The Lord Bude posted:There'll likely be a fix within a couple of weeks. I'd say just wait and do nothing. In the mean time you scarcely notice it if you have rapid mode on. So is RAPID mode something I should turn on? I got conflicting opinions in a quick search I did.
|
|
# ? Sep 20, 2014 16:38 |
|
|
# ? Jun 7, 2024 16:36 |
|
So i have a 256GB EVO that my son has been destroying as quickly as possible and I ran HD Tach which is at least 8 years old.. it reported crap numbers in places but I don't trust that software at all. Going to try some different benchmarks. [edit] So.. yeah I ran MHDD which is a low level disk tester based in DOS. It reports many many slow areas.. slow in the order of 40-60MB/s Not only that the access time is in the 30-50ms range for those areas. I am loving disgusted. For reference this 256GB EVO has 7TB of writes. redeyes fucked around with this message at 18:03 on Sep 20, 2014 |
# ? Sep 20, 2014 17:34 |
|
Mr E posted:So is RAPID mode something I should turn on? I got conflicting opinions in a quick search I did. Yes. It's the key feature that makes Samsung drives so unbeatably good. It makes the drive so fast it breaks benchmarking software.
|
# ? Sep 20, 2014 17:39 |
|
Mr E posted:So is RAPID mode something I should turn on? I got conflicting opinions in a quick search I did. Where did you get conflicting opinions?
|
# ? Sep 20, 2014 17:52 |
|
PerrineClostermann posted:Where did you get conflicting opinions? They don't seem to like it on the Overclock forums, (where the big EVO problem thread is) one guy in particular had this to say (I'm not second guessing goons here just relaying the info) quote:If you turn on CRAPID mode, you are not benchmarking your SSD anymore, you are benchmarking your RAM.
|
# ? Sep 20, 2014 18:02 |
|
For the uninitiated, what is the read speed range for say 7200RPM HHDs? Are the slower read speeds being exhibited on these SSDs still faster than a decent HHD?
|
# ? Sep 20, 2014 18:13 |
|
Massasoit posted:For the uninitiated, what is the read speed range for say 7200RPM HHDs? Are the slower read speeds being exhibited on these SSDs still faster than a decent HHD? My 2TB HDD, 7200 RPM, 64MB Cache: My 500GB 840 EVO, ~4.4TB of writes: Speeds are more consistent on my HDD than my SSD.
|
# ? Sep 20, 2014 18:23 |
|
Im seeing 150ms+ access times with my brutalized EVO. loving horrible! [edit] Doing a backup/clone finding unreadable sectors.. WHAT THE gently caress?! These EVOs are totally on my shitlist now. gently caress them. redeyes fucked around with this message at 19:41 on Sep 20, 2014 |
# ? Sep 20, 2014 18:24 |
|
Liu posted:They don't seem to like it on the Overclock forums, (where the big EVO problem thread is) one guy in particular had this to say (I'm not second guessing goons here just relaying the info) So...a guy learned how RAPID works and thinks Samsung is pulling a fast one? He's retarded. Yeah, that's how it works. The vast majority of system writes are very small, very quick. RAPID allows you to "buffer" them and finish more quickly. Unless you're commonly using that edgecase (in which case, you're putting RAPID on the wrong drive), it's not an issue. I'm not storing BD rips on my system drive, that's for damned sure. And I don't plan to have my computer's power cut during a file transfer anyway.
|
# ? Sep 20, 2014 18:38 |
|
Xenomorph posted:My 2TB HDD, 7200 RPM, 64MB Cache: How the hell are you getting those speeds off a platter drive? My 1tb WD Black drive starts at 140 and drops to 70'ish. My 840 evo is at 50'ish at worst, but most of the drive is at 400+.
|
# ? Sep 20, 2014 19:38 |
|
Maybe this is a dumb question but why doesn't windows do the sort of write caching that RAPID is doing by default? Is it the risk of data loss on crash or power failure? How does Samsung get around that?
|
# ? Sep 20, 2014 19:45 |
|
I know im posing a lot about this but the EVO in question 7TB writes, about 6 months old has more bad clusters than the drive can deal with and thus has to be RMA'd. If I didn't check it by cloning I probably wouldn't have known it was failing at all. THE SSD itself, Samsung Magician, Windows, SMART status, etc all reported the drive was happily fine and had %97 left on the wear status. This could be a fluke or more likely these EVOs are trash.
|
# ? Sep 20, 2014 20:19 |
|
Or maybe its just a bad drive
|
# ? Sep 20, 2014 20:20 |
|
go3 posted:Or maybe its just a bad drive Could be. And it would be the first SSD out of maybe a hundred I have used/sold that failed in this manner. I had a bunch of the original Sandforces brick from the firmware bugs but this is a bit different.
|
# ? Sep 20, 2014 20:21 |
|
The same thing happens with HDDs constantly. It's not even exclusive to the technology.
|
# ? Sep 20, 2014 21:28 |
|
redeyes posted:This could be a fluke or more likely these EVOs are trash. Yeah, based on your sample size of one we can clearly come to that conclusion. Are you related to that other guy?
|
# ? Sep 20, 2014 22:15 |
|
Also Windows will write out the cached data before restarting. The file won't get corrupted without instant power loss.
|
# ? Sep 20, 2014 22:28 |
|
It depends on how well the RAPID handler deals with Windows' data flushing and the restart command. That guy's problem literally should never happen, which makes it difficult to believe. If such a story could be verified, though - even once would be enough, for the same reasons the hardware threads take a 'not even once' stand with data corruption on everything else - that sounds like a legitimate reason to wait until Samsung fixed it before using RAPID. And Samsung having problems to fix wouldn't be unheard of. dont be mean to me fucked around with this message at 22:33 on Sep 20, 2014 |
# ? Sep 20, 2014 22:29 |
|
quote:If you turn on CRAPID mode, you are not benchmarking your SSD anymore, you are benchmarking your RAM. This opinion is pretty retarded. YES, RAPID mode effectively turns part of your RAM into part of your SSD, and it works by doing a lot of storing and transferring in the RAM when you command files to be read/written, then after that, RAPID transfers the data in the RAM to the SSD in the background as you keep working. But here's a question for you... who cares? You're still doing work on a computer that not only feels but actually is ludicrously fast at the cost of some extra CPU cycles on the back end you're probably not using anyway... which if you have a processor built in the last 4 or 5 years, you won't feel at all. So what's the problem? RAPID mode is basically auto-tune for hard drives. The singer may or may not be pitch perfect but guess what? The experience of listening to an auto-tuned singer as an end user is that of an absolutely flawless, pitch-perfect performance, just like the experience as a user of an SSD with RAPID on is the experience of an absolutely flawless, lightning-fast hard drive. And that's on top of an already extremely fast SSD anyway. So unless you're a giant hipster about it and need to know your favorite singer was raised by a hermit on a mountain somewhere from birth to only sing perfectly without autotune, and you need to know your SSD was designed to deliver RAM-fast speeds from the ground up with no help period and nothing else will do... it doesn't matter. GreatGreen fucked around with this message at 22:50 on Sep 20, 2014 |
# ? Sep 20, 2014 22:34 |
Also, RAPID might help combine writes better, so the SSD has to do less erasing and moving half-full blocks around. That's the main argument for it being potentially better than a generic cache, it can be tuned for specific drive parameters. And Windows does have write cache for drives, but it is probably much more conservative. You can tune it a bit from Device Manager.
|
|
# ? Sep 20, 2014 22:43 |
|
Siochain posted:How the hell are you getting those speeds off a platter drive? My 1tb WD Black drive starts at 140 and drops to 70'ish. My 840 evo is at 50'ish at worst, but most of the drive is at 400+. Probably density. They're up to 1 TB per platter now.
|
# ? Sep 20, 2014 22:57 |
|
Yeah, that last checkbox. That's about all that I could think of to make that story up-thread legit. And I'm not even sure why that exists or who would have cause to use it (or for that matter who could justify it without being labeled a space-grade doofus).
|
# ? Sep 20, 2014 23:04 |
Sir Unimaginative posted:Yeah, that last checkbox. If you have a high-end RAID controller with its own backup power supply. Then Windows can pretend writes have gone through as soon as they are in the controller's buffers even though the data haven't actually hit the hard storage yet.
|
|
# ? Sep 20, 2014 23:09 |
|
OC Forums posted:And after you read this, you will never enable RAPID again.....it will actually make your performance worse not better Also, isn't the maximum cache size with RAPID 2.0 4GB (or 25% of RAM, whichever is lower) so, if this slow-down problem gets fixed (aka all parts of your drive run at SATA-connection/driver speed limit which looks to be roughly 400MB/s or 200MB/s for SATA III and II, respectively from these benchmarks), shouldn't that be on the order of tens of seconds for it to finish actually writing to the drive anyway? That is, 4GB / 400MB/s = 10 seconds (or, with SATA II: 4GB / 200MB/s = 20 seconds). If these issues aren't resolved, that could cause mishaps (taking into the minutes to commit data to the drive; not that any average use I've ever done involves compulsive copying/restarting so chances are this is also a non-issue), but if they are: you're almost certainly winning by clipping off fractions of/whole seconds thousands of times with RAPID over the course of hours of use as opposed to losing out on having to wait 30 seconds before restarting after copying over large files. I guess it's lazy if Samsung don't have it configured to postpone a restart (surely possible), but if you avoid this one rare case you still get undeniable improvements to read/write performance. redeyes posted:So i have a 256GB EVO that my son has been destroying as quickly as possible and I ran HD Tach which is at least 8 years old.. it reported crap numbers in places but I don't trust that software at all. Going to try some different benchmarks.
|
# ? Sep 20, 2014 23:18 |
|
I am mad because if drives I have been selling end up making GBS threads the bed I gotta take care of it and that sucks. Having said that I have a 500GB original 840 that has been in use for 2-3 years and it is fine according to all the normal tests. No major drops in performance, so I am hoping this is really firmware fixable.
|
# ? Sep 20, 2014 23:34 |
|
The original 840 suffers from the same bug ostensibly. If you're using Windows 7 or lower (or possibly Linux), make sure it isn't automatically defragging your drive. Not that it would matter too much at that storage capacity.
|
# ? Sep 21, 2014 00:53 |
|
Factory Factory posted:Probably density. They're up to 1 TB per platter now. Aren't they at 1.2TB per platter now for some drives? The Western Digital Red 6TB HDD has 1.2TB per platter.
|
# ? Sep 21, 2014 02:25 |
|
Today's tech tip: when trying to figure out why the Samsung Magician software isn't recognizing your drive, try to remember that you actually bought a Mushkin because it was on sale at the time. Sometimes me are dum.
|
# ? Sep 21, 2014 02:51 |
|
Drive is less than a year old, light use, 3.64TB of writes: I broke my don't buy samsung policy on this drive too
|
# ? Sep 21, 2014 02:59 |
|
GrizzlyCow posted:The original 840 suffers from the same bug ostensibly. If you're using Windows 7 or lower (or possibly Linux), make sure it isn't automatically defragging your drive. Not that it would matter too much at that storage capacity. That may be but my 500GB non-evo HD-Tunes totally normal numbers. My 250GB EVO is a loving disaster after 6 months. That article says the 840 is affected but does not show benchmarks showing it is.
|
# ? Sep 21, 2014 03:30 |
|
Factory Factory posted:Probably density. They're up to 1 TB per platter now. Bingo. I try to stick to 2-platter drives. My last drives were the WD Black 640GB drives. Two 320GB platters. When they were new that was a pretty big deal. Less platters = more dense data, less heat, less stress on internal components, etc. When I was looking for an upgrade I wanted to stick to 2-platter drives, and the 2TB option (Toshiba) met that requirement, was cheap and had decent ratings/reviews.
|
# ? Sep 21, 2014 04:12 |
|
So the Intel SSD 530 240GB is the obvious replacement for the Samsung 840 Evo at 250GB, but what about 500GB? It's down to the Crucial MX100 512GB and the SanDisk Ultra II 480GB. I'm inclined to recommend the SanDisk because they are a better brand than Crucial, but the 4-channel Ultra II isn't actually better than the MX100, and I don't see a single review of the 8-channel version to validate an assumption that it is. At 1TB the Ultra II is the only option, but again I'm not comfortable recommending a drive because it's PROBABLY fast. This lack of good SSD options almost makes getting a Samsung 840 Evo and waiting for a firmware update seem reasonable.
|
# ? Sep 21, 2014 04:16 |
|
Where are you dumb-dumbs getting hd tach? When I run it it complains I don't have Windows 2000 or XP.
|
# ? Sep 21, 2014 04:44 |
|
You need to use compatibility mode.
|
# ? Sep 21, 2014 04:45 |
|
Aphrodite posted:You need to use compatibility mode.
|
# ? Sep 21, 2014 04:48 |
|
So, I decided to chill out and then test this 256GB EVO more. Earlier today I not only could not clone/backup the thing with Acronis but a chkdsk yielded so many bad clusters that it bombed out with a operation failed. I took the drive out of the machine and set it down on my desk. A bit ago I manually copied the important files off the drive and then attempted another Acronis backup. This time it finished with no unreadable stuff (WTF?!). So I got a full image of the system and I now have a drive to test on and/or possibly RMA. My first instinct is to secure erase it but that wouldn't be very scientific. It does bug me that the drive is acting so funky. Not sure what to think.quote:This lack of good SSD options almost makes getting a Samsung 840 Evo and waiting for a firmware update seem reasonable.
|
# ? Sep 21, 2014 05:07 |
|
redeyes posted:That may be but my 500GB non-evo HD-Tunes totally normal numbers. My 250GB EVO is a loving disaster after 6 months. That article says the 840 is affected but does not show benchmarks showing it is. No, the 840 is affected by the bug. It's just wasn't as popular as the 830 and 840 Pro when it first released, so not as many people have as, say, the 840 Evo. It most definitely suffer from the same bug. Like I said, you may have auto-defrag on your 840 if you're not seeing the issue. Or something else. fake edit: Alereon posted:So the Intel SSD 530 240GB is the obvious replacement for the Samsung 840 Evo at 250GB, but what about 500GB? It's down to the Crucial MX100 512GB and the SanDisk Ultra II 480GB. I'm inclined to recommend the SanDisk because they are a better brand than Crucial, but the 4-channel Ultra II isn't actually better than the MX100, and I don't see a single review of the 8-channel version to validate an assumption that it is. At 1TB the Ultra II is the only option, but again I'm not comfortable recommending a drive because it's PROBABLY fast. You could always recommend the SanDisk Extreme Pro at 1TB. It only has a $70 premium. As for the 500GB range, you can always fallback to the Chronos Deluxe or XLR8 Pro. Of course, the Intel 530 is nice proven drive by a reputable company, and the prices between it and the Evo isn't too huge. Though, so far, neither the MX100 nor the M550 have been horrible, so maybe we can give Crucial another chance. M4 was bad, but the M500 wasn't terrible. GrizzlyCow fucked around with this message at 06:05 on Sep 21, 2014 |
# ? Sep 21, 2014 05:20 |
|
Massasoit posted:For the uninitiated, what is the read speed range for say 7200RPM HHDs? Are the slower read speeds being exhibited on these SSDs still faster than a decent HHD? They are not faster, assuming sequential access. (Most of the performance graphs posted so far are sequential performance tests.) It's not a simple question, though. HDDs take a horrific performance hit when doing non-sequential I/O. It's possible that an EVO afflicted by this bug still handily outperforms any HDD for general purpose computing, which might explain why it took so long for people to notice an issue and start talking about it.
|
# ? Sep 21, 2014 05:45 |
|
A 7200 RPM hard drive is good for 80-150ish MB/sec in sequential reads. However, it's also good for only about 90-120 IO operations per second (IOPS). For anything other than reading/writing large files from/to an unfragmented hard drive, random IOPS matters a lot more than sequential read speed. Even the slowest SSDs are good for tens of thousands of IOPS. The average latency also drives how fast the drive "feels" -- 7-9 MS for hard drives (governed by the speed of rotation of the drive) vs. <1 MS for good SSDs. edit: Regarding Samsung's RAPID: coming from an enterprise background, the idea of using any kind of write caching without battery backup is stupid. I think I said this in the old thread, but IMO any time a manufacturer is writing filter drivers it's a huge red flag. They're probably doing something they shouldn't, and they also get a fraction of the testing that the stock Windows AHCI drivers get. Are they honoring sync IO requests properly and only accelerating async IO? Doesn't anyone remember how bad the data corruption issues with the NForce filter drivers were? SSDs are already orders of magnitude faster than spinning disk -- I tend to agree with the opinions I've seen on other forums that RAPID is not worth the risk without BBWC or UPS power. KS fucked around with this message at 06:15 on Sep 21, 2014 |
# ? Sep 21, 2014 06:02 |
|
|
# ? Jun 7, 2024 16:36 |
Alereon posted:This lack of good SSD options almost makes getting a Samsung 840 Evo and waiting for a firmware update seem reasonable. https://www.youtube.com/watch?v=tRVUOGUmxJI
|
|
# ? Sep 21, 2014 06:23 |