Madshrimps Forum Madness

Madshrimps Forum Madness (
-   WebNews (
-   -   Performance issues with OCZ/Supertalent/Patriot MLC SSDs Investigated (

jmke 9th September 2008 16:15

Performance issues with OCZ/Supertalent/Patriot MLC SSDs Investigated
Although this links to a section of their Intel X25-M review, it's worthy of a special mention for those who are looking to buy one of these affordable SSD drives, as they present cases were performance is really worse compared to classic HDDs or SLC based SSDs.
I had suspicions as to the nature of the problem based on my experience with it in my Mac Pro. The SuperTalent MLC drive in my machine would pause, most noticeably, randomly when I'd want to send an IM. What happens when you send an IM? Your logfile gets updated; a very small, random write to the disk. I turned to Iometer to simulate this behavior.

Iometer is a great tool for simulating disk accesses, you just need to know what sort of behavior you want to simulate. In my case I wanted to write tons of small files to the drive and look at latency, so I told Iometer to write 4KB files to the disk in a completely random pattern (100% random). I left the queue depth at 1 outstanding IO since I wanted to at least somewhat simulate a light desktop workload.

Iometer reports four results of importance: the number of IOs per second, the average MB/s, the average write latency and the maximum write latency. I looked at performance of four drives, the OCZ Core (Jmicron controller MLC), OCZ SLC (Samsung controller), Intel MLC (Intel controller) and the Seagate Momentus 7200.2 (a 7200RPM 2.5" notebook drive).

Though the OCZ core drive is our example, but please remember that this isn't an OCZ specific issue: the performance problems we see with this drive are apparent on all current MLC drives in the market that use a Jmicron controller with Samsung flash.

jmke 9th September 2008 16:23


This is just ridiculous. Average write latency climbs up to fifteen seconds, while max latency peaked at over thirty seconds for the JMicron based MLC drives. All this graph tells you is that you shouldn't dare use one of these drives in a server, but even at a queue depth of four the max latency is over two seconds which is completely attainable in a desktop scenario under heavy usage.
:ws: for their IOmeter tests to reveal this stunning performance issues of MLC drives with Jmicron controller and Samsung flash.

Crysis min FPS jumps up using the Intel SSD :D

phlegm 9th September 2008 17:07

I have 2 of these drives in RAID0 and they work like a dream. Vista and games load damn fast and I've had zero issues with corruption. But I made sure to read up on the drives before getting them and understand what I was getting into.

MLC is not ready for mainstream at the moment because they do require a bit of knowledge to use properly. Maybe Intel will iron it all out so that it's plug and go regardless of setup but they're definitely an enthusiast only product right now.

jmke 9th September 2008 17:15

what OS are you running? he tested OS X for daily use and registered very slow random writes; while Vista was also noticeably slower when doing multi-tasking.


they do require a bit of knowledge to use properly.
What are the tricks of the trade? :)

phlegm 9th September 2008 18:06


Originally Posted by jmke (Post 223542)
what OS are you running? he tested OS X for daily use and registered very slow random writes; while Vista was also noticeably slower when doing multi-tasking.

What are the tricks of the trade? :)

Don't use AHCI with these SSD's. The main reason is that NCQ will murder performance. A lot of higher-end controller cards will let you disable just NCQ, so do that instead and keep hot swap. What really sucks about this part is that some motherboards and laptops won't let you turn it off or don't turn it off when you tell it to. People with these boards are screwed.

Make sure you have a compatible motherboard if using the built-in controller. Some boards just refuse to play nice with them. There's no rhyme or reason to this part, just read up on it. I haven't heard of an add-in controller card having this issue yet, so who knows what's going on.

Be prepared to RMA your drive if it's acting weird. Some of these drives will just die in the first week and others will merely perform badly from the start. A lot of people on the OCZ forums will jump through hoops for days instead of just RMAing it immediately and saving a headache.

Never defrag it and disable auto-defrag on the drive.

Install the latest, greatest drivers for the built-in controller if you use it.

I remember duki also had some weird performance issues after doing the upgrade trick with Vista on his P35 board. I don't know if he ever figured out what the hell was going on with that. PM him if you're curious.

Edit: I'm using Vista Ultimate x64. Some people on Mac laptops were reporting that they couldn't disable AHCI. This may be the issue.

jmke 9th September 2008 18:08

excellent info there :ws: maybe an article detailing performance before/after these tweaks/settings would be interesting?

phlegm 9th September 2008 19:09

The only two you can really toggle are AHCI and the drivers. And I'm pretty sure the Anandtech article will get an update as soon as OCZ gets wind of it and send them a nastygram about how to set it up correctly for the Vista box. The Mac one may just be doomed unless Apple decides to fix it, not sure about that one.

I can send you an OCZ Core or two of mine if you want to do a writeup on them though.

jmke 9th September 2008 19:14

check your PM box :)

Kougar 9th September 2008 21:04

Thanks for the info Phlegm!

From what I know RAID mode on Gigabyte boards automatically includes AHCI... anyone know for sure?

jmke 11th September 2008 15:55

Settings used by OCZ, tested with Iometer, random read:

All times are GMT +1. The time now is 16:42.

Powered by vBulletin® - Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO