PCs longer than Puget Systems has been around (not by much, but a few years) and I've been using Lr since it was released.  I'm not comfortable with their assertion because it doesn't seem to marry with real-world results that I've seen, but I'm very much prepared to accept their testing if it's robust and measures real-world scenarios regardless of how uncomfortable the results may be to me because good testing is always better than general observation.  I'm always concerned about claims and tests that don't detail the methodology, though.
Logged
Phil Brown

Farmer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2848
Re: SSD - M2/PCIE vs SATA
« Reply #13 on: June 18, 2018, 04:20:14 am »

Fair comment Phil.  I don't have a huge catalog, but its still 1.6 GB plus 20GB previews.
For me uploading is not an issue - I can go away and make a coffee if needed. Exporting is more of an issue as I tend to do this one file at a time once processing is complete.  At the moment uploading 50MP files takes a couple of seconds, but its mainly HDR and panorama merge and general responsiveness that's an issue and I thought that was linked to processor speed.
To what degree does accessing the catalog require good read/write speeds?
A 240 GB M2/PCIE is only about twice the price of a SATA SSD. But then do I go with a 500GB SATA … and where does it stop?

In simple terms, it stops when you feel happy about the price/performance level.  If you're happy with what you've got, I wouldn't spend another second thinking about it or another cent in changing it.  That's where I'm at - my in no way impacts on me in any way worth even a cent to do anything about.  That will eventually change and then I'll deal with it.  I just like to know the basis of tests and claims so as to understand their real value and, unfortunately, Puget hasn't posted those.  As I just said in my previous response, regardless of what I think may or may not be correct, if their testing stacks up then I'll gladly take it on board and benefit from that knowledge - I just don't like the idea of making a decision based on methodology I can't consider because it's unknown.
Logged
Phil Brown

jrsforums

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1288
Re: SSD - M2/PCIE vs SATA
« Reply #14 on: June 18, 2018, 11:11:51 am »

I would then make an assessment of the veracity and usefulness of the test.  I've been building PCs longer than Puget Systems has been around (not by much, but a few years) and I've been using Lr since it was released.  I'm not comfortable with their assertion because it doesn't seem to marry with real-world results that I've seen, but I'm very much prepared to accept their testing if it's robust and measures real-world scenarios regardless of how uncomfortable the results may be to me because good testing is always better than general observation.  I'm always concerned about claims and tests that don't detail the methodology, though.

And I was involved in product development and product management of the PC for 20 years starting in 1983...and started with Lightroom on the first beta....but that does not make me a testing guru, nor have I performed any concise tests of my own.

However, their conclusion statement seems reasonable based on my experience in using SSDs vs HDs with LR and statements Adobe has released on performance considerations.

That statement is “...Whether the source images themselves were on an SSD or a platter drive did not appear to make all that much of a difference, but having the catalog, preview, and cache files on an SSD allowed us to convert RAW images to DNG about 2% faster and export images about 7-8% faster. Upgrading to an even faster NVMe did not further improve performance very much, however.”
Logged
John

Farmer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2848
Re: SSD - M2/PCIE vs SATA
« Reply #15 on: June 18, 2018, 07:22:56 pm »

And I was involved in product development and product management of the PC for 20 years starting in 1983...and started with Lightroom on the first beta....but that does not make me a testing guru, nor have I performed any concise tests of my own.

However, their conclusion statement seems reasonable based on my experience in using SSDs vs HDs with LR and statements Adobe has released on performance considerations.

That statement is “...Whether the source images themselves were on an SSD or a platter drive did not appear to make all that much of a difference, but having the catalog, preview, and cache files on an SSD allowed us to convert RAW images to DNG about 2% faster and export images about 7-8% faster. Upgrading to an even faster NVMe did not further improve performance very much, however.”

/shrug

You asked me what and why I wanted the info so I told you - I didn't realise this was going to turn into a pissing contest.  In what way is it unreasonable to ask someone/some company who is publishing test results to provide data about that test?  If you're happy, that's great.  I'm not unhappy, I just don't think their test is authoritative unless they provide the specifications because we all should know that benchmarking and real-world aren't necessarily the same.  At the end of the day, I want to make decisions based on the best information and I want to be passing on the best information when people ask me questions.  That means validating test results by checking the methodology.  If that goes against my expectations, that's fine - I've learned something.
Logged
Phil Brown

Joe Towner

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1365
Re: SSD - M2/PCIE vs SATA
« Reply #16 on: June 19, 2018, 02:13:19 pm »

Baseline performance differences between SSD's and HDD's doesn't line up with a less than 10% difference.  Things like preview render sizing & raw MP sizing make a huge difference, as does the overall workflow.  Fully disclosing a testing methodology is standard operating procedure.

As to performance differences between SSDs v NVMe drives, it's really all about the threading that the software supports.  NVMe drives can bypass part of the bottleneck in a chipset by using PCIe lanes, but you need a x4 slot.
Logged
t: @PNWMF

Shiftworker

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 71
Re: SSD - M2/PCIE vs SATA
« Reply #17 on: June 20, 2018, 03:16:03 am »

Despite NVME drives being faster than SSD the process of opening and saving files also involves CPU work and this becomes the bottle neck and not the drive speed. I'm just about to reconfigure my workstation and will be moving my LR cat and cache to the NVME drive and moving my C drive from that to an SSD. Booting from an NVME drive has been plagued with problems from the start on my x99 mb. As a long term LR user I have found CPU speed to be the biggest performance decider once you have set up other systems optimally i.e LR cat and cache on fastest drive, image files on separate fast drive, C on separate drive.
Logged

kers

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4388
    • Pieter Kers
Re: SSD - M2/PCIE vs SATA
« Reply #18 on: July 24, 2018, 07:13:48 am »

... As a long term LR user I have found CPU speed to be the biggest performance decider once you have set up other systems optimally i.e LR cat and cache on fastest drive, image files on separate fast drive, C on separate drive.
I agree ; opening a file on a program is more than just copying data.
I have an old mac pro with a very fast drive and opening a file in photoshop is between 50 and 330MB/sec. The drive can do more than 700mb/sec.

Logged
Pieter Kers
www.beeld.nu/la
Pages: [1]   Go Up