SSD Thread (Benchmarks, Brands, News, and Advice)

That Solid 2 a user was asking about last week seemed to be good for the cost-GB-performance. Indillix controller with slower memory than the Vertex/Agility but that controller has more memory channels so it sorta overcomes the Agility in things

Edit

Nikola Tesla would be proud of that technology

 

I just went ahead and purchased the Intel X25-m G2 160GB SSD from Amazon.com

It'll be here tomorrow. Tonight I'll prep my machine for the new toy

http://www.amazon.com/Intel-Mainstr...2?ie=UTF8&s=electronics&qid=1265906732&sr=8-2

 

Do the newer Samsung firmware revisions still have background garbage collection?

 

Yes TRIM and GC.

 

So my assumption is that, if I use a AHCI driver that does not pass TRIM commands from windows to the drive, I will still benefit from the built in firmware garbage collection. Is this accurate? I have a fully patched Corsair P128.
I would prefer to use the newer Intel drivers but having been sticking with the default MS drivers because of the lack of TRIM functionality with Intel drivers to date.

 

Yes GC should still work

 

Anyone seen/tried this yet? http://www.tomshardware.com/news/SSD-Performance-Utility-Tweak-Windows,9655.html

 

Just FYI - ecost.com has the lowest price on this...I think I paid like $455.00

Edited to Add: Now higher - about $475.00

 

Hey Cape! I noticed this (and chuckled a little when I did).

You know what they say; when you're thinking about 'tweaks' just don't do it.

Best advice ever.

It does nothing for the SSD in question, it simply cripples the O/S (to make it work slightly better with a 'bad/old' SSD.

 

well, it's the geek part of the consumer world of "call 0123456 to get released from all your pains thanks to sir lovejoy's power of the gods of polyricos" style ads running all night long on tv (between.. those.. other .. ones )

the "we can help you with some magical things to make your life better" crap industry is one of the building blocks of all our industries..

this is the geek version: system tweaking.

 

Hey tiller, yes, I thought that might be it. Have no SSD right now so did not install it. Waiting for the thrid generation, I guerss. even my i7 upgrade went bad.

I sent the Studio XPS 8100 back to Dell as it was not up to the quality I thought it should be. So, I am back to my core2 for awhile and there I shall stay. For now

 

Wow Cape, bad luck there with the XPS.

What kind of quality issues? Keyboard/touchpad/chassis? Or was the i7 not the CPU you thought it should be?

I too keep saying to myself that the G3 SSD's are what I'm waiting for - but that is only because the G2's are so scarce/pricey right now!

 

Could I work while (SSD) Intel Toolbox is running in the background? I set the toolbox to run once a week (Thursday), and last night I forgot it was running and I unzipping some files while the toolbox was running and my system was crawling...

 

I wouldn't do anything disk intensive while it was running, just set it to run at a time you are not usually working and postpone it if you need to work during that time.

I don't like to interfere with any maintainance tasks and I don't want maintainance tasks interfering with my work.

It is counter productive if routine performance maintainance causes performance issues.

 

I believe it works better if you leave the computer alone while the toolbox runs. I noticed a difference when I did that vs using programs while the toolbox was running.

 

What he said. It takes all of, what, a minute?

 

Edit: It does take less than a minute.

 

I am considering buying a SSD for my Asus G50VT-X6. Since it already has a 320GB HDD which is more than enough for data, the SSD will only need to store the OS and a few programs so 40-60 GB would be more than enough for me.

I was considering the Intel X25-V 40GB for $130 but for the same price (after rebates) I can get the OCZ Agility 60GB.

http://www.newegg.com/Product/Product.aspx?Item=N82E16820227461

Feedback from anyone who has used the drive would be great as would suggestions for other drives. My budget is no more than $150.

 

I would go for the Intel X-25V, it performs better in almost every real world benchmark except sequential writes. I also like the Intel SSD toolbox features. I have tested the X-25V and it got 7.7 WEI disk score. The Agility gets 7.1 or something like that.

 

I know the Intel is better. But compared to the OCZ, is it worth losing 20 GB of space for?

 

if you know they won't matter?

i said yes in case of my media center. i would say no in other situations. just depends if you need/care about the storage.

 

I think it is worth losing the 20gb of space because I value performance over space on my OS and apps drive. If you have apps that you need to use frequently that will exceed 80% of the Intel then you will be needing more space. It is fairly easy to juggle apps between the HDD and SSD because they install/uninstall quickly on the SSD. You can use the SSD toolbox to maintain performance after switching back and forth to keep performance up.

 

http://www.newegg.com/Product/Product.aspx?Item=N82E16820227462

Good price on the OCZ Agility

Man i want to get an ssd drive but im still undecided on which one .My hpdv6 i7 laptop is (from what i read) limiting the sata to sata1 when it supposed to be sata2 .I have been reading alot and now that the intels are better but the drive thats on sale has 40 more gigs than the intel80G2,maybe get the g2 and get a 32gig secure Digital Card later.Any thoughts

 

There should be some deals for SSD's this month as alot of enthusiasts are holding off for the Gen3's to be released. OCZ is scheduled to retail the VertexII's with Sandforce controller in March. Hopefully the Intel G2's will then come down in price. I am planning to get a new SSD in March when my ex-wifes taxes and 401k is cashed in. She kicked the bucket back in October and I get to file her taxes and cash her 401K. getting $3900 back on her taxes. :

 

Sorry about your ex .Well i can wait a bit.i just hope they release a bios update to fix the sata problem.They advertise sata II but can only get sata1 speeds..I will jusy keep checking this forum for updates.

 

If that HP DV6t quad is limited to SATA I, I would send it back and get something else but I doubt that it is limited. That HDTune benchmark that you linked from the other DV6 owner with the Vertex was over 150mb/s. It is probably a SATA driver issue or something easily fixable.

 

thanks alot man .i will wait and check the pricing in march.i also sent hp an email about it .thanks

 

One question, do you guys really think that sequential speeds really matter that little?

On the X25-M, 70-100MB/s write speeds do not seem to be enough of a hindrance to knock it off from being one of the fastest, if not the fastest SSD, but the X25-V's 40MB/s seems too much of a knock off. The sequential write is so low that it would limit random limit throughput at certain sizes.

Yes, latency is very important. But bandwidth would have to be reasonably high too. X25-M is arguably the best SSD drive out there, but its not without its flaws. Garbage Collection that happens at write time isn't fast enough to hold off not needing things like TRIM by itself, and the sequential write is too low. In fact one of the reviews concluded that the IOPS are so high that it can choke sometimes.

G3 needs
-Lower IOPS to match its capabilities to sustain the IOPS, X25-M is not a server drive
-Concurrently, a better GC to minimize degradation without using TRIM
-Sequential write speeds that are substantially higher. 250MB/s+ would be fair

From some presentations I have seen, the points will be addressed. It's not here yet though.

 

IntelUser,

Starting back in post:
http://forum.notebookreview.com/showpost.php?p=5860693&postcount=3483

I think it was concluded that Tomy B. should be more than impressed with the 40GB 35MB/s write limited Intel (vs. his Super Talent MasterDrive SX 64 GB SSD). Sure, it was slower by around 3 minutes (I think) extracting a 7z 5.43GB CS4 Suite program file, but that is on a twenty minute time span and with a sequential write speed that is more than 3 times slower than the Super Talent SSD.

I know that the sequential speeds are not was is affecting these results directly, but does show how a few simple benchmarks can really skew real world results.

I also agree with you though about bandwidth needing to be higher though for the true 'high performance' SSD's.

Actually, what I am waiting to see is the Intel version of an SSD with 'Balance' in it's performance characteristics. I hope the G3's deliver that balance. I don't care if I need to buy a new notebook (with SATA 3) to get the performance SSD's have promised for so long (and I simply have not seen in any live system - apart from one being 'benchmarked').

Yeah, its not here yet - but hopefully, this year will be the time when spending ridiculous amounts of cash (no problem with that! ) will actually give a semi-reasonable stab at getting an indication of ridiculous performance increases (compared to my 4x VRaptor setups).

This will happen (and you know, I'm only rooting/betting on Intel here) when the competition gives Intel a feeling that they are being eclipsed - until then, Intel is more than happy to dole out bite sized performance increases to the masses (and I don't blame them at all, from a business perspective).

So, with all the above in mind; here's hoping that the Indilinx II's, SandForce, CFast and all the other 'newer' contenders smoke Intel's current offerings on performance and on price - I'm comfortable to wait for Intel's answer to them (which will be once again comparable to the world vs. Core Duo in the CPU arena like when they were first introduced in 2006).

 

depends on the use cases. but in general, yes, they don't matter much.

 

I don't think those 512k sequential writes are that important either because they happen infrequently in normal usage. The X-25's standout by a wide margin in random writes at 64k and lower which happen more commonly in normal usage scenarios.

Maybe someone more knowledgeable in OS and application read and write sizes could chime in on what would be the best happy median for performance/k size. How many writes are 512k sequential and what applications utilize 512k sequential writes? If you look at the usual suspects real world benchmarks, the Intel drives usually win over drives with double the 512k sequential write speed.

 

sgilmore62,

I'm sure Intel has done those very same statistics for O/S r/w's. That is why they concentrated on boosting the 4K reads.

Did I misunderstand? (You or Intel)?

 

You mean Core 2 Duo.

Starting from 16KB already its very close to reaching sequential speeds. The write advantages diminish greatly even at that size. According to Intel, around 60% of the consumer usage is random 4KB and rest are greater than 4KB. 40%, while smaller than 60%, is still a significant amount.

I know how people got to realize highest numbers aren't the best thing, but usually people are very lopsided, they almost "team" with one result and never in the middle. Having the balance is the key, and I know people have been burned by OCZ before, but everyone makes mistakes, even companies like Intel. I have used OCZ products and while they might be iffy in some areas, I had fantastic experience with them along with excellent technical support.

If I was looking in the X25-V price range, I'd get an OCZ. 20GB more at the same price? That 2-3x higher random writes and IOPS will be about as useful as faster sequential. It's just like the CPUs. High end, Intel no doubt, but low-end, they cripple their products too much and competitors have really good pricing there.

 

well, the -V has proven to be in the ballpark with others. and still has much better numbers in about anything except sequencial writes.

so, the question is just, is the V better balanced for it's jobs than others? and i can say, now that i own one, yes. it lives up to the other intel ssds quite much, and delivers all you need except high write speeds. still, setups, installations, about anything runs very fast, as you NEVER reach high write speeds in those on other ssds, just as well.

 

Maybe Intel is too fast?! :laugh:

Everything is so obvious if You look at last four results:

Code:

> Average Read Time with Sequential Writes	ms	0,355	7,9		0,953	7,7
> Latency: 95th Percentile			ms	0,973	7,9		3,118	6,9
> Latency: Maximum				ms	14,610	7,9		3,880	7,9
> Average Read Time with Random Writes		ms	0,396	7,9		0,980	7,9

and some numbers for X25-M 80 GB:

Code:

  1.342 ms - 7.4
  2.593 ms - 7.3
281.867 ms - 7.1
  1.927 ms - 6.9

 

That still means Vertex is better at X25-V price range.

Most of you got your opinions about Intel vs. others from benchmarks. File transferring and zipping speeds will change with sequential speeds and as I said earlier, 40% of everything you do is not 4KB.

Here's the Summit drive with extremely low random write speeds(much lower than Vertex) and really high latency: http://techreport.com/articles.x/17269/4

Loading times are like the X25-M and they trade blows in other benches. How would you explain that? It's random writes are slow right? It's only one metric of performance. People say Samsung drives are terrible because of slow random writes but they did not have degradation problems(not the "break-in" degradation, but even worse after that), and the new controllers do pretty well without really standing out in random reads/writes.

 

most file transfers will be lower anyways, as most users user wlan to transfer, or even downloads or anything.

you try to create an elephant out of a mosquito. in MOST cases, the V provides great performance for it's price. and that's why it's made that way.

but it doesn't matter if you can't accept that. i can, and get great power out of my V.. and the knowledge that it works the same as my other drives in reliability, installation etc. it's just the same drive.

 

Intel is better at random writes or random reads, but it looks like Samsung is better at parallel reads and writes and, as we all know, that's what matter.

 

haha, and exactly there it fails which is why my samsungs got my system to block during the AS bench, and failed to deliver any usable numbers. none of the intels does that. they deliver about exactly X times the performance on non-parallel reads and writes, with X being 10 for 80 and 160gb, and 5 for the 40gb.

 

The Samsung drives choke on anything past queue depth 8 for parallel reads and writes while the Intels are still gaining speed past queue depth 32.

Is it most likely that the least saavy tech users would benefit the most from the Intel drives since they tend to open and run multiple applications simultaneously without deference to performance issues?

I would think that most experienced users would try to work or play with only one or a couple of intensive applications at a time and could get by with an SSD less suited for multi tasking like the Samsungs.

 

Ugh thinking of pulling the plug on the 160GB Intel X-25M. I'm getting impatient waiting around for the higher capacity Intel drives. Nothing is coming out till at least the end of this year right?

 

Probably that's a reason why I didn't feel any difference doing usual staff on notebook.

 

Yeah, I noticed something else about the Intel SSD's while benchmarking. Everybody talks about how crappy the Samsungs are at 4k random writes and how great the Intel drives are with that benchmark. The Samsung drives will give you a burst of 10-14 mb/s for a few seconds then taper off to abot 4-5mb/s by the end of the 3 minute test. The Intel drives start off at HDD 4k random write speeds of about .3mb/s for a few seconds then gradually build to 40mb/s by the end of the 3 minute test.

How often do 4k random writes last for more than a few seconds and what application usage scenarios would the average user realize a benefit from the Intel drive over the Sammy?

 

That's a very good question Sgilmore. I have an even more confusing/intriguing one to ask: if the Intel's powerhouse benchmark of random writes takes a few seconds to build to 40 mb/s, how often does normal workstation usage utilize 4k random writes in normal scenarios? and more importantly, how MANY 4k random write scenarios happen in a row, and is it this repetition of random writes in normal usage what symbolizes the delay before hitting the sweet spot in the benchmarks?

 

well. for small workloads, it doesn't matter if it's slow, as it isn't much anyways. big workloads mostly happen on installations of applications, windows updates, and similar stuff. and there, it sure helps.

but yeah, interesting.

 

I would think it would make a big difference to gamers who would be installing/uninstalling games frequently to conserve space. GTAIV takes up alot of space and takes awhile(45min) to install on my Summit. I would like to see how long it takes to install on an Intel. With the Intel, after uninstalling the game you can run the SSD toolbox and be good as new right away -- with the Samsung SSD, while it has TRIM, there is no such tool and the cleanup happens at idle time if ever.

 

http://techreport.com/discussions.x/18272

Does this mean that the upcoming 25nm drives will be cheaper, have greater capacities, sequential writes of 300mb/s and be available this year?

I can understand them not wanting to support 6gbps standard right away because there will not be any notebooks with it this year and probably only enthusiasts in the desktop markets.

 

That is exactly the question I was wondering when I was shopping around for my SSD's. Couldn't find a straight answer to that question thought, as to how do O/S's, apps and drivers distribute their reads/writes in both terms of size, and in terms of random or sequence data access. If a hammer can punch so big nails but I only use small ones, what's it worth for me buying that said super-power-hammer ?

In my case, because I couldn't afford the Intels, that was back early october 2009, and because I still wanted to switch to SSDs, and because I managed to get 2 X Sammies in RAID0 for ish ±$800CAN, using barely 96GB out of the 477GB available, I chose to go with them even if their 4K write speed was very low, not to say stihty, and even if TRIM was not in sight at that time, because, believe it or not, I swear to God, ITGC works like a charm, no degradation on my drives, I know, they're already degraded, and I got plenty of space for wear leveling to do it's work without impacting my rig's speed.

This being said, Windows boots in ± 27 ~ 33 seconds, while it used to take ± 42 ~ 58 with my previous 2X7200RPM RAID0, that's a HUGE improvement.

When I open my Outlook, it takes less than two seconds to be fully operationnal, while it used to take ± 4 ~8 seconds with my previous rig, this is HUGE improvement, my outlook.pst file is almost 400MB...

I DO INDEED install/uninstall programs on a regular basis, and it does fly compared to my previous setup. If this can help anyone here making a choice, I would be willing to test program installs and uninstalls if you want.

Cheers !

 

I would think the possible Postville refresh would bring those speeds.

 

that's for x-25v? x-25m starts at about 1m/s. and intel SSD seems to demand huge cpu usage when testing multi-thread random read/write. i put x-25v in hp netbook and found its cpu usage hits 100% when testing this.