NEW OCZ Vector

[Mfusick]

jwilliams4200 you make me laugh at your bias.

Do you mind if I ask why your so against OCZ?

 

[ocztony]

Guys its time out time...jwilliams, i must apologise to you as it looks like we were singling you out, you are outspoken about OCZ, me...and our drives and the beta testers are somewhat loyal so i have taken steps to give you an easier life here on the Anandtech forum.

Its hard for me to answer your questions as to do so would reveal things I can not, if this makes you feel I lower the standard of OCZ support then I am sorry. I am not actually support, I'm marketing, hence I was focused on you not fully understanding how our drives work and nothing else. I have asked we try to release some educational material that will help people understand how our drives work better..this is being discussed now.

It seems obvious that vector and vtx4 are not drives you like, can we call a truce on this situation?

Thanks

Tony

 

[jwilliams4200]

I have asked we try to release some educational material that will help people understand how our drives work better..this is being discussed now.

I think that would be a positive development for OCZ. There is no reason to keep the basic functionality of the SSDs secret. I can certainly understand why the details of implementation would be a trade secret. But I don't think anyone is asking for that level of detail.

I am only saying that it would be good for those who are interested to be able to know enough about the basic operation of the SSD so that it is possible to predict when and under what conditions there will be a slowdown. Just saying that "as long as you use it normally" does not really cut it. Not everyone uses their SSDs in the same way.

P.S. I appreciate the apology. I have no problem with OCZ employees or beta testers vigorously promoting OCZ's products. But I do think it looks bad for OCZ when those same people accuse me (or others) of irrational hatred of OCZ when I have repeatedly explained the reasons why I am wary of OCZ products. It can only improve OCZ's image if OCZ's employees and associates are more careful with their facts and accusations when they post in threads like this. And if I made any unfounded accusations, I certainly apologize for doing so.

 

[ocztony]

under any circumstances the ONLY time you will ever see a slow down is when you write 100% LBA in 1 go...so you are in fact inducing the slow down by looking for it.

This is why even Anand says:

At 50% capacity there's an internal reorganization routine that's triggered on Vector, similar to what happens on the Vertex 4. During this time, all performance is impacted, which is why you see a sharp drop in performance just before the 135GB mark. The re-org routine only takes a few minutes. I went back and measured sequential write performance after this test and came back with 380MB/s in Iometer. In other words, don't be startled by the graph above - it's expected behaviour, it just looks bad as the drive doesn't get a chance to run its background operations in peace.

And remember he just wrote 100% LBA with a benchmark which induces this slowdown.

Also if the drive is still reorganising data and you ask it to write again the write speed will be reduced a little, this is what i feel you mean by steadystate speed, if the drive is allowed to complete its reorg routine the drives return to full speed. Let us also not forget in normal use TRIMs will be issued that also reset the drive..so again in normal use away from benchmarking the drive is continually being reset and readied for subsequent writes being issued.

Please also remember the 512's do not do this as i showed you earlier today.


Any further info now will have to come from OCZ directly...

 

[jwilliams4200]

under any circumstances the ONLY time you will ever see a slow down is when you write 100% LBA in 1 go...so you are in fact inducing the slow down by looking for it.

I find that very hard to believe.

I strongly suspect that there can be a slowdown when you write less than 100% of LBAs at once, for example, if the SSD is already 80% full and you (over)write 40% of LBAs at once, I bet that can trigger the consolidation / special GC. Possibly even if the SSD is 80% full and you write more than 10% at once.

One theory that I have heard (from people who have thoroughly tested Vertex 4 SSDs) is that whenever you use up 50% of the remaining free space (even if you started with the SSD already filled up more than 50%), that there will be a slowdown / consolidation period. I'm not sure if that explanation is exactly correct, but it makes a lot more sense than the claim I quoted above.

By the way, does OCZ still say that there is a "performance mode" and a "storage mode" as was once claimed about the Vertex 4?

 

[Mfusick]

From what I am seeing and reading it looks like the Vector handles many of the tough things with aplomb, and outperforms the majority of the SSD's on the market.

So why criticize or look for negatives with such an active effort?

Isn't that like being ultra critical of Tom Brady/Drew Brees/Aaron Rogers for throwing one incomplete pass.

I mean it's great to look for a negative but lets keep in perspective the positives, and manage our expectations.

So far from what I have read- I would happily trade my MAX IOPS Vertex3 for one of these bad boys- and my drive is almost a year old only.

If you had a 600 horsepower corvette that dominated on the race track, and drag strip- had best brakes, handling, acceleration and top speed-- But- It could not go around a hair pin turn at full throttle because the tires would give out and cause a spin out do you know what logical people or the MFG might say ???

Don't go around hairpin corners at full throttle.

It doesn't mean the product is any less impressive.

manage your expectation and stop looking so hard for a negative. That's my advice Jwilliams4200.

I think the Vector is a very impressive drive. You don't have to agree but you should let it go.

 

[bradley]

OCZ forum threads are better than the company itself, well at least they are more amusing. Pump and dump.

 

[beginner99]

I am not actually support, I'm marketing, hence I was focused on you not fully understanding how our drives work and nothing else.

That would be a first. A marketing or sales guy in the tech world actually understanding what he is marketing or selling.

 

[ocztony]

That would be a first. A marketing or sales guy in the tech world actually understanding what he is marketing or selling.

I am the longest serving employee with OCZ, with over 10 years in now, i was brought on board to be the end user, so an inhouse end user/overclocker who would test/develop new product and if pleased the product would be released. It has not happened that way so much with SSD's but when we did ram I worked on many products in DDR1 and DDR2, VX and EB, were mine...there are more but we don't do ram now so its not a big deal.

So what I am saying is I am a little different to a normal marketing guy, im more of an engineering guy who helps promote the product we make.

 

[Mfusick]

Send me a Vector and I will promote the hell out of it.... Just Sayin'

 

[cytg111]

Send me a Vector and I will promote the hell out of it.... Just Sayin'

well said too.

 

[josephjpeters]

One guy was able to get max IOPS of 172k for the Vector, compared to 127k for the Samsung 840 Pro... interesting. I know these are Max values so they don't mean all that much but that's quite the gap though.

http://www.xtremesystems.org/forums...OCZ-s-(INDILINX-Barefoot-3)-Controller/page2&

 

[jwilliams4200]

One guy was able to get max IOPS of 172k for the Vector, compared to 127k for the Samsung 840 Pro... interesting. I know these are Max values so they don't mean all that much but that's quite the gap though.

Either those results are bogus, or else they are 512B reads or writes. If the latter, then it is basically useless, since who runs workloads that are sustained 512B reads or writes at QD32 on an empty SSD?

 

[Hellhammer]

Either those results are bogus, or else they are 512B reads or writes. If the latter, then it is basically useless, since who runs workloads that are sustained 512B reads or writes at QD32 on an empty SSD?

65.11MBps / (512B/1000^2) = 127,168 IOPS

Yup, the IO size is 512B. I'm guessing those are writes because OCZ has always liked to cache a lot to the DRAM (that's why the Agility 4 gets okay write speeds, but read speeds are horrible).

 

[Mfusick]

65.11MBps / (512B/1000^2) = 127,168 IOPS

Yup, the IO size is 512B. I'm guessing those are writes because OCZ has always liked to cache a lot to the DRAM (that's why the Agility 4 gets okay write speeds, but read speeds are horrible).

The agility uses the same controller as vertex3 right?

Buy it has cheaper NAND and that's why it's slower ?

 

[Hellhammer]

The agility uses the same controller as vertex3 right?

Buy it has cheaper NAND and that's why it's slower ?

Agility 4 uses Indilinx Everest 2 (i.e. the same as in Vertex 4) but the NAND is asynchronous whereas Vertex 4 uses synchronous NAND.

 

[Mfusick]

Agility 4 uses Indilinx Everest 2 (i.e. the same as in Vertex 4) but the NAND is asynchronous whereas Vertex 4 uses synchronous NAND.

I was thinking the Agility 3. I always thought the Agility 3 was slower than vertex3 because of the NAND.

Looks like they do the same thing with Vertex4 and Agility 4.

OCZ seems to be distancing itself from Sandforce.

 

[JellyRoll]

Rofl. LBA page mapping is typically held in the DRAM, not user data There have been very few who have tried holding user data in DRAM.

 

[rob.L]

Either those results are bogus, or else they are 512B reads or writes. If the latter, then it is basically useless, since who runs workloads that are sustained 512B reads or writes at QD32 on an empty SSD?

The point of his post was to point out how much more powerful and capable the Barefoot 3 controller really is.

 

[jwilliams4200]

The point of his post was to point out how much more powerful and capable the Barefoot 3 controller really is.

And the point of my response was to point out that it does not matter how many 512B QD32 IOPS an SSD can do, since no one cares.

 

[rob.L]

New review of the 128GB Vector.

http://www.tweaktown.com/reviews/5110/ocz_vector_128gb_ssd_review/index.html

 

[Mfusick]

You don't see many 128GB reviews anymore... I wonder why ?

 

[josephjpeters]

You don't see many 128GB reviews anymore... I wonder why ?

EDIT: This is for one particular drive and it may not apply across all categories (ie. Value SSD's)

I'm not sure when the change occurred but there are more 256GB drives being sold now than 128GB drives.

256GB > 128GB > 512GB

I think $300 unit price is key for 512GB. We'll probably get there next year when everything is 20nm MLC/TLC.

 

[Coup27]

Generally speaking 256GB drives are the fastest in the range (the range being 64, 128, 256 and 512GB) so once one manufacturer started sending 256GB drives for review they all had to.

 

[jwilliams4200]

I'm not sure when the change occurred but there are more 256GB drives being sold now than 128GB drives.

How do you know?