Who says 2 X 74GIG Raptors isnt better than one? Anandtech?

[GotNoRice]

One thing I see a lot is that someone will take their existing drive configuration, change it, have to do a re-install (because they changed their boot drive config), and then attempt to compare the performance of the PC with a completely fresh install to how it was prior. Last week I was cleaning up and putting XP on an old Pentium2 450Mhz that had a 17.4g 5400rpm hard drive. I was totally blown away by the speed in which this OLD system installed XP. Only thing I could figure is that I had just installed a new CD-Rom drive and it seemed to copy files from the disc fast, which normally takes up a lot of time. After XP was installed everything including the boot times was snappy. As long as the system has a good amount of ram and a fast CD-Rom this is usually my experience with most computers I do installs regardless of other hardware. It seems kinda silly to draw conclusions based on that.

I had two 36g raptors in Raid-0 prior to going with 15k scsi. Raid-0 did seem slightly faster than just using one single drive. What seemed even faster however, was when I used both drives but ran them separate. I made one my OS/apps drive and my 2nd one just a Games drive. Having that 2nd drive there, with nothing ever asked of it other than games seemed to give a larger performance boost than when I had both drives in Raid-0. Maybe someone could do a dual raid-0 setup where one array is for their OS and another array is for their games and perhaps get the best of both worlds?

For comparison purposes, here is the HDtach from my new Fujitsu 15,000rpm SCSI drive. I can say without a doubt it is faster than any configuration I ever had my raptors in, but I’d love to get a raptor 150 or a pair of 74’s and do some real comparisons.

 

[Jibly]

Here's my two 74's in raid 0, 32k stripe:

 

[dominick32]

I appreciate everyones comments in this thread and I really hope that it does steer away from a heated discussion and stay on the point of being a pure analysis from my own "personal perspective" of raid. I say personal, because my own perspective will most definitely differ from many of you and therefore the so called "real world feeling" results will differ again from one individual to the next.

However, I do own the best video card on the market right now, basically the fastest retail dual core AMD processor on the market besides the FX-60 (x2 4800+ @ 2.6Ghz), 2 gigs of ram with excellent overclockability, and basically a top of the line configuration, if you want to call it that.

Back to my post thread statement: "Raid is for the dektop user that wants the maximum possible performance squeezed out of his machine. The overclocker, the hardware fanatic, the geek. "

I use that statement because technically the only thing bottlenecking my computer was the hard drive. Heck, the only bottleneck to most non raid systems today is the hard drive. I am the individual that wants the maximum possible performance out of my PC even if it means spending the extra amount of money on 2 raptors for a raid array, or buying the latest and greatest video card every 3 months. Theoretically even if my 2 raptors performed the exact same as a single raptor and displayed a 2% to 5% improvement over a non raid system. That 2% to 5% means a heck of a lot to a person like me that wants maximum performance out of my PC. If that was the scenario and it was a small gain, was it worth the money? To some of you, it isnt. To me, it is.

If you also remember, I started a thread a couple of days ago that can be read here: http://www.ocforums.com/showthread.php?t=441758

To the individual that was telling me numbers and benchmarks matter and the actual "feeling" of a difference does not. You can quote me on saying "I think I wasted my money on a single Raptor drive." Coming from a regular Western Digital SATA 7200rpm drive and moving up to the Single Raptor 74gig was a waste of money in my eyes. I did not notice the incredible improvement that I noticed when I bumped up to 2X Raptors. This is why I hope you can use my analysis as not being biased and most definitely valid. One raptor didnt show any improvement over my sata drive, but when I moved to a Raid 0 array of 2 raptors the improvement was dramatic. I apologize if I came off a little too "hurray hoorah, wow!!" in the comments about real world performance. Yes, if you psycho analyze my comments that may have sounded overly exaggerated to many of you, but you need to trust me on this one that I am not being biased here. I simply wanted to display the exact way I was feeling by using a keyboard and sometimes thats hard to do.

Simple math would be:

1 X WD2500 = Baseline SATA hard drive. Good performance, 7200rpm
1 X Raptor = No noticeable performance increase, or very little.
2 X Raptor (Raid 0) = Incredible gains realized, noted in every aspect of usage.

By looking at that simple equation I dont see how a biased opinion of PC performance can come out of it. Especially when I reviewed a single raptor one day before installing a second for the raid setup.

Anyhow, I hope this thread continues as non violent and as an analysis rather than a "Raid sucks, Raid Rules" argument.

Regards,
Dom

 

[GotNoRice]

I think that when a lot of people talk about Raid-0 not being worth it, they are talking from the perspective of a new drive purchase. For less than the cost of two 74g raptors you could get a raptor 150 that would most certainly outperform any configuration of dual 74g raptors. But if you already have one 74g raptor, getting a 2nd could be a real option.

 

[Jibly]

^ That's exactly what I did. I had my original one for about two years now. Just last week I ordered a second one. Definitely a worth while purchase.

 

[dominick32]

I took it upon myself to follow up on your claim. It looks as though I almost exactly specified the differences. lol As you can see, two 74 Gig Raptors in Raid trample the sustained performance of a single 150, not to mention the 150 has a slower random access time. Here are the results:

 

[GotNoRice]

Seek time is important; however the differences between the drives are very little to the point of insignificance. STR is an interesting metric that can be useful in understanding how the drive might perform under certain circumstances – normal desktop usage not being one of them. Higher STR really does not usually translate into real world performance increase.

The raptor 150 is still new enough that not all sites have had times to do full reviews against raid, etc. StorageReview, which I would certainly trust more regarding storage than a site like anandtech or gamepc, is currently working on their article but the Administrator of that site was kind enough to share some preliminary benchmarks.

http://forums.storagereview.net/ind...=221874&mode=threaded&show=&st=0&#entry221874

The raptor 150 not only prevails over dual 74g raptors, but quad 74g raptors as well. Note, this is using "real world" games and apps and not synthetic benchmarks.

I certainly await the completion of that article with a great deal of anticipation

 

[dominick32]

Seek time is important; however the differences between the drives are very little to the point of insignificance. STR is an interesting metric that can be useful in understanding how the drive might perform under certain circumstances – normal desktop usage not being one of them. Higher STR really does not usually translate into real world performance increase.

The raptor 150 is still new enough that not all sites have had times to do full reviews against raid, etc. StorageReview, which I would certainly trust more regarding storage than a site like anandtech or gamepc, is currently working on their article but the Administrator of that site was kind enough to share some preliminary benchmarks.

http://forums.storagereview.net/ind...=221874&mode=threaded&show=&st=0&#entry221874

The raptor 150 not only prevails over dual 74g raptors, but quad 74g raptors as well.

I certainly await the completion of that article with a great deal of anticipation

I would really like to see some substantial evidence of this posted by more tech review domains because I have yet to see or hear any evidence of this. First off, he is posting his results in his own forum, before the main article is released. I find it hard to believe that a 150 Gig Raptor can outperform 4 X 74's in Raid. I think the author is focusing strictly on i/o benchmarks for his review and for the last week or two, every article that I have read has differed from his results. 2 X 74 Gig Raptors in Raid have always performed better than one single 150 raptor.

You gave me a picture of an I/O Benchmark performing better on a single raptor 150 drive than 4X74 Raptors in raid which I think is not correct or stated properly in that gentlemans review. Now I will go ahead and give you an I/O benchmark of 2 X 74 gig raptors performing better than a single 150. Do you see where I am getting with this? We can go on for days comparing spreadsheets. LOL

Now I will give you charts in all of the benchmarks, including I/O. It can be clearly seen that 2 X 74 gig raptors outperform 1 single 150 in every aspect.

(This last picture is irrelevant to this thread but included for total review)

I truly do not believe a single 150 to outperform 2 X 74's in raid 0. Until I can see substantiated evidence backing up that gentlemans claim. As well as various other hardware review articles backing it up also. Heck, if and when I do see this I will be the first one to sell my raptors and hop on the 150 X 2 bandwagon. I might even buy you a bottle of champagne and pat you on the back. But until then, I just dont believe it.

Dom

 

[GotNoRice]

There are a few problems with the benchmarks you posted. First of all, you are still only posting synthetic benchmarks. You are trying to compare benchmarks using a synthetic tool (IOMeter) to real world game measurments. StorageReview also uses IOMeter in their testing and their results do not disagree with the ones you posted, it just simply does not reflect real world performance since it is not a real world app. More STR and transfer rate graphs, which once again don’t translate into real world performance. The seek times are comparable as .5ms is not significant enough to make a huge impact. The last graph with the burst transfer rate is completely irrelevant as they used a card with a huge (I believe 128mb) buffer which is obviously what that is measuring, not the drive itself. The article itself mentions this.

The fact that a raptor 150 outperforms a raptor 74g even in raid configurations is not surprising considering that the raptor 150 is brand new where as the 74g raptor is approaching something like 2+ years of age. I still believe that getting a 2nd raptor if you already have one is a great investment, but if you are starting from scratch it is hard to justify it compared to the raptor 150. That is not to say that dual 74g raptors isn’t a nice setup.

 

[dominick32]

There are a few problems with the benchmarks you posted. First of all, you are still only posting synthetic benchmarks. You are trying to compare benchmarks using a synthetic tool (IOMeter) to real world game measurments. StorageReview also uses IOMeter in their testing and their results do not disagree with the ones you posted, it just simply does not reflect real world performance since it is not a real world app. More STR and transfer rate graphs, which once again don’t translate into real world performance. The seek times are comparable as .5ms is not significant enough to make a huge impact. The last graph with the burst transfer rate is completely irrelevant as they used a card with a huge (I believe 128mb) buffer which is obviously what that is measuring, not the drive itself. The article itself mentions this.

The fact that a raptor 150 outperforms a raptor 74g even in raid configurations is not surprising considering that the raptor 150 is brand new where as the 74g raptor is approaching something like 2+ years of age. I still believe that getting a 2nd raptor if you already have one is a great investment, but if you are starting from scratch it is hard to justify it compared to the raptor 150. That is not to say that dual 74g raptors isn’t a nice setup.

I understand your point and where you are coming from with this, as yes indeed I am using synthetic benchmarks. In every single synthetic benchmarking application (HDTach, HDTune, IOMeter, Diskbench, ATTO) 2X74G Raptors trample a 150G Single drive. Does all of this information mean nothing?

(I know that Burst Speed has no relevance but I figured I would post a picture of every synthetic benchmark available from that review. Including the irrelevant Burst Speed)

What I dont agree on is that Reading and Writing to the hard disk is not "real world performance". I posted up a pic of write and read performance benchies.

When you install software on your computer do you not read and write from the hard disk? Especially if installing executables off of it. Is this not considered real world performance?

When you go to read a 10 Gigabyte mpeg4 file and edit that file in real time (processor and ram do account for much of this) does this not read and write to your hard drive? Is this not real world performance?

When you read off of your HDD when booting up Windows XP and you have that higher sustained read does this not equate to real world performance?

To me, real world performance is having that incredible sustained read/write speed to my hard drive. It is very notable to me in everyday computer usage and as you can see the "Synthetic" benchmarks that measure these read/write speeds show that it is also extremely notable as well.

However. I do see your point and where you are going with this. You could be absolutely correct about a single raptor 150 beating 74gig raptors in Raid on certain "real world benchmarks" but work with large file intensive tasks and see what comes out on top, raid or the single 150?

We will just have to wait and see on the final article. I will be anticipating this article as well, because if true I will most definitely be considering the purchase of a 150 X 2 raid0 array. hehe

I appreciate your response and information added to this thread. All aspects are welcome here and it is great to have all of the pros and cons thrown together in one thread.

Kind regards,
Dom

PS - Keep me posted through the PM system about updates on this article.

 

[billb]

Here's my two 74's in raid 0, 32k stripe:

Same Here.

 

[Jibly]

What causes mine to spike so much, processes?

 

[dominick32]

What causes mine to spike so much, processes?

Did you run the 32mb zone benchie or the 8mb zone?

Dom

 

[Jibly]

8mb

 

[dominick32]

8mb

Sometimes the 8mb benchy has a tendency to skew results. Try the longer 32mb test and post back. Lets see what changes.

Dom

 

[Jibly]

Big spikes?

 

[dominick32]

Big spikes?

You are perfectly in line with a normal HD Tach3.0 benchie for 2 Raptors. It looks good! hehe

129 sustained, 8.0 ms... looks fine

 

[Jibly]

Cool

 

[dominick32]

Cool

Well,
How does it feel compared to an old single 7200rpm setup?
I am extremely interested in this. Especially boot times, game loading times, general windows xp operating system usage, etc? etc?

Thanks man.

Dom

 

[Jibly]

Well, my main rig has had a single 74g raptor in it since the summer of 2004. I have other rigs with 7200 8mb sata's but I don't game on them. Overall, it's just as I expected, everything seems/feels faster. For windows, I really dont do much besides surf and play music. Gaming wise I only play BF2 atm, and beta test a game called Vanguard. Loadup and map changes is where the raid shines. I really wish I would have set this up years ago. I guess I got caught up in all the single drive hype. What a great addition this was to my rig, now I can't really think of anything else to upgrade....Dare I say, I'm done?