Intel Enterprise SSD's -- S3610 480GB ($89.99) -- S3520 960GB ($159.99)

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,625
2,043
113
I got some of the 480gb variant... work fine, but limited drive info since they're HP Firmware.

They've been on there for months, slowly dropping prices too fwiw


The other drive, IMO at $160 is not a deal.
 

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,625
2,043
113
The drive is an Intel, but it's made for HP and has HP Firmware on it.
(There may be guides to change it to Intel firmware, I know we had 1 on NVME on the forum.)

Thus, unless your'e using it in an HP system or happen to use an HP HBA\Raid card with HP firmware still running you won't be able to see certain things such as how many hours the drive has been, how much data has been written to it, etc, etc.
 

james23

Active Member
Nov 18, 2014
441
122
43
52
if you can run SAS drives, i woudl skip those and get these (i just posted):

https://forums.servethehome.com/index.php?threads/480g-hgst-ssd-sas3-60-75-offer-accepted.23727/

much better performance / $ (IF YOU can run sas, not sata)
 

Tiberizzle

New Member
Mar 23, 2017
25
11
3
124
Does anyone know why the power consumption on certain (I want to say recent but honestly have no idea what the pattern is) SSDs is so high?

The elderly Intel 1500 Pro 480GB is specced 125mW idle 195mW active.

The S3610 480GB is specced at 600mW idle and 4.3W active, which makes the idle power about 4x the Pro 1500 series's active power and the active power like 20x the Pro 1500's.

The 1500 Pro at the same capacity is actually specced at like 4x the random write performance of the S3610 -- aren't block erase and to a lesser extent block write the "power hungry" operations in NAND? Interestingly, they are also both specified to use 20nm MLC...

The HP SAS drives above are 2.2W idle 11W active -- the active power spec is higher than the seek power of many 15K drive and even approaching the spin-up power of some!?

This seems to be something that never really gets brought up in these SSD deal threads, but in e.g. the case of 24x2.5" chassis the power difference between inefficient SSDs and efficient SSDs would amount to potentially 250+ watts. This is higher than the total system power on low load of most dual E5-2600v2 builds and further an extra 250 watts up front would definitely affect considerations like whether or not you can safely/reasonably hack a GPU into a chassis which does not officially support it.

Do these higher power consumption SSDs justify the higher consumption with some obscure feature support or performance advantage under specific access patterns or is it just cost cutting / apathy on the part of manufacturers to see orders of magnitude difference in power consumption for SSDs with similar specified capacity and performance?
 

Evan

Well-Known Member
Jan 6, 2016
3,346
598
113
Well sure some SSD’s are less efficient than others and different NAND types and production nodes have an effect as well but in general SAS needs a good deal more power than SATA, as 12G uses more power than 6G and dual port also has an impact.
Higher power is also needed for more performance.

But your point is completely valid, sometimes the power is really high, look at NVMe early drives as well, super high power, again with more performance.

If you only storing a few files the good old intel SATA SSD would be a good start, on the other hand active VM’s live some NMVe goodness.
 

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,625
2,043
113
Does anyone know why the power consumption on certain (I want to say recent but honestly have no idea what the pattern is) SSDs is so high?

The elderly Intel 1500 Pro 480GB is specced 125mW idle 195mW active.

The S3610 480GB is specced at 600mW idle and 4.3W active, which makes the idle power about 4x the Pro 1500 series's active power and the active power like 20x the Pro 1500's.

The 1500 Pro at the same capacity is actually specced at like 4x the random write performance of the S3610 -- aren't block erase and to a lesser extent block write the "power hungry" operations in NAND? Interestingly, they are also both specified to use 20nm MLC...

The HP SAS drives above are 2.2W idle 11W active -- the active power spec is higher than the seek power of many 15K drive and even approaching the spin-up power of some!?

This seems to be something that never really gets brought up in these SSD deal threads, but in e.g. the case of 24x2.5" chassis the power difference between inefficient SSDs and efficient SSDs would amount to potentially 250+ watts. This is higher than the total system power on low load of most dual E5-2600v2 builds and further an extra 250 watts up front would definitely affect considerations like whether or not you can safely/reasonably hack a GPU into a chassis which does not officially support it.

Do these higher power consumption SSDs justify the higher consumption with some obscure feature support or performance advantage under specific access patterns or is it just cost cutting / apathy on the part of manufacturers to see orders of magnitude difference in power consumption for SSDs with similar specified capacity and performance?
The Intel 1500 Pro is a consumer drive, most consumer drives consume MUCH LESS power than enterprise counterpart.

Look at the Intel S3700, S3500 for power consumption as those are the same era as the Intel 1500, their power usage is much more.

It's not brought up because it's understood performance = more power.

They justify the higher power by higher performance... by a huge margin over low power consumer drives.
 

Tiberizzle

New Member
Mar 23, 2017
25
11
3
124
The Intel 1500 Pro is a consumer drive, most consumer drives consume MUCH LESS power than enterprise counterpart.

Look at the Intel S3700, S3500 for power consumption as those are the same era as the Intel 1500, their power usage is much more.

It's not brought up because it's understood performance = more power.

They justify the higher power by higher performance... by a huge margin over low power consumer drives.
Erm.. as I stated in the post you quoted, the Intel 1500 Pro has 4x the specified write performance and virtually the same specs in every other way as the S3610, except at 20x lower specified power.

The S3610 manages not much more than 10% advantage over the 1500 Pro in any benchmark for its 20x higher TDP rating and is in fact >30% slower in benchmarks that include a random write component.

The theory does not hold water based on specified performance or benchmark I have seen and why the obvious hypothesis is such a consistently terrible fit for the experimental data is pretty much the question I intended to pose.

The SAS3 drives have a clear advantage from the faster interface and multipath capability and that's not really the phenomenon I'm trying to draw attention to here.

Rather, it's the huge range of the power specifications for the SAS2 / SATA3 drives that cluster at the interface limits.

I am interested in knowing specifically what gains come from the orders-of-magnitude additional power budget afforded to the "Enterprise" drives in the SAS 2 / SATA 3 interface-limited cluster, and which benchmarks I can see these gains on.

I haven't seen anything to justify the power consumption and to me that implies we are being marketed an inferior product at a premium.
 

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,625
2,043
113
Erm.. as I stated in the post you quoted, the Intel 1500 Pro has 4x the specified write performance and virtually the same specs in every other way as the S3610, except at 20x lower specified power.

The S3610 manages not much more than 10% advantage over the 1500 Pro in any benchmark for its 20x higher TDP rating and is in fact >30% slower in benchmarks that include a random write component.

The theory does not hold water based on specified performance or benchmark I have seen and why the obvious hypothesis is such a consistently terrible fit for the experimental data is pretty much the question I intended to pose.

The SAS3 drives have a clear advantage from the faster interface and multipath capability and that's not really the phenomenon I'm trying to draw attention to here.

Rather, it's the huge range of the power specifications for the drives that all pretty much cluster at the SAS2 / SATA3 interface limits.

I am interested in knowing specifically what gains come from the orders-of-magnitude additional power budget afforded to the "Enterprise" drives in the SAS 2 / SATA 3 interface-limited cluster, and which benchmarks I can see these gains on.

I haven't seen anything to justify the power consumption and to me that implies we are being marketed an inferior product at a premium.

You're 100% incorrect, no "theory" needed here :D I have experience with both those drives and many more as do many others here who will tell you the same thing.

I suggest you read the forums more about comparing consumer drives to enterprise drives or go look at the very few sites that run steady-state performance of enterprise and also include some consumer for comparison (very hard to find).

Most of those "FAST" consumer drives drop below 10,000 iops, a lot below 5,000 and certain ones once out of cache will drop to 900-1500.

If you don't believe me try out 12 consumer drives vs 12 s3610 and come back and share your experiences... you need to test them appropriately too, consistently for hours. Or, put them to 'real use' and report back in a month or two how slow they've gotten.

If you're running a 4GB or 8GB test then yeah, that's bad results\data that's useless.


The answer to your question is performance and QOS require more overhead, and that = power and performance = power. The cache on consumer = less power to hit high IOPs where the enterprise require higher power to provide this consistently 100% across the entire capacity while also maintaining low latency\consistency. (This is also why samsung while similar iop rating perform worse than Intel enterprise. From what I've seen Samsung has improved on this, but I haven't gotten any of their new drives to compare and test with yet.)
 
Last edited:

Dreece

Active Member
Jan 22, 2019
503
160
43
T_Plus is correct.

Throw in an enterprise drive, throw in a consumer 'phoar-thats-fast' drive... load them silly with parallel disk benchmarking, not just sequentials but also the IOP calculators too... afterwards crunch the numbers, don't be surprised when you see how consistent the enterprise drive's performance is in comparison to the consumer drive where it mostly tails off into abysmal low numbers.

IOPS and consistent performance is where enterprise drives really excel, one won't need what enterprise drives offer if loading windows and browsing the web and playing Battlefield is about as much load as the drive will experience... burst cache performance is great for consumers, but not much use on a single enterprise drive, well let us not forget that enterprise would normally have a fair few dozen of these drives all working together serving multiple VMs etc... also where raid is used, drive caches are routinely turned off anyway.

Consumer drives are fast for consumers, enterprise drives are fast for the enterprise, comparing the two is pointless really, the way the technology works between the two is fairly different, it isn't just about MLC or 3D NAND or whatever... the controller and its algorithms all come into play depending on the workload.
 

Patriot

Moderator
Apr 18, 2011
1,450
789
113
I love my samsung drives... but I get the enterprise variants because of the tuning for consistency ... and plp

Enterprise drives are meant to have consistent performance at the cost of peak ... consumer drives are meant to have breaks to run GC and cleanup for the next burst.
 
  • Like
Reactions: Dreece

Peanuthead

Active Member
Jun 12, 2015
839
177
43
44
@T_Minus and @Patriot are 100% correct. I run Intel enterprise in my home lab for overkill, consistency (closer to real world), would never get my $ back out of them, and why not with all others being the same?
 

i386

Well-Known Member
Mar 18, 2016
4,218
1,540
113
34
Germany
The S3610 manages not much more than 10% advantage over the 1500 Pro in any benchmark for its 20x higher TDP rating and is in fact >30% slower in benchmarks that include a random write component.
Do you have 1500 Pros and windows as os?
Can you run the following benchmark and post or pm me the results?
Code:
diskspd -b4K -c20G -d120 -L -o8 -r -Sh -t4 -w20 testfile.dat
 

Dreece

Active Member
Jan 22, 2019
503
160
43
I love my samsung drives...
Same.

Been with Samsung on desktop side since the very beginning of SSD arrival in consumer scene, currently sporting a 2tb 970 pro, I wouldn't even consider throwing in a sas ssd/spinner into my desktop, considering I've got limited pcie lanes as it is and then the additional waste of energy ie heat from a controller card which the 10g nic is already happily fulfilling... forget it.

Having built probably in the region of around 25 odd video-editing suites for medium to big budget clients, only one ever went for a sas config, and that was only because he was working with a red 8k setup whence dealing with very large files and needed a fairly large expandable disk enclosure on the side (D3700).

Personally I will always stick with Samsung NVME and whatever better tech comes out of their research labs for us enthusiast consumers, reliability and top performance for workstation use-cases is their forte.

ps. I grabbed the 2tb off of ebay for 215 quid just before xmas... one should never rush into purchasing... there's always a deal round the corner, my tip > filter for 'New Listings' on ebay, sometimes a newbie just throws something up for pennies and you need to get in there and pinch it before he/she realises they could have sold it for double or before those pain in the backside resellers come along and buy it all up and then sell it back on ebay for double (they annoy the hell out of me) but it's a free world I guess.
 
Last edited:

Tiberizzle

New Member
Mar 23, 2017
25
11
3
124
You're 100% incorrect, no "theory" needed here :D I have experience with both those drives and many more as do many others here who will tell you the same thing.

I suggest you read the forums more about comparing consumer drives to enterprise drives or go look at the very few sites that run steady-state performance of enterprise and also include some consumer for comparison (very hard to find).

Most of those "FAST" consumer drives drop below 10,000 iops, a lot below 5,000 and certain ones once out of cache will drop to 900-1500.

If you don't believe me try out 12 consumer drives vs 12 s3610 and come back and share your experiences... you need to test them appropriately too, consistently for hours. Or, put them to 'real use' and report back in a month or two how slow they've gotten.

If you're running a 4GB or 8GB test then yeah, that's bad results\data that's useless.


The answer to your question is performance and QOS require more overhead, and that = power and performance = power. The cache on consumer = less power to hit high IOPs where the enterprise require higher power to provide this consistently 100% across the entire capacity while also maintaining low latency\consistency. (This is also why samsung while similar iop rating perform worse than Intel enterprise. From what I've seen Samsung has improved on this, but I haven't gotten any of their new drives to compare and test with yet.)
I have somewhere in the ballpark of 100 assorted Intel consumer and enterprise SSDs in my lab and have tested hundreds of drives from pretty much every major manufacturer.

While it is true that many low end consumer SSDs are unable to sustain their specified performance for very long, it's equally true that many are cache-less designs which are effectively PLP and sustain steady state performance until the drive is full or exhausts P/E cycles.

I have referred to specifications from the manufacturer's own website and supporting benchmarks for the example I began the discussion with.

I am not particularly interested in discussing this further if will continue to proceed without data or specific implementation details that directly and logically relate the power consumption with any observable quantity.
 

Tiberizzle

New Member
Mar 23, 2017
25
11
3
124
Do you have 1500 Pros and windows as os?
Can you run the following benchmark and post or pm me the results?
Code:
diskspd -b4K -c20G -d120 -L -o8 -r -Sh -t4 -w20 testfile.dat
I do have 1500 Pros and I can provide data from them but I'm afraid I do not have Windows available. What metrics are you interested in or do you have e.g. fio config I can run?