Jump to content
Sign in to follow this  
Angry_Games

hard drives & raid - benchmark and compare!

Recommended Posts

brakezone:

 

No need to be upset, when I'm talking in general, I am talking in general and not about you as a person or your personal opinions about what is or isn't hardware raid. We are all here to learn, aren't we?

 

(Close your eyes and take 10 deep breaths)

 

You quoted me all wrong. I will let someone more intelligent answer your questions, but yes, nvraid is hardware. (Not the thing I think about when using the term hardware raid, but yes it is hardware since it is on the motherboard). Most people would call it hardware raid. I would not, and I pointed that out in my post. And Nvidida most definatly is not lying - if they did, well they would be in alot of trouble.

Share this post


Link to post
Share on other sites

well, i'm sorry if I gave the impression that I was upset. I also want to appologize if I misunderstood your post. The actual theme of my post was that I did not want to be misundesrtood.

 

 

So, how does your 4X raptor setup working out?

 

I think they say that the raid is hardware accellerated because nforce3 has something in it that accellerates the raid; however, some of the process may be carried out by the CPU. If that were true, that would explain why they say "hardware accellerated" instead of saying "hardware raid". On the other hand, I don't know how it is all being carried out, it may very well be full hardware or full software based upon what I currently know.

Share this post


Link to post
Share on other sites

Technically speaking, I'd say that nvidia is not lieing. The controller itself is hardware, and built into the motherboard. It is acellerated by hardware, which is your cpu and ram. For that matter, any raid can say it is hardware acellerated. Even software raid is "hardware acellerated". I think the key word here is "acellerated". They are not saying it is completely hardware, but saying it is acellarated by some hardware. But to be a "true" hardware raid, it would have to have it's own cpu and ram, and no on-board solution has this. I have 4 drives in my array, and the benchmark says it's using 11% cpu. I've gained 50% in overclocking, so giving up 11% here to help the bottleneck out, is worth it.

Share this post


Link to post
Share on other sites

brakezone:

 

Ok, no problem :)

 

It's working great... I just wished I could get it to accept 16kb as my stripe size when I set the whole array up. Somehow it was just not recognized as an raid array when set to 16kb. However setting it to 64kb worked without any problems... even tried to low-level format the drives and writing zeros to them before making the array with 16kb stripe size, nothing worked.. but then when I made the array with 64kb, voila, it works.

 

So it is 64kb stripe size with 4kb cluster size... the speed is great, but not what I wanted for small files... I think the speed would be better for smaller files (OS) with a 16kb stripe size.. but yeah it's OK.

 

(from earlier in this thread)

 

This is my RAID Array:

 

4x74GB Raptors in RAID 0 with Stripe size 64kb and cluster size 4kb, and I'm getting 113Mb/s in sequential read and 227MB/s burst read.

 

hdtach4x74gbraptors6444iopagel.th.jpg

 

atto4x74gbraptors6444iopageloc.th.jpg

Share this post


Link to post
Share on other sites

well yeah every single little bit or byte on a computer is hardware accellerated, but by convention, when you ask nvidia if something is done with hardware or software you consider hardware to mean that it has a specialized chip that handles it, taking the load off the cpu; on the other hand, you mean software if the cpu is doing it (because the cpu is more of a multipurpose device, it can be programmed to carry out other functions). Now obviously, any hardware device that is connected to a motherboard which has a device driver requires some sort of service from the cpu; however, the driver acts as an interface between the hardware and the operating system and allows the user to control the hardware is quite a bit different from when the device is emulated by a cpu in order to provide the same function, which i've never seen anybody ever claim that it was not software at this point.

 

If they are saying its hardware because its accellerated by the cpu, then they would be totally going against a common convention and should be slapped for it.

 

I saw that post earlier with your setup, it looks pretty good.

Share this post


Link to post
Share on other sites

Thanks. Yes I think you are right, slap on :)

 

I think this was a way for the mobo/chip manufacturers to get "ordinary" people to use raid puttin it in a onboard solution. Before the onboard solutions you had to buy an controller card and probably you had to do some research before you bought anything, first of all to get interested in raid, but also to know what to buy and what not to buy. (market was for few entusiast and video/music editors etc)

 

Now there is alot of mobo's with onboard raid solutions, and when you have it onboard anyways people want to try it out if they have more than 2 drives lying around, since the already got raid on the mobo. So the more people trying it out and experiementing with it, the more people will get there eyes on how fabiolus raid really is (I know there are haters, but we love you anyways :) ).

 

When demand on the market raises, more people will get intersted in the more expensive, high-performance raid controller cards (with all fancy CPU and dedicated MEMORY). So price go down, they sell more raid chips, more money gets spent into development = better hardware chips in the future. Everyone is happy. Everyone lives happy ever after.

 

The end. :nod:

Share this post


Link to post
Share on other sites

I would like to get a dedicated raid card, but $300 for a decent one is at all attractive to me.

 

 

Anyone noticed that obviously when u use both the sil controller and the nvidia one that avg. read speads go down a tad?

Share this post


Link to post
Share on other sites

anyone notice the transfer rates of hardware cards? last time i looked, we were right there with em. I know i could very well be wrong, as im just going by what i saw on newegg.

Share this post


Link to post
Share on other sites
much better results after i botched my install and had to redo:

 

HDTach 8mb

8mbcopy7mc.th.jpg

HDTach 32mb

32mb3wm.th.jpg

 

ATTO Bench showing my chipset's wall

untcopy5ip.th.jpg

 

i've noticed people with Raptors getting 200+mb/s burst speeds. why is mine so much higher than there's? 300+mb/s.

Share this post


Link to post
Share on other sites
Delirious, i'm using 2 of these: Seagate 7200.9 . Review . i think the single platter design makes up for the lack of not having 16mb of cache. i also think my 1.5gb/s interface is holding it back some.

 

 

Im using the same drives and was getting about the same score till i added another array. then it droped to about 100mbs using 16k stripe default cluster.

 

What size stripe are u using?

Share this post


Link to post
Share on other sites

i'm using 16/4 for my raid setup as reading through here seemed to suggest the 4:1 ration was best. didn't have the time to try larger mixes like 32:8, etc. wouldn't think any overhead would make you drop. matter of fact i always thought that a larger RAID0 array would render better speeds. maybe you need to try a larger stripping mix. some guys are using 64:16 with nice results from my reading around.

 

at first i was thinking the density of my platters was getting me some good results but you have the rmp speeds like crazy so that should keep you way ahead of me.

Share this post


Link to post
Share on other sites
@inphared

 

Two Raptors in RAID-0 is the absolute best performance you can buy.

 

Two Hitachi 80GB SATA 3.0Gbps drives in RAID-0 is the second best you can buy.

 

Just about any drive will get the job done for storage.

 

I got 90% of the Raptor performance from my Hitachi drives for 1/3rd the cost. I thought that was a pretty good deal since I didn't need the "absolute best".

 

Game/level load times are dependent upon the weakest link in the chain. If you run RAID-0 on a weak controller, chipset, CPU or less than 1GB RAM, you won't see much improvement.

 

The onboard RAID in the nF4 chipset, A64 processors and at least 1GB of RAM combine to give you significant performance gains in game/level load times.

 

 

Just wondering, would 4 Hitachi Sata2 drive in Raid0 beat 2 raptors?

 

Are there any results for 4 Sata2 hitachis in this thread? Angry did 3hitachis sata2 plus a regular, but i couldnt find 4 sata2 drives...

 

 

And would you recomend

 

4xdrives in Raid0 (for os and apps) plus my 250gb WD for music and vids

 

or

 

2Raid0's with 2drives in each array (one raid for windows the other for apps) plus the 250gb for music and vids

 

or

 

some other mix??

 

Thanks Snook

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this  

×
×
  • Create New...