Buyer beware: OCZ Octane "S2" is slow

[SickBeast]

Yesterday I bought a 128gb OCZ Octane S2 SSD after reading reviews of the regular octane. I didn't realize that the SATA 2 version would be so much slower than the SATA 3 one. It also does not have the new 1.13 firmware upgrade which doubles the performance of the SATA 3 drive.

I returned the drive today.

I just wanted to post this because there are no reviews of the S2 drives yet, and I want to let people know that it is *much* slower than the SATA 3 version.

 

[iamlilysdad]

Could that be because of the interface?

 

[SickBeast]

No. They must have used cheaper NAND chips or else they crippled the drive badly in other ways.

The interface is only going to cap the large sequential transfers. I was getting slow speeds in random 4k reads and writes.

 

[iluvdeal]

Did you run any benchmarks (ex. HD Tune) to quantify just how slow it was? What were the numbers?

 

[SickBeast]

I ran Crystal Disk Mark, but my results are gone as they were saved to the SSD before I returned it.

The drive got a 7.5 in the Windows Experience Index.

It seemed marginally slower than my old OCZ Vertex 2.

 

[groberts101]

Nice to know. I almost bought one to test the other day too. Then I thought.. "why?.. I'll never use another sata2 drive anyways". lol

 

[Charlie98]

Just FYI, I had a problem with write times on my SATA3 Agility... until I turned off my C3/C6 states.

 

[exdeath]

No. They must have used cheaper NAND chips or else they crippled the drive badly in other ways.

The interface is only going to cap the large sequential transfers. I was getting slow speeds in random 4k reads and writes.

The interface will put a cap on the small reads/writes even worse as the protocol overhead for thousands of inefficient uncombined random reads/writes needs command packets/headers for each little transfer and it ends up consuming a large portion of useable SATA bandwidth with command traffic. A drive that can physically do 200 MB/sec randoms on SATA 3 really needs 300-400 MB/sec so on SATA 2 it won't get anywhere near 200 MB/sec even though it appears that 200 MB/sec data alone is under the 300 MB/sec cap.

This is the biggest difference between USB 2 and USB 3 as well when it comes to many small transfers.

 

[bryanW1995]

Nice to know. I almost bought one to test the other day too. Then I thought.. "why?.. I'll never use another sata2 drive anyways". lol

It's unbelievably annoying that the best option for high speed data (intel CPU + Sata 6 gb/s ssd) is still limited to only two good onboard connectors. I'm not even a power user, but I'm worried that my IVB build this fall will still be stuck with only 2 connectors. Have you heard anything on this? When/if intel plans to give us MOAR bandwidth???

 

[groberts101]

Last I read we will still be looking at 2 x 6G ports for IB with a "chance"(however slight it may be) for additional 6G ports being introduced on IB-E.

As for total PCH max throughput?.. doubtful that will change much in relation to the current chips 1300-1400MB/s limits.

IMO, besides the processor gains to be had?.. the newer platforms are more about massive PCI-E lane implementation to give those running major amounts of cards less concerns for bandwidth hogging. Now you will be able to run 3 gfx cards AND a riadcard with arrays of SSD's without major losses. We couldn't do that before these newest boards came along and we had to divy up what we had available.

 

[slow_poke]

groberts101, can you tell me how you determined the PCH max bandwidth? thanks

 

[=Wendy=]

On P67/Z68/H67/X79/Z77 PCH you have 8 PCIe2 lanes with each lane having 10Gbps of synchronous bandwidth, so you have a total of 80Gbps of synchronous bandwidth for the PCH. That's without overheads, and those 8 PCIe2 lanes and bandwidth are shared among all the peripherals on the PCH.

If Z77 PCH was to be a PCIe3 device, then you would have had twice the bandwidth, so certainly 4 SATA 6Gbps ports would have been possible. It looks like Z77 PCH will be a PCIe2 device though.

In any case, the PCH has to connect to the CPU via DMI, which is only 20Gbps on P67/Z68/H67/X79, so that's all you really have. Perhaps it's the bandwidth on DMI that is the bottleneck.

Maybe DMI might have more bandwidth on Z77. We will have to wait and see.

 

[Shmee]

Good to know about the S2 octane. Would this also apply if a Sata 3 octane was used on a sata 2 controller?

This is certainly worthy of further investigation

 

[groberts101]

Thanks to Wendy on that one. I botched it up good that time on my quick "fly by" and should have read "sata max throughput". lol

And yeah.. we can only hope that the divy'ing up of said resources allows greater allocation of sata bandwidth on future platforms. I'm not holding my breath though, as I simply don't believe that Intel wants us pushing 6 drive SSD arrays over 2GB/s on their chips.

@Shmee.. only losses would be sequential as the sata2 chip will not bottleneck the 6G Octane anyways.

 

[gramboh]

Z77 only has 2 Intel SATA3 connectors + 4 SATA2 (same as P67/Z68). Lame as hell I know, was hoping to see more.

 

[Elixer]

I ran Crystal Disk Mark, but my results are gone as they were saved to the SSD before I returned it.

The drive got a 7.5 in the Windows Experience Index.

It seemed marginally slower than my old OCZ Vertex 2.

Curious, where did you buy the SSD from?

 

[zijin_cheng]

I have a question about this, there's the Octane S2 128Gb on sale for $65, and I'm gonna be moving from mechanical to SSD, you said its gonna be slow, I'm sure I'll still notice a difference when I use it and even after it gets to its slowest point right?

 

[Coup27]

Definitely. You will see a big difference moving from a mechanical drive to that SSD. At $65 you can't really argue and have to accept that it won't be up to the standard of the leading drives but then it costs a fraction of the price.

 

[theshowman]

Just for completeness of this thread, here is my benchmark results for the Octane S2.

Dell Vostro V130 Laptop
Intel(R) 5 Series 6 Port SATA AHCI Controller
Mobile Intel® HM57 Express Chipset

-----------------------------------------------------------------------
CrystalDiskMark 3.0.1 x64 (C) 2007-2010 hiyohiyo
Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]

Sequential Read : 180.758 MB/s
Sequential Write : 106.422 MB/s
Random Read 512KB : 100.342 MB/s
Random Write 512KB : 88.334 MB/s
Random Read 4KB (QD=1) : 8.155 MB/s [ 1991.0 IOPS]
Random Write 4KB (QD=1) : 26.672 MB/s [ 6511.6 IOPS]
Random Read 4KB (QD=32) : 63.355 MB/s [ 15467.6 IOPS]
Random Write 4KB (QD=32) : 35.581 MB/s [ 8686.7 IOPS]

Test : 1000 MB [C: 53.1% (55.9/105.4 GB)] (x5)
Date : 2012/09/05 7:44:25
OS : Windows 7 Ultimate Edition SP1 [6.1 Build 7601] (x64)


The 4K reads are definitely sub-par. I did make a note of it in the OCZ forums, but the support personnel weren't terribly alarmed by it:

http://www.ocztechnologyforum.com/f...chmark-results-seem-weird&p=742332#post742332

 

[alexruiz]

I installed this SSD in a system I built (yes, it was a killer price, hence why I got it). While the benchmarks show it as slow, I would be hard pressed to say that I could feel it slower compared to another system where I have a Sandisk Extreme 240GB, arguably the fastest SSD available right now.

I think the issue here is that under certain scenarios, the octane S2 will feel slower. What boggles my mind is why it is still priced at where it is when the demonstrably faster solutions are almost the same price.

 

[Hellhammer]

The 4K reads are definitely sub-par. I did make a note of it in the OCZ forums, but the support personnel weren't terribly alarmed by it:

That's what you get with a mediocre controller and asynchronous NAND. It's nothing out of the ordinary.