I like the OCZ solid state drive myself. Their support staff I found to be concern with actually helping their customers. When I tried moving their hard drive to another computer and the computer would not recognize it for what ever reason. I sent them back the drive and they replaced it with a new drive. When I got that drive it also, was not recognized. The only way I could make it recognize was to place the drive in the machine and use a Windows 7 disk to load the OS. Once the OS loaded I could then put it on the USB bus and use a clone program to transfer the OS program on the original disk to the OCZ disk. Later I found out that he original disk was being run in Compatibility Mode which is IDE mode. This is the way I bought the machine from an authorized Widows refurbish-er. I believe because the original OCZ disk was not recognized in this mode. I went on the Microsoft site in the forum section and was trying to find how I could switch the mode back to AHCI. There people said one had to completely re-format the disk to switch it. I look further on the web and found a software person who explained how to change the registry to do this without re-formating. When I talk to the tech at OCZ he agree with me that it would had been nice if Microsoft had detected this and ask the operator if they wanted to change or leave the format the same. Also, the OCZ tech told me that Microsoft is suppose to have a program that will do this without have the operator go into the registry. I guess if one was able to find that on their site it would be good. Also, it would be good if their techs writing on the forum knew this.
-
-
Re: HDD issues.
This mode just makes the drive look like PATA. It is a function of the SATA controller, not the drive. The drive still operates in SATA mode, but without certain AHCI features. I can't see how this could have resulted in a problem.Originally posted by keeney123 View Post
BTW, OCZ are/were the least reliable SSDs by far. Some models had failure rates in excess of 30%.
Components returns rates:
http://web.archive.org/web/201309180...s-rates-7.html
- 40.00% for the OCZ Petrol 64 GB
- 39.42% for the OCZ Petrol 128 GB
- 30.85% for the OCZ Octane 128 GB SATA II
- 29.46% for the OCZ Octane 64 GB SATA II
- 9.73% for the OCZ Vertex 2 120 GB 3.5"
- 9.59% for the OCZ Vertex 2 120 GB
- 6.73% for the OCZ Vertex 2 60 GB
- 5.43% for the OCZ Agility 3 240 GB
- 5.12% for the OCZ Vertex Plus 128 GB
"With such rates, we can justly classify such models [OCZ Petrol and Octane SATA II] as defective and it is shameful that such products have remained on sale in stores!" -
Re: HDD issues.
Well it does with have an effect if the operating system sees it also, want mode it is runs in is controlled by the registry's not the drive which are in the OS. I have a vector 150 and it runs like a top. Also, they stand behind their word on repairs. I had no problem with them.Originally posted by fzabkar View PostComment
-
Re: HDD issues.
IDE mode is native to Windows. It does not require any special drivers. It just works.Originally posted by keeney123 View PostComment
-
Re: HDD issues.
Actually, I had to switch from "IDE" mode to AHCI mode after installation with an XP machine I think it was? I too found the registry article you're talking about. It was very nice. The tutorial I was following wasn't the best. The gave a download link for shareware software that could allow me to edit the registry via a bootable disk. This shareware version didn't work though, wanted me to register. Thankfully, though, I found away to just use regedit to load the keys via a recovery console. That was nice. I might of made a bootable XP live disc, I don't remember. I really didn't want to reinstall the OS, that's for sure! After all the work I did getting it up and running. I knew something was wrong though with how long it was taking. I just thought it was slow.Originally posted by keeney123 View Post
So, I was thinking of getting a solid state. When I checked into them when they first came out though, I heard they didn't have a very long life time. Is this the case? Or will the last longer than a "normal" drive under "normal" circumstances? I mean, can I expect to get more reads and writes from a solid state than, lets say a SATA 3.0 drive? There's also that M.4 or whatever. I think that's the future.-- Law of Expanding Memory: Applications Will Also Expand Until RAM Is FullComment
-
Re: HDD issues.
On the system I was playing with, the drive's performance was hugely impacted. Everything was soooo slow. Granted, this was a much older system. Maybe things have changed since then? Are you saying that if the BIOS has the drive configured as a PATA drive, it would still perform as fast as if it was configured for AHCI? I'm not questioning you here, just trying to understand this all.Originally posted by fzabkar View Post-- Law of Expanding Memory: Applications Will Also Expand Until RAM Is FullComment
-
Re: HDD issues.
These drives I am talking about are hybrid drives they are part sold state and part hard drive have you read about them?Comment
-
Re: HDD issues.
Hybrid drives incorporate a NAND flash based read cache on the PCB, typically about 8GB. Otherwise they function just like regular HDDs.Originally posted by keeney123 View Post
WD's Black-2 hybrid is a little different than the others. It combines a regular rotating HDD with a 120GB SSD. AFAICT, the two drives are connected to a SATA port multiplier which then connects to the host via a single SATA data cable. Personally I don't see the point of it, except in those applications where there is no room for two separate drives. Otherwise it would be far cheaper to purchase the individual drives.Comment
-
Re: HDD issues.
Yes that is but one type of hybrid drive and this cache is to speed up the access time by storing information and utilizing repeated information like a stack reg. would. Also, a solid state drive will make a mistake every 200,000 operations, a hard drive makes a mistake every 10^14 operations. It is a way of have the best of both worlds. They are more dynamic as to were the information is to be written to and how to retrieve that information. A program is set up to optimize the disk and the way in which the disk is cleared and written to. This program can interact with the OS for this operation, but it does not have to. A "example" that is similar would be IVp4. router. and the addressing routine. Because IVp4 did not consider how many addresses it would have to write there was a limited number. Solution was to separate inside address and outside address. The inside address go to the users equipment and the outside address go to the world wide web. The web can address a certain number address, but that address does not equal the direct address to the inside computers. The router can be set up in a static mode or dynamic mode. In the dynamic mode there is no specific address for the inside equipment because the router is constantly assigning and reassigning addresses in a stack that it remembers. This is the best way I can explain how one can get out of direct addressing a disk drive. Also, as I remember the dynamic address is more common in a hybrid drive, but can also be used in hard drive's firmware with a regular hard drive. It has been sometime since I have study this and the very last sentence is a belief that I remember this instead of a yes I know it.Originally posted by fzabkar View PostLast edited by keeney123; 09-18-2015, 04:21 PM.Comment
-
Re: HDD issues.
Quote: "... a solid state drive will make a mistake every 200,000 operations ..."
I don't know where your figure of 1 in 200,000 operations comes from, but AFAIK the HDD manufacturers typically specify an Uncorrectable Bit Error Rate (UBER) of less than 1 bit in 10E14 bits read for consumer grade HDDs while SSD manufacturers such as Intel and Samsung specify 1 in 10E16 for consumer grade SSDs and 1 in 10E17 bits for enterprise class products.
According to this Kingston article, the JEDEC standard specifies less than 1 bit in 10E15 bits read for consumer grade SSDs, and 1 in 10E16 bits for enterprise class:
http://www.kingston.com/en/ssd/enter...sus_client_ssd
The Kingston article goes on to say that Redundant Array of Independent Silicon Elements (R.A.I.S.E.) technology reduces the UBER to 1 in 10E29 bits. RAISE stripes data across several NAND chips.
The Kingston article also lists the raw NAND Bit Error Rate (BER) for TLC (1 in 10Kbits), MLC (1 in 10Mbits), e-MLC (1 in 100Mbits) and SLC (1 in 1Gbit) flash types.
FYI, if you have an SSD based on the SandForce SF-2000 controller series, the SMART report will include actual error rate data:
Last edited by fzabkar; 09-18-2015, 05:10 PM.Comment
-
Re: HDD issues.
Here is an article explain the problem. http://blogs.marvell.com/2014/10/err...-state-drives/Originally posted by fzabkar View Post
I will look further to find the 1 in 200,000Comment
-
Re: HDD issues.
Here is a more complex explanation of the problem
https://anturis.com/blog/ssd-reliability-in-2014/
Also, I did not say that the 1 in 200,000 was and un-correctable bit error I said that they created and error 1 in 200,000 operations. They now stack these cells to duplicate the data and the above article in this message states it is 1 in 100,000 writes to a cell.Last edited by keeney123; 09-18-2015, 07:27 PM.Comment
-
Re: HDD issues.
The figure of 1E14 is the UBER for HDDs. It makes no sense to compare a UBER with a BER. In any case the BER for TLC NAND flash is 1 bit in 10Kbits which corresponds to 1 bit error in 2.5 sectors. For MLC flash, the BER is 1 bit in 2500 sectors.
As for those two articles, neither supports your statement. The first just reads like a marketing blurb and essentially just says that modern flash has higher BER because there are more bits per cell and more cells on the same silicon. The author then goes on to say that this requires a more robust ECC strategy.
The second article talks more about write endurance and essentially just says that BER increases as cells wear out, and that this wear is the result of a limit on the number of program/erase cycles for the different NAND technologies. However, the author goes on to say that the UBER is unaffected until the max P/E cycles have been seriously exceeded. For example, a TLC based Samsung SSD that was rated for 1000 P/E cycles did not begin to show degradation until it hit 3000 cycles.Comment
-
Re: HDD issues.
So the write cycle of 1 in 100,000 write cycles error does not support my statement of creating and error in 1 in 200,000 operations?Last edited by keeney123; 09-19-2015, 02:52 PM.Comment
-
Re: HDD issues.
So, if one where to buy a solid state drive, they'd want one based off of the MLC flash instead of the TLC? I remember seeing motherboards coming out that supported a new hard drive, a hard drive on a chip, something called M.4 I think or something weird like that. It was supposed to be much faster than SATA 3 (6.0Gbps, not 3.0Gbps). At the time though, I think only a few motherboards truly supported it. It took up a bunch of PCI-E lanes for the 4x speed and the board needed something like a PCI-E Multiplexer or something. This was a limit of the CPUs and sockets at the time I believe, not necessarily the motherboards. I believe even with the multiplexor though, speed wasn't really an issue.Originally posted by fzabkar View Post
I also want to say I remember something like you could use one of the new hard drives but maybe could only use 2 PCI-E video cards running at 16x. Without using one of the new hard drives, you could use 4 PCI-E running at 16x or something crazy like that. Do you know what I'm talking about? Wish I could remember more.-- Law of Expanding Memory: Applications Will Also Expand Until RAM Is FullComment
-
Re: HDD issues.
The author of that article has a fundamental misunderstanding of the concept of a "die". He believes that a die and a cell are the same thing. In fact a die refers to the piece of silicon on which the entire NAND flash is fabricated, not individual cells.Originally posted by keeney123 View Post
https://en.wikipedia.org/wiki/Die_(integrated_circuit)How SSDs work
Each memory cell in SSD storage is a die that can contain one or more bits. Its physical properties are such that each die can handle about 100,000 writes before it wears out.
The author's article cites the same Kingston article that I cited in an earlier post. The figure of 100,000 writes (aka Program/Erase cycles) refers to SLC flash. In any case your statement that "a solid state drive will make a mistake every 200,000 operations" is a totally different thing, even if it were true. The first number is an endurance figure whereas your number is an error rate. In simple terms, the author is saying that an SLC NAND cell will begin to make mistakes after 100,000A die in the context of integrated circuits is a small block of semiconducting material, on which a given functional circuit is fabricated. Typically, integrated circuits are produced in large batches on a single wafer of electronic-grade silicon (EGS) or other semiconductor (such as GaAs) through processes such as photolithography. The wafer is cut ("diced") into many pieces, each containing one copy of the circuit. Each of these pieces is called a die.
writes whereas the Kingston article states that a newly manufactured SLC cell will have an inherent read error rate of 1 bit in one billion bits.Comment
-
Re: HDD issues.
TLC based flash drives have much stronger ECC algorithms than SLC based ones, so you'll never see the effects of the much higher raw BER. At least that's what the endurance tests seem to be saying.Originally posted by Spork Schivago View Post
Also, because the capacity increases as the max P/E cycles decrease, the SSD will still last a very long time before it technically wears out.
For example, if you were to write 10GB of data per day over a period of 10 years, then that amounts to 36.5 TB. A 500GB SSD with perfect wear levelling and 1:1 write amplification would rack up only 73 P/E cycles.
I don't know anything about M.2 or M.4. Sorry.Comment
-
Re: HDD issues.
I could not find the exact article that states what I have stated, either it is not on the web anymore or because I am using a public computer, now, it will not search the same as there is site controls on this computer. I did not say that they are less reliable then hard drives as I was stating one aspect. I think one of the most complete comparisons is found here https://en.wikipedia.org/wiki/Solid-state_drive. One time I called up Seagate and talk to their Engineering Technical Department. I ask them how long will a hard drive last. The answer I got is they can fail at any time so if you are storing something important you should have as many back ups as you can, at least 3. That is when I researched and found that magnetic reel to reel tape last the longest of all of the storage devices. That is why movie film has lasted so long. So if one whats to retrieve some information it would be best to load it on magnetic tape and store it. If you store it in climate control conditions you might get upwards of 100 years out of it. The only time you would use it is when you have lost all other copies. You could load it up and record it to another device and then put it back into storage. I do not have anything that is that important and most people don't, but for those who are in the extreme mind this might work. The M.2 and M.4 are suppose to be rivaling the solid state drives for speed because they are using solid state components in a parallel bus.Comment
-
Re: HDD issues.
Loved following this thread up to about post 51.
Let me know if you want me to split the discussion of SSD drives into a new thread following that post, to make it easier to follow.
EDIT: Discussion about SSD's in the old thread has been split into this new one
Also some information for you:
AHCI is an improvement for IDE, both support the TRIM command for SSD's.
This is in no way related to the physical interface which can also be IDE or SATA.
The IDE interface is limited to 133MB/sec, SATA 6Gbps is limited to 600MB/sec minus overhead which translates to ca 525MB/sec in the real world.
That is to say a 6Gbps SSD using the physical interface SATA can still have the IDE driver installed and reach the same performance.
Only some things like NCQ and power saving features will be missing.
SSD's have been limited by the SATA 6Gbps interface for years now.
As you can see in the Anandtech review linked above they all run into the same 525MB/sec bottleneck.
That is where the M.2 interface (formerly NGFF) comes into play. (Nothing called M.4 exists btw).
It introduces a new physical interface: PCI Express.
But it can also support SATA as a physical interface for backwards compatibility.
When running on the SATA physical interface M.2 is limited to the same 6Gbps as regular SATA is.
But when running in PCI Express mode the SSD can get up to a PCI Express x4 3.0 interface which translates to 32Gbps or 4000MB/sec.
As you can understand this is quite an improvement over SATA!
The PCI Express interface can also support a new logical interface: NVMe.
It's a driver like IDE or AHCI but developed for SSD's.
Remember AHCI was developed for mechanical HDD's so it has allot of unnecessary overhead for SSD's.
Also worth mentioning is that there was another successor planned for SATA.
It is called SATA Express but basically no devices have been released for it and the standard is dead.
The reason is the clunkiness of it's connector and the fact that it's a x2 PCI Express interface so only half the bandwidth that M.2 offers.
But of course a cabled interface is required, and this is where U.2 comes into play. (Formerly SFF-8639).
Basically it's a M.2 interface with a cable attached to it, even passive such adapters exist but also active."The one who says it cannot be done should never interrupt the one who is doing it."Comment
-
Re: HDD issues.
I agree with everything you said except there is a M.4 you just didn't look hard enough.Comment
Comment