Lowest storage option lower read/write speed?

[LoveToMacRumors]

iPhone 7 had this. Anyone think the iPhone 8 will also have this "problem"?

The pricier model can write data at 341 MBps, while its cheaper cousin can only achieve 42 MBps.

https://www.tomsguide.com/us/iphone-7-32-slower,news-23701.html

In a benchmarking test, Hilsentenger found that the iPhone 7 models with the smallest and largest amounts of storage have similar storage read speeds — 686 MBps for the 32GB, 856 MBps for the 256GB — but the write speeds were drastically different. (That's megabytes per second, not megabits per second, if you were wondering.)



The pricier model can write data at 341 MBps, while its cheaper cousin can only achieve 42 MBps. That’s an eight-fold difference, in case your low-end iPhone calculator is taking too long to do the math.

 

[rui no onna]

iPhone 7 had this. Anyone think the iPhone 8 will also have this "problem"?

64gb version slower than the 256/512gb options? for the upcoming iPhone 8?

https://www.tomsguide.com/us/iphone-7-32-slower,news-23701.html

That's pretty standard for solid state drives (SSD). Think of the 256GB as 4x 64GB in RAID-0. Some mask it by allocating 5-30GB of pseudo-SLC cache but once you run out of cache, then it goes back to regular steady state performance.

Day to day use, you won't notice much difference between 32GB or higher capacity models.

 

[joeblow7777]

Yeah, this is the kind of thing that tech hypochondriacs worry about, but if no one told you about it, you would never notice it on your own.

 

[Applejuiced]

It's the end of the world
And off course some will claim they can tell the difference...

 

[44267547]

Another example of write speeds making little, to no discernible difference the user can be concerned with. Placebo effect for the majority.

 

[LoveToMacRumors]

The pricier model can write data at 341 MBps, while its cheaper cousin can only achieve 42 MBps.

 

[joeblow7777]

The pricier model can write data at 341 MBps, while its cheaper cousin can only achieve 42 MBps.

Benchmarks don't always reflect reality. It's like how there's theory and then there's practice. I very much doubt that you will observe a 256GB phone functioning 8 times faster than a 32GB one.

 

[Shirasaki]

Even if it is true, as more and more companies pushing their code base AWAY from client and embrace cloud, I seriously doubt this “huge” disadvantage would really matter. It would only be noticeable when transferring like 10GB of files on two models.

 

[rui no onna]

Even if it is true, as more and more companies pushing their code base AWAY from client and embrace cloud, I seriously doubt this “huge” disadvantage would really matter. It would only be noticeable when transferring like 10GB of files on two models.

Raw performance difference is true. That's just the nature of the technology and is how SSDs work.

That said, as you mentioned, it's largely irrelevant. It's only helpful for multi-GB file transfers. Not with a computer though. Unless Apple includes USB 3.0 slave support on the new iPhone, transfer from the computer caps out at ~40MB/s. Besides, users who regularly deal with multi-GB files are also likely to go for higher storage capacities.

 

[MacUse-R]

My MacBook Pro writes about 30GB a day to the SSD if i dont do anything at all on the MacBook Pro, and since a SSD has a limited amount of data that can be written to it then why wouldnt it be a good idea to minimize the spontaneous writings to the SSD, those are the writings im trying to minimize.

Somebody wrote in an earlier post that petabytes can be written to SSD's, but my Samsung 850 EVO 500GB has i think 150TBW specified, so thats far from Petabytes, dont know if the real world amount is higher than the guaranteed amount stated by the manufacturer, but i guess it cant be that big of a difference?

 

[rui no onna]

My MacBook Pro writes about 30GB a day to the SSD if i dont do anything at all on the MacBook Pro, and since a SSD has a limited amount of data that can be written to it then why wouldnt it be a good idea to minimize the spontaneous writings to the SSD, those are the writings im trying to minimize.

Somebody wrote in an earlier post that petabytes can be written to SSD's, but my Samsung 850 EVO 500GB has i think 150TBW specified, so thats far from Petabytes, dont know if the real world amount is higher than the guaranteed amount stated by the manufacturer, but i guess it cant be that big of a difference?

Did you post on the wrong thread? Anyway...

Petabytes might have been the case for SLC/MLC NAND or higher capacities. It's less on smaller lithography and TLC. Don't know what the rated P/E cycles is for 3D NAND, though.

The 150TBW figure quoted by Samsung includes write amplification (WA) and is a minimum, I expect. WA's a bit more variable depending on task and free space and I reckon Samsung and Intel are fairly conservative there (maybe 5-10x WA). Quick calculation based on S.M.A.R.T. data for my Samsung 840, WA on that particular drive given my usage is 2.3x. Doesn't do anything terribly heavy although AV software manages several GB worth of writes per day.

That said, I don't think iOS (or Android for that matter) does anywhere near 30GB writes a day to the SSD. Otherwise, we would've seen NAND failures on 8-16GB models.

All things considered, trying to minimize writes is unnecessary with typical desktop usage. You'll probably want to replace the SSD with something cheaper and higher capacity before you run out of P/E cycles. One of my oldest SSDs, an Intel X25-M G2 120GB (circa 2009/10), is still going strong. Iirc, last I checked I've only used less than 5% of the rated P/E cycles.

 

[MacUse-R]

Did you post on the wrong thread? Anyway...

Petabytes might have been the case for SLC/MLC NAND or higher capacities. It's less on smaller lithography and TLC. Don't know what the rated P/E cycles is for 3D NAND, though.

The 150TBW figure quoted by Samsung includes write amplification (WA) and is a minimum, I expect. WA's a bit more variable depending on task and free space and I reckon Samsung and Intel are fairly conservative there (maybe 5-10x WA). Quick calculation based on S.M.A.R.T. data for my Samsung 840, WA on that particular drive given my usage is 2.3x. Doesn't do anything terribly heavy although AV software manages several GB worth of writes per day.

That said, I don't think iOS (or Android for that matter) does anywhere near 30GB writes a day to the SSD. Otherwise, we would've seen NAND failures on 8-16GB models.

All things considered, trying to minimize writes is unnecessary with typical desktop usage. You'll probably want to replace the SSD with something cheaper and higher capacity before you run out of P/E cycles. One of my oldest SSDs, an Intel X25-M G2 120GB (circa 2009/10), is still going strong. Iirc, last I checked I've only used less than 5% of the rated P/E cycles.

Yes you are right, apperently i posted in the wrong thread.
I have seen many people online having around 30GB writes to their SSD's and many people answers that thats perfectly normal for a Mac OS system.
Maybe there is a big difference with iOS which you are talkning about (and which this thread seems to be about), sorry my post ended up in the wrong thread somehow.

Then i guess it is of value trying to minimize writes to the SSD by disabling functions that does a lot of writes to the drive and which one is not using anyway or that one can be without, even though it seems thats not what most people thinks online.
[doublepost=1521580177][/doublepost]

Yes you are right, apperently i posted in the wrong thread.
I have seen many people online having around 30GB writes to their SSD's and many people answers that thats perfectly normal for a Mac OS system, dont know if that is correct though?
Maybe there is a big difference with iOS which you are talkning about (and which this thread seems to be about), sorry my post ended up in the wrong thread somehow.

Then i guess it is of value trying to minimize writes to the SSD by disabling functions that does a lot of writes to the drive and which one is not using anyway or that one can be without, even though it seems thats not what most people thinks online.

 

[rui no onna]

Yes you are right, apperently i posted in the wrong thread.
I have seen many people online having around 30GB writes to their SSD's and many people answers that thats perfectly normal for a Mac OS system.
Maybe there is a big difference with iOS which you are talkning about (and which this thread seems to be about), sorry my post ended up in the wrong thread somehow.

Then i guess it is of value trying to minimize writes to the SSD by disabling functions that does a lot of writes to the drive and which one is not using anyway or that one can be without, even though it seems thats not what most people thinks online.

I expect those numbers are pretty normal for a desktop OS. The Windows 7 desktop I'm typing this from averages 20GB host writes per day.

At 30GB writes a day, you're looking at 14 years at a 150TBW rating. Realistically, minimum NAND writes for a 500GB TLC SSD is 500GB * 1,000 P/E cycles = 500TBW. At 500TBW, you're looking at 45 years. The writes are really are a non-issue.