Kobalt Q9650 vs Q9550/X3360 and OCZ Core II SSD Benchmarking - Single 128GB, RAID5 (3 x 128GB) vs Single WD250BE

OK, so this isn't going to be the most enthralling read ever posted on the nbr forum but I have been doing a lot of testing and know that people have been asking about the performance of various components. The question I keep being asked at work is "how much faster are Solid State Drives", well here's the answer!!

Before the figures I'd just like to say something about these OCZ SSDs. I know that Sager have pulled them from their options but here's our position - we have had no issues with single drives but currently there seems to be a niggle with 128GB SSDs and RAID5. I must stress that it isn't the OCZ drives that are at fault, other brands act the same but because of our companies relationship with OCZ I have been testing their drives and passing the information onto them (UK and US) in order to see if we can improve compatability.

I used the Passmark default HDD benchmarks for this which are designed to emulate different HDD scenarios, they do not show the drives maximum outright speeds but instead MB/s figures under different conditions. These (in my eyes) are more important than maximum MB/s in an artificial file transfer bench as it actually shows real world useage benefits. Take a look at the Read / Write and Sequential / Random percentages in each test - you will notice that the RAID 5 is especially beneficial in the Webserver Benchmark but in others it's pretty much what you would see with 2 SSDs in RAID 0 or 1.

System Spec

Kobalt Comanche SLI (D901C) Intel965/ICH8R
Intel X3360 CPU
4GB (2 x 2GB) DDR2 800MHz OCZ RAM
Windows Vista Ultimate 64
2 x 8800M GTX (obviously will have no effect on HDD results!!)

HDDs with fresh install (slipstream so the 3 installs are identical):

1 x Western Digital 250GB 7,200rpm Black Edition (WD2500BE)

1 x OCZ Core 128GB SSD

3 x OCZ Core 128GB SSD in RAID 5









Results are pretty conclusive apart from the Workstation Benchmark which I'm pretty sure is a spurious result on the RAID5, I didn't notice it at the time so didn't re-test but I have included the chart to show at least the WD2500BE vs single OCZ SSD.


Q9650 vs Q9550/X3360

The speed benefit pretty much speaks for itself, 2.83GHz vs 3.0GHz = 6% clock increase which will give 6% faster benchmark results. It is actually an X3360, think we were out of stock of Q9550s at the time.

I did some wPrime benchies on this about 3 weeks ago ago and the main aim of testing was stability/heat as we knew exactly what the performance increase would be. Each screenshot is the same spec system as the HDD benchmarks but with 1 x WD2500BE.

I just ran each test once for these screenshots but 10 runs of the 32 test pretty much level out at the 1024 max temps which level out under load at 67C max. RealTemp was running from the start so the max temp is what the CPU reached at the end of the test. If anyone if unfamiliar with wPrime it's a program that runs all cores at 100% load and lets you know when each thread finishes - an easy way to spot core throttling.

X3360 wPrime32 Speed Test


Q9650 (ES) wPrime32 Speed Test


Q9650 (ES) wPrime1024 Stability Test


(Intel Core 2 Quad 3000MHz (ES) is an Engineering Sample of a Q9650, I did this testing a while ago but they'll be pretty much identical to the boxed retail versions)

You'll notice that the temps are pretty much the same as a desktop with a stock Intel HSF and bearing in mind that an Intel Server Motherboard doesn't ramp the CPU fan up to 100% until the CPU hits 70C they are nicely within decent limits. This is getting close to the extreme of CPU burning in and so the max temps realy are the absolute max temps you'll see with the chassis on a flat table top.

 

Wow, the ssds are fast. Any thoughts on power consumption comparisons?

Its good to know that the Q9650 actually does work without many issues. Hope to see them offered soon.

 

The Q9650 is now offered from Sager, if not already from Kobalt as well

 

Haven't got power consumption figures I'm affraid though I expect I can get some from OCZ next week.

I expect the Q9650s will be up on the website over the weekend - I believe stock is about a week away and they have to sort out what happens to existing orders, I expect we will do what Sager has done and fre upgrade Q9550 to Q9650 etc etc

 

I think those wPrime scores might be off... since you might not have set the threads to 4 (for the quad-core) in Advanced Settings.

 

With Wprime 1.61 I got 15.4 s at 2.66 GHz and 22.3 s at 2.0GHz on XP. There is quite a big difference between Vista and XP :-s.

PL

 

I forgot to mention.

ONLY use wPrime Version 1.55

... its the most recent "Stable Build"

 

In favor of which, _Vista or XP?

 

Yup 1.55 is classed as the most stable build but that's just because the later versions aren't comparable speed wise. So the results I posted are spot on for 1.6 but they aren't comparable to 1.55 results, probably should have mentioned it at the time.

 

Can you sell me those SSDs that you are testing to me for half price?

 

i am confused on the hdd benchmarks because i only use hd tune. is it possible for you to do an hd tune benchmark for the raid 5 array? i am curious how well it stacks up against a twin hdd raid 0 array like mine.

 

So wait, there arent problems with ocz hdds in our systems unless they are in raid? How hard would it be to install one ourselves?

 

With 3 SSDs in RAID 5 it will blitz 2 x 320GBs, if I get time I'll get hdtune up and running. Got a couple of RAID 5 320GB builds at the end of the week so I'll put them up too on Friday - that'll be a more sensible comparison!

 

wow those results really make me wanna wait a lil and see how SSDs do by the end of the year. they really make 7,200 rpms look like yesterday technology(faster, quieter, more stable, longer life-span...)

 

hey what about Raid 0... I want to see the difference.. I want all out performance.. I back my machine up nightly......

I bet the Raid 0 is even more impressive..

Can I get 3 drives today?If so pm me.

Joe

 

Thank you very much Neil for taking the time to do these tests. With the prices dropping on SSDs, they are starting to look really good. I may have to get a couple soon

 

My [email protected] gets better scores

 

3 people asking for SSDs - are there supply issues in the US?

 

I doubt there is a supply issue since you can buy them from newegg.com

I think the cheapest 34Gb SATAII are about 160 dollars now, and the 64GB SATAII are about 220 dollars. Though from the reviews I have been reading, those cheaper SSD seem to freeze from time to time when multiple programs running at the same time.

http://www.newegg.com/Product/Produ...actory=1550&SubCategory=636&SpeTabStoreType=0

 

I have read so many issues with these drives lately...

Does anybody have 3 of them in a raid 0 and what is the performance...

How come you haven't posted anymore details lately Neil?

JOe

 

they are a bit busy building laptops at the moment (including mine).

 

I've posted elsewhere on the forums that SSDs make a very noticeable real world performance improvement. Furthermore, if, as we do, you develop software that requires multiple heavy weight tools running simultaneously the difference in elapsed time for common complex operations can be can be 3X over a similiar Sager with HDs. I use MTRON server SSDs (SLC technology) in RAID 0 in both my laptop and desktops (built by OverdrivePC).

I personally recommend waiting for the Intel MLC SSDs due to some issues with other SSDs (like OCZ). See here for the best overall article on the technologies as well as pros and cons:

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3403


I doubt SSDs matter very much in straight gaming. My 3DMark06 scores for the OPC built quadcore with SSDs aren't really very different from other Clevo 901c machines. But I don't have a high end graphics card as I don't game so maybe I'm wrong.

Also, if your integrator/reseller stands behind the SSD (performance not just malfunction, mine does) then you might want to take the plunge but they are expensive and not always easy to integrate, more so with Vista.

I won't buy another machine without high performance SSDs.

Phil Schaadt

 

The problem is not with specific brands of SSDs, it's the way they interact with the RAID controller and vice versa. I have not carried out extensive testing on RAID 0 but did have issues with RAID 5, as have other companies which is why you won't find Sager offering them in RAID either.

If OverdrivePC stand by them in RAID then that's fair enough, maybe they are having better luck with RAID 0 or the MTRON SSDs though they are server SSDs and 4x the price of OCZ drives!

 

Since we have a team of 8 to ten people working on a fixed price contract to deliver software to a global corporation, buying higher priced gear that we know to be reliable and cuts elapsed time makes sense for our firm. We buy multiple custom performance systems from Velocity/OPC for specific development jobs and pay OPC for what we need.

This is not a specific plug for OPC and it is definitely not a knock at Kobalt or anyone else. Note that I recommended against anyone buying SSDs just for fun at this time. In a year, I'm sure my recommendation will be to only buy SSDs for performance.

Phil Schaadt