OCZ to come out with new VERTEX SSD,S equiped with 64MB OF CACHE starting at $129

[ThugsRook]

Well I dont know what i'm going to get now. I want to get an SSD just as my OS drive...

dont let these super RAID guys concern you too much. as a single OS drive (WXPSP2) 30gb V2 runs perfectly fine with the simple reg/OS tweaks and FAT32. you dont need a ramdrive or nothing like that.

my 30gb V2 is about 40-50% faster then my 74gb Raptor according to HDTach.

[alfaunits]

Your words about HDTach show exactly how much you have a clue about hard drives - please leave any storage discussions mate, you're not helping

[ThugsRook]

Your words about HDTach show exactly how much you have a clue about hard drives - please leave any storage discussions mate, you're not helping

im sorry, you misunderstand me. im not saying HDtach is my proof, im just saying that for a single OS drive it runs very well. no quirks, no problems, much noticable improvment over my raptor, im very happy with my V2.

[Halk]

It WAS a priority and it was MARKETED as a high IOPS drive but look what happened. They might fix it after all. I mean, I am sure Intel faced the same decision when they designed the X25-M of high seq writes versus high IOPS and they went the IOPS route. Intel isn't filled with idiots last time I checked.

Are you implying that OCZ is? No company is perfect. Intel haven't always been right!

[RealTelstar]

Just a quick note: 2x X25-Es in raid 0 was the same performance as 1 for me in all apps. Just in general raid 0 gives little benefit to performance (read the raid sticky if you want to find out why) and in this case it shows even more because single drives are already so fast. I only got two because I needed 64GB of space. If you are fine with 32GB then save your money and get just one; performance is the same.

Yeah... But i need 150+GB and I have no choice, if not waiting for 32nm SLC or some more creative MLC version.

Another quick note, besides the cache, the adaptec 5405 I have actually SLOWS DOWN the drives versus onboard. You will be perfectly fine with onboard. In fact, I am selling my 5405 and moving to onboard because it is faster in most things and I dont have to deal with 45sec being added to my boot time due to the controller.

I also agree with this. I'm hoping to NOT need a controller. On the other side, one with 2+GB of cache will help handling big video files that i have on the platter disks.

Also overall real world write performance degrades about 20% after a lot of use. The only way to get it back is to format the drive. I reformat my array after every month; I have a system set up now that takes an image of my boot array and writes it to one of my storage HDDs then I wipe then I stick the image onto the array again; takes 1.5h-2h overall to do a "refresh".

Was this supposed to be fixed in the newer batches of units?

Thanks for the OT

[alfaunits]

Well, I cannot read your mind, what you type is all I can go with, and what you typed was exactly as I interpreted it, there is no place for mistakes.
That it runs good for you? Sure... does that mean that it is better than Raptor even in any way? Hell no, my old 13GB drive runs good, but I sure wouldn't use it nowadays!

im sorry, you misunderstand me. im not saying HDtach is my proof, im just saying that for a single OS drive it runs very well. no quirks, no problems, much noticable improvment over my raptor, im very happy with my V2.

[Gilhooley]

Are you implying that OCZ is? No company is perfect. Intel haven't always been right!

The point I think was that, Intel's drives have been kicking ass for 5 months in desktops tasks and benchmarks due to their high IOP/S. So fiddling around at this time with a delayed product.. Well. you decide Then again OCZ is just a sales company, you can afterall start your own right away if you want to. So they have in all likelihood nothing to do with the delay.

[Halk]

due to their high IOP/S.

How can we tell it's because of the high IOPS? As far as I'm concerned the reason the Intel drives are so good is because of their ability to write/read small sizes fast. The IOPS may or may not be related, people are jumping to that conclusion but I'd like to see some evidence.

[One_Hertz]

How can we tell it's because of the high IOPS? As far as I'm concerned the reason the Intel drives are so good is because of their ability to write/read small sizes fast. The IOPS may or may not be related, people are jumping to that conclusion but I'd like to see some evidence.

LOL is all I have to say at that post. What do you think IOPS means? Inputs/Outputs per second. In other words, the ability of the drive to handle small files quickly, sequentially or randomly. I am going to be laughing at you a little all day long. Thanks, I needed that this Friday

[Griff805]

LOL is all I have to say at that post. What do you think IOPS means? Inputs/Outputs per second. In other words, the ability of the drive to handle small files quickly, sequentially or randomly. I am going to be laughing at you a little all day long. Thanks, I needed that this Friday

Can you prove that IOPS means Inputs/Outputs per second? I want to see some evidence.. Who told you that? I think you're jumping to conclusions-

[Gilhooley]

....

[Halk]

LOL is all I have to say at that post. What do you think IOPS means? Inputs/Outputs per second. In other words, the ability of the drive to handle small files quickly, sequentially or randomly. I am going to be laughing at you a little all day long. Thanks, I needed that this Friday

Heh. That's not quite what I meant :P

I'm meaning if there's a spike in a benchmark and it reports 4,000 IOPS that doesn't mean the best drive would be one that can sustain 4,000 IOPS.

[One_Hertz]

Heh. That's not quite what I meant :P

I'm meaning if there's a spike in a benchmark and it reports 4,000 IOPS that doesn't mean the best drive would be one that can sustain 4,000 IOPS.

I can agree with that sustaining high IOPS for too long is probably unnecessary for most users. It has to be able to spike high though. I see 20k+ IOPS (reads) spikes when loading games but the loading times are quite short and when the game is already running its always sub 100.

LOTS of lulz to be had in that OCZ thread. For example someone said

"IOPS is huge for many of us. That's specifically why I use RAID0."

Of course he is right if he is talking about operations with files that are larger than the stripe size, but I don't think he is lol.

~~~~~~~~~~

http://www.ocztechnologyforum.com/fo...&postcount=156

Why did sequential small writes take such a big dump? Tony?

[Griff805]

I can agree with that sustaining high IOPS for too long is probably unnecessary for most users. It has to be able to spike high though. I see 20k+ IOPS (reads) spikes when loading games but the loading times are quite short and when the game is already running its always sub 100.

LOTS of lulz to be had in that OCZ thread. For example someone said

Of course he is right if he is talking about operations with files that are larger than the stripe size, but I don't think he is lol.

~~~~~~~~~~

http://www.ocztechnologyforum.com/fo...&postcount=156

Why did sequential small writes take such a big dump? Tony?

That thread you linked to is very interesting... I thought they might actually get the Vertex to work well, but looking at those numbers is not very encouraging.

Yeah, the IOPS are up, but the speeds are way down. I wonder if that controller is up to the task-

[Halk]

I see 20k+ IOPS (reads) spikes when loading games but the loading times are quite short and when the game is already running its always sub 100.

That's exactly what I'm saying we need more evidence about before we leap to any conclusion.

Sure a drive that drops a lot of sequential speed to run higher IOPS is going to chew through that particular usage example. But when the game is running and it's 100 IOPS then the drive with higher sequential speed would do better.

We just don't know which drive will do better over all. I'm not meaning Intel vs Vertex here, I'm meaning Vertex Firmware vs. Vertex Firmware.

Everybody has leapt aboard this IOPS bandwagon with the assumption that higher IOPS means higher overall performance, but there doesn't exist any proof.

The Intel drive isn't proof of it, since the Intel drive might well achieve such good results because it's simply a better drive, and the higher IOPS capability might just be icing on the cake, or incidental.

I think the only way we're going to see any real evidence one way or the other is if OCZ bench the Vertex with the superfast throughput firmware, and also bench the Vertex with more emphasis on IOPS. Tony did say the superfast throughput firmware did do better on a simulated Windoes load benchmark, but I'd be looking for a bit more of a deeper look. What I suspect is that what one firmware gives, it also takes away, and we'll not see a huge difference overall.

You did mention stripe size, and when my drives arrive I'll be looking at putting a smaller stripe size than 128K on, just to see if it helps with the smaller numbers.

Ultimately though the real world performance of any drive is a blend of how it performs with reads, with writes, with small reads, with small writes, with small writes while it's being asked to do big reads, with small reads while it's being asked to do big writes, with small writes while it's doing small writes somewhere else, etc. Just looking at an ATTO benchmark isn't really going to show that.

[Brahmzy]

LOTS of lulz to be had in that OCZ thread. For example someone said

Of course he is right if he is talking about operations with files that are larger than the stripe size, but I don't think he is lol.

Hey Brainiac, that was my post and YES I meant files larger than my stripe size. What about my statement is so surprising to you?

Have you ever watched a Crysis level load before and after you RAID0'd your storage? Or, boot times? Or massive disk to disk / LAN backups? Ok then, STFU.

[ThugsRook]

...does that mean that it is better than Raptor even in any way?

yes, it runs as good or better then a single Raptor.

...im very happy to be off the mechanical monsters

[HDCHOPPER]

NEWS for the vertex ... new firmware will have MAX IO performance read post #192
http://www.ocztechnologyforum.com/fo...ad.php?t=51364

[Griff805]

It looks like OCZ might actually have a good drive with the Vertex if they use that high IO firmware. However, they've only had a very short amount of time to test the firmware and find any issue. I'd be leery of jumping on the Vertex boat right at launch. Might want to wait and see how they actually perform when people get their hands on them.

[One_Hertz]

Have you ever watched a Crysis level load before and after you RAID0'd your storage? Or, boot times? Or massive disk to disk / LAN backups? Ok then, STFU.

Crysis - no, but all these games: AOE3, left 4 dead, warcraft 3, fallout 3, fear, doom 3, oblivion, silent hill 4 showed no improvement in any loading times in R0 versus single drive. Boot times are consistently a tiny bit slower in R0 versus one drive in windows xp 64 bit and windows 7.

[headala]

Did they mention a ship date?

[Griff805]

For reference...

It looks like OCZ might have realized the error of their ways- Things are looking a little brighter for the Vertex:



Of course, this is just one benchmark. I'm interested to see the others. Good thing Halk ain't running things over there, or else they'd have a 1 IOPS drive that does 800MB/s...

[TimoneX]

LOL

Yeah you try checking your email while virus scanning and the file system pops up a window and says GET OUT AND PUSH!

[Griff805]

LOL

Yeah you try checking your email while virus scanning and the file system pops up a window and says GET OUT AND PUSH!

Actually, I don't know if it'd be able to even have the Pop up window come up till the virus scan ended...

[Brahmzy]

Crysis - no, but all these games: AOE3, left 4 dead, warcraft 3, fallout 3, fear, doom 3, oblivion, silent hill 4 showed no improvement in any loading times in R0 versus single drive. Boot times are consistently a tiny bit slower in R0 versus one drive in windows xp 64 bit and windows 7.

Wow man, you've got somethin' SERIOUSLY wrong with your R0 setup then. FO3, L4D, FEAR 1 & (2 I'm sure as well) (only games I've tried that you listed) FC2 etc., all are almost halved in load times. I remember going from a 2 VR R0 setup to a 3 VR setup and I specifically remember all of my games showing about a 1/3 decrease in load times. Don't know what you're doin' wrong, but R0 lives up to it's benches when it comes to large file reads and writes.