I have the one with the 845D chipset...did you go to mwave.com? that's where I got mine..so far I've built two of these systems for myself and one very similar for jetoney...
so far so good except fir jetoney's machine is acting up...keeps rebooting itself? ...we're troubleshooting it right now...we'll let you know
Opus

Opus - Thanks for the clearity :)

No. Didn't check them out. I'll do it now though. I did win the chip auction. $160 for the 1.6a Now if I can keep the board price to about $120 I should be golden.

Hope you don't mind the thread title. I figured this would be a good time to start a new thread to trade ideas specific to your machine build.

Each day I just get more jazzed about seeing this machine for myself.

Thanks again,

CustomProd

I like the thread title!! lol
You will see an amazing display of power once you get the machine up and running....I promise you that!!!
Opus

Hey, is this gonna be the thread ? Great!
I have mine up and running @ 2.1

So far everything was smooth, the hardware monitor showed Vcore at 1.47, so I bumped the voltage to the CPU one click up. The stock Intel CPU fan is pretty quiet @ only 2500 rpm.

My DAW rig so far:
ASUS P4B266 revision 2.01
P4 1.6a @ 2.1
2x256 Crucial DDR
Maxtor 40GB 7200
Lite-On 24x10x32 CDRW
Toshiba 16xDVD
Matrox G450
Realtek 10/100 network card
Seasound Solo 24/96
Powmax/Leadman 400watt P/S
Black 17" IBM Trinitron and KDS 19"
Win2000 Pro
no case, just the old steel frame of an Enlight laying sideways for now.

notes: I formatted the drive fat32 instead of NTFS because I need to move files across my home network with some computers that will be running Win98SE. Opus, do you think I should have used NTFS? I don't know the benefits for a DAW. Thanks.

EDIT: Intel Application Accelerator takes away all DMA choices from the usual place(device manager), but running a report from the IAA program shows all is good. I changed my DDR cas settings from 2.5 to 2. I don't plan on running my PCI bus out of spec, but I'll try turbo1 and turbo2 and see what Sandra reports.

Tommy P.

Allright!!! Nice job!
In my BIOS setting all I did was disable the Serial ports, change the LPT port to EPP and set the cpu speed to 2133Mhz..I didnt change any CAS settings or multiplier settings..it's working like a charm..
400Watt psu eh? that'll cover just about anything you put to it!!
You dont need to format to Fat32 if you're just doing some network transfering..networking will see NTFS drives even from a 98 source machine. Going FAT32 means that you can pull the drive out and put it into a 98 machine and see it..98 wont see an NTFS drive physically inside the machine.
NTFS is better since now you can shut the machine down with the power switch and not worry about the file data being corrupted..it's a better stream anyways with NTFS..
Opus

Thanks,I'll probably re-format to NTFS or get some removable drive bays to keep my pro rig separate from the home stuff.

My Sandra benchmarks were slightly less than what you had posted for your set-up. I did reach similar or better when I chose turbo2 mode for the RAM, did you use those settings also? ;)

EDIT: its a confirmed bug, the BIOS temp readings get stuck at 40C. The chip seems much cooler, barely warm to the touch. No fix from ASUS yet. I'm runnung BIOS ver 1003, I can't get into any ASUS BIOS websites
EDIT2: the CPU voltage reading is stuck too @ 4.75V.
Finally got into the ASUS BIOS download area. Looks like ver 1005 is available and ver 1005b004( a beta). Which one? :confused:

Indeed I do see the same thing about the temp in the BIOS setting..My system is working great..I aint touching no BIOS upgrade..that's a mishap just waiting to happen! lol
If it works dont try and fix it..ya know what I mean!!!
I'm not too worried about getting the latest BIOS updates..
If you're not getting the same results as me then you tweaked too much on your BIOS..as I said the ONLY things I did in the BIOS was to set the speed of the CPU from manual to 2133Mhz and also disable the Serial ports...pretty much it.
Good luck on the SDFC setup!! lol
Opus

P.S...my machine is doing 16 tracks of 24/96 with CPU topping out at 25%..if that
:D

Hey Opus...
Is there a particular reason you got the ASUS P4B266 (plain)?
I've got the ASUS P4B266-E with the RAID option. I have good a good experience with the Promise Fasttrak100 PCI-Raid, which I believe is integrated in the ASUS P4B266-E. Is there any other differences between the "E" and the "plain"?
I've also got a silent NoiseControl chassis with a Zalman 300w (silent) powersupply. I haven't put any of this together yet and I still have the option of changing parts.
Isn't 300w enough if you stuff the computer with drives and PCI-cards?

Thankful for your opinions...
Regards,
Fredrik Lidin

I went with the non Raid version because of boot up time and I like to keep onboard peripherals to a minimum if I can help it. Plus when I was building the P4S333 system I needed a mainboard quick and found that one at Fry's Electronics..I grabbed it on the whim but was fully aware there is a RAID version. I would rather install my own SCSI or ATA controller than be stuck with the onboard one.
300Watt will be sufficient enough to handle what you throw at it.
I dont know of any major differences other than the RAID between the plain and the E version..there's also an SE version of the P4B266 which I believe I have since it's the 845D chipset..if I'm correct the first run of these boards only had the 845 and not the D version..havent really found out what the differences are..probably just revision and erratta issues
Opus

Thank you for that input...
Now, isn't it possible to discard the onboard RAID-function if I decide to just use the onboard connectors as normal IDE-channels? And possibly create a RAID-array from a PCI-card instead. Maybe for better performance in the future?

Or are the connectors always functioning as RAID-arrays? Is there some drawback (other than boot-time) to use these RAID-connectors as "normal" IDE?

Appreciate your (or anybody's) thoughts on this.

Best to you,
Fredrik Lidin

I just moved the whole rig into a lightweight all- aluminum silver rack mount 4U. I've never encountered this kind of quality in a case before. I drove a hefty 60 miles round trip to get the last one. Check it out:
http:// :D

I'll be getting some removable drive bays when I get the cash

Fredrik
You can use the RAID as normal IDE connectors with no problems..not sure what the settings for it would be but I know you can do it...not sure if you can disable the onboard RAID from the BIOS or not..I think it's just there permanently...Never got a system with one built on so I wouldnt know!
The onboard Primary and Secondary controllers are not RAID configurable..using a PCI card for RAID at that point would be redundant..if you have the RAID built on, use it!

Tommy...that's the exact case I got from Mwave..it's a generic shell that people slap their compant name on!! Nice aint it?!!
Opus

Originally posted by Fredrik:
Now, isn't it possible to discard the onboard RAID-function if I decide to just use the onboard connectors as normal IDE-channels? And possibly create a RAID-array from a PCI-card instead. Maybe for better performance in the future?

The answer is yes. In most cases the setting is located under the Integrated Periphials section of the BIOS and its called Onboard ATA/RAID Device, you simply select Disable and it no longer funtions as a RAID or ATA controller.

However, this really isn't required as pluging in yet another controller just gives you more resources and keeps the ATA function of the extra 2 connectors intact. But the option is there if you so desire to use it.

I think you're right. I'm gonna keep the "E" version of the P4B266-board. Use the RAID function, and if desired, add another RAID controller to the system.

Here's another question: In the manual it says that there's an option of a setting that lets you control all the DIP-switches via software BIOS. Why shouldn't I use that option?
It definately seems more handy to control the CPU-clocking and such in the software. I thought this was a default preset on all modern mobos?

Why on earth would I want to open the chassis and press some DIP-switch or change a jumper-setting when configuring the computer?

Regards,
Fredrik Lidin

Fredrik - You are right about setting the board up to use bios settings instead of dipswitches. Also, J.Scott is at least half right about setting the raid functions from bios. (J, judging from the accuracy of your posts I'm probably flogging a non-existent horse, sorry if that's the case...)I haven't downloaded the manual for your MoBo, so this may not apply to yours at all, but the 4 different MoBo's with built-in raid controllers that I DO have manuals for ALL use both jumpers on the board AND bios settings to control all the functions of the raid. The way all the boards I've looked into do it is this: a jumper on the MoBo sets whether the raid controller works AT ALL - If you enable it with this jumper, THEN you have the option in BIOS to decide whether it is just two more IDE channels, or raid 0, 1, or raid 0+1. So far I haven't seen a Promise embedded raid chip that allows 0+1 (striping AND mirroring of drives) but all the HiPoint embedded raid chips I've seen DO allow all 3 modes.

Since I have not looked at the particular MoBo you're using, I could easily be 'way off base. I only mentioned this in case you can't get things to work by just using BIOS settings - there may be a jumper on the board. I would strongly recommend downloading the .pdf manual for any MoBo and studying it before you buy. That has saved me from ordering two different MoBo's so far. In fact, if Soyo ever gets off their ass and sends me the manual they promised me 2 weeks ago, hopefully my quest will be over... Steve

OK!
Thank you. Your conclusion regarding the jumper-enabling of RAID is correct.
What I want to do is to have one disk for OS and apps and two additional disks striped for performance. On the striped pair is where my audio-files and GigaStudio-samples will be.

That should be possible, shouldn't it?

Anyway, thanks for your input... :)

Regards,
Fredrik Lidin

If you have a moderate amount of audio and a good number of Gigastudio tracks playing back at the same time, you may actually get better performance by leaving each to it's own non-striped drive. That way, you can potentially double the amount of playback. You won't get that kind of performance increase with raid. Plus, striping by itself is not a very safe way of doing things- lose one of the two drives, lose everything.

Just my two cents!

Originally posted by Fredrik:
OK!
Thank you. Your conclusion regarding the jumper-enabling of RAID is correct.
What I want to do is to have one disk for OS and apps and two additional disks striped for performance. On the striped pair is where my audio-files and GigaStudio-samples will be.

That should be possible, shouldn't it?

Anyway, thanks for your input... :)

Regards,
Fredrik Lidin

Jon, looks like we're in total agreement on the raid thing - that's one of the reasons it's taking me so long to find a MoBo - I've had to narrow it down to the ones with the HiPoint raid chip, due to the 0+1 thing. I am not sure I want to use raid at all, but if I do I want all the options I can get, so I can make a fact-based decision on whether to go 0+1, just 1 with daily backups to a separate drive, or? The way I look at the math, if one drive has a MTBF of 10,000 hours, then 4 drives in raid 0 would have a MTBF of 2500 hours at best (sorry, at goodest; that was a bad pun even for me...) So, depending on benchmarks with real audio tracks, there's a good chance I will end up using the raid function as 4 separate drives. (Looking at a Soyo Fire Dragon MoBo, if the turkeys ever get me a manual) Hopefully, I can get hold of one this week sometime, and end up with a floor full of parts like you...

Sometimes I think I should use all the time it's taken on computer crap to get a part-time job and use the money to buy a used SSL and Ampex 1200 24 track. (Don't remember the audio version's model #, I used to work on the video 2" machines - same basic transport except for electronics and head stacks)

Hope your "PC Hell" is starting to gel... Steve

Originally posted by knightfly:
J, judging from the accuracy of your posts I'm probably flogging a non-existent horse, sorry if that's the case

Based upon the information I have, I believe you can accomplish all you need to by using the settings in the bios, no mobo jumper changes will be required.

Nonetheless, no offense taken, I'm as interested in being accurate as possible, and maybe there is something else I'm unaware of? I based my comments on the requirements of the Gigabyte 8IRXP, and a few others I've looked at.

Jon,
Why wouldn't a striped pair of drives increase read/write performance? Do you mean that two drives sharing the workload between Giga-samples will perform better than a striped pair?

I'm not gonna use the PC-drives for tracking. I will just use them for playing back samples and loops. I do all my audio-tracking in Soundscape...
So I don't really need the safety (mirrored) option on my machine since I'll be using samples that are backed up on our network server.

Regards,
Fredrik Lidin

Maybe a better understanding of the RAID available on these onboard controllers needs to be disscussed?

RAID appears at first glance to offer a substantial advantage for throughput. It does in most cases provided its configured right and you choose wisley. However, boards with on-board RAID offer several "not so appearent" features, not so much for the RAID ability, but due to the added channels they can provide which can eliminate or lessen IRQ conflicts. When concepts are combined they offer several other solutions to DAW problems.

For example, it can offer an advantage to those with slower 5400 rpm drives and allow them to out-perform even the fastest single IDE drives.

Drives require striping, which is the function of telling the drives what size the sectors need be. For server desktop applications, the value is usually 8Kb whereas for Audio its 64Kb up to 1024Kb. If you choose the wrong stripe sector size, performance will suffer.

RAID 0 can be comprised of up to 4 drives with these MB's. When using 2 drives, you can read and write data at about twice the speed of the slowest drive. With 4 drives, you can get about 3 times the speed of the slowest drive. So you could save some bucks buy adding 5400 rpm drives, but capacity is limited to the smallest of the drives times the number of drives.

With RAID 1, this is a true mirror, and is therefore redundant, however, the writes are equal to the slowest drive whereas the reads are twice as fast as the slowest drive. RAID 1 requires even pairs of drives and capacity is equal to the smallest of drives.

In both cases, the reads will out perform a simgle drive. BUT, If you put everything on one RAID group, the heads have to move between program calls and data reads thereby slowing down the function of your only drive group.

So I would think, if you really want to maximize the abilitiy of the RAID configuration, you are still better off with programs on one drive or RAID group and data on another. So two stripped arrays of RAID 0 will give maximum performance and storage for both reads and writes wheras two stripped arrays of RAID 1 will have lesser storage and offer a performance enhancement of the READS while still offering an increased ability to write over just one array, but not the speed of RAID 0. This is all assuming your controller will allow you to concurently read and write to 2 arrays?

If not, you can come close to this by using both the onboard IDE and RAID controllers leaving programs to the IDE channel and DATA to the RAID configuration. Then Jon's comment still has merit.

Again, I think its important to stress the difference between normal server business app needs vs. DAW where constant streaming is required rather than the random reads/writes that are very acceptable for business apps.

You just have to be creative about your needs to determine if data should be on RAID 1 for security, or RAID 0 for speed, or combine concepts.

When you stripe a pair of drives, you theoretically could double read/write performance, but as I understand it it's more like a 30-40% increase. I think that if you keep seperate libraries on each drive, and just let each drive spit out it's own samples, you're going to be able to take care of the maximum read/write of each drive without the overhead associated with RAID configuration. It may help you organizationally, as well- drive A is orchestral loops and samples, drive B is other types, whatever. With massive single chunks of audio, data, or video, then yeah, there are more advantages to two drives looking like a single volume, but for your application, you're going to have little bits of audio spread throughout a sample library anyway, so just make it two drives. Simpler is better!

Originally posted by Fredrik:
Jon,
Why wouldn't a striped pair of drives increase read/write performance? Do you mean that two drives sharing the workload between Giga-samples will perform better than a striped pair?
Regards,
Fredrik Lidin

I agree about RAID potentially helping out with *large* track counts of high res audio, but in this case the random read/write actions of a business type app is basically what's going to be happening anyway, with sample libraries. I think it makes more sense to let each hard drive head operate autonomously, and find/play the samples it's got, while ignoring the other ones.

Originally posted by jscott:

Jon - I'm not a "gigastudio" guy to comment either way. My thread was purley informational and intended in a theoretical sort of way to help try and educate on how the RAID thing works and possible solutions it can provide? I'm lost a bit on the needs of sampling programs as I don't use any. I would have assumed streaming would have been a bigger deal.

In any event, like I said, I think one just has to try and understand concepts and apply them to solve the needs of the work they do.

No, I think you're right about raid and streaming.

Of course, now that I think about it, wouldn't it be faster to just save half your audio tracks on one drive, and half on another? It'd probably create nightmares somewhere else in the process, not to mention some really interesting file organization, but it's an interesting thought.

:eek: WOW DID THIS THREAD GET TANGLED OR WHAT?