Comment on "New PC advice"

Comments

Softy wrote on 3/3/2004, 2:16 PM
Making it the master rather than the slave won't do the trick. Get a Promise controller card and hook it to that, assuming you have the available resources. Being on the same channel as the DVD-rw drive is most likely what's compromising its performance.
CDM wrote on 3/3/2004, 2:33 PM
wow - that's so interesting. I've never seen ANYTHING about that!
Rednroll wrote on 3/3/2004, 2:40 PM
I noticed on my motherboard when I put my DAW together, (this is when EIDE66 was the fastest), that with the 2 IDE channels channel 1 was EIDE66, while channel 2 was IDE33. So it seems that the 1st channel is intended for hard drives and the 2nd channel is intended for slower speed Cdrom/DVDrom devices. The thing to make a note of though, is when you put a hard drive and a Cdrom/dvdrom device on the same channel, that channel will goto the slower speed. So in your instance Charles, the DVD-rw is most likely an IDE33 device, where your harddrive is probably a faster EIDE device, but by having them on the same channel the harddrive has to use the IDE33 transfer speed. With my system, I have my EIDE66 harddrive on channel 1, as a master. On the second channel is my DVD-rw device. Then in addition to this I have a PCI SCSI card with a 10K RPM scsi hard drive for recording all my audio too. Even at this time with this configuration using a 700Mhz Athlon processor, I am getting great performance and able to run 32 track projects with plugins 8in/20out I/O with no problem. I also have 1.1 Gig of RAM. I pushed this system to the limit to see how many simultaneous tracks I could record and playback without any skips and static. I was able to record 72 simultaneous tracks and then playback was 88 simultaneous tracks. After this, the system would crash, but it pretty much told me where my limits are. The other thing to make sure of is the DMA is enabled on all the hard drives on the IDE buses.
CDM wrote on 3/3/2004, 3:10 PM
thanks Red -
here's the other weird thing... I can't find settings for pio or DMA anywhere in my current setup. If I go to the properties of each IDE channel in device manager, there is no advance settings tab, which there used to be (which would tell me how each device was running). I know for a fact that I was able (before) to set a CD-RW drive on the SAME channel as a Hard Drive to PIO, while the drives was at UDMA 5.

What baffles me is I just did a test with Sandra that showed that my E: drive (the one sharing the secondary channel with the DVD drive) scored a lot higher than my d: drive, which share the primary bus with my system drive.

This is all very puzzling. I really don't feel I should have to by a separate controller card for my drives... I mean, that's what all the options are for on the MB.... in theory...
CDM wrote on 3/3/2004, 3:16 PM
and what about S.M.A.R.T? should I be enabling this???
Softy wrote on 3/3/2004, 3:33 PM
No (to enabling SMART).

And as for the extra controller card, it's simply a matter of how much performance you want. There's nothing your motherboard's ATA controller channel can do about the fact that ATA devices will negotiate the lowest common denominator speedwise, when connected on the same channel.
Rednroll wrote on 3/3/2004, 3:39 PM
"there is no advance settings tab"

That is strange. You're running WinXP right? I usually change the Widows Themes in XP to "Windows Classic", just because I know where to find everything after learning on Win95 and Win98. Maybe it's being hidden, due to the theme you're using. It sounds like you're doing everything right. Just go into the hardware devices control panel, click on the IDE/ATA ATAPI Controlers icon and expand it. Then right click on the Primary IDE channel, select properties and you should see the advanced settings tab, do the same for the Secondary IDE channel.
CDM wrote on 3/3/2004, 3:47 PM
yeah, that's what I'm doing and there's no advanced settings tab.

weird.

I'm in Classic mode too.
drbam wrote on 3/3/2004, 5:45 PM
This master/slave issue is something that has conflicting opinions by a lot of pretty savy folks. I've been inquiring about this very thing on several forums and with private individuals for a couple of years. Some will say that the OS drive should be master 1 w/CD rom as slave and the data streaming drive master 2. Others (as in this thread) say the OS should be master 1 with the data drive as slave and CD rom master 2 or use a IDE pci card. Since these conflicting opinions come from what I consider VERY knowledgable people, getting a definitive answer seems impossible. Personally I've tried both configurations and can't really see a performance difference. However, I seldom have projects that run over 20+ tracks or so at mixdown so perhaps I'm never pushing the system hard enough to determine this. Since Charlesdem is having some problems that seem related, perhaps he'd be inclined to perform an A/B test and report back??

Confused,

drbam
tmrpro wrote on 3/3/2004, 8:51 PM
One way to see the differences if your using XP is to check your CPU usage window and create a benchmark from the same .veg file with the different configurations.

The method I've shown will give the best results concerning multitracking's biggest drawback -- CPU usage.

This will hold true with any of the multitrack programs that we talk about in this forum and will be consistant from one system to another. Maybe not consistant from the standpoint of the specific amount of CPU usage, but you will see the same approximate differentials between configs.

IOW, if you try the same benchmarks on a P3, a P4, a celeron, a duron or an opteron..... you will see the a similar and consistant processor differential between IDE configurations with the options that I pointed out as being the methods for least CPU usage.
Softy wrote on 3/4/2004, 4:20 AM
There are differences between how these things work, depending on the MB in question. There are some MBs (mostly older ones) with say ATA66 primary and ATA33 secondary controller channels. I haven't seen any newer MBs with say ATA133 primary and ATA66 secondary, but then I haven't really done a survey on that. All of the MBs I currently work with have identical primary and secondary channels (ATA133, with backward compatibility to ATA100, ATA66 and ATA33) and some have in addition, SATA, RAID, etc. A quick perusal of the manual for your MB will tell you what the spec's of your disk controller channels are. Keep in mind though, that if you put an ATA133 hard drive on the same channel as an ATA33 CD-ROM drive, both devices will run at ATA33 speed, since the lowest common denominator will be negotiated.

By convention, people generally put their boot drives on the primary controller channel, configured as master. In ancient times, this was because certain OSs (if you could call things like MS-DOS and CP/M OSs) would be configured to boot from that device, and were hardwired to do so. Nowadays, we have much more elaborate BIOS code, to allow us to configure in CMOS settings, things like boot order, etc.

Generally speaking, the "master/slave" nomenclature is misleading. It's really only a device select address, not a priority mechanism. Even if you somehow had a controller that gave a higher interrupt priority to the master, the performance degradation caused by having a slower device on the same channel would be far more significant.

MarkWWW wrote on 3/4/2004, 2:10 PM
That's because you are using a motherboard with an Intel chipset and are using Intel's IDE drivers for it - these go by a variety of names like Intel Application Accelerator and similar. When using these drivers the DMA settings for the drives do not appear in the usual place where you have been looking.

Instead, you need to use the application that accompanies the Intel IDE drivers (or Intel Application Accelerator, or whatever), an item that is usually called Intel Companion or something similar. Fire up the Companion and you will see that it list what drives, etc, are currently available on the system, gives details of their capablities (PIO modes, DMA modes, etc) shows how they are currently set up (DMA modes, etc) and allows you to change these if they are not what you want - normally you want everything to be running at the highest UDMA mode it is capable of).

Mark
pwppch wrote on 3/5/2004, 8:47 AM
>>Kontakt and Giga both use a RAM based preload for streaming and functionality. It works below windows and they determine the best bussing methods outside anything else that is going on in your system
<<
How does it work "below windows"?
tmrpro wrote on 3/5/2004, 9:47 AM
You know Peter, it's pretty sad when I feel like I have to defend myself against an employee of a company that I represent.

Your answers:

A PC computer is divided in to two very different regions. "User Level" is the part of the computer that you, as an end user, interface with. The Windows operating system gets loaded at User Level, as do all of the applications (Word, DAW software, Internet stuff, etc) that sit on top of the OS. The main issue with trying to run audio applications at User Level is the "hands on" involvement, or "governing" that Windows performs on tasks that occur here. Windows controls the game, regardless of the hit on system performance.

Kernel Level is the part of the computer that Windows communicates with after processing the tasks at the User Level. Hardware drivers are found here and Windows will reach down to Kernel Level to complete a specific task. Kernel Level is a very important part of the process, but still very basic in the amount of code that exists here. Engineers often refer to writing code at this level as "coding close to the metal".

The audio driver for GigaStudio is written at Kernel Level. The result is an ultra-efficient audio engine that performs its tasks out of the reach of Windows. The end result? Incredibly low latency specs (because Windows doesn't process the audio created in GigaStudio) and increased polyphony because GigaStudio is running at a very low section of the PC platform. The only part of GigaStudio that exists in User Level is the User Interface (where you load instruments, push buttons and other basic functions). For all of the performance considerations, the UI communicates with the Kernel Level audio engine.