Which SSD is best?

samsung 840 pro should be the best.

The pro uses a different type of NAND. The Pro uses MLC which is older but faster. The regular 840 and the Evo use TLC which is newer and cheaper to produce.
Also, the Pro's MLC NAND is supposed to last much longer. The 2nd article estimates that MLC's expected lifespan is 60 years vs 19 years for TLC. If you do not plan on using the SSD for more than a decade, this is trivial.
Depending on how close the prices are, the Pro is the way to go. If the Evo 840 is significantly cheaper, it is not bad at all.
If you can find a regular 830 get that, it also uses MLC. The difference between the Evo and regular is that the Evo only uses 2 "bits per cell" while the regular uses 3 bpc. This makes the Evo faster.

General Overview as Currently only Samsung gives tlc technology.


http://www.anandtech.com/show/7173/samsung-ssd-840-evo-...
http://www.ubergizmo.com/2013/05/samsung-840-pro-review...

If I could only choose from the solid state drives you listed, then I would recommend the Samsung 840 Pro. Samsung has a proven track record of excellent performance and reliability. I normally recommend Samsung anyway.

I maintain the ssd database listed in a sticky at the very top of this forum section. Here is the link:

http://www.johnnylucky.org/data-storage/ssd-database.html

Scroll down to the brands and models you are interested in and follow the links to the technical reviews.

They are not necessarily faster in real world situations. There is a very big problem with synthetic benchmarks. There is a reason those benchmarks are called synthetic. They are not real. They were specifically designed to grossly exaggerate very minor differences between ssd's. They are very misleading. Companies like to manipulate the benchmark settings so that their ssd's are presented in the most favorable light. As a result over 90% of the synthetic benchmarks do not represent real world use. Consider it an advertising gimmick.

A common benchmark used in ssd advertising is Iometer which was originally developed by Intel in 1998. The benchmark has since become an open source project. The benchmark measures input/output operations per second or "IOPS". It is used on the enterprise side of the market two ways. First, the benchmark is used to determine data storage requirements for network servers. Typically Iometer is installed, settings are adjusted, and the benchmark is set to run for several days. Second, the benchmark is used to generate a series of work loads to measure the maximum IOPS an ssd is capable of. The combined results serve as a guide for data storage planning. For example, a company that had a server and 1,000 desktop pc's would set up a data drive array for 100 of the pc's and measure the IOPS during actual real world use for an entire week. The results would be used to determine data drive array and IOPS requirements for the server that fed all 1,000 pc's.

Early on Intel actually conducted research to determine IOPS requirements for consumer ssd's. Intel determined that an ssd capable of 20,000 IOPS was just about right for consumers. AnandTech, a well respected website that publishes technical reviews of ssd's, also suggested 20,000 IOPS as a practical upper limit. Results posted by actual consumers confirmed Intel's and AnandTech's estimates. The big surprise was AnandTech's gaming workload tests. Two thirds of the ssd's that were benchmarked made up a very tight performance cluster. The measurements were between 309 IOPS and 325 IOPS. That was the average IOPS during use, not the burst or peak IOPS. Other reports indicate the peak IOPS which only last for a very very brief period of time are about 4,000 IOPS.

Beginning in 2009 articles and white papers were published indicating the IOPS benchmarks were being abused. Company advertising aimed at the emerging consumer market focused on maximum IOPS instead of actual requirements for maximum optimal performance. As with other benchmarks there are numerous settings. SSD manufacturers adjusted settings and quoted a best case IOPS. None of them ever quoted a worst case scenario or a well balanced scenario. In addition, the ssd manufactures failed to disclose sufficient information explaining how the benchmarks were conducted. In that respect the IOPS are useless. The advertising does not always reflect real world performance.

With the emergence of modern 3rd generation solid state drives all of the current top rated ssd's are capable of IOPS that are far beyond the needs of multi-tasking power users and hardcore gamers. In fact, modern ssd's would be ideal for systems running several virtual pc's.

I recommend Samsung 840 Pro.

Google translation:


De er ikke nødvendigvis raskere i virkelige situasjoner. Det er et veldig stort problem med syntetiske benchmarks. Det er en grunn til disse benchmarks kalles syntetisk. De er ikke ekte. De ble spesielt utviklet til grovt overdrive svært små forskjeller mellom SSD-tallet. De er svært misvisende. Selskaper som å manipulere standardverdiinnstillinger slik at deres SSD-er er presentert i den mest gunstige lys. Som et resultat over 90% av de syntetiske standarder ikke representerer virkelige verden bruk. Anser det som en reklame gimmick.

En vanlig målestokk som brukes i SSD-annonsering er Iometer som opprinnelig ble utviklet av Intel i 1998. Referanseindeksen har siden blitt en åpen kildekode-prosjekt. Referanseindeksen måler input / output operasjoner per sekund eller "IOPS". Den brukes på bedriftens side av markedet to måter. Først er benchmark brukes til å bestemme datalagring krav til nettverkstjenere. Vanligvis Iometer er installert, blir justert, og referanseindeksen er satt til å kjøre i flere dager. Det andre er benchmark brukes til å generere en rekke belastninger å måle maksimal IOPS en SSD er i stand til. De kombinerte resultatene tjene som en guide for datalagring planlegging. For eksempel vil et selskap som hadde en server og 1000 stasjonær PC sette opp en data-stasjon array for 100 av pc-er og måle IOPS under selve virkelige verden bruk for en hel uke. Resultatene vil bli brukt til å bestemme data drive array og IOPS krav til serveren som matet alle 1000 PC-er.

Tidlig på Intel faktisk gjennomført undersøkelser for å fastslå IOPS krav til forbrukeren ssd-tallet. Intel fastslått at en ssd i stand til 20000 IOPS var omtrent rett for forbrukerne. AnandTech, en godt respektert nettsted som publiserer tekniske vurderinger av ssd-tallet, også foreslått 20 000 IOPS som en praktisk øvre grense. Resultater lagt ut av faktiske forbrukere bekreftet Intels og AnandTech estimater. Den store overraskelsen var AnandTech gaming arbeidsbelastning tester. To tredjedeler av ssd-er som ble testet gjort opp et svært stramt ytelse klynge. Målingene var mellom 309 IOPS og 325 IOPS. Det var den gjennomsnittlige IOPS under bruk, ikke briste eller peak IOPS. Andre rapporter indikerer toppen IOPS som bare vare i en veldig veldig kort periode er ca 4000 IOPS.

Fra og med 2009 artikler og white papers ble publisert indikerer IOPS benchmarks ble misbrukt. Selskapet reklame rettet mot det voksende forbrukermarkedet fokusert på maksimal IOPS i stedet for faktiske behov for maksimal optimal ytelse. Som med andre referanseverdier det finnes mange innstillingsmuligheter. SSD-produsenter justeres innstillinger og siterte et best case IOPS. Ingen av dem noen gang siterte en worst case scenario eller en godt balansert scenario. I tillegg produserer SSD unnlot å avsløre tilstrekkelig informasjon som forklarer hvordan benchmarks ble gjennomført. I så måte er IOPS er ubrukelig. Reklame ikke alltid gjenspeiler reell ytelse.

Med fremveksten av moderne tredje generasjon solid state drives alle av dagens topp karakter ssd-er er i stand til IOPS som er langt utover behovene til multi-tasking avanserte brukere og hardcore gamere. Faktisk ville moderne ssd oss ​​være ideell for systemer som kjører flere virtuelle PC-er.


I got a question for you, why did you tell me this? I just wanted to know whats fastest, that will have fastest boot time, load games fastest etc. So please tell me, can you write "top 5 best SSD's"? So I got more then only one too choose between?

Define best . Let me use an example....what is the best tool in your toolbox.... the hammer, the screw driver or the pliers ? I think you'll agree that the answer depends on whether ya wanna bang in a nail, put in a screw or pull off the fingernails of the guy who built ya last store bought PC.

Highest performance in sequential reads ?
Highest performance in Random 4k reads ?
Highest performance in ATTO ?
Highest "bang for the buck" ?

There's so little difference between the top tier 1 SSD's that the "best" one to buy is the one from the list above that's on sale the day of your purchase. I have installed several vertex 3 Max IOPS, Muskin Chronos Deluxe, Samsung 840 Pro and can't say I in any way prefer one of the other.