Samsung 840 EVO SSD: Tested At 120, 250, 500, And 1000 GB

[Guest]

I can't believe how fast SSD prices are dropping

 

[Tanquen]

How about the fact that the 840 Evo read speeds go way down after you have data on the drive?

Vs.

 

[Someone Somewhere]

One of those is an 830.

 

[Tanquen]

One of those is an 830.

Yes and it has a lot of data on it but the read speed across the dive is still the same but the 840 Evo has poor read speeds were you have data on the drive.

 

[stoatwblr]

I really wish that Samsung would put some onboard power failure protection on these drives. That's the only thing missing which would make them a giant killer, even if it meant it might cut into enterprise drive sales a little.

 

[elmo2006]

I've got the the Samsung 840 PRO 256GB and the drive is exceptional. Now with the latest Magician update, I have enabled RAPID mode and boy the R/W's have doubled.
I'll continue to purchase these products if Samsung would resolve their Canadian Warranty issues.

 

[Psyintz]

Does anybody know why my SSD would be performing so poorly compared to all of the other results I've seen posted?

Processor: AMD FX 6100
RAM: 8GB (Corsair DDR3)
OS: Windows 8.1
SSD: Samsung SSD 840 EVO (250GB)

My SSD Benchmark Results: http://postimg.org/image/ni5unmfrx/

All of the other screenshots I've seen have both the Sequestial Read/Write and the Random Read/Write meters shooting through the roof. However, my Random Read/Write scores seem to be quite poor compared to theirs.

Can anybody help point me in the right direction in regards to why I'm scoring so low on these particular benchmarks?

 

[Psyintz]

Does anybody know why my SSD would be performing so poorly compared to all of the other results I've seen posted?

Processor: AMD FX 6100
RAM: 8GB (Corsair DDR3)
OS: Windows 8.1
SSD: Samsung SSD 840 EVO (250GB)

My SSD Benchmark Results: http://postimg.org/image/ni5unmfrx/

All of the other screenshots I've seen have both the Sequestial Read/Write and the Random Read/Write meters shooting through the roof. However, my Random Read/Write scores seem to be quite poor compared to theirs.

Can anybody help point me in the right direction in regards to why I'm scoring so low on these particular benchmarks?

 

[mbarbantini]

Glad this review also refutes the perception of low TLC write endurance in normal desktop workloads. Been using a 128GB 840 in a daily-use desktop for eight months now and the endurance counter hasn't decremented at all. I'll have replaced the entire rig long before it wears out.

Just checked your specs and you're living off a 160 GB SSD. Do you have another external drive or is that it?

 

[mbarbantini]

Does anybody know why my SSD would be performing so poorly compared to all of the other results I've seen posted?

Processor: AMD FX 6100
RAM: 8GB (Corsair DDR3)
OS: Windows 8.1
SSD: Samsung SSD 840 EVO (250GB)

My SSD Benchmark Results: http://postimg.org/image/ni5unmfrx/

All of the other screenshots I've seen have both the Sequestial Read/Write and the Random Read/Write meters shooting through the roof. However, my Random Read/Write scores seem to be quite poor compared to theirs.

Can anybody help point me in the right direction in regards to why I'm scoring so low on these particular benchmarks?

Usually, benchmarks are influenced by other components. Would you mind posting your specifications here or messaging me?
For example, poor CPU performance will obviously slow down everything else.

 

[Psyintz]

Does anybody know why my SSD would be performing so poorly compared to all of the other results I've seen posted?

Processor: AMD FX 6100
RAM: 8GB (Corsair DDR3)
OS: Windows 8.1
SSD: Samsung SSD 840 EVO (250GB)

My SSD Benchmark Results: http://postimg.org/image/ni5unmfrx/

All of the other screenshots I've seen have both the Sequestial Read/Write and the Random Read/Write meters shooting through the roof. However, my Random Read/Write scores seem to be quite poor compared to theirs.

Can anybody help point me in the right direction in regards to why I'm scoring so low on these particular benchmarks?

Usually, benchmarks are influenced by other components. Would you mind posting your specifications here or messaging me?
For example, poor CPU performance will obviously slow down everything else.

Sure will! What's the best method for doing so? Do you prefer the spec log from a specific application, or? I apologize. I'm quite new here.

 

[mbarbantini]

Does anybody know why my SSD would be performing so poorly compared to all of the other results I've seen posted?

Processor: AMD FX 6100
RAM: 8GB (Corsair DDR3)
OS: Windows 8.1
SSD: Samsung SSD 840 EVO (250GB)

My SSD Benchmark Results: http://postimg.org/image/ni5unmfrx/

All of the other screenshots I've seen have both the Sequestial Read/Write and the Random Read/Write meters shooting through the roof. However, my Random Read/Write scores seem to be quite poor compared to theirs.

Can anybody help point me in the right direction in regards to why I'm scoring so low on these particular benchmarks?

Usually, benchmarks are influenced by other components. Would you mind posting your specifications here or messaging me?
For example, poor CPU performance will obviously slow down everything else.

Sure will! What's the best method for doing so? Do you prefer the spec log from a specific application, or? I apologize. I'm quite new here.

Are all your drivers up to date?
Is your PSU powerful enough?
What about your GPU?
For the specs; don't you know them already? I understood that upgraded your pc right? You didnt list your gpu, psu and cpu cooler

 

[stoatwblr]

Sure will! What's the best method for doing so? Do you prefer the spec log from a specific application, or? I apologize. I'm quite new here.

For starters: The benchmark program must be the ONLY thing running on the computer. Leaving _anything_ else running is likely to impact benchmark performance.

Secondly: What is the actual state of the drive itself?

Most published benchmarks are for "fresh" (or freshly erased) SSDs. Some are run at "50% full" (whewre the definition varies). If you have a nearly full drive then performance will often falter badly.

Published testing methodology is important and it must be repeatable

Thirdly: do you have "fresh" drive data to compare against? Every time you change the hardware configuration you need to re-benchmark (before/after comparisons).

This is important as a lot of sata controllers are pretty rotten and become major bottlenecks for a SSD.

It's good that Toms are doing long-term tests, but you do need to take into account that they erase the drives for each test cycle to ensure they're in a known state (this is the only way to run long term tests). What that means is that "your milage may vary", especially if your drive is low on available blocks.

"real world" tests are often very different to lab ones. It's hard to replicate a real world environment and real world loading in labs.

That means: For all intents and purposes, a benchmark is merely a nod in the direction of whtat you'll see in service - and running benchmarks on a "real world" system often gives results at odds with the actual performance seen for any given application (sometimes benchmarks say it should be great and it's not, or vice vsersa)

_Everything_ is about compromise.

As a case in point:

I admin academic scientific cluster systems with thousands of Tb of shared storage and hundreds of CPUs. As loads get cranked up, they often fail in ways that the software authors never even dreamed of, let alone devised tests for. This makes deploying new technology a white knuckle ride, long after the equipment has bedded in - and of course people deploying new software seldom bother checking with us beforehand, so the first we know of a new failure mode is alarms and/or angry complaints that XYZ doesn't work

A lot of the time we find that what they're trying to do works on a single server/drive system but scales to larger sizes or multiuser systems spectacularly badly - so we have to teach them how to optimize for what's effectively cloud computing and storage. Having gone through that optimization process you may find that performance on the original testbed system is better - but generally it's significantly worse, because the software no longer monopolises all available resources on the smaller system.

On a single-user machine running batch processing, noone cares if interactive performance suffers, but if 30 other people are using the systems and interactive performance suffers (or other processes stop completely) it's a different matter. SImlar things apply to disks - both spinning rust and SSD

 

[IceA_Few]

I bought the Samsung 840 Evo 1tb ssd. The cloning software did not work for me. My ssd would only be seen as a NON OS & would not boot. Samsung does not support this type of issues, only Samsung OEM PCs & was directed back to place of purchase. I had to use third party cloning software outside of my OS, (another computer), to get my ssd to work as a boot drive. Now that it works, I love the speed of it in my gaming PC. I also installed a WD Black dual drive in my laptop at the same time, it was effortless in comparison. WD offers tons of support including install videos & contact info in case something goes wrong.

 

[IceA_Few]

I bought the Samsung 840 Evo 1tb ssd. The cloning software did not work for me. My ssd would only be seen as a NON OS & would not boot. Samsung does not support this type of issues, only Samsung OEM PCs & was directed back to place of purchase. I had to use third party cloning software outside of my OS, (another computer), to get my ssd to work as a boot drive. Now that it works, I love the speed of it in my gaming PC. I also installed a WD Black dual drive in my laptop at the same time, it was effortless in comparison. WD offers tons of support including install videos & contact info in case something goes wrong.