Can nothing beat the 840 Pro?

Ketchup

Elite Member
Sep 1, 2002
14,559
248
106
Just looking at the review for the new Kingston SSDNow V300. As always, I am a little surprised with the results. SSD was gaining speed and space by leaps and bounds, then coming to a sudden halt. The Samsung 840 Pro review was back in September, and it still handily beats everything out there.

What gives?
 

mrpiggy

Member
Apr 19, 2012
196
12
81
The reason there is no leaps and bounds in speed anymore is that there is no SATA4 spec on the close horizon. The current crop of SSD's are already approaching close to maximum with SATA3. Hence, most companies, rather than dump a lot of money and research into getting another 5-10% on an already close to max interface are looking into other solutions like PCIe with much more bandwidth available for speedier products.

Samsung hit a home run with the 840 pro's performance, but it's still a small specialty market that demands speed above all else. It's not the type of SSD product that the giant OEM market (or even Joe Average cares about) is generally interested in, but that's where all the large volume profits sales are made. In these markets, price and then size of the SSD is king as long as it performs "decently" in speed.
 

Ketchup

Elite Member
Sep 1, 2002
14,559
248
106
The reason there is no leaps and bounds in speed anymore is that there is no SATA4 spec on the close horizon. The current crop of SSD's are already approaching close to maximum with SATA3. Hence, most companies, rather than dump a lot of money and research into getting another 5-10% on an already close to max interface are looking into other solutions like PCIe with much more bandwidth available for speedier products.

Samsung's Pro is more than 5-10% faster than many other drives, so I don't really see that as an excuse. The Pro is about 40% faster than the Kingston in yesterday's Anandtech Storage Bench section, a difference that would not be seen if limited by SATA3.

Samsung hit a home run with the 840 pro's performance, but it's still a small specialty market that demands speed above all else. It's not the type of SSD product that the giant OEM market (or even Joe Average cares about) is generally interested in, but that's where all the large volume profits sales are made. In these markets, price and then size of the SSD is king as long as it performs "decently" in speed.

SSD was really developing for the first few years, and the adaption rate has grown significantly since then (which is basically from zilch). They are being found in more OEM machines than ever, so I'm not sure where you are coming from here.

http://www.bestbuy.com/site/Asus+-+1...&skuId=6869315
 

Eeqmcsq

Senior member
Jan 6, 2009
407
1
0
Maybe SSD controller development has plateaued because they've used up all of the easy tricks to increasing sequential and random read/write speeds.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
The reason there is no leaps and bounds in speed anymore is that there is no SATA4 spec on the close horizon. The current crop of SSD's are already approaching close to maximum with SATA3. Hence, most companies, rather than dump a lot of money and research into getting another 5-10% on an already close to max interface are looking into other solutions like PCIe with much more bandwidth available for speedier products.

Not sure if you know about this or not, but SATA I/O decided not to release a SATA 12 Gb/s (aka "SATA 4"). Instead we will be getting "SATA Express".

http://www.sata-io.org/documents/SATA_Express_Article_2012.pdf

Additional info here on SATA Express ---> http://www.anandtech.com/show/6294/breaking-the-sata-barrier-sata-express-and-sff8639-connectors

Pretty much all high-end client SSDs have no issues saturating the current 6Gbps SATA interface. We've talked about a move to PCIe based SSDs but these are the connectors that will do it.
 

Ice_Dragon

Senior member
Nov 17, 2011
236
0
71
Samsung has good reliability too generally. The 470 was good, the 830 was good, and now the 840 Pro seems to be good.
 

mrpiggy

Member
Apr 19, 2012
196
12
81
Samsung's Pro is more than 5-10% faster than many other drives, so I don't really see that as an excuse. The Pro is about 40% faster than the Kingston in yesterday's Anandtech Storage Bench section, a difference that would not be seen if limited by SATA3.

And so what? SSD maker's should spend millions, simply to have the fastest SSD to sell in a smaller market at higher cost? Especially considering everyone says the traditional PC market is shrinking rapidly.

The burgeoning market for SSD's is to Average Joe coming from a slow hard drive, not Mr. enthusiast who already had a slower model SSD and needs more bragging rights. If a SSD producer can make $10 per unit selling ten million to average Joe and OEM's with a decent performing SSD and lower back-end overhead (R&D, manufacturing, etc) or they can make $20 per unit selling 50,000 to enthusiasts, and having higher back-end expenses, which do you think they will cater to? Average Joe (and OEM's) cares far more about 5-10% in price difference + larger capacity than 5-10% in pure performance (hence, why HD's still sell so many).

Don't get me wrong, I like performance too. But don't delude yourself that the majority of people care about 5-10% of anything beyond price point and size when it comes to PC storage.

SSD was really developing for the first few years, and the adaption rate has grown significantly since then (which is basically from zilch). They are being found in more OEM machines than ever, so I'm not sure where you are coming from here.

http://www.bestbuy.com/site/Asus+-+1...&skuId=6869315

Have you taken a look at the SSD's that come in OEM machines? They are "not" the high-end SSD's on the market. They are mid-range, price-is-right models that OEM's will buy tons of. The "bread and butter" of sustainable profits for component suppliers. This is a far more stable profit model than catering to the SSD-of-the-day enthusiast crowd.

Notice how the spec sheet of your example only states "256GB SSD". It does not say "256GB Samsung 840 Pro", and I will bet you hard cash absolutely no other similarly priced model laptop from any other manufacturer will have a Samsung 840 Pro in it either despite it being the fastest consumer SSD.
 
Last edited:

Ketchup

Elite Member
Sep 1, 2002
14,559
248
106
Notice how the spec sheet of your example only states "256GB SSD". It does not say "256GB Samsung 840 Pro", and I will bet you hard cash absolutely no other similarly priced model laptop from any other manufacturer will have a Samsung 840 Pro in it either despite it being the fastest consumer SSD.

What's you point? Intel's current Extreme Edition chips are the fastest you can buy, you would be hard pressed to find one in a consumer desktop, but we are going to see something faster from them in a couple months.

And I have never seen a retailer tell you the make and model of the hard. You are just getting silly at this point, not sure why.

As SSD's are new to the market (compared to CPU's), one would think the speed and capacity increases would have kept going a bit longer than they have.
 

Hellhammer

AnandTech Emeritus
Apr 25, 2011
701
4
81
The burgeoning market for SSD's is to Average Joe coming from a slow hard drive, not Mr. enthusiast who already had a slower model SSD and needs more bragging rights. If a SSD producer can make $10 per unit selling ten million to average Joe and OEM's with a decent performing SSD and lower back-end overhead (R&D, manufacturing, etc) or they can make $20 per unit selling 50,000 to enthusiasts, and having higher back-end expenses, which do you think they will cater to? Average Joe (and OEM's) cares far more about 5-10% in price difference + larger capacity than 5-10% in pure performance (hence, why HD's still sell so many).

And that's why Samsung has the 840 and 840 Pro. The Pro can be sold at a higher price (and profit) because it's faster and there are people who are willing to pay a bit extra for the best-in-class performance.

Usually the enthusiast or otherwise special market is where the profit is made. The high-volume mainstream market has more players and less product differentiation and is hence closer to a perfect competition state (market share is divided between multiple players and one player cannot affect pricing). Of course, perfect competition is just a theoretical model and it does not exist in real world but the enthusiast market is more oligopoly-like as only a few players have the ability to compete there (others don't have products competitive enough).

Have you taken a look at the SSD's that come in OEM machines? They are "not" the high-end SSD's on the market. They are mid-range, price-is-right models that OEM's will buy tons of. The "bread and butter" of sustainable profits for component suppliers. This is a far more stable profit model than catering to the SSD-of-the-day enthusiast crowd.

Samsung SSDs are actually very popular among OEMs. For example Apple uses them and quite a few Ultrabooks have Samsung's mSATA SSDs.

OEMs care about much more than just price, steady supply being one of the most important ones. It's no use to choose a supplier that's slightly cheaper if there's no certainty that they'll be able to meet the demand. SSD manufacturers without fabs are not popular among OEMs because their supply is dependent on the supply of NAND. If there's a slowdown is NAND production, the 3rd party manufacturers will be the ones suffering as the NAND manufacturers will prioritize their own products and ship the leftover NAND to others. Lately there's been a NAND shortage and that has affected many of the smaller OEMs.
 

mrpiggy

Member
Apr 19, 2012
196
12
81
What's you point? Intel's current Extreme Edition chips are the fastest you can buy, you would be hard pressed to find one in a consumer desktop, but we are going to see something faster from them in a couple months.

And I have never seen a retailer tell you the make and model of the hard. You are just getting silly at this point, not sure why.

As SSD's are new to the market (compared to CPU's), one would think the speed and capacity increases would have kept going a bit longer than they have.

You asked for reasons why no one is making "leaps and bounds" in speed anymore for standard SSD's. I gave you several reasons both technical, like lack of a SATA spec faster than current SATA3 and market conditions. All you come back with is "but SSD's are new", or "but Intel keeps coming out with faster processors, so SSD's should too" or other stuff that has nothing to do with "SSD" markets, production, research costs, or anything about NAND in general.

Using Intel processors as as reason SSD's should be faster is erroneous. Intel also requires you to purchase a new motherboard every time they come out with a new processor in a different socket configuration. SSD makers don't make you upgrade everything else just to run a SSD.

Sorry I can't provide counter to your emotional argument of "but that's the way it should be".
 
Last edited:

mrpiggy

Member
Apr 19, 2012
196
12
81
And that's why Samsung has the 840 and 840 Pro. The Pro can be sold at a higher price (and profit) because it's faster and there are people who are willing to pay a bit extra for the best-in-class performance.

Usually the enthusiast or otherwise special market is where the profit is made. The high-volume mainstream market has more players and less product differentiation and is hence closer to a perfect competition state (market share is divided between multiple players and one player cannot affect pricing). Of course, perfect competition is just a theoretical model and it does not exist in real world but the enthusiast market is more oligopoly-like as only a few players have the ability to compete there (others don't have products competitive enough).

Actually, while there may be a slightly higher profit per unit on the pure enthusiast side of the SSD consumer market, the lack of volume sales and higher back-end production costs means these high-end "consumer" SSD profits are far lower than mid-range high volume, but lower margins per unit sales profits.

Just like GM may make more profit per unit sale on a Corvette than a Camaro (or other lower tier model), they will sell far more Camaros which generally makes the overall product profit higher for the lower tier vehicle. The top-tier Corvette has a lot more back-end R&D/production costs than the simpler Camaro so when you include back end cost + lower sales volumes, even though it probably makes more per each sale of Corvettes, Gm is making much more overall with the Camaro. (FYI: the two cars just chosen at random for this analogy. I don't know the actual figures for these vehicles).

The actual push of SSD's producers is into the OEM "enterprise" markets where the profit margins are MUCH higher than the cutthroat consumer market. Only the big players even get to enter this Enterprise-level playground as the back-end costs are VERY high for testing and validation.


Samsung SSDs are actually very popular among OEMs. For example Apple uses them and quite a few Ultrabooks have Samsung's mSATA SSDs.

I never stated that Samsung's are not popular with OEM venders. I stated top-of-the-line "Samsung 840 Pros" are not popular with OEM venders. Simply because it will hurt the PC maker OEM's razor-thin profit margins significantly with any, even-small, price increases at the component level. Chances are with an OEM laptop,with an SSD, it will be an 830 or maybe a regular 840 if lucky or a similarly cheap Toshiba unit. It's "not" going to be a enthusiast high-end "Pro" model SSD.

OEMs care about much more than just price, steady supply being one of the most important ones. It's no use to choose a supplier that's slightly cheaper if there's no certainty that they'll be able to meet the demand. SSD manufacturers without fabs are not popular among OEMs because their supply is dependent on the supply of NAND. If there's a slowdown is NAND production, the 3rd party manufacturers will be the ones suffering as the NAND manufacturers will prioritize their own products and ship the leftover NAND to others. Lately there's been a NAND shortage and that has affected many of the smaller OEMs.

While I disagree with you on the overall profits of the enthusiast level SSD's compared to the lower tiers, I do agree completely with this last part. I forget mention that when I discuss OEM or VAR market SSD's I don't even consider a non-NAND producer SSD maker as a real long-term player in the OEM market anymore. They are too subject to the whims of NAND supply/pricing.
 
Last edited:

Hellhammer

AnandTech Emeritus
Apr 25, 2011
701
4
81
Actually, while there may be a slightly higher profit per unit on the pure enthusiast side of the SSD consumer market, the lack of volume sales and higher back-end production costs means these high-end "consumer" SSD profits are far lower than mid-range high volume, but lower margins per unit sales profits.

I'm only posting what I've heard and seen from manufacturers. Usually (not always, of course) the enthusiast-level products are the biggest source of profit.

While you're right that R&D costs for high-end products are often higher, keep in mind that R&D is a fixed cost - the cost stays the same no matter how many units you sell. When you sell millions of units, the R&D cost per unit becomes rather small.

The attraction of designing a high-end product is that you actually get a mainstream product with very little extra R&D. Like in the case of Samsung SSDs, the same controller is used in the Pro and non-Pro. Basically, you spent a little more on R&D and got two products for two different markets. Had you cut down R&D, you would only have gotten a mainstream product but now you get both.

The actual production cost differences between mainstream and enthusiast products are rather small too. Each wafer has different qualities of NAND, which are separated by binning. By having low and high-end products, you can use the best chips in the high-end products where their performance and endurance may actually matter, whereas the lower quality chips can go in the mainstream model as they are still "good enough". Hence the production costs are pretty much the same - if you only had a mainstream product you would end up shoveling the high-end chips in there too, without actually taking advantage of their higher quality in terms of higher price.

If the mainstream market was more profitable, then everyone would stay there because why invest more when it's going to return less profit. OCZ and Nokia are great examples of this. OCZ prior strategy was to be in the mainstream market and compete in price. What happened was that they got into a big financial trouble because the margins were cut down to a state where they no longer covered expenses. They faced the laws of economics: When volumes go up, more players enter the market and prices (and profits) go down. With very little product differentiation, none of the players can grab a major market share and hence the volumes are also moderate (okay, uni entrance exams are making me crazy as all I see is economics everywhere, but it does apply here). In the high-end market, products can be differentiated more easily and buyers will also pay the premium for the differentiation.

Same happened to Nokia. They still sell dozens of millions of phones but most of them are low-profit feature phones. The real money is made in the smartphone business because margins are much, much higher.

Chances are with an OEM laptop,with an SSD, it will be an 830 or maybe a regular 840 if lucky or a similarly cheap Toshiba unit. It's "not" going to be a enthusiast high-end "Pro" model SSD.

SSD 830 is still available for B2B so that's what's found inside most. However, it wasn't cheap back in the day (before the sales started).
 
Last edited:

Torn Mind

Lifer
Nov 25, 2012
12,078
2,772
136
Well, OCZ also is in the dumps now because their unreliability alienated a lot of repeat buyers and potential buyers when they did have the price lead.

Upping the price does work up to the point elasticity remains inelastic or unit elastic. But once it enters the elastic part of the demand curve, sales will drop off real fast.
 

Ketchup

Elite Member
Sep 1, 2002
14,559
248
106
Well, OCZ also is in the dumps now because their unreliability alienated a lot of repeat buyers and potential buyers when they did have the price lead.

Upping the price does work up to the point elasticity remains inelastic or unit elastic. But once it enters the elastic part of the demand curve, sales will drop off real fast.

They really shot themselves in the foot, in my eyes. They had some drives with really good performance, but I would NEVER buy one, even though some of their models seem to be fine.

How many times can one say "It's better now."?
 

cheez

Golden Member
Nov 19, 2010
1,722
69
91
OP: That's right. Nothing will EVER beat the 840 pro. Ever. Not in a million years.
This is the exact type of sarcasm we don't need. Garbage worthless post.


OP, I like your post and I respect for your thread. Good job.


cheez
 

lilrayray69

Senior member
Apr 4, 2013
501
1
76
Doesn't SATA3 go up to about 750 MB/s? The 840 Pro says it's max is 530 MB/s, and that's of course it's ideal which you probably won't see in actual performance
 

Cerb

Elite Member
Aug 26, 2000
17,484
33
86
Doesn't SATA3 go up to about 750 MB/s? The 840 Pro says it's max is 530 MB/s, and that's of course it's ideal which you probably won't see in actual performance
No. The cable's singal is going at 750MB/s, but for every 8 bits of command or data, 10 bits of signal are sent, encoded so as to reduce DC bias, by evening out the number of 1s and 0s (at MHz, DC is AC, and sending 010 as 0101 keeps things more balanced over time), which is one of many methods used to keep pushing signals faster. You can safely assume that every PC interface in which the Gbps speed and GBps bandwidth are off by a factor of 10 use this same scheme.
 

Makaveli

Diamond Member
Feb 8, 2002
4,976
1,571
136
RAID. Bottleneck removed.

Agreed aslong as you do proper back ups :)

However if you raid 0 two of these drives and your doing 1000/1000 you need a source that you can write or read to that is equally as fast if your transfering data.
 

Cerb

Elite Member
Aug 26, 2000
17,484
33
86
What would it take to increase single queue 4k speeds?
Faster controller, faster flash, and lots of emphasis on 4KQD1 performance. At QD1, more NAND channels can't do too much, except provide bandwidth. Every write or read must complete before the next one can begin.

Take the Crucial C300. Best 4KQD1 read scores around, right up until the Vertex 4 and 840 Pro. But, practical performance kept going up. Even desktops and notebooks can take advantage of NCQ, even though they will rarely go to depths above maybe 4 or 5. More emphasis has been put on improving performance with overlapping IO, than purely synchronous IO, mainly because (a) synchronous IO as a limitation represents a minority of users, (b) near-linear gains from utilizing multiple IO channels are realistic for sequential transfers, and (c) with queuing and those multiple NAND channels, the chances are good that you may have some channel that can be read from or written to right now, even if others are busy (FI: busy writing to channel 2, reading from 3, a new read's data may exist on 5 (idle), and a new read that must wait on 2's write: not counting SATA overhead, that's 4 requests in the time of 2), resulting in better average service times and bandwidth, even if the implementation of that may reduce synthetic/peak performance scores.

Writes take much longer than reads, so I suspect the ever-higher write performance has been to try to balance out the actual time spent writing, v. time on any channel available for reading.
 

Torn Mind

Lifer
Nov 25, 2012
12,078
2,772
136
They really shot themselves in the foot, in my eyes. They had some drives with really good performance, but I would NEVER buy one, even though some of their models seem to be fine.

How many times can one say "It's better now."?

Well, I personally was traumatized in perhaps half an hour to 45 minutes. I was just fixing up a netbook's keyboard connector and the OCZ drive booted up twice with no issue. I forgot to take out the battery on the third try, but regardless, after I finally fixed the keyboard and mouse hinge, the thing wouldn't boot even though it was detected in BIOS. I don't know if static caused a surge or something while the battery was plugged in, but for a drive to be this undurable (disk was on the other side), well, I won't support such s***, ever.