If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#1
|
|||
|
|||
Hard disk partitions and optimization
I've got two supposedly identical WD 40GB drives, but the new one comes up a
bit slower in the SiSoft hard disk benchmark tests. Could that be due to the fact that it is set up as the slave drive (the other one is the Master)? Does it make sense to put all heavy video work on the Master drive to improve the performance time? I mean, some of these video vob files are 1 GB each. Or is the master slave difference pretty much irrelevant for disk file reads and writes? (Any mpg video editing and transcoding takes quite a bit of time with an 800 MHz machine!) I also remembered something else. I think I told ya it took about an hour and a half to COPY a nearly full 20 GB partition. But I had forgotten to enable DMA on that drive, which I have now done. Would that have made any difference (at the level that BING runs at)? |
#2
|
|||
|
|||
Just wanted to jump in for a second and mention other options, is that you can
use a PCI card so that you can run both your HD's as Master's, and as well have them on their own separate feed, which also makes room for you to run your CDRW and CD/DVD on their own separate feeds as well each as master too. Besides the advantages of that, using the Controller PCI Card has the known inherent increase of performance of your whole system running through this setup can give you up to a 25% boost in the real-time processing of data, I've seen sometimes even 35% ..this increase is similar to what the application accelerator software can do - but in this case is so much better as hardware - it's a no contest; application accelerator software applications are the pits anyway. This increase is realized especially with higher than 1GHz and even moreso with 2 GHz + machines is the best increase, but even for yours @ 800 MHz would see a respectable gain too. ...just talking outloud about it... I would not run a W98 system without one, and for that matter W2K/WXP You can use whichever brand you like, but the best is a Promise Ultra133 TX2 Controller Card and here's just one place you can get them http://www.newegg.com/app/ViewProduc...102-007&depa=0 Rick Bill in Co. wrote: I've got two supposedly identical WD 40GB drives, but the new one comes up a bit slower in the SiSoft hard disk benchmark tests. Could that be due to the fact that it is set up as the slave drive (the other one is the Master)? Does it make sense to put all heavy video work on the Master drive to improve the performance time? I mean, some of these video vob files are 1 GB each. Or is the master slave difference pretty much irrelevant for disk file reads and writes? (Any mpg video editing and transcoding takes quite a bit of time with an 800 MHz machine!) I also remembered something else. I think I told ya it took about an hour and a half to COPY a nearly full 20 GB partition. But I had forgotten to enable DMA on that drive, which I have now done. Would that have made any difference (at the level that BING runs at)? |
#3
|
|||
|
|||
You haven't mentioned the SiSoft figures, so it's hard to tell whether it's
worth keeping everything on the faster drive, but I would guess that fragmentation will have a bigger effect than any difference in drive performance, and that will be much easier to control on the slave drive. Also, keeping your files spread across two drives (eg, application, temp and swap on one drive, video files on the other) will allow the system to make maximum use of the drive cache. Enabling DMA would make a very large difference to drive performance in WINDOWS. AFAIK BING would use DMA if it's available, but you would need to confirm that with the BING documentation. -- Jeff Richards MS MVP (Windows - Shell/User) "Bill in Co." wrote in message ... I've got two supposedly identical WD 40GB drives, but the new one comes up a bit slower in the SiSoft hard disk benchmark tests. Could that be due to the fact that it is set up as the slave drive (the other one is the Master)? Does it make sense to put all heavy video work on the Master drive to improve the performance time? I mean, some of these video vob files are 1 GB each. Or is the master slave difference pretty much irrelevant for disk file reads and writes? (Any mpg video editing and transcoding takes quite a bit of time with an 800 MHz machine!) I also remembered something else. I think I told ya it took about an hour and a half to COPY a nearly full 20 GB partition. But I had forgotten to enable DMA on that drive, which I have now done. Would that have made any difference (at the level that BING runs at)? |
#4
|
|||
|
|||
Jeff Richards wrote:
You haven't mentioned the SiSoft figures, so it's hard to tell whether it's worth keeping everything on the faster drive, but I would guess that fragmentation will have a bigger effect than any difference in drive performance, and that will be much easier to control on the slave drive. Both drives (which are identical) were defragmented (and I do this often) If you can believe the Sisoft figures, here they are (even the help file says they may not be accurate): Old drive (master): Drive Index = 37000, Sequential read& write = 56 Mbps New drive (slave): Drive Index = 23000, Sequential read& write = 33 Mbps Buffered read and write rates are about the same for both drives (70-80 Mbps) I'm still thinking that the master/slave thing enters into this, but I'm not sure. I get similar SMART values for both drives when I run the Western Digital "SMART" hard disk information utilities. Also, keeping your files spread across two drives (eg, application, temp and swap on one drive, video files on the other) will allow the system to make maximum use of the drive cache. Enabling DMA would make a very large difference to drive performance in WINDOWS. Yeah, I knew that much from the video stuff I had been doing. It would have helped, however, if I remembered to check that for the new drive! AFAIK BING would use DMA if it's available, but you would need to confirm that with the BING documentation. Next time I'm in BING at least I'll have that option. Frankly I'm surprised that when I installed the second drive that the DMA was not automatically selected, so I had to go and do that. I had completely forgotten to do that, until I saw it flagged in Sisoft (THANK GOODNESS!!!) -- Jeff Richards MS MVP (Windows - Shell/User) "Bill in Co." wrote in message ... I've got two supposedly identical WD 40GB drives, but the new one comes up a bit slower in the SiSoft hard disk benchmark tests. Could that be due to the fact that it is set up as the slave drive (the other one is the Master)? Does it make sense to put all heavy video work on the Master drive to improve the performance time? I mean, some of these video vob files are 1 GB each. Or is the master slave difference pretty much irrelevant for disk file reads and writes? (Any mpg video editing and transcoding takes quite a bit of time with an 800 MHz machine!) I also remembered something else. I think I told ya it took about an hour and a half to COPY a nearly full 20 GB partition. But I had forgotten to enable DMA on that drive, which I have now done. Would that have made any difference (at the level that BING runs at)? |
#5
|
|||
|
|||
Now that is an interesting idea. I think if I end up using this machine
pretty heavily, I should probably consider it. Rick Chauvin wrote: Just wanted to jump in for a second and mention other options, is that you can use a PCI card so that you can run both your HD's as Master's, and as well have them on their own separate feed, which also makes room for you to run your CDRW and CD/DVD on their own separate feeds as well each as master too. Besides the advantages of that, using the Controller PCI Card has the known inherent increase of performance of your whole system running through this setup can give you up to a 25% boost in the real-time processing of data, I've seen sometimes even 35% ..this increase is similar to what the application accelerator software can do - but in this case is so much better as hardware - it's a no contest; application accelerator software applications are the pits anyway. This increase is realized especially with higher than 1GHz and even moreso with 2 GHz + machines is the best increase, but even for yours @ 800 MHz would see a respectable gain too. ...just talking outloud about it... I would not run a W98 system without one, and for that matter W2K/WXP You can use whichever brand you like, but the best is a Promise Ultra133 TX2 Controller Card and here's just one place you can get them http://www.newegg.com/app/ViewProduc...102-007&depa=0 Rick Bill in Co. wrote: I've got two supposedly identical WD 40GB drives, but the new one comes up a bit slower in the SiSoft hard disk benchmark tests. Could that be due to the fact that it is set up as the slave drive (the other one is the Master)? Does it make sense to put all heavy video work on the Master drive to improve the performance time? I mean, some of these video vob files are 1 GB each. Or is the master slave difference pretty much irrelevant for disk file reads and writes? (Any mpg video editing and transcoding takes quite a bit of time with an 800 MHz machine!) I also remembered something else. I think I told ya it took about an hour and a half to COPY a nearly full 20 GB partition. But I had forgotten to enable DMA on that drive, which I have now done. Would that have made any difference (at the level that BING runs at)? |
#6
|
|||
|
|||
An update to one part below. I just tried AIDA32's disk benchmarking
plugin (at least for the read operations), and got nearly the same results for both disks. Also, the new hard disk is nearly full (but just temporarily) with a bunch of junk in both partitions, unlike the old drive, so that might have been a factor, too. Who knows.. Bill in Co. wrote: Jeff Richards wrote: You haven't mentioned the SiSoft figures, so it's hard to tell whether it's worth keeping everything on the faster drive, but I would guess that fragmentation will have a bigger effect than any difference in drive performance, and that will be much easier to control on the slave drive. Both drives (which are identical) were defragmented (and I do this often) If you can believe the Sisoft figures, here they are (even the help file says they may not be accurate): Old drive (master): Drive Index = 37000, Sequential read& write = 56 Mbps New drive (slave): Drive Index = 23000, Sequential read& write = 33 Mbps Buffered read and write rates are about the same for both drives (70-80 Mbps) I'm still thinking that the master/slave thing enters into this, but I'm not sure. I get similar SMART values for both drives when I run the Western Digital "SMART" hard disk information utilities. Also, keeping your files spread across two drives (eg, application, temp and swap on one drive, video files on the other) will allow the system to make maximum use of the drive cache. Enabling DMA would make a very large difference to drive performance in WINDOWS. Yeah, I knew that much from the video stuff I had been doing. It would have helped, however, if I remembered to check that for the new drive! AFAIK BING would use DMA if it's available, but you would need to confirm that with the BING documentation. Next time I'm in BING at least I'll have that option. Frankly I'm surprised that when I installed the second drive that the DMA was not automatically selected, so I had to go and do that. I had completely forgotten to do that, until I saw it flagged in Sisoft (THANK GOODNESS!!!) -- Jeff Richards MS MVP (Windows - Shell/User) "Bill in Co." wrote in message ... I've got two supposedly identical WD 40GB drives, but the new one comes up a bit slower in the SiSoft hard disk benchmark tests. Could that be due to the fact that it is set up as the slave drive (the other one is the Master)? Does it make sense to put all heavy video work on the Master drive to improve the performance time? I mean, some of these video vob files are 1 GB each. Or is the master slave difference pretty much irrelevant for disk file reads and writes? (Any mpg video editing and transcoding takes quite a bit of time with an 800 MHz machine!) I also remembered something else. I think I told ya it took about an hour and a half to COPY a nearly full 20 GB partition. But I had forgotten to enable DMA on that drive, which I have now done. Would that have made any difference (at the level that BING runs at)? |
#7
|
|||
|
|||
It's not possible to keep the drive with the operating system properly
defragged for any length of time, as the system does so much writing of its own. But you can isolate the other drives from OS activity and they will stay defragged for much longer. I am not aware of any factor in the Master/Slave arrangement of IDE controllers that would create such a speed difference as SiSoft reports unless there were differences in the drives themselves or the way they were configured. Therefore, I am inclined to believe the Aida results and doubt the SiSoft results. Whether or not the disk is nearly full should not affect a benchmark (although it does significantly affect ordinary operations). -- Jeff Richards MS MVP (Windows - Shell/User) "Bill in Co." wrote in message ... An update to one part below. I just tried AIDA32's disk benchmarking plugin (at least for the read operations), and got nearly the same results for both disks. Also, the new hard disk is nearly full (but just temporarily) with a bunch of junk in both partitions, unlike the old drive, so that might have been a factor, too. Who knows.. Bill in Co. wrote: Jeff Richards wrote: You haven't mentioned the SiSoft figures, so it's hard to tell whether it's worth keeping everything on the faster drive, but I would guess that fragmentation will have a bigger effect than any difference in drive performance, and that will be much easier to control on the slave drive. Both drives (which are identical) were defragmented (and I do this often) If you can believe the Sisoft figures, here they are (even the help file says they may not be accurate): Old drive (master): Drive Index = 37000, Sequential read& write = 56 Mbps New drive (slave): Drive Index = 23000, Sequential read& write = 33 Mbps Buffered read and write rates are about the same for both drives (70-80 Mbps) I'm still thinking that the master/slave thing enters into this, but I'm not sure. I get similar SMART values for both drives when I run the Western Digital "SMART" hard disk information utilities. Also, keeping your files spread across two drives (eg, application, temp and swap on one drive, video files on the other) will allow the system to make maximum use of the drive cache. Enabling DMA would make a very large difference to drive performance in WINDOWS. Yeah, I knew that much from the video stuff I had been doing. It would have helped, however, if I remembered to check that for the new drive! AFAIK BING would use DMA if it's available, but you would need to confirm that with the BING documentation. Next time I'm in BING at least I'll have that option. Frankly I'm surprised that when I installed the second drive that the DMA was not automatically selected, so I had to go and do that. I had completely forgotten to do that, until I saw it flagged in Sisoft (THANK GOODNESS!!!) -- Jeff Richards MS MVP (Windows - Shell/User) "Bill in Co." wrote in message ... I've got two supposedly identical WD 40GB drives, but the new one comes up a bit slower in the SiSoft hard disk benchmark tests. Could that be due to the fact that it is set up as the slave drive (the other one is the Master)? Does it make sense to put all heavy video work on the Master drive to improve the performance time? I mean, some of these video vob files are 1 GB each. Or is the master slave difference pretty much irrelevant for disk file reads and writes? (Any mpg video editing and transcoding takes quite a bit of time with an 800 MHz machine!) I also remembered something else. I think I told ya it took about an hour and a half to COPY a nearly full 20 GB partition. But I had forgotten to enable DMA on that drive, which I have now done. Would that have made any difference (at the level that BING runs at)? |
#8
|
|||
|
|||
Interesting. I just cleaned up the drives (removed half of the stuff) and
reran Sisoft, and this time the results were similar for both drives. So who knows. At any rate, I feel better that the results are about the same now for both drives. Maybe when I get up in the morning it will all be different if I were to check it again with Sisoft - (I wouldn't be totally surprised :-) Jeff Richards wrote: It's not possible to keep the drive with the operating system properly defragged for any length of time, as the system does so much writing of its own. But you can isolate the other drives from OS activity and they will stay defragged for much longer. I am not aware of any factor in the Master/Slave arrangement of IDE controllers that would create such a speed difference as SiSoft reports unless there were differences in the drives themselves or the way they were configured. Therefore, I am inclined to believe the Aida results and doubt the SiSoft results. Whether or not the disk is nearly full should not affect a benchmark (although it does significantly affect ordinary operations). -- Jeff Richards MS MVP (Windows - Shell/User) "Bill in Co." wrote in message ... An update to one part below. I just tried AIDA32's disk benchmarking plugin (at least for the read operations), and got nearly the same results for both disks. Also, the new hard disk is nearly full (but just temporarily) with a bunch of junk in both partitions, unlike the old drive, so that might have been a factor, too. Who knows.. Bill in Co. wrote: Jeff Richards wrote: You haven't mentioned the SiSoft figures, so it's hard to tell whether it's worth keeping everything on the faster drive, but I would guess that fragmentation will have a bigger effect than any difference in drive performance, and that will be much easier to control on the slave drive. Both drives (which are identical) were defragmented (and I do this often) If you can believe the Sisoft figures, here they are (even the help file says they may not be accurate): Old drive (master): Drive Index = 37000, Sequential read& write = 56 Mbps New drive (slave): Drive Index = 23000, Sequential read& write = 33 Mbps Buffered read and write rates are about the same for both drives (70-80 Mbps) I'm still thinking that the master/slave thing enters into this, but I'm not sure. I get similar SMART values for both drives when I run the Western Digital "SMART" hard disk information utilities. Also, keeping your files spread across two drives (eg, application, temp and swap on one drive, video files on the other) will allow the system to make maximum use of the drive cache. Enabling DMA would make a very large difference to drive performance in WINDOWS. Yeah, I knew that much from the video stuff I had been doing. It would have helped, however, if I remembered to check that for the new drive! AFAIK BING would use DMA if it's available, but you would need to confirm that with the BING documentation. Next time I'm in BING at least I'll have that option. Frankly I'm surprised that when I installed the second drive that the DMA was not automatically selected, so I had to go and do that. I had completely forgotten to do that, until I saw it flagged in Sisoft (THANK GOODNESS!!!) -- Jeff Richards MS MVP (Windows - Shell/User) "Bill in Co." wrote in message ... I've got two supposedly identical WD 40GB drives, but the new one comes up a bit slower in the SiSoft hard disk benchmark tests. Could that be due to the fact that it is set up as the slave drive (the other one is the Master)? Does it make sense to put all heavy video work on the Master drive to improve the performance time? I mean, some of these video vob files are 1 GB each. Or is the master slave difference pretty much irrelevant for disk file reads and writes? (Any mpg video editing and transcoding takes quite a bit of time with an 800 MHz machine!) I also remembered something else. I think I told ya it took about an hour and a half to COPY a nearly full 20 GB partition. But I had forgotten to enable DMA on that drive, which I have now done. Would that have made any difference (at the level that BING runs at)? |
#9
|
|||
|
|||
"Bill in Co." wrote in message
... I've got two supposedly identical WD 40GB drives, but the new one comes up a bit slower in the SiSoft hard disk benchmark tests. Could that be due to the fact that it is set up as the slave drive (the other one is the Master)? Does it make sense to put all heavy video work on the Master drive to improve the performance time? I mean, some of these video vob files are 1 GB each. Or is the master slave difference pretty much irrelevant for disk file reads and writes? (Any mpg video editing and transcoding takes quite a bit of time with an 800 MHz machine!) I also remembered something else. I think I told ya it took about an hour and a half to COPY a nearly full 20 GB partition. But I had forgotten to enable DMA on that drive, which I have now done. Would that have made any difference (at the level that BING runs at)? Its impossible for the PC to read/write on one hard drive and read/write to another hard drive on the same ide ribbon cable at the SAME time. Some performance software may monitor performance, and write monitoring results and assessments while doing the reads/writes on another hard drive. See above and make your conclusions. DMA usage tasks the DMA controller to move the data and does not call the control processor unit to utilize physical memory during this task. If the hard disk controller is operating correctly, data flow will move in 32 bit format vs PIO mode 16 bit format. DMA use allows the control processor unit to concentrate on video data transcoding vice doing both that and moving the data to a physical location (HD). Using an alternate physical ide HD controller (add-on) would allow simultaneous reads/writes when doing video transcoding and editing, if the other physical hard drive is connected to this alternate controller. This also makes for faster imaging files of other hard drives and their partitions. An add-on ide controller uses a technique similar to SCSI for moving data in windows mode. Has little impact, similar to DMA in this respect, on the control processor unit. It will revert back to a slower mode when used in dos real mode as data cannot be moved simultaneously (reads/writes) in dos real mode. DMA does not work in dos real mode either. And, the data will move in 16 bit PIO mode in dos real mode. |
#10
|
|||
|
|||
Bill in Co. wrote:
Jeff Richards wrote: You haven't mentioned the SiSoft figures, so it's hard to tell whether it's worth keeping everything on the faster drive, but I would guess that fragmentation will have a bigger effect than any difference in drive performance, and that will be much easier to control on the slave drive. Both drives (which are identical) were defragmented (and I do this often) If you can believe the Sisoft figures, here they are (even the help file says they may not be accurate): Old drive (master): Drive Index = 37000, Sequential read& write = 56 Mbps New drive (slave): Drive Index = 23000, Sequential read& write = 33 Mbps Buffered read and write rates are about the same for both drives (70-80 Mbps) I'm still thinking that the master/slave thing enters into this, but I'm not sure. I get similar SMART values for both drives when I run the Western Digital "SMART" hard disk information utilities. So swap them around and be sure. -- dadiOH ____________________________ dadiOH's dandies v3.05... ....a help file of info about MP3s, recording from LP/cassette and tips & tricks on this and that. Get it at http://mysite.verizon.net/xico |
Thread Tools | |
Display Modes | |
|
|