Spotted: Samsung's 850 EVO SSD
Tags:
-
SSD
-
Samsung SSD
-
Samsung
Last response: in News comments
N.Broekhuijsen
September 9, 2014 1:17:49 PM
More about : spotted samsung 850 evo ssd
-
Reply to N.Broekhuijsen
jasonelmore
September 9, 2014 1:31:47 PM
Related resources
- 1TB Samsung EVO 850 PRO's RAPID - Forum
- 1TB Samsung EVO 850 Pro vs 1TB Samsung EVO 840 - Forum
- Samsung 850 PRO vs 840 EVO - Forum
- Difference between Samsung 840 evo vs 850 pro - Forum
- Samsung 850 pro or 840 evo. - Forum
turkey3_scratch
September 9, 2014 1:37:56 PM
dovah-chan
September 9, 2014 1:46:03 PM
Amdlova
September 9, 2014 2:12:48 PM
kinggremlin
September 9, 2014 2:20:44 PM
alidan
September 9, 2014 3:06:31 PM
Quote:
Here's a prediction you can take to the bank. It will be imperceptibly faster than its predecessor. SSD performance "innovation" has pretty much ground to a halt. There isn't anything that can really be done at this point to change the user experience for the better.use 2 sata ports and software that sets it up as a raid 0, there, just increased the speed by two... could probably have 4 way without to much extra cost too.
-
Reply to alidan
m
-1
l
Mac266
September 9, 2014 3:33:41 PM
littleleo
September 9, 2014 3:43:21 PM
dovah-chan
September 9, 2014 3:50:46 PM
kinggremlin
September 9, 2014 4:02:51 PM
Quote:
use 2 sata ports and software that sets it up as a raid 0, there, just increased the speed by two... could probably have 4 way without to much extra cost too.
You have potentially doubled throughput, you have not doubled performance. What makes the SSD user experience so much better than traditional hard drives is not the higher throughput but the much quicker access time. Access time with SSD's is to the point it is nearly impossible to improve performance by a perceptible amount for every day computing.
-
Reply to kinggremlin
m
3
l
Tanquen
September 9, 2014 4:27:09 PM
alextheblue
September 9, 2014 5:00:52 PM
Quote:
Here's a prediction you can take to the bank. It will be imperceptibly faster than its predecessor. SSD performance "innovation" has pretty much ground to a halt. There isn't anything that can really be done at this point to change the user experience for the better.
-
Reply to alextheblue
m
5
l
childofthekorn
September 9, 2014 5:22:06 PM
Quote:
Quote:
Here's a prediction you can take to the bank. It will be imperceptibly faster than its predecessor. SSD performance "innovation" has pretty much ground to a halt. There isn't anything that can really be done at this point to change the user experience for the better.
stop typing out this smut.
-
Reply to childofthekorn
m
-3
l
sewalk
September 9, 2014 5:39:15 PM
Quote:
we need sata 4 or sas... sata 3 is too slow!I dare you to actually sit down and document the difference in the user experience between using a Samsung 800-series SSD on a SATA-II port vs SATA-III. Then you can decide if we want Samsung to focus on developing a faster interface or if we'd rather have them focus on increasing density and economy so we can have TB SSDs as cheap as platter drives.
-
Reply to sewalk
m
4
l
BulkZerker
September 9, 2014 11:25:03 PM
Quote:
Quote:
we need sata 4 or sas... sata 3 is too slow!I dare you to actually sit down and document the difference in the user experience between using a Samsung 800-series SSD on a SATA-II port vs SATA-III. Then you can decide if we want Samsung to focus on developing a faster interface or if we'd rather have them focus on increasing density and economy so we can have TB SSDs as cheap as platter drives.
Something about PCIx interface based SSDs...
-
Reply to BulkZerker
m
-1
l
nitrium
September 10, 2014 1:04:56 AM
Since the vast majority of transfers on Windows are random reads of small files with queue depths of 1, we are seeing increasingly small real world differences between old SSDs and the latest models INCLUDING PCIe/M2 based drives. Is there any drive capable of delivering even a pathetic 50-100 MB/sec for random 4K reads at QD1?
-
Reply to nitrium
m
0
l
xyriin
September 10, 2014 5:37:32 AM
dovah-chan
September 10, 2014 6:32:13 AM
The reason the 840 EVO slows down is because it runs out of turbo cache space and then lowers the transfer speed. The 840 EVO isn't the best long term transfer performer and TLC is inherently less hardy or long lasting as MLC. Really turbo cache is both a blessing and a curse.
But the reason the drive became so popular is because of its cheap price, good overall performance, and nice feature set. Mix in that familiar Samsung brand name which naturally draws consumers towards your products and you've got a hot seller.
Also all drives generally reduce in speed after being over 50% full. Well now that might be a myth as I don't think no one has done any testing on that notion for SSDs but it's generally true for HDDs.
But the reason the drive became so popular is because of its cheap price, good overall performance, and nice feature set. Mix in that familiar Samsung brand name which naturally draws consumers towards your products and you've got a hot seller.
Also all drives generally reduce in speed after being over 50% full. Well now that might be a myth as I don't think no one has done any testing on that notion for SSDs but it's generally true for HDDs.
-
Reply to dovah-chan
m
-2
l
Tanquen
September 10, 2014 1:09:07 PM
Quote:
The reason the 840 EVO slows down is because it runs out of turbo cache space and then lowers the transfer speed. The 840 EVO isn't the best long term transfer performer and TLC is inherently less hardy or long lasting as MLC. Really turbo cache is both a blessing and a curse. But the reason the drive became so popular is because of its cheap price, good overall performance, and nice feature set. Mix in that familiar Samsung brand name which naturally draws consumers towards your products and you've got a hot seller.
Also all drives generally reduce in speed after being over 50% full. Well now that might be a myth as I don't think no one has done any testing on that notion for SSDs but it's generally true for HDDs.
I’m not 100% sure as I’m old and forgetful but… No, the EVOs slow down the moment you put data on them, some nonsense about the 3bit cells. The only time I got the 400MBs-ish reads was on the parts of the drive with no data. The over 50-70% thing is an SSD thing, hard disc drives don’t have that issue.
So, you are paying way more but:
It’s really slow with small files, it slows down the moment you put data on it, It’s un realistically slow after you’ve used it for a time and or it get 50%+ full, it’s not a real TB, even though you have to or should overprovision it and make it even smaller. So your 1TB drive is not 1TB it’s 900GB before you overprovision and you can’t use more than half of it without it slowing to a crawl. Then people think the SATA ports are too slow. Not even.
They lie (about all HDs) and say a GB is 1000MB, it is not! It is and always will be 1024MB. Even though Apple caved in a recent OS update as they were tired of support calls from people wanting to know why their 500GB drive was not 500GB. So your 1TB drive is not 1TB it’s 900GB. I’m old enough to remember when they lied (made up their own rules) and said something like, well a MB is 1024K until 100MBs or some such BS and then it’s 1000k and that’s why when you by RAM you actually get 32MB when you buy 32MB. Oh really, so when the file is in RAM it’s 1MB but when I save it to the HD it’s now 1.024MB. Right!
-
Reply to Tanquen
m
-1
l
xyriin
September 11, 2014 7:19:20 AM
Tanquen said:
I’m not 100% sure as I’m old and forgetful but… No, the EVOs slow down the moment you put data on them, some nonsense about the 3bit cells. The only time I got the 400MBs-ish reads was on the parts of the drive with no data. The over 50-70% thing is an SSD thing, hard disc drives don’t have that issue.So, you are paying way more but:
It’s really slow with small files, it slows down the moment you put data on it, It’s un realistically slow after you’ve used it for a time and or it get 50%+ full, it’s not a real TB, even though you have to or should overprovision it and make it even smaller. So your 1TB drive is not 1TB it’s 900GB before you overprovision and you can’t use more than half of it without it slowing to a crawl. Then people think the SATA ports are too slow. Not even.
They lie (about all HDs) and say a GB is 1000MB, it is not! It is and always will be 1024MB. Even though Apple caved in a recent OS update as they were tired of support calls from people wanting to know why their 500GB drive was not 500GB. So your 1TB drive is not 1TB it’s 900GB. I’m old enough to remember when they lied (made up their own rules) and said something like, well a MB is 1024K until 100MBs or some such BS and then it’s 1000k and that’s why when you by RAM you actually get 32MB when you buy 32MB. Oh really, so when the file is in RAM it’s 1MB but when I save it to the HD it’s now 1.024MB. Right!
Some mixed information in here.
First off normal hard drives do slow down the closer to capacity they reach. The reason for this is because of the way conventional hard drives seek to read and write information on the platters. When the drive is empty they can simply write one bit right after the other. When the drive is closer to getting filled you may be able to write a few bits before you have to skip over existing written data to find the next empty spot. Similarly, when reading this information later you have to make the same skips in the sequential path resulting in wasted time. This is also why defragmenting your hard drive results in faster read speeds. Additionally, if you want to be able to defragment a hard drive you have to leave empty space as well. Additionally, hard drives experience further slow down the more full they are because as the information ends up further out on the platter, the seek takes longer to get out there to even start reading or writing your information even if you don't have a fragmented drive.
SSDs don't suffer the seek slowdown that hard drives do when they hold more information. What SSDs do suffer from is wear leveling. What causes this slowdown is when a SSD attempts to read information with worn NAND if it fails the first time it attempts to read again, which takes additional time. However, modern SSDs eliminate this issue with over provisioning, wear leveling, and garbage collection. While SSDs don't need to defrag, they still need this same space for a different reason and that is to perform all the maintenance tasks like garbage collection. It's also key to note that NAND wear is caused by write operations. Unless you're using SSDs in a server capacity or intensive disk writing like maybe video production your never going to see appreciable wearout of the NAND in your drive. Casual use or gaming should never experience SSD slowdown (inside of a 10-15 year period) as long as you have a quality SSD with the firmware and an OS that properly supports the maintenance functions.
So in summary to maintain top SSD speed you need to leave the 10% or so free for the maintenance operations and you're good to go with no slowdown. For a hard drive you need to leave the 10% or so free for defragging and you still get to deal with slowdowns as the drive is filled to capacity. End result both storage mediums need 10% overhead (neither wins) but conventional hard drives still experience slowdown.
As for listed drive size that is just marketing BS taking advantage of the 1k to 1024 difference. It's long been practice to manufacturers to list physical space and not formatted capacity.
-
Reply to xyriin
m
0
l
dovah-chan
September 11, 2014 8:19:29 AM
http://www.anandtech.com/show/7173/samsung-ssd-840-evo-...
Here is a refresher on why the 840 EVO suffered from a slowdown during large file transfers. It is true in part that is due to TLC and that is the core issue.
Here is a refresher on why the 840 EVO suffered from a slowdown during large file transfers. It is true in part that is due to TLC and that is the core issue.
-
Reply to dovah-chan
m
0
l
xyriin
September 11, 2014 9:17:54 AM
dovah-chan said:
http://www.anandtech.com/show/7173/samsung-ssd-840-evo-...Here is a refresher on why the 840 EVO suffered from a slowdown during large file transfers. It is true in part that is due to TLC and that is the core issue.
That is perfectly valid information however you'll notice that it only affects write performance. For normal users writes are a very insignificant portion of typical drive use. Additionally, this is a slowdown of the 'turbo' mode not in inherent slowdown of the NAND itself. Think of this as an artificial technology like RAID striping where you increase effective speed but still don't exceed the speed of the actual medium. It's just like a buffer on any other drive. If you read the graph you'll also see that even on the 120GB model you are able to write for approximately 5 seconds at 500MB/s. How often are you doing sequential WRITES on a drive with more than 2.5GB? The answer is never unless you're doing video production work, in which case you're using a higher capacity SSD anyway. Those doing that case specific video production work using the same technology 1TB model can perform 11GB of sequential writes before capping the buffer.
The only file sizes a normal user will ever see that could exceed that buffer capacity would be game installs or transfer of HD video. Let's say you install a 20GB game via Steam. Doesn't matter, what is your internet speed? 25/5, 50/25, 100/50? Even 100Mb/s internet means you download at a peak of 12.5MB/s...which will never reach the capacity of your SSD. How about gigabit internet? Again you peak at 125MB/s still well below the buffer. Installing from a disc won't matter either as your optical drive stupidly slow compared to the SSD. Want to transfer files across your local network? Again same scenario as even ethernet has you running at 100/1000Mb/s (12.5/125MB/s). Have a large movie on your computer that you want to transfer from one drive to another? Well if you're transferring from a normal hard drive it doesn't matter, the speed cap on a mechanical disk won't exceed the buffer. Now going from one SSD to another SSD would present a potential slowdown but how large is your file? Are you writing enough sequential data to cap the buffer...probably not.
The 840 EVO slowdown is mainly a theoretical problem. As the article clearly states the buffer size varies from 2.5% of 120GB model to 1.2% of the 1TB model. So to maintain performance on a mechanical drive for defragging you need to save approximately 10%, garbage collection plus a turbo write buffer is going to come in well under that. So a SSD needs less overhead and still doesn't have the read slowdowns like a mechanical drive. Sure a SSD could experience a drop in write speed in extreme benchmark situations but it's not a normal scenario.
-
Reply to xyriin
m
1
l
littleleo
September 12, 2014 3:26:06 PM
Tanquen
September 13, 2014 2:28:03 PM
xyriin said:
Some mixed information in here.
First off normal hard drives do slow down the closer to capacity they reach.
Tanquen said:
Not if you defrag regularly.
This is also why defragmenting your hard drive results in faster read speeds. Additionally, if you want to be able to defragment a hard drive you have to leave empty space as well.
Tanquen said:
Again, not if you defrag regularly.
Additionally, hard drives experience further slow down the more full they are because as the information ends up further out on the platter, the seek takes longer to get out there to even start reading or writing your information even if you don't have a fragmented drive.
Tanquen said:
It's not that bad, one drive I have peaks at a real 160MBs and slowly goes down to 125MBs at the end of the drive. Where as my EVO goes to thrashing 25-400MBs the moment I put data on it. Only getting a stable 425MBs at the end of the drive with no data on it.
SSDs don't suffer the seek slowdown that hard drives do when they hold more information.
Tanquen said:
They just seem to have it all the time as most files are small and you normally need to read many small files.
What SSDs do suffer from is wear leveling. What causes this slowdown is when a SSD attempts to read information with worn NAND if it fails the first time it attempts to read again, which takes additional time. However, modern SSDs eliminate this issue with over provisioning, wear leveling, and garbage collection. While SSDs don't need to defrag, they still need this same space for a different reason and that is to perform all the maintenance tasks like garbage collection. It's also key to note that NAND wear is caused by write operations. Unless you're using SSDs in a server capacity or intensive disk writing like maybe video production your never going to see appreciable wearout of the NAND in your drive. Casual use or gaming should never experience SSD slowdown (inside of a 10-15 year period) as long as you have a quality SSD with the firmware and an OS that properly supports the maintenance functions.
Tanquen said:
Samsung don’t have quality firmware? My 1TB drive is 907-ish GB because of lies. Then you have to overprovision down to 800GB or so depending on who you talk to and then it slows down anyway as you well, try to use it. I had to bring my drive back down to 500GB to get back the original 25-400MBs thrashing speed vs the 20-75MBs thrashing speed because the drive was more than 50% full.
So in summary to maintain top SSD speed you need to leave the 10% or so free for the maintenance operations and you're good to go with no slowdown. For a hard drive you need to leave the 10% or so free for defragging and you still get to deal with slowdowns as the drive is filled to capacity. End result both storage mediums need 10% overhead (neither wins) but conventional hard drives still experience slowdown.
Tanquen said:
I think it’s a little more than 10% for SSDs, closer to 50%. Just the overprovisioning recommendation is like 10-15%. If you keep your HDD drive defragged as you use it there will be little to nothing to defrag when you’re at 90% and you’ll still have good speeds. Not so with the SSDs. You can find articles were they tell you to copy everything off the drive reformat and then copy everything back just to get the read speeds back up. Maybe just some other SSDs have this issue but again I had to delete stuff on my 1TB… oh I mean 907-ish GB… well 838GB after overprovisioning, Evo down to 520GBs or so to keep the speeds up and they still are not that great. Listing 400-550MBs reads speeds on these SSDs is really pushing it. After some average and reasonable use you’ll never see those numbers. The moment you put data on the EVO you’ll not see it save for some random spikes.
More importantly, this information is not brought up enough when they advertise and even review SSDs.
As for listed drive size that is just marketing BS taking advantage of the 1k to 1024 difference. It's long been practice to manufacturers to list physical space and not formatted capacity.
Tanquen said:
I know just tired of it and the 60" Class HDTVs. There shouldn’t need to be a law that they tell you what the useable space is. They know what we use them for and what we want out of them but they still can’t resist making it sound like they are offering more than they really are. I’m going to start selling cars that get 158.4MPG. What the rubes don’t know is that I think a mile is 1000 feet not that 5280 feet BS. Then when they complain that it only gets 30MPG I’ll say no that’s 30MiPG you stupide consumer, the car gets 158.4MPG, go away.
-
Reply to Tanquen
m
-1
l
BlackAle
September 13, 2014 6:52:19 PM
Quote:
Additionally, hard drives experience further slow down the more full they are because as the information ends up further out on the platter, the seek takes longer to get out there to even start reading or writing your information even if you don't have a fragmented drive.Yes, mechanical HD's slow the more they are full, but you have it backwards. Considering that data density is equal across the platter and the fact HD's read from the edge inwards, seek times will be higher as you move further inwards, though more crucially for mechanical HD's the transfer rate will drop substantially.
-
Reply to BlackAle
m
-1
l
xyriin
September 13, 2014 8:22:33 PM
Ok, information time again.
Yes, hard drives slow down the more full they get even if they are defragmented. This is why seek time for hard drives isn't a static value, it is in fact a changing value. If you need further proof look up the definitions of track to track seek and full stroke seek. The TLDR is that it takes longer for the head to reach the the furthest point from the head. The only thing defragmenting does is limit the back and forth seek time on a single file. It doesn't stop back to forth seeking between multiple files. So yes, the more full the drive is the more back and forth seek time you waste between different files. Additionally, think about a disc (like a hard drive platter) and spin it at a constant speed. Now imagine tracking the drive head across that platter from one point to another.
If your EVO is thrashing the moment you put data on it you've got your OS misconfigured, a junk controller, or a bad drive. I've got one EVO 840 500GB, one EVO 850 1TB, and two Crucial m500 960GB. None of them have significant slowdowns even at 90%+ capacity. The reason SSDs have a stable seek time unlike mechanical drives is due to their architecture. Your comments about reading lots of small files don't fly either. SSDs excel at operations like that. In fact it's why they are also rated in IOPS in addition to just read and write speed and also are preferred for servers.
Everything you're describing sounds 100% like a TRIM issue which is not a problem with SSD or NAND technology but instead an OS or SATA controller/driver issue. Your experiences of overprovisioning also don't match any hardware reviews either. Specifically AnandTech has a test they perform during their SSD testing where they test write speeds after disabling TRIM. In this engineered failure condition they do get minimum speeds around 30MB/s, however after one TRIM pass full performance is immediately restored. But again, read speed never dropped, only write speed.
If you're getting exactly the conditions you're describing it is definitely a configuration issue with your system.
Yes, hard drives slow down the more full they get even if they are defragmented. This is why seek time for hard drives isn't a static value, it is in fact a changing value. If you need further proof look up the definitions of track to track seek and full stroke seek. The TLDR is that it takes longer for the head to reach the the furthest point from the head. The only thing defragmenting does is limit the back and forth seek time on a single file. It doesn't stop back to forth seeking between multiple files. So yes, the more full the drive is the more back and forth seek time you waste between different files. Additionally, think about a disc (like a hard drive platter) and spin it at a constant speed. Now imagine tracking the drive head across that platter from one point to another.
If your EVO is thrashing the moment you put data on it you've got your OS misconfigured, a junk controller, or a bad drive. I've got one EVO 840 500GB, one EVO 850 1TB, and two Crucial m500 960GB. None of them have significant slowdowns even at 90%+ capacity. The reason SSDs have a stable seek time unlike mechanical drives is due to their architecture. Your comments about reading lots of small files don't fly either. SSDs excel at operations like that. In fact it's why they are also rated in IOPS in addition to just read and write speed and also are preferred for servers.
Everything you're describing sounds 100% like a TRIM issue which is not a problem with SSD or NAND technology but instead an OS or SATA controller/driver issue. Your experiences of overprovisioning also don't match any hardware reviews either. Specifically AnandTech has a test they perform during their SSD testing where they test write speeds after disabling TRIM. In this engineered failure condition they do get minimum speeds around 30MB/s, however after one TRIM pass full performance is immediately restored. But again, read speed never dropped, only write speed.
If you're getting exactly the conditions you're describing it is definitely a configuration issue with your system.
-
Reply to xyriin
m
0
l
BlackAle
September 13, 2014 9:31:05 PM
xyriin said:
Ok, information time again.Yes, hard drives slow down the more full they get even if they are defragmented. This is why seek time for hard drives isn't a static value, it is in fact a changing value. If you need further proof look up the definitions of track to track seek and full stroke seek. The TLDR is that it takes longer for the head to reach the the furthest point from the head. The only thing defragmenting does is limit the back and forth seek time on a single file. It doesn't stop back to forth seeking between multiple files. So yes, the more full the drive is the more back and forth seek time you waste between different files. Additionally, think about a disc (like a hard drive platter) and spin it at a constant speed. Now imagine tracking the drive head across that platter from one point to another.
You've still got it backwards. mechanical hard drives start at the edge of the disc not the centre. The seek time will be lower at the edge, as the data density over the platter(s) stays the same, so the head(s) require less movement further from the centre.
-
Reply to BlackAle
m
-1
l
xyriin
September 14, 2014 7:35:05 AM
BlackAle said:
You've still got it backwards. mechanical hard drives start at the edge of the disc not the centre. The seek time will be lower at the edge, as the data density over the platter(s) stays the same, so the head(s) require less movement further from the centre.I don't think you read what you quoted...
"hard drives experience further slow down the more full they are because as the information ends up further out on the platter, the seek takes longer to get out there to even start reading or writing your information"
"The TLDR is that it takes longer for the head to reach the the furthest point from the head."
I think both those are pretty clear. It doesn't matter where the head starts, inner or outer. Data is written starting at the head location and then moving to the furthest location. When the drive is empty your read and write is where the head is starting. As the drive gets filled you now have to travel further from the starting location in order to write to a blank space (even if defragmented). Again, the more full the drive is the greater time between file seeks as you have to travel over more of the disk to get what you need. You seem to be under the false impression that a hard drive can ignore seek time and the only delay is read/write time.
In an empty disc you don't have to seek far at all to write new data. Again reading data takes no time at all as the seek time is the shortest.
Closest X - - - - - - - - - - - - - - - - - Furthest
In a fragmented disc your seek time for starting write may actually be shorter than a defragmented disk. However, the read time seek is increased because you have to pass over empty space.
Closest X X X - X - X X X X - - X X - X - X Furthest
In a defragmented disc your seek time for writing is going to be the maximum possible. However, the read time seek will be the minimum possible. Keep in mind though that to get to the outer part of the disk it will still take more time as the head has to move further from the starting location.
Closest X X X X X X X X X X X X - - - - - - Furthest
It doesn't matter what the physical arrangement of the head is. As you have to pass over more blocks of data to get what you want it takes longer. It's simply physics. Hard drives can't magically jump to the bit they need to read, it's why they call it a seek time.
-
Reply to xyriin
m
0
l
Tanquen
September 14, 2014 8:52:24 AM
Ok, the real none SSD lover information time again! Come on, I’m just talking about my experience. SSDs have issues, the 840 EVO more so.
Your information is not quite right. Yes, what you are describing can happen but your saying that everything you install will always need to read all parts of the drive. If I keep the drive defragged and it’s almost full and then install a program, most all of the programs data will be together. Not all randomly spread across the drive platters. The speed drop is not that bad across the HDD drive. You have to benchmark to see it. Especially compared to the EVO slowdowns for putting data on the drive and having it too full.
Well, it’s an Intel X70 SATA controller and I've had an ASRock and 2 ASUS X79 boards do the same thing. I would love to know the magic (your holding it wrong) setting that is the cause. Windows 7 with TRIM enabled, AHCI mode enabled, the Samsung recommended 10% over provisioning, OS Optimization settings on and correct 4k alignment. I’ve even tried the manual Samsung TRIM, Performance Optimization.
Below is what happens when the drive gets close to 90%. You don't need to benchmark to feel this. From 90% to 100% you can see the speeds when the drive has no data in that area. From 75% to 90% you can see the speeds once you put data on the drive. Format the drive and put Windows on it and that is what you see were there is data. I changed no setting but removed data until the rest of the drive with data all looked like the 75% to 90% area. Still not ideal but not crap. I had to remove around 300GB for that to happen. I and others saw the same when we went from our 830 drives to the 840EVO.
![]()
Maybe, but others saw the same thing when the 840EVO first came out, so they are all bad?
Sorry, but that is what happens. Yes each SSD seek is very fast and stable but when you go to read the files it’s not the whole story. SSDs may benchmark better but when you try to read a large number of small files the reads drop down to 10-20MBs on many drives like my EVO. Like folks that go nuts over 4 channel DDR3 – 3000 RAM and look at all the crazy benchmarks and the you run you apps and see very little (can only see it when you bench it) improvement.
Your information is not quite right. Yes, what you are describing can happen but your saying that everything you install will always need to read all parts of the drive. If I keep the drive defragged and it’s almost full and then install a program, most all of the programs data will be together. Not all randomly spread across the drive platters. The speed drop is not that bad across the HDD drive. You have to benchmark to see it. Especially compared to the EVO slowdowns for putting data on the drive and having it too full.
xyriin said:
If your EVO is thrashing the moment you put data on it you've got your OS misconfigured, a junk controller… Below is what happens when the drive gets close to 90%. You don't need to benchmark to feel this. From 90% to 100% you can see the speeds when the drive has no data in that area. From 75% to 90% you can see the speeds once you put data on the drive. Format the drive and put Windows on it and that is what you see were there is data. I changed no setting but removed data until the rest of the drive with data all looked like the 75% to 90% area. Still not ideal but not crap. I had to remove around 300GB for that to happen. I and others saw the same when we went from our 830 drives to the 840EVO.

xyriin said:
a bad drive.xyriin said:
The reason SSDs have a stable seek time unlike mechanical drives is due to their architecture. Your comments about reading lots of small files don't fly either.-
Reply to Tanquen
m
0
l
xyriin
September 15, 2014 5:30:54 AM
Your first statement sums it up. It's your experience, not the normal experience. It's also not the experience of a single hardware reviewer. In fact the only mention of the issue you are describing was manufactured by reviewers who disable TRIM to check TRIM performance. The only time I see it mentioned is your anecdotal evidence.
Thank you for admitting there is a speed drop for mechanical hard drives even though you claim it's 'not that bad'. Not every program you install will not need to read all parts of the drive. However, your mechanical hard drive can't start reading where you install Office, Photoshop, or the latest game. Every time you run one of those programs your mechanical drive has to seek to where that software is installed. The first program you install on a fresh drive will have the best seek time. But install a program when your drive is almost full and you have the max seek time even if you're defragged when you install.
Additionally, if you want to talk about wasted time...defragging itself is the LARGEST waste of time. First you can't do anything when you're defragmenting the drive. Second, even if you scheduled it overnight you're wasting a lot of power in a futile attempt to keep your hard drive operating at close to peak levels. Also, the more full your drive is, the longer the defragmenting process takes. Sure, SSDs do garbage collection to maintain performance which is the equivalent of defragmenting but they do it in parallel unlike a hard drive in which defragmenting is a 100% operation.
Based on professional hardware reviews of the EVO 840 and your descriptions of what you're experiencing one of two things is true. Either you're lying or you have a fault in your configuration. Because quite frankly I don't believe you more than dozens of reviews from professionals or my own personal experience. So remove my results, remove yours, and the professional reviews still match what I'm saying.
As for my personal configuration of SSDs I use them out of the box and I ensure exactly three things: TRIM enabled, latest SSD firmware, latest SATA controller drivers. I don't even have to use the manufacturer software, and I definitely never have to keep it running. Only time I use the software is if I want to check advanced SMART data on the SSD which may be manufacturer specific. I've used an Intel X-25M, Crucial C300, Samsung 830 Pro, 2x Samsung EVO 840, and 2x Crucial m500 and I've never had a slowdown with any of them. Additionally, I've never enabled over-provisioning and never had one drop below 99% NAND life. This of course is my own anecdotal evidence, however it is backed up years of SSD research, data, and reviews.
Also, update your HD Tune. You're running v2.55. They made major changes after version 4 specifically for SSDs. You're not even using a valid piece of software to benchmark your SSD. Next, verify you have the latest BIOS update for your motherboard. Then, verify you have the latest firmware update for your SSD. It also wouldn't hurt to try out some other benchmarking software as well since you may have a software conflict on your system with HD Tune as well. Crystal DiskMark and AS SSD Benchmark are also valid testing tools. If you're still having issues look at your SATA controllers. I've had a lot of discussions during benching and testing with others using similar hardware and in at least half the cases performance abnormalities were related in some fashion to the controller (typically driver). It's funny how drive performances tend to become identical once all the variables are removed due to poor configuration.
The others who saw the same thing as yourself were end users not the professional reviewers. If you dig into the support forums most of them fixed or at least improved their drive close to the 'normal' performance by evaluating and applying the fixes I've mentioned above.
As for small files which you keep mentioning the following test was performed...
"The test simulated a real life access to a database, with a transfer request size of 8 kB, 35/65% of random/sequential distribution; 65%/35% of read/write distribution and 64 outstanding IOps. The test ran for 5 minutes and was repeated 6 times per disk. Additionally, the test were conducted over the last 10% of each drive because it is the slowest area in the disks, that means that it is a limiting factor of performance hence a representative (an more real) parameter of the behavior of the disks in a critical condition, that is, with the drives almost full."
http://www.8088.net/images/IOps/IOps_mean_comparison_EN...
So you can say SSDs fail at small files but the factual reality is that they crush mechanical hard drives at those operations even when the drive is close to being filled. In fact it isn't even close. So if you're struggling with small files transfers on a SSD compared to a mechanical drive there is an issue present besides the technology.
Thank you for admitting there is a speed drop for mechanical hard drives even though you claim it's 'not that bad'. Not every program you install will not need to read all parts of the drive. However, your mechanical hard drive can't start reading where you install Office, Photoshop, or the latest game. Every time you run one of those programs your mechanical drive has to seek to where that software is installed. The first program you install on a fresh drive will have the best seek time. But install a program when your drive is almost full and you have the max seek time even if you're defragged when you install.
Additionally, if you want to talk about wasted time...defragging itself is the LARGEST waste of time. First you can't do anything when you're defragmenting the drive. Second, even if you scheduled it overnight you're wasting a lot of power in a futile attempt to keep your hard drive operating at close to peak levels. Also, the more full your drive is, the longer the defragmenting process takes. Sure, SSDs do garbage collection to maintain performance which is the equivalent of defragmenting but they do it in parallel unlike a hard drive in which defragmenting is a 100% operation.
Based on professional hardware reviews of the EVO 840 and your descriptions of what you're experiencing one of two things is true. Either you're lying or you have a fault in your configuration. Because quite frankly I don't believe you more than dozens of reviews from professionals or my own personal experience. So remove my results, remove yours, and the professional reviews still match what I'm saying.
As for my personal configuration of SSDs I use them out of the box and I ensure exactly three things: TRIM enabled, latest SSD firmware, latest SATA controller drivers. I don't even have to use the manufacturer software, and I definitely never have to keep it running. Only time I use the software is if I want to check advanced SMART data on the SSD which may be manufacturer specific. I've used an Intel X-25M, Crucial C300, Samsung 830 Pro, 2x Samsung EVO 840, and 2x Crucial m500 and I've never had a slowdown with any of them. Additionally, I've never enabled over-provisioning and never had one drop below 99% NAND life. This of course is my own anecdotal evidence, however it is backed up years of SSD research, data, and reviews.
Also, update your HD Tune. You're running v2.55. They made major changes after version 4 specifically for SSDs. You're not even using a valid piece of software to benchmark your SSD. Next, verify you have the latest BIOS update for your motherboard. Then, verify you have the latest firmware update for your SSD. It also wouldn't hurt to try out some other benchmarking software as well since you may have a software conflict on your system with HD Tune as well. Crystal DiskMark and AS SSD Benchmark are also valid testing tools. If you're still having issues look at your SATA controllers. I've had a lot of discussions during benching and testing with others using similar hardware and in at least half the cases performance abnormalities were related in some fashion to the controller (typically driver). It's funny how drive performances tend to become identical once all the variables are removed due to poor configuration.
The others who saw the same thing as yourself were end users not the professional reviewers. If you dig into the support forums most of them fixed or at least improved their drive close to the 'normal' performance by evaluating and applying the fixes I've mentioned above.
As for small files which you keep mentioning the following test was performed...
"The test simulated a real life access to a database, with a transfer request size of 8 kB, 35/65% of random/sequential distribution; 65%/35% of read/write distribution and 64 outstanding IOps. The test ran for 5 minutes and was repeated 6 times per disk. Additionally, the test were conducted over the last 10% of each drive because it is the slowest area in the disks, that means that it is a limiting factor of performance hence a representative (an more real) parameter of the behavior of the disks in a critical condition, that is, with the drives almost full."
http://www.8088.net/images/IOps/IOps_mean_comparison_EN...
So you can say SSDs fail at small files but the factual reality is that they crush mechanical hard drives at those operations even when the drive is close to being filled. In fact it isn't even close. So if you're struggling with small files transfers on a SSD compared to a mechanical drive there is an issue present besides the technology.
-
Reply to xyriin
m
-1
l
Tanquen
September 15, 2014 9:28:48 AM
I think it is the normal experience, it happens over time not a 4 day review session.
Like I said TRIM is enabled.
I stated that before. I admitting nothing.
“Every time you run one of those programs your mechanical drive has to seek to where that software is installed.” It gets to that general area and stays there. Again, it’s not like it has to go all over the drive all the time.
Yes defragging is a waste of time. Again, I’m not trying to say SSDs are worthless just that they (in the real world) have issues too, that get ignored by the companies that make them and the people that review them. There is stuff all over the internet about why my SSDs is so slow and that’s just the folks that notice. That is part of my point. Even if you get the 400-550MBs read, it’s not really like everything loads and everything you do is 4x faster. People don’t notice because in day to day use the speed drop from the 400MBs to the real 150-ish MBs don’t really matter, more so if you never really got 400MBs read to begin with. Most folks aren’t creating macros to open 10 copies of their word processor and then timing it. Even when I had my 830 with a clean install, it stated Windows a little faster than my old PC with an old icky slow poorly defragged almost full HDD. Mostly the clean install really helps but still it was only a few seconds faster. Then you time games loading up or loading new levels and the old PC with an HDD and slower RAM and CPU and GPU is right there with the new one with faster 4 channel RAM and faster CPU and faster GPUs. There are other things going on when you load a game or program. If you really got the 400-500MBs all the time then maybe you would notice more.
I am not lying, you are! No, I’m not lying. I even posted a pic. The fault in my configuration (for the extra bad SSD stuff) was plugging in an 840EVO, as far as I can tell. Again, I’d love to know the magic setting.
“Also, update your HD Tune. You're running v2.55.” That version works just fine. It did not show me anything I was not experiencing. All other drive and flash drive and RAM drive show correctly. The EVO has to read extra data that is not the data you want and I think that is why only it reads like this after data is on the drive. You are kinda trying to have it both/all ways. I’m lying, don’t know what I’m doing, have a bad drive and or the EVO needs special testing software that don’t show its bad read speed.
I have the latest BIOS. As I did on the last 3 X79 motherboards.
“The others who saw the same thing as yourself were end users not the professional reviewers.” Ok but they are people too and miss stuff and more over do not tend to use the equipment more than a few weeks.
I’d do some small file testing but you say it’s only my SSD. Worse on HHDs sure but as good as they say, not that I've seen.
“they crush mechanical hard drives at those operations” Yes and the 850 Pro is “One Giant Leap Forward” I guess Samsung did not read the reviews.
Like I said TRIM is enabled.
I stated that before. I admitting nothing.
“Every time you run one of those programs your mechanical drive has to seek to where that software is installed.” It gets to that general area and stays there. Again, it’s not like it has to go all over the drive all the time.
Yes defragging is a waste of time. Again, I’m not trying to say SSDs are worthless just that they (in the real world) have issues too, that get ignored by the companies that make them and the people that review them. There is stuff all over the internet about why my SSDs is so slow and that’s just the folks that notice. That is part of my point. Even if you get the 400-550MBs read, it’s not really like everything loads and everything you do is 4x faster. People don’t notice because in day to day use the speed drop from the 400MBs to the real 150-ish MBs don’t really matter, more so if you never really got 400MBs read to begin with. Most folks aren’t creating macros to open 10 copies of their word processor and then timing it. Even when I had my 830 with a clean install, it stated Windows a little faster than my old PC with an old icky slow poorly defragged almost full HDD. Mostly the clean install really helps but still it was only a few seconds faster. Then you time games loading up or loading new levels and the old PC with an HDD and slower RAM and CPU and GPU is right there with the new one with faster 4 channel RAM and faster CPU and faster GPUs. There are other things going on when you load a game or program. If you really got the 400-500MBs all the time then maybe you would notice more.
I am not lying, you are! No, I’m not lying. I even posted a pic. The fault in my configuration (for the extra bad SSD stuff) was plugging in an 840EVO, as far as I can tell. Again, I’d love to know the magic setting.
“Also, update your HD Tune. You're running v2.55.” That version works just fine. It did not show me anything I was not experiencing. All other drive and flash drive and RAM drive show correctly. The EVO has to read extra data that is not the data you want and I think that is why only it reads like this after data is on the drive. You are kinda trying to have it both/all ways. I’m lying, don’t know what I’m doing, have a bad drive and or the EVO needs special testing software that don’t show its bad read speed.
I have the latest BIOS. As I did on the last 3 X79 motherboards.
“The others who saw the same thing as yourself were end users not the professional reviewers.” Ok but they are people too and miss stuff and more over do not tend to use the equipment more than a few weeks.
I’d do some small file testing but you say it’s only my SSD. Worse on HHDs sure but as good as they say, not that I've seen.
“they crush mechanical hard drives at those operations” Yes and the 850 Pro is “One Giant Leap Forward” I guess Samsung did not read the reviews.
-
Reply to Tanquen
m
0
l
xyriin
September 15, 2014 10:13:00 AM
Tanquen said:
It gets to that general area and stays there. Again, it’s not like it has to go all over the drive all the time.It does stay there while it's reading a particular file and then it goes back. The only time a mechanical hard drive isn't wasting time seeking is when it's performing a sequential read or write on a defragmented drive. This condition only occurs during a benchmark test or file copy operation. So yes, it does go all over the drive all the time. Even while web browsing you're jumping around back and forth writing to the swap file, writing browser temp files, and accessing Windows system files...not to mention accessing files for anything else you have running in the background.
You are correct in that most steady state operations don't benefit from a SSD. What is affected are program load times, boot times, and any operations involving a lot of file operations. Personally, the only steady state operations I see a noticeable difference in are in-game load times, torrent operations, and video encoding. In-game load time obviously reduces wait time. For torrents it's pretty easy to reach a disk overload condition with a mechanical drive...and this ties into the IOPS performance. With a SSD I don't have to place hard caps on concurrent torrent connections to prevent a disk overload. For the video encoding there isn't too much of a difference on the actual encoding however the SSD allows other operations at the same time which the mechanical drive would never be able to do.
While your version of HD Tune may be 'fine' it probably isn't fully supporting the latest SSDs. From the HD Tune website...
HD Tune 4.60:
- Improved support for SSD
- Added support for more SSDs
- Random access
- Maximum access time is shown
- Added random seek 4 KB test
- Added align options
HD Tune 5.50:
- Added support for more SSD drives
As SSD technology evolves the benchmark software has to as well. HD Tune 2.55 was released in early 2008, almost half a year before somewhat affordable SSDs first hit the mass consumer market. AS and CrystalMark are free, and you can get a free trial of the newer versions of HD Tune as well.
Don't settle for some kind of conflict making your hardware perform below it's potential. Eliminate the conflict and get your money's worth out of your hardware.
-
Reply to xyriin
m
0
l
Tanquen
September 15, 2014 2:40:48 PM
"You are correct in that most steady state operations don't benefit from a SSD. What is affected are program load times, boot times, and any operations involving a lot of file operations."
Even then it’s iffy. My friend and I play games almost every weekend and the load times are almost always with in a second or two of each other. The computers are side by side and you can see when they are both ready. With all the other things going on or wait states SSDs are just not a big deal. That’s with everything being older or slower on my old PC and that poor old HDD seeking all over the place. Either one hosting.
SSDs are best when upgrading an older or very slow power saving 4200-something RMP laptop HDD drives. You see a nice bump there. More so if the laptop has a newer controller. I put my old 830 in my laptop and it makes a difference even though it limited to SATA2 or 150MBs(?). I think the drive can give close to that limit most of the time where the old HDD it had was like maybe 40-50MBs. If the laptop has a faster controller I think you’d only see the benchmarks change in day to day use and I could toggle it from SATA 2 and 3 and you’d not notice.
Yes but all the files for the program should be close by. I think that’s why you just don’t see the big jump in load times. I don’t think the swap file is doing much these days. It’s there and gets used but most folks are not running 512MB of RAM anymore.
I see what you are saying about HD Tune having updates. Its just that the calls to the SATA drive should not need any hacks to get good read on one type of SSD. I’m just thinking they are maybe hacking HD Tune so the tests look good. If you are really thinking that the updated HD Tune would benchmark differently. I think I did even try that back when I go the 840EVO. Could try it again I guess. But that is not going to change how slow it got after being more than 50% full. Hardware venders do that kind of thing all the time, tweaks and hacks to make their hardware look good in benchmark software. Why would you need to change HD Tune to improve and add support for SSDs? Are they SATA drives or not? But at the same time I don’t know if there really is an issue with just my EVO or all EVOs. All my other drives, flash, RAM, disc, USB all look as expected but not the EVO. I’ve talked to others in other forums that saw the same thing with the EVOs and EVOs do read differently (having 3-bit cells) than other SSDs, even the 830 it replaced. So I just don’t think I’m holding it wrong.
So, what else to check and don’t forget that no other drive I have, including the old 830 has the read thrashing. Don’t know about the 50% thing but lots of people seem to have the issue with SSDs.
Points to note:
Windows 7 with TRIM enable. Even tried the manual Samsung TRIM. Still the same. It was like this from day one. Should not be anything for TRIM to do.
AHCI mode is on.
It was noticeably slower when the drive got over 50%. After removing 300GBs of data the speed was better and the HD Tune benchmark showed the same before and after.
New ASUS X79 board with latest drivers. Two other X79 boards did the same thing.
Using the X79 Intel SATA ports. I’ve see time and time again that they are the best to use. The other ports with the add-on controllers always seem to be slower and have extra issues.
Using the Samsung recommended 10% over provisioning.
Using the Samsung OS Optimization settings.
Correct 4k alignment.
Even then it’s iffy. My friend and I play games almost every weekend and the load times are almost always with in a second or two of each other. The computers are side by side and you can see when they are both ready. With all the other things going on or wait states SSDs are just not a big deal. That’s with everything being older or slower on my old PC and that poor old HDD seeking all over the place. Either one hosting.
SSDs are best when upgrading an older or very slow power saving 4200-something RMP laptop HDD drives. You see a nice bump there. More so if the laptop has a newer controller. I put my old 830 in my laptop and it makes a difference even though it limited to SATA2 or 150MBs(?). I think the drive can give close to that limit most of the time where the old HDD it had was like maybe 40-50MBs. If the laptop has a faster controller I think you’d only see the benchmarks change in day to day use and I could toggle it from SATA 2 and 3 and you’d not notice.
Yes but all the files for the program should be close by. I think that’s why you just don’t see the big jump in load times. I don’t think the swap file is doing much these days. It’s there and gets used but most folks are not running 512MB of RAM anymore.
I see what you are saying about HD Tune having updates. Its just that the calls to the SATA drive should not need any hacks to get good read on one type of SSD. I’m just thinking they are maybe hacking HD Tune so the tests look good. If you are really thinking that the updated HD Tune would benchmark differently. I think I did even try that back when I go the 840EVO. Could try it again I guess. But that is not going to change how slow it got after being more than 50% full. Hardware venders do that kind of thing all the time, tweaks and hacks to make their hardware look good in benchmark software. Why would you need to change HD Tune to improve and add support for SSDs? Are they SATA drives or not? But at the same time I don’t know if there really is an issue with just my EVO or all EVOs. All my other drives, flash, RAM, disc, USB all look as expected but not the EVO. I’ve talked to others in other forums that saw the same thing with the EVOs and EVOs do read differently (having 3-bit cells) than other SSDs, even the 830 it replaced. So I just don’t think I’m holding it wrong.
So, what else to check and don’t forget that no other drive I have, including the old 830 has the read thrashing. Don’t know about the 50% thing but lots of people seem to have the issue with SSDs.
Points to note:
Windows 7 with TRIM enable. Even tried the manual Samsung TRIM. Still the same. It was like this from day one. Should not be anything for TRIM to do.
AHCI mode is on.
It was noticeably slower when the drive got over 50%. After removing 300GBs of data the speed was better and the HD Tune benchmark showed the same before and after.
New ASUS X79 board with latest drivers. Two other X79 boards did the same thing.
Using the X79 Intel SATA ports. I’ve see time and time again that they are the best to use. The other ports with the add-on controllers always seem to be slower and have extra issues.
Using the Samsung recommended 10% over provisioning.
Using the Samsung OS Optimization settings.
Correct 4k alignment.
-
Reply to Tanquen
m
0
l
xyriin
September 15, 2014 7:49:15 PM
Part of the testing is calls, but a larger part is how the data is called. When SSDs started being used they actually had to create new tests (and specifically larger tests) because SSDs could hit performance levels not dreamed of on mechanical drives. HD Tune even added updates to support 2TB and 4TB hard drives so changes aren't exclusive to just SSDs. As for benchmarks I wouldn't trust a manufacturer benchmark, but that is why so many 3rd party tools like HD Tune, AS, and CrystalMark exist...for an impartial test.
As for USB flash drives and the like it's important to remember that those types of technology are very static. Firmware is a large part of SSD performance, as evidenced by the fact there are a very small handful of actual NAND manufacturers out there. Most of the SSDs you see have identical NAND in them, the only appreciable difference is firmware and that causes sometimes drastic performance differences.
As for other steps double check the firmware revision on the SSD and verify you have the latest controller drivers.
For settings in the Samsung software I don't use over provisioning and don't use the automatic OS optimization.
If you'd like to copy my exact 'OS Config' settings starting my applying the Max Performance settings, then go to the Advanced tab and deactivate Hibernation. Then Apply All settings.
Also, I recommend running AS SSD benchmark at least once as a really useful thing it does in addition to benching is that it displays the storage controller being used in the upper left. Often people will be plugged into the Intel port but still be on the Microsoft AHCI driver. (msahci for the Microsoft default driver vs. iastor for the Intel driver)
As for USB flash drives and the like it's important to remember that those types of technology are very static. Firmware is a large part of SSD performance, as evidenced by the fact there are a very small handful of actual NAND manufacturers out there. Most of the SSDs you see have identical NAND in them, the only appreciable difference is firmware and that causes sometimes drastic performance differences.
As for other steps double check the firmware revision on the SSD and verify you have the latest controller drivers.
For settings in the Samsung software I don't use over provisioning and don't use the automatic OS optimization.
If you'd like to copy my exact 'OS Config' settings starting my applying the Max Performance settings, then go to the Advanced tab and deactivate Hibernation. Then Apply All settings.
Also, I recommend running AS SSD benchmark at least once as a really useful thing it does in addition to benching is that it displays the storage controller being used in the upper left. Often people will be plugged into the Intel port but still be on the Microsoft AHCI driver. (msahci for the Microsoft default driver vs. iastor for the Intel driver)
-
Reply to xyriin
m
0
l
Tanquen
October 17, 2014 1:25:09 PM
Oh, what is this all about then? I guess I was not holding it wrong! *&@&%!!!!
I'll have to try it out as for me it slows down the moment you write data to the drive not after 30 days.
http://www.kitguru.net/components/ssd-drives/matthew-wi...
I'll have to try it out as for me it slows down the moment you write data to the drive not after 30 days.
http://www.kitguru.net/components/ssd-drives/matthew-wi...
-
Reply to Tanquen
m
0
l
xyriin
October 17, 2014 2:34:22 PM
Tanquen
October 17, 2014 3:23:35 PM
Well, no. I had a bad experience for almost a year and I don’t know if trust their fix. Is it rewriting the data to keep it readable and degrading the drive? The 850 is the one with the longer warranty. The good thing about it is knowing more about what the hell was going on and being rid of all the “you did something wrong” vibe from everyone.
It was also funny how the Samsung Magician benchmark showed 522MB read and 526MB writes when it was nowhere near that. What a joke. It must be a peek and not average.
It was also funny how the Samsung Magician benchmark showed 522MB read and 526MB writes when it was nowhere near that. What a joke. It must be a peek and not average.
-
Reply to Tanquen
m
0
l
xyriin
October 17, 2014 3:34:32 PM
It could still be something else for you. This has been isolated to a very specific issue of old data (minimum of a month) on the drive.
Personally, I never saw the issue even on benchmarks. Still applying the fix though as newer firmware rarely slows drives down. The firmware flash took about 10 seconds, system shutdown, and the restoration process commenced one the system booted back up. The restoration does take a little while but if it isn't your system drive you can just let it run in the background. For someone performing the update on a system drive, I'd recommend running it overnight.
I'll be running it on the 500GB torrent drive after I get done with the current one.
As for benchmarks I never trust the in house tools, I always go third party.
Personally, I never saw the issue even on benchmarks. Still applying the fix though as newer firmware rarely slows drives down. The firmware flash took about 10 seconds, system shutdown, and the restoration process commenced one the system booted back up. The restoration does take a little while but if it isn't your system drive you can just let it run in the background. For someone performing the update on a system drive, I'd recommend running it overnight.
I'll be running it on the 500GB torrent drive after I get done with the current one.
As for benchmarks I never trust the in house tools, I always go third party.
-
Reply to xyriin
m
0
l
Tanquen
October 17, 2014 4:09:51 PM
Yea, overnight is the way to go. I was really worried until the PC booted back up. It seems to have worked as the trusty old HD Tune 2.55 now looks like it should. My before pic is above in this tread but my before and after look just like the ones in the Kit Guru article. So again, it’s not me or just my drive. I think the 30 day thing is BS and it can affect any of their drives at various times. I’m sure they think telling folks it was only really old data and that only one or two people had the issue is better than saying it was lots of drives and many had the issue the moment you put data on the drive while others could take months or years to show the issue.
http://www.kitguru.net/components/ssd-drives/matthew-wi...
“As for benchmarks I never trust the in house tools, I always go third party.”
Well the Samsung one is outright lying or is uselessly giving the peek read and write speeds.
http://www.kitguru.net/components/ssd-drives/matthew-wi...
“As for benchmarks I never trust the in house tools, I always go third party.”
Well the Samsung one is outright lying or is uselessly giving the peek read and write speeds.
-
Reply to Tanquen
m
0
l
xyriin
October 17, 2014 5:05:06 PM
Well it kind of makes sense.
First, if you're accessing regular things day after day or at least weekly you'd never notice it. Typical OS files, commonly used applications, etc. In other words the stuff you use most of the time wouldn't be affected. If there was no time limit on the bug then everyone would have experienced the bug from day one including the couple hundred tech reviews when the product was launched.
Complicating matters, if someone thought something was wrong and just wiped their system and reinstalled their OS the issue would have disappeared as well. The logical thought process in this case would be someone thinking their OS install was borked.
As it's a firmware fix the same issue applied to all drives, at least those on the latest firmware prior to this update. I haven't seen any testing related to this bug and different firmware revisions. It is also possible one of the post launch firmware updates resulted in this bug which could also explain why some users never saw the issue if they were on an older firmware revision. That said, as there is only one hardware revision of the 840 EVO (to my knowledge) this bug would have been present on every drive out there.
I'm actually a bit surprised I didn't experience this issue. My 500GB EVO is a torrent drive so there is pretty much no old data on it. However, the 1TB EVO is my game install drive. I've got a good 600-700GB installed on it and obviously some of those games I don't touch for months on end, yet I never had a slow down on the drive.
Based on the number of drives out there and the difficulty that was had tracking down the exact issue, I tend to think it really is obscure as it's made out to be. I don't think it's as simple as any data greater than 30 days, but it's probably some specific scenario involving untouched data.
First, if you're accessing regular things day after day or at least weekly you'd never notice it. Typical OS files, commonly used applications, etc. In other words the stuff you use most of the time wouldn't be affected. If there was no time limit on the bug then everyone would have experienced the bug from day one including the couple hundred tech reviews when the product was launched.
Complicating matters, if someone thought something was wrong and just wiped their system and reinstalled their OS the issue would have disappeared as well. The logical thought process in this case would be someone thinking their OS install was borked.
As it's a firmware fix the same issue applied to all drives, at least those on the latest firmware prior to this update. I haven't seen any testing related to this bug and different firmware revisions. It is also possible one of the post launch firmware updates resulted in this bug which could also explain why some users never saw the issue if they were on an older firmware revision. That said, as there is only one hardware revision of the 840 EVO (to my knowledge) this bug would have been present on every drive out there.
I'm actually a bit surprised I didn't experience this issue. My 500GB EVO is a torrent drive so there is pretty much no old data on it. However, the 1TB EVO is my game install drive. I've got a good 600-700GB installed on it and obviously some of those games I don't touch for months on end, yet I never had a slow down on the drive.
Based on the number of drives out there and the difficulty that was had tracking down the exact issue, I tend to think it really is obscure as it's made out to be. I don't think it's as simple as any data greater than 30 days, but it's probably some specific scenario involving untouched data.
-
Reply to xyriin
m
0
l
Related resources
- Samsung 850 Pro worth the Premium ove 840 Evo? Forum
- SolvedWill the Samsung 840 Evo patch continue working after I reformat SSD? Forum
- Samsung 850 evo. Forum
- SolvedSamsung 840 EVO SSD update Forum
- SolvedTrouble with getting new Samsung 840 EVO SSD detecting SATA3 (6gbps) connection - SOLVED Forum
- SolvedWhich SSD Should I Get? Samsung 850 Pro or the M.2 Plextor M6e? Forum
- Solvedwhich one is a better ssd samsung evo pro or sandisk extreme pro ?? Forum
- SolvedSamsung 840 EVO SSD low benchmark Forum
- SolvedHow to best use the Samsung 840 EVO SSD? Forum
- SolvedUrgent - Samsung 840 EVO or Intel 730 series SSD Forum
- SolvedSamsung 840 EVO SSD USB 2.0 connection? Forum
- Difference between Samsung 840 evo vs 850 pro Forum
- SolvedSamsung EVO SSD Benchmarks 3000 Mb/s Forum
- SolvedSamsung EVO series SSD's - model numbers? Forum
- SolvedSamsung SSD EVO 840 120 GB Real copy speed so slow Forum
- More resources
!