LinkBack URL
About LinkBacksTo be fair, they are a lot more than that. But they're still expensive for what they are.Originally Posted by pp05
To be fair, they are a lot more than that. But they're still expensive for what they are.
Flicking through this thread focus seems to be mainly on OCZ ones, is there a particular reason for that? I am picking a SSD for my system so would be interested in any thoughts, I was mainly looking at Intel ones since they seem to have the best read/write MB/s speeds (out of SATA SSDs anyway, I am undecided whether I'm going to go for one of the PCI ones since they have 3x+ the speed of SATA ones it seems)
Don't see why you'd notice the speed difference with a PCI one rather than SATA, and last time I checked the PCI's weren't bootable, which is a pain in the neck.
OCZ and the like were traditionally cheaper than the Intel equivalents, although not quite as good, so that's why they were in the list. With the new G2 cheaper Intel drives I'm not sure whether that's still the case, I haven't really been keeping track of the pricing.
http://www.bit-tech.net/news/hardwar...ches-z-cards/1
SATA is the bottleneck apparently, PCI-E bypasses that for the faster speeds
Corsair Extreme Series X256 GB ("240MB/s sequential read and 170MB/s sequential write speed") @ £533.57
OCZ OCZSSDPCIE-ZDP84256G Z-Drive P84 256GB ("Read: Up to 770MB/s Write: Up to 640MB/s") @ £1149.99
double the price but more than triple the read+write speeds of SATA...
http://hothardware.com/Articles/Fusi...Match/?page=10Fusion-io's SSD technology circumvent the looming SATA bottleneck
[..]the SATA interface itself might very well be approaching its twilight years.Yeah, that's the big question really:
http://www.cdrinfo.com/Sections/News...x?NewsId=25975
google.com/search?q=boot+OR+bootable+pci-e+ssd+-eee+-"mini+pcie"
In the other thread someone said they have a delay before being recognised by the system which could slow boot times, but I'm not sure, it says plug and play in the article - they might be out of date too. I'm all a bit confused it's hard to know where to get reliable information, I wouldn't bother asking OCZ because I figure they'd just downplay any issues :/ Not many impartial reviews/testing that I can see... They really should send a couple free ones to reputable websites/magazines
Last edited by Perfectionist; 01-11-2009 at 09:05 PM.
I'm well aware that PCI-E allows a faster connection, but my point still stands - where would you notice that speed increase?
I installed my Intel X25-M this weekend and found while the boot time was improved, the main benefit was the response time between asking the machine to do something and it starting to do it. It really makes the whole PC *feel* much faster and it is certainly a worthy upgrade over my 150gb Raptor (now for sale if anyone is interested).
Can I justify the £167 I paid for it? No, probably not. Effectively it is just a toy.
PS. I found it a bit unnerving not hearing any HDD noise to start with, especially when installing W7!
Hmmmm. Raptor ?
How much ? how old ?
Society's to blame,
Or possibly Atari.
When I started this thread, the drive with Indilinx the controller was just out, and it was the only real alternative to the Intel (very good, but expensive). Okay, Samsung were fine too at the time, but I found it hard to differentiate their models, and I don't think it was on par with the Indilinx one anyway. I dismissed drives using the JMicron controller due to the infamous stutter issue. And I viewed (and still view) things like Fusion-io as a nice tech-demo of things to come, but not as something I, or the average enthusiast would buy (high price, non-bootable).
So Indilinx it was. And OCZ was the first one to come with a drive using that controller (Vertex). As months went by, other drives using the Indilinx came appeared the market. The G.Skill Falcon was one of them, and it was also quite consistently cheaper than the Vertex. So I switched my focus to that drive. Later on, Intel introduced the G2, which was priced a lot more competitively. So I started tracking those too.
Ultimately this is something I decided to do mostly for myself, so I picked product I was most likely to buy. I decided to post it here as a semi-permanent record so that I can compare the price of the same product months ago, and help other people interested in similar product.
I'm probably looking for £40 plus postage costs (you can pick how it's sent). I bought it in March 08 and it has been my system drive until this weekend. It's not a Velociraptor I should add and I must warn you that Raptors are louder in operation than most drives. I should really make a FS thread on this to avoid this post going horribly off topic!
Anything that uses the hard disk?Searching, loading, you name it? The PCI-E MB/s speeds are more than 3x SATA SSDs...
They are bootable, that one might not be but most seem to be:
http://www.cdrinfo.com/Sections/News...x?NewsId=25975
google.com/search?q=boot+OR+bootable+pci-e+ssd+-eee+-"mini+pcie"
That's wasn't available when the thread was started. I actually remember that there were talk of a bootable Fusion-io, be it rumour or work in progress, but I quickly looked away when I saw the price tag. At the current price, OCZ's offering may be significantly than the Fusion-io 6 months ago, but it's still as much as I would spend on an entire PC.. with an SSD drive. But if you can afford that, or are interested to see how price for those drives change over time you are more than welcome to use this thread for that purpose.
Still, you did get me curious and I decided to look up the OCZ Z-Drive. What interests me the most is 4k Random read/write. The reason why SSD is deemed more 'snappy' is not because of those 250+ MB/sec numbers you see plastered all over the place. Sure they are relevant, but SSD are more responsive due to the much lower response time (but there are no big differences between them) and MUCH faster random write of very small files (it's the sort of activity that occurs in the background while you are using your PC normally). 20MB/sec may not sound much, but keep in mind that a Velociraptor can't even manage 1MB/sec in such tests if I remember correctly. Actually, reading the rest of the review, the drive's main benefit is sequential transfers.
The 500+MB/sec of sequential speed doesn't interest me as much. I don't shift massive files between multiple drives that often. It's nice to have it, but I wouldn't pay over the top for it. If I *needed it* (say, I do lots of HD encoding), a PCI-E SATA RAID controller with a couple of TB drives would probably be more suitable due to the capacity requirement (you'll fill that SSD in 10 minutes). For my money, I would definitely stick with a single 160GB Intel G2 (the G2 also does better than the Z-Drive for gaming and media application).
Last edited by TooNice; 02-11-2009 at 04:06 PM.
No, you wouldn't notice it when searching or loading.Anything that uses the hard disk?
They are bootable, that one might not be but most seem to be:
http://www.cdrinfo.com/Sections/News...x?NewsId=25975
google.com/search?q=boot+OR+bootable+pci-e+ssd+-eee+-"mini+pcie"
As TooNice pointed out, the only time that you might is when you're transferring a 20GB file. And that's not really typical usage for an SSD. So I still don't really get what you're talking about.
Ok, so I did some more research (I didn't know about the IOPS factor, as I've said elsewhere not had a SSD drive before), think you may be right here
120GB OCZ Vertex
4 KB random IOPS performance: 25.75 MB/s read / 12.10 MB/s write
120GB OCZ Vertex Turbo
4 KB random IOPS performance: 28.41 MB/s read / 67.81 MB/s write (someone said before the turbo version wasn't worth getting, hmm!)
(note, the text in benchmark review says a different number of IOs, but in the picture it shows 17358 - it says it is 17358 IO writes for vertex turbo in the benchmarks for the below corsair drive too)
256GB Corsair X256 CMFSSD-256D1 (Scan use this by default in their most expensive builds, there's no option for anything not Corsair)
4 KB random IOPS performance: 29.08 MB/s read / 63.45 MB/s write
160GB Intel X25-M G2
4 KB random IOPS performance: 58.50 MB/s read / 34.50 MB/s write
64GB Intel X25-E
4 KB random IOPS performance: 55.90 MB/s read / 31.70 MB/s write
500GB OCZ Z-Drive
4 KB random IOPS performance: 17.63 MB/s read / 28.33 MB/s write
(From the pictures attached further down the thread)
It doesn't say whether it is the the e84 (fastest) "SLC Enterprise Series", p84 "MLC Performance Series" or m84 "MLC Mainstream Series" though, but interestingly, those forum benchmarks work out to 4.5K IOPS - the official information on OCZ's site gives no data for m84, "16k maximum" for e84 and "10k maximum" for e84 which of course I was wary of when I read since "maximum" in advertising-speak is the same as "up to" meaning it could well be less than half in actual use, with no differences between the sizes of the disk(s)
So yeah, looks like Intel is the best for general use (using read more than write), hopefully they've sorted out that firmware thing now, reading around it seems like that goes back to July
However if those OCZ ones are poor, it seems using other disks in a SATA 2 cage works pretty damn well:
such as 2 x Intel X25-Ms or 4x (2 seems to give double the performance, 4 drives about 3x the IOPs)
Last edited by Perfectionist; 02-11-2009 at 06:52 PM.
HD tune random numbers are just no where near a useful benchmark I'm afraid, other than showing the massive difference between mechanical and ssd, the numbers are near useless. IOPS numbers and HD Tune benchmarking is completely incomparable to numbers from Crystaldiskmark.
http://www.myce.com/review/intel-x25.../Benchmarks-4/
For instance the numbers you gave for the intel from a personalised benchmark in IOmeter from Anandtech don't match Crystaldisk results at all, varying drives slightly(maybe didn't check think the ones I linked are first revision of the drive) but basically you can't compare numbers from different benchmarks at all, identical benchmarks and even then often from the same reviewer so the same testing style will be the only way to really compare.
Frankly all the ocz vertex based drives won't be far off on crystal disk mark results, and the intel's, all of them, will have about those numbers in the link I showed in that benchmark.
At this point in time the choice is realistically Indilinx based drive, cheapest you can get one the better, or an Intel and no matter what you get you'll be very hard pressed to tell the difference once in your system under any normal home use and most other ways to use hard drives frankly.
The intel only really shines in very specific area's and not many that people frequently use in any circumstances.
Well, that sucks. I had to search other sites because I couldn't find one reviewer that had done them all. Someone should really do that...
HEXUS?
Hehe, yes, hard drive benchmarking is hard to compare because they aren't as widely benchmarked so harder to see comparisons on the same site in the same review. While any gfx card on most sites in the same review they show results in the same test done the same way for anything from 5-10 different gpu's.
The biggest issue really though is, hdd benchmarking is very , odd. While games have game benchmarks that give you a rough idea what you'd get in that game. But you don't have say an encoding app with a nice little hdd benchmark to compare real world performance, just these artificial benchmarks which tend to always go after very specific types of data transfer which rarely if ever reflect real world performance.
IOmeter isn't bad as you can record your own usage pattern, so you can set it too monitor the io requests when made say, loading windows, opening a game, playing for a while, closing, play another game, close surf, download etc, etc. But few sites use iometer and those that do don't use standard testing, its all what they consider normal usage, which is different from user to user and also incomparable from review to review.
The oddest thing is, while benchmarks scores keep improving, realistically theres a certain cut off in performance with hard drives where diminshing returns become a huge issue. I can't see any difference whatsoever between a single 64gb crucial ssd, and 2 in raid 0, benchmarks are WAY ahead, almost perfectly scaled to double the performance, even in the key 4kb random writes(goes from 15mb/s to 30mb/s in crystal mark). However, in real world performance i simply don't see that performance difference, loading games, windows, pretty much anything its almost identical. The few things that would benefit from the higher sequential speeds would be unrar'ing downloads or moving lots of files. But I download to a bigger mechanical HDD to prevent excessive use of the SSD's anyway.
There are currently 1 users browsing this thread. (0 members and 1 guests)
SCAN.care@HEXUS
Posting Permissions
Reply With Quote