Vertex 4 Reviews and Thoughts

[Nizzen]

http://www.anandtech.com/show/5741/o...-based-silicon

LOL!

So there is no OCZ controller :p

[Highendtoys]

Most of us already knew this but didn't have any solid proof. That article went live without sharing the proof.

[DooRules]

You got to be joking.

[lm358]

http://www.anandtech.com/show/5741/o...-based-silicon

TBH I didn't believe you when you first mentioned it, considering the fanfare they made about their "Indilinx controller", but wow. I guess that proves it. Quite interesting how the performance characteristics between the V4 and M3P are quite different -- I suppose this really highlights how much impact firmware can have on SSD performance.

I generally assumed most firmware would be fairly similar and, say, could account for maybe +-10%, but clearly they really do some unique things.

Thinking about it a bit further, I guess the actual "controller" is probably nothing more than a couple of ARM(?) cores with some hardware extensions relevant to disk processing.. SATA, flash controller interface, encryption module etc. So I suppose that probably answers why vendors (read: Intel, OCZ) feel that there is nothing much to be gained in the hardware.

[johnw]

Quite interesting how the performance characteristics between the V4 and M3P are quite different

The Plextor M3P uses the Marvell 88SS9174 controller, but the Vertex 4 is using a controller that is virtually the same as the new Marvell 88SS9187. Also the Plextor is using toggle flash, while the Vertex 4 is using IMFT ONFI flash. So the difference is more than just firmware.

So, of the three differences

1) Controller: V4 > M3P

2) Flash: M3P > V4
toggle flash wins on performance

3) Firmware: M3P > V4
The Vertex 4 suffers from the same problem as the m4 and earlier Plextor's did, which is poor sequential reads at low QD. But Crucial and Plextor fixed that issue with later firmwares. The Vertex 4 firmware still has the problem.


So the M3P wins 2 out of 3. Anyway, I expect Plextor will soon release a new model with the Marvell 88SS9187 controller, then Plextor will be 3 for 3.

[johnw]

Most of us already knew this but didn't have any solid proof. That article went live without sharing the proof.

Anand apparently got confirmation from OCZ before posting the article. He was vague in the actual article, but he posted a comment where he said so explicitly.

[Christopher]

This isn't really surprising, nor is it a big deal to anyone other than OCZ and their shareholders.

[ChinStrap]

http://www.anandtech.com/show/5741/o...-based-silicon

Christopher? called shot?

[Gilgamesh]

The Plextor M3P uses the Marvell 88SS9174 controller, but the Vertex 4 is using a controller that is virtually the same as the new Marvell 88SS9187. Also the Plextor is using toggle flash, while the Vertex 4 is using IMFT ONFI flash. So the difference is more than just firmware.

So, of the three differences

1) Controller: V4 > M3P

2) Flash: M3P > V4
toggle flash wins on performance

3) Firmware: M3P > V4
The Vertex 4 suffers from the same problem as the m4 and earlier Plextor's did, which is poor sequential reads at low QD. But Crucial and Plextor fixed that issue with later firmwares. The Vertex 4 firmware still has the problem.


So the M3P wins 2 out of 3. Anyway, I expect Plextor will soon release a new model with the Marvell 88SS9187 controller, then Plextor will be 3 for 3.

Waiting for Crucial M5 - Marvell 88SS9187 - 512MB cache - nand sync IMFT 20nm and 256 GB @ 200-250$

[Anvil]

I heard the rumors some time ago, a bit surprised but why shouldn't they, the Marvell controller is a great platform.

The new one (9187) should be very interesting.

[Nizzen]

LOL

I posted the info with the Marvell controller on the OCZ forums in the vertex 4 thread. Now it is gone, also with some oter answers to that.

OCZ do not want people to know...

[RyderOCZ]

Please do not spread false information: http://www.ocztechnologyforum.com/fo...-based-Silicon

If you read my post in there, I merged the 2 threads, we only need 1 on the topic.

Thank you.

Here is my post explaining that: http://www.ocztechnologyforum.com/fo...l=1#post723385

Here is your post: http://www.ocztechnologyforum.com/fo...l=1#post723360

All made part of the 1 thread.

[johnw]

Please do not spread false information

That is a very funny comment coming from OCZ.

I think spreading false information is OCZ's corporate strategy.

[Nizzen]

Please do not spread false information: http://www.ocztechnologyforum.com/fo...-based-Silicon

If you read my post in there, I merged the 2 threads, we only need 1 on the topic.

Thank you.

Here is my post explaining that: http://www.ocztechnologyforum.com/fo...l=1#post723385

Here is your post: http://www.ocztechnologyforum.com/fo...l=1#post723360

All made part of the 1 thread.

Sorry, did not see the new thread. Only used direct shortcut to "Vertex 4 performance and analysis" thred.

BTW, Vertex 4 is fast So far so good

[Royteane]

Noob question. Due to m3p is unavailable in my country, if I buy vertex4 for Os, can I update firmware without any configuration changed? And how to?

M3s is available here, is it a better choice than vt4?

[johnw]

M3s is available here, is it a better choice than vt4?

Yes. I'd get a Plextor M3S or a Samsung 830 if you cannot get an M3P.

[Christopher]

The 830 and M3's are both excellent. The M3S has a longer warranty, while the 830 is more interesting in my opinion, if not necessarily faster. Get whichever one is cheaper where you are located, unless there is some over-riding concern of if the price difference is small.

[Ao1]

Ao1:
You can ask OCZ on the OCZ forums.
I have spammed to much there already :p
http://www.ocztechnologyforum.com/fo...d=1#post722642

I think I have worked it out. It’s the implementation of a new SATA 3.0 feature that enables host (rather than device) control over NCQ streaming. I believe you need an active partition to do that. Should be easy to test it out. EDIT: Run the same benchmark on an un-partitioned drive and a partitioned drive and then maybe verify by running some benches on a SATA 3.0 port using an xfer size that does not saturate SATA 2.0 and then repeat the benchmark on a SATA 2.0 port.

There is info on how NCQ Streaming works here: http://www.serialata.org/documents/S...bs-to-6gbs.pdf

This is a Marvell based feature that will be appearing in new Marvell based SSD’s that should be appearing in a couple of months once validation testing is completed.

If OCZ would like to explain how they managed to get a V4 to load a game in half the time of a V3 I'm all ears.

[Christopher]

I think I have worked it out. It’s the implementation of a new SATA 3.0 feature that enables host (rather than device) control over NCQ streaming. I believe you need an active partition to do that. Should be easy to test it out. EDIT: Run the same benchmark on an un-partitioned drive and a partitioned drive and then maybe verify by running some benches on a SATA 3.0 port using an xfer size that does not saturate SATA 2.0 and then repeat the benchmark on a SATA 2.0 port.

There is info on how NCQ Streaming works here: http://www.serialata.org/documents/S...bs-to-6gbs.pdf

This is a Marvell based feature that will be appearing in new Marvell based SSD’s that should be appearing in a couple of months once validation testing is completed.

If OCZ would like to explain how they managed to get a V4 to load a game in half the time of a V3 I'm all ears.

Is this how you play with "le scheduler"?

[Marc HFR]

If OCZ would like to explain how they managed to get a V4 to load a game in half the time of a V3 I'm all ears.

Yes, strange results. Here on game loading (BF3) no difference at all between V3 and V4 and on other app launch (Windows or 3d studio max) V4 is a bit slower than V3

[Nizzen]

Same problem with Vertex 4 as M4 with Areca 1880ix and other lsi sas2 controllers?
If this is the case... Dooooooohhhh. My 4 V4s arrives tomorrow and the plan is to use it on mye Areca :S

http://www.ocztechnologyforum.com/fo...sting-Vertex4s

[Royteane]

How about Corsair performance pro vs m3s ? Bot available in my country

[Nizzen]

Now I can confirm problems with 4x Vertex 4 128gb R0 with Areca 1880ix. Suprise

Tony OCZ:

we have tested Vtx4 on various raid cards, however it looks like issues are dependent on the number of drives used, with 2 or 3 you may see no issues, with 4 or higher we may have an issue. We are looking into this right now and do talk with Areca etc. If you see issues, you are going to have to wait for us to work thru them, if we swap drives you will probably see the same issue if the FW is the same etc.

Please be aware of this and I will report where the fix is needed (either FW on the drives or bios on the raid card)

Looks like a fix is coming. But is it coming soon enough? In Norway there is 14 days return with refund on things bought on the internet... I am hoping that OCZ is convincing me to keep the ssds. I getting some great numbers But other numbers is just strange

[JR.]

Hi,

My first post here

I thought folks may like to see what 2 of OCZ's latest Vertex 4s can achieve in a P67 rig.

Benchmarks run with Intel RST 11.1, high performance power plan and CPU at 4800MHz (64K stripe).





Kind regds, JR

[AMD_Freak]

Now I can confirm problems with 4x Vertex 4 128gb R0 with Areca 1880ix. Suprise

Tony OCZ:



Looks like a fix is coming. But is it coming soon enough? In Norway there is 14 days return with refund on things bought on the internet... I am hoping that OCZ is convincing me to keep the ssds. I getting some great numbers But other numbers is just strange

yea who would have thought it was your hardware and not the drives get your jumping shoes on your going to need them for all the hoops