Samsung 860 EVO SATA III with older motherboard

[TitusTroy]

earlier today I bought a Samsung 860 EVO 1TB 2.5 Inch SATA III Internal SSD as the price was too good to pass up ($130)...but I'm still using a motherboard from 2010 (Asus P6X58D Premium)...the mobo is native SATA II (3.0 Gb/s) but it does have 2 special SATA III (6.0 Gb/s) ports...but the SATA 3 ports use a Marvell controller...what am I losing by using the Marvell port?...will it still be substantially faster then a non-SSD mechanical drive in the SATA II port?…or should I just hook it up to the native SATA II ports and not even bother with the Marvell?

is there anything I need to do in the BIOS to max out the performance of the Marvell port?...the motherboard manual mentions something about setting it to 'Standard IDE mode'...I attached a pic from the motherboard manual to illustrate it...also is it best to hook it up to the 6.0_1 or 6.0_2 port?

 

[Insert_Nickname]

earlier today I bought a Samsung 860 EVO 1TB 2.5 Inch SATA III Internal SSD as the price was too good to pass up ($130)...but I'm still using a motherboard from 2010 (Asus P6X58D Premium)...the mobo is native SATA II (3.0 Gb/s) but it does have 2 special SATA III (6.0 Gb/s) ports...but the SATA 3 ports use a Marvell controller...what am I losing by using the Marvell port?...will it still be substantially faster then a non-SSD mechanical drive in the SATA II port?…or should I just hook it up to the native SATA II ports and not even bother with the Marvell?

You'll want to connect it to the native Intel ports, even if they're only SATA2. You'll loose a bit of sequential R/W performance, but random performance should be better. That's what make an SSD feel significantly faster then any HDD.

is there anything I need to do in the BIOS to max out the performance of the Marvell port?...the motherboard manual mentions something about setting it to 'Standard IDE mode'...I attached a pic from the motherboard manual to illustrate it...also is it best to hook it up to the 6.0_1 or 6.0_2 port?

Avoid IDE mode. That'll kill performance, and is only really required for XP compatibility. Vista and newer have native AHCI drivers.

Also, remember to set the Intel ports to either AHCI or RAID mode before you install the OS.

 

[TitusTroy]

You'll want to connect it to the native Intel ports, even if they're only SATA2. You'll loose a bit of sequential R/W performance, but random performance should be better. That's what make an SSD feel significantly faster then any HDD.

Avoid IDE mode. That'll kill performance, and is only really required for XP compatibility. Vista and newer have native AHCI drivers.

Also, remember to set the Intel ports to either AHCI or RAID mode before you install the OS.

thanks for the info...yes I always use AHCI prior to installing the OS even on my mechanical hard drives...so the Marvell ports on the board aren't true 6.0 Gb/s SATA III?...it's more like a fake imitation of the real thing?...any idea the performance differences between the Marvell SATA III and native Intel SATA II port on this board?

will it harm my SSD if I run a bunch of speed/performance benchmarks on it?...meaning will it degrade performance in any way using the built-in Samsung benchmark or another 3rd party one?

 

[VirtualLarry]

will it harm my SSD if I run a bunch of speed/performance benchmarks on it?...meaning will it degrade performance in any way using the built-in Samsung benchmark or another 3rd party one?

As long as you're not running them multiple times a day, every day, no.

Intel SATA2 (native) ports will have lower latency, but lower sequential speeds.
Marvell (third-party) SATA ports will have higher sequential, but also slightly higher latency (lower 4K IOPS).

 

[Insert_Nickname]

As long as you're not running them multiple times a day, every day, no.

Intel SATA2 (native) ports will have lower latency, but lower sequential speeds.
Marvell (third-party) SATA ports will have higher sequential, but also slightly higher latency (lower 4K IOPS).

^^what VL said. The Marvell controller is also limited by it's PCIe 2.0 x1 interface, which means you'll never get -true- SATA3 performance out of it anyway. The PCIe interface tops out at ~450MB/s (Intel SATA2 ~270MB/s), so the actual difference is negligeble.

You might be asking a -why bother then- at this point. This is fine if you're connecting a pair of HDDs to it, but SSDs are another matter. HDDs generally don't push more then 1-1.5MB/s random R/W (which translates to ~100 IOPS), but can use the higher sequential speeds if you're RAID'ing them.

If you want higher performance, you'd need a PCIe expansion card featuring something like the Marvell 9230 PCIe 2.0 x2 controller. Which has enough PCIe bandwidth to support a couple of SATA3 SSDs. I think startech.com makes one.

will it harm my SSD if I run a bunch of speed/performance benchmarks on it?...meaning will it degrade performance in any way using the built-in Samsung benchmark or another 3rd party one?

Other then putting a bit of more-or-less useful writes on it that wont do anything. Degrading SSDs are a thing of the remote past.

 

[TitusTroy]

I received my Samsung EVO today and was wondering how important it was to update the firmware on a regular basis?...do I need to update it every time a new one is released or can I just keep using the version it shipped with for the life of the drive?

 

[VirtualLarry]

I received my Samsung EVO today and was wondering how important it was to update the firmware on a regular basis?...do I need to update it every time a new one is released or can I just keep using the version it shipped with for the life of the drive?

The real answer? "It depends."

Sometimes, firmware "fixes" cause more bugs than they fix (SandForce 2nd-gen controller-based SSDs come to mind).

And sometimes, if you're not having issues, you might not want to change anything.

But then, sometimes, there are definite bugs that you need a solution or workaround for, such as the "1000 hours bug" in Crucial M4 SSDs. (They would go offline every hour on the hour, after they had accumulated 1000 hours runtime. Weird bug.)

So, for the general user, I would say, read the firmware "release notes", if any are provided, and decide for yourself, if it's necessary to fix bugs that you think that you would encounter.

Might also want to wait a few weeks to a month, to make sure that there aren't any show-stopping new bugs in the new firmware, bricking people's drives that updated.

I mean, I generally believe in updating mobo BIOSes, as that helps to adapt to new technology coming out. (Especially with AM4 boards like mine, new memory support, etc.)

But SSDs? Unless there is a definite identified bug in the current firmware, that the update definitely fixes, I tend to leave well enough alone, especially with the budget drives. With Intel, Crucial, and Samsung, I might be more tempted to update, as they generally are larger companies with bigger R&D and testing budgets, and generally test their stuff better.

 

[TitusTroy]

But SSDs? Unless there is a definite identified bug in the current firmware, that the update definitely fixes, I tend to leave well enough alone, especially with the budget drives. With Intel, Crucial, and Samsung, I might be more tempted to update, as they generally are larger companies with bigger R&D and testing budgets, and generally test their stuff better.

thanks...I think I'll update it to the latest firmware before using it for the first time but after that I'll leave it be unless there's a specific issue

one more question...should I enable Rapid Mode?...looking online it seems opinions are mixed...I currently have 12GB of system RAM but plan on building a new rig soon with 16GB...also when benchmarking the SSD should Rapid Mode be disabled?

 

[VirtualLarry]

thanks...I think I'll update it to the latest firmware before using it for the first time but after that I'll leave it be unless there's a specific issue

That's a good plan.

one more question...should I enable Rapid Mode?...looking online it seems opinions are mixed...I currently have 12GB of system RAM but plan on building a new rig soon with 16GB...also when benchmarking the SSD should Rapid Mode be disabled?

It's up to you, but honestly, my impression is just that "Rapid Mode" is made for "cheating" / scoring high on benchmarks, and therefore has little real-world utility, unlike actual tiered caching programs. Also, Windows' itself already does RAM-caching, so that by dedicating RAM to caching with RAPID, you are reducing Windows' own sophisticated file-caching mechanisms.

Basically, I tend to let Windows manage it's own virtual-memory and RAM-caching.

 

[TitusTroy]

I've always used HDDErase to reformat/erase my mechanical hard drives prior to re-installing Windows...is HDDErase still the best way to secure erase modern SSD's like the 860 EVO?...is the manufacturer provided 'Secure Erase' tool preferred over it (for example Samsung's Magician Software toolkit)?...I hear a lot of reports about people's computers being in a frozen/locked state after using the Samsung Secure Erase tool

 

[TitusTroy]

I recently bought a Samsung 860 EVO 1TB drive as the price was too good to pass up ($130) but I'm still using an older motherboard with SATA II connections (it does have 2 special SATA III (6.0 Gb/s) Marvell ports but I heard it's best not to use those)...I did a benchmark run with my new Samsung 860 EVO 1TB SSD and wanted to know if the numbers I'm getting are what they're supposed to be (considering I'm using a SATA II connection)...

 

[VirtualLarry]

Yeah, that's what you'll get in a SATAII port.

 

[TitusTroy]

Yeah, that's what you'll get in a SATAII port.

so the numbers are exactly what they should be and it's not any slower (or faster)?...just for fun I compared to a benchmark run using my previous Western Digital Black 640GB mechanical hard drive...

 

[VirtualLarry]

Just use your drive and enjoy, man. There's nothing apparently from that benchmark, that would indicate anything non-functional. Sure, benchmarks can vary from run to run, and from system to system, but yours are in the ballpark for a SATA6G SSD in a SATA2 port.

 

[Boo_hoo]

I realise this is an old thread but I recently bought a Samsung EVO 860 1TB and have a P6X58D-E motherboard and thought I would share some data for possible future googlers. I connected the SSD to both the marvell 9128(SATA 3) and intel(SATA 2) ports and recorded benchmarks using Samsung Magician and CrystalDiskMark. (It should be noted that Samsung Magician could not detect the interface when the SSD was connected to the marvell ports.)

Sequential Read (Q= 32,T= 1) : 414.047 MB/s
Sequential Write (Q= 32,T= 1) : 250.679 MB/s
Random Read 4KiB (Q= 8,T= 8) : 110.476 MB/s [ 26971.7 IOPS]
Random Write 4KiB (Q= 8,T= 8) : 187.117 MB/s [ 45682.9 IOPS]
Random Read 4KiB (Q= 32,T= 1) : 115.599 MB/s [ 28222.4 IOPS]
Random Write 4KiB (Q= 32,T= 1) : 107.321 MB/s [ 26201.4 IOPS]
Random Read 4KiB (Q= 1,T= 1) : 27.487 MB/s [ 6710.7 IOPS]
Random Write 4KiB (Q= 1,T= 1) : 41.178 MB/s [ 10053.2 IOPS]

Sequential Read (Q= 32,T= 1) : 284.961 MB/s
Sequential Write (Q= 32,T= 1) : 272.952 MB/s
Random Read 4KiB (Q= 8,T= 8) : 215.188 MB/s [ 52536.1 IOPS]
Random Write 4KiB (Q= 8,T= 8) : 203.489 MB/s [ 49679.9 IOPS]
Random Read 4KiB (Q= 32,T= 1) : 200.189 MB/s [ 48874.3 IOPS]
Random Write 4KiB (Q= 32,T= 1) : 170.095 MB/s [ 41527.1 IOPS]
Random Read 4KiB (Q= 1,T= 1) : 31.895 MB/s [ 7786.9 IOPS]
Random Write 4KiB (Q= 1,T= 1) : 62.762 MB/s [ 15322.8 IOPS]

Test : 32768 MiB [E: 4.1% (38.4/931.5 GiB)] (x5) [Interval=5 sec]

As can be seen the single advantage of the marvell port is the Sequential read speed. The intel port has much higher IOPS. The results are also confirmed in Samsung magician shown below:

marvell(SATA3):


intel(SATA2)




It is generally agreed that IOPS are a better indicator of typical use. It is also interesting to see from the data that the sequential write speed is 20MB/s faster on the intel ports. It still bothers me that I have to compromise but I guess that is the cost of using an ancient motherboard.

 

[Rifter]

I had a P6X58D-E in my last system, I can confirm the marvell ports are a total waste of space.

 

[Insert_Nickname]

I had a P6X58D-E in my last system, I can confirm the marvell ports are a total waste of space.

They certainly beat no ports at all. But are really only usable with HDDs.

Back when that kind of controller was designed SSDs were very much a (expensive) novelty item. I don't think they ever realised something would come out that could completely max out their controller...