Intel vs OCZ

Cronyx

Limp Gawd
Joined
Mar 24, 2005
Messages
146
These are the two SSDs I'm looking at.

OCZ Vertex Series 60GB SATA II
http://www.newegg.com/Product/Product.aspx?Item=N82E16820227394

Intel X25-M G2 Postville 80GB
http://www.newegg.com/Product/Product.aspx?Item=N82E16820167018

Now other than the size, is there any reason to go with the Intel? It's a lot more expensive, and the speeds are better on the OCZ. This is going to be a boot drive and "at boot application load" drive. Games and media will go on other drives. So basically I'm looking for reliability and speed here. These two were recommended by 4chan /g/ and I'm just double checking. (When they aren't suggesting installing gentoo linux, they're usually pretty helpful)
 
Intel is faster where it counts small random reads / writes. Also it seems to be more reliable.

Yup, agreed. But IMNSHO its not enough faster to be worth the near 50% price premium. Also, OCZ supports garbage collect - which Intel does not - so if you are running it in Raid where you don't have support for Trim OCZ win hands down.
 
maybe maybe not, depends upon your usage patterns. the intel is superior though its random reads are leaps and bounds better than the vertex. for raid though you cant beat its GC features. if going single device go intel. if raid go ocz.
 
The intel speed advantage seem big in bench but in real life usage you wont notice the difference. One thing you do notice is on copy of large files the OCZ are faster and that is apparent in real life usage when you do copy large files and not only in bench.

That is why I chose a OCZ Agility 120. Very fast and don't seem slower than my friend Intel G2 80GB in real life scenarios.
 
The intel speed advantage seem big in bench but in real life usage you wont notice the difference. One thing you do notice is on copy of large files the OCZ are faster and that is apparent in real life usage when you do copy large files and not only in bench.

That is why I chose a OCZ Agility 120. Very fast and don't seem slower than my friend Intel G2 80GB in real life scenarios.

Pfft, not on the new SF controller based drives. They're piss ass slow unless it's highly compressable benchmark data being written to the disk.

Intel all the way, it's worth it to not support OCZ at all.
 
Not like you're copying large files on an SSD anyway. People can't seem to differentiate system disks with mass storage.
 
I've got the Intel 80GB G2 in my laptop and a 120GB Vertex in my desktop and built a system recently for a friend with the Intel 160GB G2 and can't tell the difference in normal everyday surfing/gaming etc. etc.

That being said. I think that the Intel drives seem to have a higher "perception" of reliability. I don't "hear" of people having consistent problems with them.

OCZ, though, has a great forum with a dedicated group of people who seem eager to help with any problem you might have which can be a great thing. But, this also means that their problems are much more visible.

Both brands have been rock solid for me. I too am drawn by the mystique of the Intel drives (thats why I went with them for my laptop after a perfectly fine experience with OCZ) but I have to say knowing what I know now I would have saved my money or gone for more capacity.
 
I own 3 of the intel based Dane Elec X18M Gen I drives and also use OCZ drives in other situations. I use the older non Trim support intel drives in my work laptop on WinXP, and in a Raid 0 on my main system at home. On a Win7 based laptop, I use an OCZ Agility drive that supports Trim. I also use a 4 drive Raid 0 going into a Perc5 controller for my play around system, although that one is down right now as I have been to busy with work and my gf to put it back together.

Anyways, the real truth is that I do not notice any real difference in performance in any of the setups. Any of the new drives will work fine for you. The only reason I bought the Intel drives was there were cheap at the time.

Don
 
Don I'm in a similar situation with X25-M (and X18-M) G1's, G2's, X25-V's, and a bunch of OCZ Vertex/ Agility drives. Real world usage, I don't notice the 4K random reads/ writes (by the time you get to 128K it is a lot closer even in benchmarks) as much.

An interesting note since I couldn't explain the huge benchmark deltas versus real world usage deltas, I summed the count of files on my main Win7 Ultimate's C drive and then assumed that each file was 4K. Then I simply did a (total files * 4K) / OCZ 4K speed and got something like 56s. It was something like 140GB (2x OCZ Vertex Raid 0) worth of files, and I keep all media on spindle drives in servers.

I think that's basically the key though. The throughput may look pathetic, but in reality it is a ton of files so when you hit the real-world usage scenarios you may gain a handful of seconds on a boot operation, but it isn't like you wait an extra minute for an Indilinx over an Intel.

I will say I bit on the Newegg Agility 2 120GB deal yesterday just because I wanted a SF controller around again for fun.
 
I use both the Vertex SSDs and the G2 intels all day every day, every day - I can tell ZERO difference between the 2 with heavy multitaksing. Same exact supporting hardware, just different drives. Only time I notice a diff is when i copy large multi-gig .iso's around on the desktops, that's when i notice the Vertex are a bit faster - enough to notice.

Everyone pisses n moans about the 4K advantages with the intels - I don't experience this advantage in what I do on a daily basis.

Of course these are probably the same people that are convinced there's no noticable difference when using RAID0. They're making statements not based on personal experience.
 
the ocz actually outperform the intels in heavy multitasking because they write faster. the intel controller has an inherent problem with reading WHILE writing. it does not handle both at the same time very well at all. while writing its read speeds suxors. look at the heavy trace runs from anandtech:

http://www.anandtech.com/show/2865/8
 
the ocz actually outperform the intels in heavy multitasking because they write faster. the intel controller has an inherent problem with reading WHILE writing. it does not handle both at the same time very well at all. while writing its read speeds suxors. look at the heavy trace runs from anandtech:

http://www.anandtech.com/show/2865/8

That makes sense. I will also say I have had a few moments when I've seen things stutter on the intels - especially FMV while things are running in the background - I have NEVER seen that on the Vertexes. I can reproduce it on my coworker's machine as well who is using the G2 intels.

The intel SSD fanboi "end-all-be-all" stuff is rather amusing and has been since before these things even hit the market. They are not perfect and when put to the test on a day to day basis (even with running the toolbox, latest fw etc.), I have seen some less than stellar results. Great SSDs, but there's room for improvement, as always.
 
the ocz actually outperform the intels in heavy multitasking because they write faster. the intel controller has an inherent problem with reading WHILE writing. it does not handle both at the same time very well at all. while writing its read speeds suxors. look at the heavy trace runs from anandtech:

http://www.anandtech.com/show/2865/8

My ocz drive was all benchmarks and nothing else. They basically lied, it was a sandforce drive (vertex le), ocz claimed 235mb sustained write, not the upto speed but sustained, I got 160mb tops and it would go down to 110mb after a while, my kingston value drive is better than that. I returned it to newegg. This test was done with real files, and not little ones but big ones 512mb to 2gb.
 
Of course these are probably the same people that are convinced there's no noticable difference when using RAID0. They're making statements not based on personal experience.

I can tell my 2x 120GB Vertex drives in Raid 0 are faster than one in benchmarks... but they aren't THAT much faster real world usage. I am fairly sure that if you gave me a random system with a single 120GB drive and then added a second in raid 0 a week later, I wouldn't easily tell the difference. Then again, I needed more than 120GB also :)

@munkle I returned my 100GB Vertex LE too, but for $270 (-$30 rebate if I remember) the Agility 2 isn't that bad. All of my compressed media sits on the network so I don't get terrible performance.
 
the ocz actually outperform the intels in heavy multitasking because they write faster. the intel controller has an inherent problem with reading WHILE writing. it does not handle both at the same time very well at all. while writing its read speeds suxors. look at the heavy trace runs from anandtech:

http://www.anandtech.com/show/2865/8
Don't see anything you write back in that article. The only thing the article shows it that Anandtech likes doing a lot of sequential benchmarks, captured by their so called real-world traces. So this is more of a sequential write benchmark to the Intels; the reads are so fast they come for free. Do a trace with only reads (like when playing World of Warcraft) and you would see a different pattern.

But even with the heavy write traces i can't say Intel is slow. We also must remind ourselves that these benchmarks are done on cleanly installed systems, which means most I/O will be sequential. On real systems over time with all kind of installation/deinstallation, the percentage of sequential I/O drops, as files become more scattered throughout the logical disk space.
 
Everyone pisses n moans about the 4K advantages with the intels - I don't experience this advantage in what I do on a daily basis.
Interesting. Though isn't this just a side effect from removing the disk-bottleneck; therefore making the CPU/MEM subsystem play a much larger role in performance? In other words, the access times on even cheap SSDs is about one hundred times lower than a mechanical drive; that means you eliminate a lot of bottlenecks straight away.

The sequential speeds you notice as you copy stuff. But copying data on a small SSD that serves only as system disk, is not that common. It might be for bigger SSDs which users also use for large files; a mixed or general storage device. However i feel that it is best to separate the mass-storage data from the system disk, and use SSDs only to facilitate the Operating System and installed applications.

The 4K scores likely will become more important as you continue to use your system with SSD over time, and as SSDs get higher sequential speeds around Christmas this year when Intel releases its third generation of SSDs. Then it should never be a bottleneck when doing sequential I/O as well; as i think 70MB/s is kind of low indeed. But that's not why i bought my Intel SSDs.

There's also the problem of not noticing performance improvements. How would you notice common tasks you do actually are a bit faster? The real times I/O performance counts is when you're waiting on it; most I/O you don't even notice as it happens in the background without interference. Generally you need 30-50% performance increase to notice anything, and with SSDs being so fast already the disk bottleneck is already too small to make such a big gain possible; the CPU still is a bottleneck for single threaded apps and kernel tasks.

But perhaps i prioritized random I/O too much. Still, i believe a system disk shouldn't see much sequential writes at all.
 
Don't see anything you write back in that article. The only thing the article shows it that Anandtech likes doing a lot of sequential benchmarks, captured by their so called real-world traces. So this is more of a sequential write benchmark to the Intels; the reads are so fast they come for free. Do a trace with only reads (like when playing World of Warcraft) and you would see a different pattern.Reply With Quote

well thats exactly the point im making. its performance goes down while writing. of course a trace with only reads would be different. that would not highlight the issue im speaking of. Not saying intels arent fast, of course they are awesome and very fast at random read, but they do have a weakness, as all controllers/ssd's do. they just arent the answer to every usage pattern. and as others make very good points, in real world usage it is very hard to tell, the differences are minuscule in real world usage to a normal person.
one more thing anand's tests arent sequential at all, read the description above the graph on the heavy trace. that is hardly a sequential only benchmark.
 
Back
Top