Overclock.net › Forums › Components › Hard Drives & Storage › SSD › PSA About the Kingston V300 SSD(Probable foul play by Kingston)
New Posts  All Forums:Forum Nav:

PSA About the Kingston V300 SSD(Probable foul play by Kingston)

post #1 of 165
Thread Starter 
Edit: Kingston has in fact confirmed switching over to asynchronous NAND. Through people posting benchmarks on various sites I have gathered that the synchronous part numbers are v505A, v505B, v505C, v507, and v520. Likewise, the asynchronous versions are v506 and v521. Kingston's official response can be found here: http://media.kingston.com/support/downloads/V300_Benchmark_Brief_MKF_586.pdf

Edit 2: I've seen some benchmarks that make it look like there are v521/v506 that were shipping with synchronous NAND so the part # convention might be out the window. Their new v526 also doesn't seem to have a convention for the type of flash used. Looks like they are just shipping out whatever they can get there hands on for cheap.

As many of you probably know, the Kingston V300 is regularly one of the cheapest SSDs you can buy. The 120GB version is currently on Amazon, Newegg, and Tiger Direct for $70. For the last 6 months or so it has regularly been the cheapest Sandforce based drive with synchronous flash. However, with their latest revision of the drive there have been numerous accounts of people getting much lower benchmark scores than they should and they are all version 506. Here are a few examples:

http://www.overclock.net/t/1380526/kingston-v300-help

http://www.tomshardware.com/answers/id-1762558/kingston-ssdnow-v300-low-read-performance.html

http://forums.anandtech.com/showthread.php?p=35676298

Many Newegg Reviews:

http://www.newegg.com/Product/Product.aspx?Item=20-721-107&SortField=0&SummaryType=0&Pagesize=10&PurchaseMark=&SelectedRating=-1&VideoOnlyMark=False&VendorMark=&IsFeedbackTab=true&Keywords=%28keywords%29&Page=1#scrollFullInfo

Most people think that it is the firmware causing this and others think that its the location of the drive's production(Taiwan vs China). If you look over the other threads you'll see people are experiencing the problem on drives from both countries though. As for the firmware I'm confident that's not what is causing the slower speeds. Sandforce is very strict with the extent that SSD manufacturers can alter their controller's firmware. They can't really do that much, hence why all Sandforce drives perform comparably if they use the same type of NAND.

I'm led to believe that Kingston has changed the type of NAND in the V300 v506 to asynchronous flash...Which is a huge deal. The benchmarks that people are showing on v506 drives match up to what other asynchronous Sandforce drives perform at.

The most compelling evidence is from what Kingston itself said:


Quote:
Hello, I'm Jewel with Kingston Technical Support and would like to offer some clarification. Our V300 Solid State Drive uses the LSI SandForce 2281 SATA 6GB/s controller and MLC NAND from various manufacturers. The performance target for these drives is 450MB/s Read and Write using ATTO 2.41 with the SSD connected to a SATA 3.0 / 6Gb port. The SandForce 2281 controllers also employ a feature called DuraWrite™. In simple terms, this enables data compression and minimizes the data written to the SSD. DuraWrite™ extends the drives endurance and overall drive life with a byproduct of increased performance.
There are a number of benchmark utilities publicly available that you can use however Kingston uses ATTO 2.41 which we feel best reflect a real world workload. ATTO is also easy to obtain and run. Other benchmarking utilities may use fully incompressible data as part of their standard workload. Incompressible data can negatively affect benchmark performance on both the read and write speeds of our V300 drive depending on which NAND is being used. This is irrespective of the firmware revision of the drive.
Our products bring to market a real world workload as advertised under the conditions outlined in our documentation. All Kingston V300 SSDs bring an excellent user experience to our customers namely faster boot and shutdown, faster application loading, extended battery life verses HDDs, faster disk performance overall and most importantly a quality SSD product. For further assistance, please call us at 1-800-435-0640 (USA and Canada only) M - F 6am - 6pm PT and I or another available Technician will assist you. Please be sure to have the part in question on hand when you call.

Thank you for selecting Kingston as your upgrade partner.

The fact that they are using ATTO to benchmark drives isn't at all surprising. Every company that makes a Sandforce drive advertises ATTO scores because Sandforce is so much better at Compressible data. This also allows them to advertise their asynchronous drives with very high speeds. This is why I always advise against paying attention to manufacturer specs.

What IS interesting is that Kingston says "Incompressible data can negatively affect benchmark performance on both the read and write speeds of our V300 drive depending on which NAND is being used". They then go on to say "This is irrespective of the firmware revision of the drive".

It looks very much like Kingston has switched from synchronous to asynchronous flash in these drives. If this is the case then Kingston is doing something very shady and unethical. While their advertised speeds may be correct, the NAND changing in the drive is huge for performance.

If you have a Kingston V300(every case I've seen so far has been the 120GB capacity) check the revision you have and benchmark it. If you have a v506(product code-506ABBF0) and are willing to open it up and take a picture of the PCB/NAND modules I can probably confirm/deny this. I understand if you aren't comfortable doing that though.
Edited by okp11 - 4/23/14 at 2:06am
post #2 of 165
You are no doubt correct about the type of NAND used in the lower performing versions of the V300. I posted the same thing in another thread weeks ago. I also agree about the firmware, which rarely affects performance very much if at all.

Manufactures will sometimes obscure the model numbers on chips, literally wiping off the printing with a solvent. Would I be surprised if that was done in this case? Not at all.

One possible way to check would be to find a very recent review of the V300, but we know which model would be sent for reviewing, don't we?
post #3 of 165
yea, it is shady imo as well. I've noted it when I first seen reports about it.

Someone here reported a Mushkin Chronos Deluxe with async NAND before, when it should have had sync NAND. Mushkin however quickly RMA'd the drive for a proper one with sync NAND.
Le Main Rig
(17 items)
 
Test Bench
(8 items)
 
Terrorbyte V2
(17 items)
 
CPUMotherboardGraphicsRAM
i7-6700K Asus Sabertooth Z170 Mark 1 Gigabyte GTX 960 WindForce 2X OC G.SKILL Ripjaws V  
Hard DriveHard DriveHard DriveHard Drive
1050GB Crucial MX300 1050GB Crucial MX300 960GB OCZ RD400 1TB Micron M600 
Hard DriveCoolingOSMonitor
1TB WD Blue Noctua NH-U12S Windows 10 Pro 64-Bit BenQ GW2765HT 
MonitorKeyboardPowerCase
BenQ GW2765HT FUNC KB-460 Corsair RM650 Fractal Design R5 
Mouse
Logitec G500 
CPUCPUMotherboardRAM
Intel Xeon L5520 Intel Xeon L5520 SuperMicro X8DTi-F 48GB Nanya 1066MHz ECC RDIMM 
Hard DriveHard DriveHard DriveHard Drive
128GB Samsung 830 480GB SanDisk Extreme II 2TB WD Red 2TB Samung Spinpoint F4 
Hard DriveHard DriveHard DriveHard Drive
2TB Hitachi Deskstar 7K3000 1TB Samsung Spinpoint F3  3TB Toshiba DT01ACA300 Seagate Barracuda 7200.14 
OSPowerCaseOther
Windows Server 2012 R2 Datacenter  Corsair 650TX  Norco RPC 4224 LSI 9261-8i 
Other
Intel RES2CV240 
  hide details  
Reply
Le Main Rig
(17 items)
 
Test Bench
(8 items)
 
Terrorbyte V2
(17 items)
 
CPUMotherboardGraphicsRAM
i7-6700K Asus Sabertooth Z170 Mark 1 Gigabyte GTX 960 WindForce 2X OC G.SKILL Ripjaws V  
Hard DriveHard DriveHard DriveHard Drive
1050GB Crucial MX300 1050GB Crucial MX300 960GB OCZ RD400 1TB Micron M600 
Hard DriveCoolingOSMonitor
1TB WD Blue Noctua NH-U12S Windows 10 Pro 64-Bit BenQ GW2765HT 
MonitorKeyboardPowerCase
BenQ GW2765HT FUNC KB-460 Corsair RM650 Fractal Design R5 
Mouse
Logitec G500 
CPUCPUMotherboardRAM
Intel Xeon L5520 Intel Xeon L5520 SuperMicro X8DTi-F 48GB Nanya 1066MHz ECC RDIMM 
Hard DriveHard DriveHard DriveHard Drive
128GB Samsung 830 480GB SanDisk Extreme II 2TB WD Red 2TB Samung Spinpoint F4 
Hard DriveHard DriveHard DriveHard Drive
2TB Hitachi Deskstar 7K3000 1TB Samsung Spinpoint F3  3TB Toshiba DT01ACA300 Seagate Barracuda 7200.14 
OSPowerCaseOther
Windows Server 2012 R2 Datacenter  Corsair 650TX  Norco RPC 4224 LSI 9261-8i 
Other
Intel RES2CV240 
  hide details  
Reply
post #4 of 165
I'm one of the poor saps who bought one of these SSD's.

I received a Taiwan 506 and its ridiculously slow. ASSSD shows around 180 mb/s read speeds on fresh install win 7.

Personally I don't believe Kingston switched the memory chips and the fault lies in the firmware. I think the problem is just complete lack of caring and incompetence at Kingston, not duplicity.

The V300 uses a custom version of the sandforce firmware, its not the same. I figure messed up the 506 version and they are just too lazy to write new custom version for whatever reason using the 507 firmware update sandforce put out there.

Most people aren't going to realize the SSD is a lemon because they are ignorant and dont know any better. Look at recent review for the v300, most people who recently got one think its a great deal and dont have clue about how insanely slow it is. To them its better than their old HDD. Most people will never run any benchmarks, nor have they ever installed windows on a ssd before as this is most likely their first SSD. The only benchmark they will see is the Windows Performance Index which rates this SSD at 7.8 on fresh install. To them thats good enough.
post #5 of 165
Quote:
Originally Posted by Lex Luger View Post

I'm one of the poor saps who bought one of these SSD's.

I received a Taiwan 506 and its ridiculously slow. ASSSD shows around 180 mb/s read speeds on fresh install win 7.

Personally I don't believe Kingston switched the memory chips and the fault lies in the firmware. I think the problem is just complete lack of caring and incompetence at Kingston, not duplicity.

The V300 uses a custom version of the sandforce firmware, its not the same. I figure messed up the 506 version and they are just too lazy to write new custom version for whatever reason using the 507 firmware update sandforce put out there.

Most people aren't going to realize the SSD is a lemon because they are ignorant and dont know any better. Look at recent review for the v300, most people who recently got one think its a great deal and dont have clue about how insanely slow it is. To them its better than their old HDD. Most people will never run any benchmarks, nor have they ever installed windows on a ssd before as this is most likely their first SSD. The only benchmark they will see is the Windows Performance Index which rates this SSD at 7.8 on fresh install. To them thats good enough.
Im 99.99% sure the issues lies in the NAND rather than firmware.
Le Main Rig
(17 items)
 
Test Bench
(8 items)
 
Terrorbyte V2
(17 items)
 
CPUMotherboardGraphicsRAM
i7-6700K Asus Sabertooth Z170 Mark 1 Gigabyte GTX 960 WindForce 2X OC G.SKILL Ripjaws V  
Hard DriveHard DriveHard DriveHard Drive
1050GB Crucial MX300 1050GB Crucial MX300 960GB OCZ RD400 1TB Micron M600 
Hard DriveCoolingOSMonitor
1TB WD Blue Noctua NH-U12S Windows 10 Pro 64-Bit BenQ GW2765HT 
MonitorKeyboardPowerCase
BenQ GW2765HT FUNC KB-460 Corsair RM650 Fractal Design R5 
Mouse
Logitec G500 
CPUCPUMotherboardRAM
Intel Xeon L5520 Intel Xeon L5520 SuperMicro X8DTi-F 48GB Nanya 1066MHz ECC RDIMM 
Hard DriveHard DriveHard DriveHard Drive
128GB Samsung 830 480GB SanDisk Extreme II 2TB WD Red 2TB Samung Spinpoint F4 
Hard DriveHard DriveHard DriveHard Drive
2TB Hitachi Deskstar 7K3000 1TB Samsung Spinpoint F3  3TB Toshiba DT01ACA300 Seagate Barracuda 7200.14 
OSPowerCaseOther
Windows Server 2012 R2 Datacenter  Corsair 650TX  Norco RPC 4224 LSI 9261-8i 
Other
Intel RES2CV240 
  hide details  
Reply
Le Main Rig
(17 items)
 
Test Bench
(8 items)
 
Terrorbyte V2
(17 items)
 
CPUMotherboardGraphicsRAM
i7-6700K Asus Sabertooth Z170 Mark 1 Gigabyte GTX 960 WindForce 2X OC G.SKILL Ripjaws V  
Hard DriveHard DriveHard DriveHard Drive
1050GB Crucial MX300 1050GB Crucial MX300 960GB OCZ RD400 1TB Micron M600 
Hard DriveCoolingOSMonitor
1TB WD Blue Noctua NH-U12S Windows 10 Pro 64-Bit BenQ GW2765HT 
MonitorKeyboardPowerCase
BenQ GW2765HT FUNC KB-460 Corsair RM650 Fractal Design R5 
Mouse
Logitec G500 
CPUCPUMotherboardRAM
Intel Xeon L5520 Intel Xeon L5520 SuperMicro X8DTi-F 48GB Nanya 1066MHz ECC RDIMM 
Hard DriveHard DriveHard DriveHard Drive
128GB Samsung 830 480GB SanDisk Extreme II 2TB WD Red 2TB Samung Spinpoint F4 
Hard DriveHard DriveHard DriveHard Drive
2TB Hitachi Deskstar 7K3000 1TB Samsung Spinpoint F3  3TB Toshiba DT01ACA300 Seagate Barracuda 7200.14 
OSPowerCaseOther
Windows Server 2012 R2 Datacenter  Corsair 650TX  Norco RPC 4224 LSI 9261-8i 
Other
Intel RES2CV240 
  hide details  
Reply
post #6 of 165
You are 99.99 percent sure its the NAND even though you basically have no proof and we can directly measure the differences in the firmware versions which are massive. I'm not saying you are wrong, but I dont see you could possible be that sure of yourself unless you physcially took apart several drives.
post #7 of 165
I have one that runs at new release/reviewed speeds in my server.

I'd update the firmware from 5.05 to 5.06, but the tool says there are no updates.

It is much more plausible that kingston used cheaper async NAND to save $ or because supply was low because performance results are similar to other sandforce drives that have cheaper async NAND. Rather than, that they tried to tweak the firmware to where performance would be similar to other SandForce drives with slower Async NAND when they used more expensive Sync NAND in the drive.
Le Main Rig
(17 items)
 
Test Bench
(8 items)
 
Terrorbyte V2
(17 items)
 
CPUMotherboardGraphicsRAM
i7-6700K Asus Sabertooth Z170 Mark 1 Gigabyte GTX 960 WindForce 2X OC G.SKILL Ripjaws V  
Hard DriveHard DriveHard DriveHard Drive
1050GB Crucial MX300 1050GB Crucial MX300 960GB OCZ RD400 1TB Micron M600 
Hard DriveCoolingOSMonitor
1TB WD Blue Noctua NH-U12S Windows 10 Pro 64-Bit BenQ GW2765HT 
MonitorKeyboardPowerCase
BenQ GW2765HT FUNC KB-460 Corsair RM650 Fractal Design R5 
Mouse
Logitec G500 
CPUCPUMotherboardRAM
Intel Xeon L5520 Intel Xeon L5520 SuperMicro X8DTi-F 48GB Nanya 1066MHz ECC RDIMM 
Hard DriveHard DriveHard DriveHard Drive
128GB Samsung 830 480GB SanDisk Extreme II 2TB WD Red 2TB Samung Spinpoint F4 
Hard DriveHard DriveHard DriveHard Drive
2TB Hitachi Deskstar 7K3000 1TB Samsung Spinpoint F3  3TB Toshiba DT01ACA300 Seagate Barracuda 7200.14 
OSPowerCaseOther
Windows Server 2012 R2 Datacenter  Corsair 650TX  Norco RPC 4224 LSI 9261-8i 
Other
Intel RES2CV240 
  hide details  
Reply
Le Main Rig
(17 items)
 
Test Bench
(8 items)
 
Terrorbyte V2
(17 items)
 
CPUMotherboardGraphicsRAM
i7-6700K Asus Sabertooth Z170 Mark 1 Gigabyte GTX 960 WindForce 2X OC G.SKILL Ripjaws V  
Hard DriveHard DriveHard DriveHard Drive
1050GB Crucial MX300 1050GB Crucial MX300 960GB OCZ RD400 1TB Micron M600 
Hard DriveCoolingOSMonitor
1TB WD Blue Noctua NH-U12S Windows 10 Pro 64-Bit BenQ GW2765HT 
MonitorKeyboardPowerCase
BenQ GW2765HT FUNC KB-460 Corsair RM650 Fractal Design R5 
Mouse
Logitec G500 
CPUCPUMotherboardRAM
Intel Xeon L5520 Intel Xeon L5520 SuperMicro X8DTi-F 48GB Nanya 1066MHz ECC RDIMM 
Hard DriveHard DriveHard DriveHard Drive
128GB Samsung 830 480GB SanDisk Extreme II 2TB WD Red 2TB Samung Spinpoint F4 
Hard DriveHard DriveHard DriveHard Drive
2TB Hitachi Deskstar 7K3000 1TB Samsung Spinpoint F3  3TB Toshiba DT01ACA300 Seagate Barracuda 7200.14 
OSPowerCaseOther
Windows Server 2012 R2 Datacenter  Corsair 650TX  Norco RPC 4224 LSI 9261-8i 
Other
Intel RES2CV240 
  hide details  
Reply
post #8 of 165
There are several problems with the "firmware is the cause of lower performance" theory:

What is the most costly component used in a SSD? The NAND data storage chips.

What component used in SSDs has decreased in price, causing SSDs to be cheaper? The NAND data storage chips.

What component used in SSDs has evolved from using an asynchronous (one data transfer per clock cycle) data transfer at a 66MHz clock rate, to a synchronous (two data transfers per clock cycle, as in DDR memory, Dual Data Rate) data transfer at 100MHz, 133MHz, and 200MHz clock rates (the latest types up to 400MHz)? NAND data storage chips.

What component used is SSDs was said to have a shortage of availability in 2013? NAND data storage chips.

What would be the easiest way to save money when purchasing NAND chips? Buy old stock asynchronous NAND chips running at 66MHz that virtually no one else wants.

The argument that many people won't know the difference between a poor and normal performing SSD is a good point, but the poor performance could easily be caused by asynchronous NAND. IMO it is much more likely to be the type of NAND used rather than the firmware.

One of the major selling points of the SandForce controller, as sold to SSD manufactures, is the manufacture has very little (if any) design and engineering work to do themselves. That is, the manufacture uses the "reference" board for the SSD, with the SF controller and associated components already engineered and tested, including working firmware (that's another story). All the SSD manufacture must do is choose the NAND and adjust a few things for each type of NAND, probably using the SF design documents.

Very few SSD manufactures using SF controllers modify the firmware much if at all, and I doubt that Kingston is one of them. The main one we know that has modified the firmware is Intel, and AFAIK the other is SanDisk. If Kingston is using "5.06" as their firmware version number, that exactly fits the official SandForce 5.0 firmware release line, which was 5.0.0, 5.0.1, etc. IIRC, SF firmware 5.0.3 finally fixed the malfunctioning TRIM bug introduced in the new SF 5.0 firmware line.

Assuming that Kingston modified the SF firmware, and their 5.06 firmware causes poor performance, why not just go back to the 5.05 firmware? Why would they even need a new version of the firmware, if the only difference is the location of the factory assembling their SSDs? One answer to that question is to accommodate different NAND chips, like asynchronous.

Do we have examples of very similar SSDs that differ in performance due to the NAND chips they use? Yes, the OCZ Agility 3 and Vertex 3.

IMO, the OP virtually proved his point (and mine) with the quote from Kingston stating that the NAND rather than firmware would cause a difference in performance. To many details add up add up to the use of asynchronous NAND in some production runs of the V300.

I also see a different clue in the OP's statement that, "every case I've seen so far has been the 120GB capacity". With the shrinking of the NAND to the 2x nm level and less, it's possible to put 60GB of user space on one NAND chip, or possibly even more. The potential problem with that is the number of channels that NAND chip has to connect to the controller does not increase, due to the physical design and standards of the chip package. Larger capacity SSDs with more NAND chips can use all the channels the controller has, but the 120GB V300s may be limited to the use of four channels instead of the usual eight. That is pure conjecture on my part.

Believe me, I would hate it if I had purchased one of the "5.06" V300s, and I get no satisfaction from writing this. I wish it was a firmware issue, but IMO it isn't.
post #9 of 165
Thread Starter 
Quote:
Originally Posted by Lex Luger View Post

You are 99.99 percent sure its the NAND even though you basically have no proof and we can directly measure the differences in the firmware versions which are massive. I'm not saying you are wrong, but I dont see you could possible be that sure of yourself unless you physcially took apart several drives.

Yeah I don't think you understand how little the firmware affects performance. Especially on a Sandforce SSD where every manufacturer is extremely limited in the scope that they can actually change it. There is a reason why every Sandforce drive performs more or less identically.
post #10 of 165
So, I opened my taiwan 128 gb 506 firmware v300 and the chips were wiped clean with solvent. There's no real information. Kingston then printed their label on top of the chips as if they actually make the chips themselves. I bet no other Sandforce SSD does this. If you are using quality NAND, they would be no reason to wipe off the original label.

I'm now a believer. They must of switched the chips like you said okp11.
Edited by Lex Luger - 1/13/14 at 7:56am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: SSD
Overclock.net › Forums › Components › Hard Drives & Storage › SSD › PSA About the Kingston V300 SSD(Probable foul play by Kingston)