RAID 0 Experiment
RAID 0 Experiment
No matter what I try, Vista will not boot or install with the Gigabyte RAID on. Now I give up! At least I have three drives now so I can keep the OS, programs and audio on separate drives as Valis suggested. A good RAID card would be expensive and could be problematic. I will get a Raptor after I pay off my bills.
Re: RAID 0 Experiment
onboard raid is nasty. just get fast drives.
if you have seperate drives make a small partition on the outside of your fastest non OS drive and put the swap file there. make it a fixed size, the less windows has to "think" the better since it still cant even do a progress bar
if you have seperate drives make a small partition on the outside of your fastest non OS drive and put the swap file there. make it a fixed size, the less windows has to "think" the better since it still cant even do a progress bar

Re: RAID 0 Experiment
I don't think the swap file matters because I have 8 gigs of RAM.
Re: RAID 0 Experiment
Swapfile still gets used by windows regardless of how much ram you have installed. Vista will page data to disk that it believes isn't essential, and attempt to use large portions of your ram for 'prefetch' data, in addition to swap.
As for RAID mode, you will need the proper drivers and you will need to configure the raid controller properly as well. Before enabling RAID, how many harddrive 'slots' do you see in your bios POST screen? (4 or 6)
As for RAID mode, you will need the proper drivers and you will need to configure the raid controller properly as well. Before enabling RAID, how many harddrive 'slots' do you see in your bios POST screen? (4 or 6)
Re: RAID 0 Experiment
Installing the drivers is the problem. It doesn't like them. It either reboots or one of them causes the blue screen of death. It can have up to 6 hard drives. Striping the drives is fairly easy.
All the drives are the same speed so I guess cache doesn't matter.
All the drives are the same speed so I guess cache doesn't matter.
- kensuguro
- Posts: 4434
- Joined: Sun Jul 08, 2001 4:00 pm
- Location: BPM 60 to somewhere around 150
- Contact:
Re: RAID 0 Experiment
my only experience with striping is very bad... I think I had a raid fail 3 times, once a 2 disk raid I did myself, and twice on an editing machine at work.. all 3 times it totally wiped everything because of striping. Why do you need such server level bandwidth? I sort of doubt disk streaming sample libs will push it that far... or does it. If so, sample sizes are just getting ridiculous I think.
Re: RAID 0 Experiment
braincell, are you using an existing windows install? If so, did you change the drivers before changing the BIOS? Intel chipset (Intel Matrix RAID or some ICH flavors can use an adaptec RAID driver as well) or Nvidia, etc? Vista sp1? (has native AHCI & possibly native Intel ICH7/9 RAID support)
Ken, you perhaps missed our lengthy segue into RAID in this thread. Most of us agreed with you. Some thought & care into how you spread your data/workload out over the drives that are available yields tangible results for most of us. Unless you're feeding contiguous files (single long streams of data), drives are limited by their ability to seek, which sort of limits striping's effectiveness since you stripe to increase throughput.
There are also other ways to improve performance for what he's trying to do, as he recently found out by increasing the preload 'disk streaming' cache setting for his rompler.
Ken, you perhaps missed our lengthy segue into RAID in this thread. Most of us agreed with you. Some thought & care into how you spread your data/workload out over the drives that are available yields tangible results for most of us. Unless you're feeding contiguous files (single long streams of data), drives are limited by their ability to seek, which sort of limits striping's effectiveness since you stripe to increase throughput.
There are also other ways to improve performance for what he's trying to do, as he recently found out by increasing the preload 'disk streaming' cache setting for his rompler.
Re: RAID 0 Experiment
Reading the book, I think I missed a step. Stripping the drives is not the same as create array. It's a totally different step and command to enter it. I will try it again now since I missed a vital part. I try "repair" Vista and if that doesn't work, I will install Vista again. I think Repair works as it asks for the driver.
Note: Windows XP requires you to have a floppy drive to install a RAID. I always hated that. Vista will see any drive including memory stick which is yet another reason of the thousands of reasons why Vista is better.
Kensuguro,
I put a lot of demand on my drive. When simply playing back audio tracks in the sequencer, there is no problem. I stopped counting after 300 mono tracks. Playing the QL pianos very fast with sustain at the highest quality with multiple microphones is a different story. It works most of the time but there are a few clicks. They recommend a 7500 RPM drive but "faster is better". My drives are 7500.
I should buy the velociraptor but I can not afford that right now. My feeling is I built this very fast computer and I should be able to do anything on it. I had to compromise a lot on all my previous computers when making music. Not everyone experiences failures with RAID. Some people use them for years with no problem.
Note: Windows XP requires you to have a floppy drive to install a RAID. I always hated that. Vista will see any drive including memory stick which is yet another reason of the thousands of reasons why Vista is better.
Kensuguro,
I put a lot of demand on my drive. When simply playing back audio tracks in the sequencer, there is no problem. I stopped counting after 300 mono tracks. Playing the QL pianos very fast with sustain at the highest quality with multiple microphones is a different story. It works most of the time but there are a few clicks. They recommend a 7500 RPM drive but "faster is better". My drives are 7500.
I should buy the velociraptor but I can not afford that right now. My feeling is I built this very fast computer and I should be able to do anything on it. I had to compromise a lot on all my previous computers when making music. Not everyone experiences failures with RAID. Some people use them for years with no problem.
- kensuguro
- Posts: 4434
- Joined: Sun Jul 08, 2001 4:00 pm
- Location: BPM 60 to somewhere around 150
- Contact:
Re: RAID 0 Experiment
ah, QL pianos. makes sense. multiple mic positions at the same time can definitely take up lots of bandwidth. Seems very overkill tho. Actually, it seems stupid that the user has to go through such lengths to get the thing to work right. It's like the opposite of engineering.
Re: RAID 0 Experiment
I don't use the "room" mic but I use the "close" and "player". It's useful for me.
I found out that other RAID setup was for the Gigabyte RAID rather than the Intel. This board has two RAIDS. I may try the Gigabyte RAID next. It is said to be slower. Meanwhile I am going to install new firmware in my hard drives. I should just give up on this.
I found out that other RAID setup was for the Gigabyte RAID rather than the Intel. This board has two RAIDS. I may try the Gigabyte RAID next. It is said to be slower. Meanwhile I am going to install new firmware in my hard drives. I should just give up on this.
Re: RAID 0 Experiment
BC before you give up totally I understand the reason for doing this.
The trouble with samples is the seeking, not the data throughput.
There's a good reason that a dedicated RAID controller asks a preium price.
In a RAID 1 set up using 2 x HDD's you actually get 2 heads seeking the same content, it actually provides redundancy while boosting polyphony which is the main reason that you are having difficulty. Using the sustain pedals, and mic positions demand high polyphony.
Keep in mind also the smallest HDD you are using is your maximum size in whatever array you configure.
The first Intel RAID that was stock on a motherboard actually gave Promise Tech and other developers somewhat of a shock as they concentrated on 0, 0+1, 5 and 10.
Meanwhile RAID 1 on the motherboard was showing excellent results and low CPU usage.
Try RAID 1 and I think you'll agree, especially if you are using 3 or 4 drives.
Just imagine for every note you hit, you will have 4 heads seeking content.
As you add sustain pedals, etc. the other heads are hard at work finding their data.
If that doesn't work then placing the folders that contain your IR's one one drive, Mic positions on another, sustains on another, etc.
When Giga 3 came out it was the first developer to add the IR's and the first suggestion they had was to keep the Impulses seperate. That was back when Raptors showed a big difference.
If your seek times on the HDD's are 9 msec's or less those spec's are fine, even at 7200k.
With samples the cache size acts as another memory subsystem buffer, which is a step up from the 64k RAM buffers. The third and final area for memory being your HDD's.
Check It Out Before You Give Up...
The trouble with samples is the seeking, not the data throughput.
There's a good reason that a dedicated RAID controller asks a preium price.
In a RAID 1 set up using 2 x HDD's you actually get 2 heads seeking the same content, it actually provides redundancy while boosting polyphony which is the main reason that you are having difficulty. Using the sustain pedals, and mic positions demand high polyphony.
Keep in mind also the smallest HDD you are using is your maximum size in whatever array you configure.
The first Intel RAID that was stock on a motherboard actually gave Promise Tech and other developers somewhat of a shock as they concentrated on 0, 0+1, 5 and 10.
Meanwhile RAID 1 on the motherboard was showing excellent results and low CPU usage.
Try RAID 1 and I think you'll agree, especially if you are using 3 or 4 drives.
Just imagine for every note you hit, you will have 4 heads seeking content.
As you add sustain pedals, etc. the other heads are hard at work finding their data.
If that doesn't work then placing the folders that contain your IR's one one drive, Mic positions on another, sustains on another, etc.
When Giga 3 came out it was the first developer to add the IR's and the first suggestion they had was to keep the Impulses seperate. That was back when Raptors showed a big difference.
If your seek times on the HDD's are 9 msec's or less those spec's are fine, even at 7200k.
With samples the cache size acts as another memory subsystem buffer, which is a step up from the 64k RAM buffers. The third and final area for memory being your HDD's.
Check It Out Before You Give Up...

Re: RAID 0 Experiment
I could have them on separate drives by loading two instances of the QL Pianos but that is too complicated for editing.
I always heard RAID 0 is faster and that makes sense to me since it is not duplicating anything.
Update: I DID IT!
Valis, you don't install the drivers while installing Windows anymore. Things have changed.
1. Pick AHCI in BIOS (not RAID)
2. Install Windows (now Windows is RAID ready)
3. Change BIOS to RAID
4. Set up RAID array
5. Change BIOS back to AHCI
6. Boot Windows and install RAID software
7. Administrative Tools - Allocate and format
That took me 3 days to figure out!
Tomorrow I begin the testing phase of QL Pianos!!!
As you can see, this isn't the same hardware that people have been ragging on. I am confident it will hold up. Generally I find technology keeps getting better.
I always heard RAID 0 is faster and that makes sense to me since it is not duplicating anything.
Update: I DID IT!
Valis, you don't install the drivers while installing Windows anymore. Things have changed.
1. Pick AHCI in BIOS (not RAID)
2. Install Windows (now Windows is RAID ready)
3. Change BIOS to RAID
4. Set up RAID array
5. Change BIOS back to AHCI
6. Boot Windows and install RAID software
7. Administrative Tools - Allocate and format
That took me 3 days to figure out!
Tomorrow I begin the testing phase of QL Pianos!!!
As you can see, this isn't the same hardware that people have been ragging on. I am confident it will hold up. Generally I find technology keeps getting better.
Re: RAID 0 Experiment
You can do it either way, and this has been the case since forever. You only need to provide drivers during install if you're going to be building a RAID volume that windows is installed to during the installation process.braincell wrote: Valis, you don't install the drivers while installing Windows anymore. Things have changed.
Re: RAID 0 Experiment
it is indeed fast, especially if you put the data on more spindles. the more disks you use for striping, the faster it is supposed to get.braincell wrote:I always heard RAID 0 is faster and that makes sense to me since it is not duplicating anything.
RAID1 is not necessarily slower than RAID0 (using 2 drives) when it comes to reading, because data can be read from both drives in parallel, if the controller and/or the software driving it supports it.
writing to RAID1 won't benefit from that, of course.
if you're ok with the fact that RAID0 doesn't exactly fit the characteristics of common RAID (redundant) and keep in mind that the failure of a single drive will void the content of the whole array, well, then use it.
RAID0 is typically used for temporary data that needs throughput for writing. it is not really recommended for long time storage. if you can restore your libraries (and whatever content) quickly from optical storage or backup, it may still fit your needs. I've already seen disk mirrors dying at the same time, so decide for yourself what your priorities are.
having a dedicated RAID controller can, nevertheless, reduce the (noticable) difference in speed, since most of them are equipped with more and more cache memory that can be (mostly) used for reading and writing, provided you use a battery backup. I felt a remarkable speed jump when I switched on the write cache on my controller.
having more ports can even enable you to benefit from both advantages by going RAID10 (high spindle count _and_ redundancy).
-greetings, markus-
--
I'm sorry, but my karma just ran over your dogma.
I'm sorry, but my karma just ran over your dogma.