P
PCR
Guest
Re: A hack, anyone, to turn on dma ? SOLVED ???
Re: A hack, anyone, to turn on dma ? SOLVED ???
Franc Zabkar wrote:
| On Sun, 6 Jul 2008 18:20:51 -0400, "PCR" <pcrrcp@netzero.net> put
| finger to keyboard and composed:
|
|>Franc Zabkar wrote:
|>| On Sun, 06 Jul 2008 07:26:37 +1000, Franc Zabkar
|>| <fzabkar@iinternode.on.net> put finger to keyboard and composed:
|>|
|>|>On Sat, 05 Jul 2008 10:16:44 -0300, Shadow <sh@dow> put finger to
|>|>keyboard and composed:
|>|
|>|>> Does it write any faster ?
|>|>
|>|>The maximum write speed (2x) is limited by design, so whether the
|>|>drive is in PIO mode or DMA mode would make no difference. One would
|>|>expect an improvement in read performance, though.
|>
|>Remember that article...?...
|>http://support.microsoft.com/?kbid=159560
|>DMA Check Box Does Not Remain Checked
|>.......Quote......................
|>MORE INFORMATION
|>
|>DMA (also referred to as bus mastering) reduces CPU overhead by
|>providing a mechanism for data transfers that do not require
|>monitoring by the CPU. The transfer rate for a particular data
|>transfer event will not noticeably increase. However, overall CPU
|>overhead should be reduced using DMA mode.
|>.......EOQ.........................
|
| Thanks, that makes sense, although PIO mode 3 is spec'ed for 11.1 MB/s
| and MW DMA 2 for 16.6 MB/s.
|
|
http://en.wikipedia.org/wiki/Atapi#ATA_standards_versions.2C_transfer_rates.2C_and_features
Alright. That article does state there is some variance in what speed
may actually be attained for any of the modes, due to clock cycle
considerations & bus congestion. But those were probably much the same
for your PIO 3 vrs. MW DMA 2 test. As far as whether the CPU was busy
(which the MS article implies gives DMA the edge because the IDE device
will do the transfer)... you do say you had no apps running. (Naturally,
the OS itself is always running, but CPU usage from that alone can often
show up as zero in System Monitor.)
Nevertheless -- because these various DMA modes have different speed
ratings (although that could be because devices got quicker themselves
by the time the UDMA modes came out)-- I'm beginning to wonder whether a
busy CPU is all that should make DMA quicker. I'm tempted again to turn
mine off & see what happens-- but I see PIO 4 & MW DMA 2 have the same
16.6 MB/s rating. So, that won't help!
Anyhow, let me start Process Explorer. That has made CPU Usage jump to
the 70's! With DMA still on, let me copy that 300 MB's again to see
whether it still takes only about 8 minutes....
YIKES, it took 19 minutes that time!!! Why???
(Well, I normally won't have a CPU that busy during CD-ROM usage,
anyhow.)
|>... I copied about 300 MB from
|>my "IDE-CD R/RW 4x4x24" to HDD in about 8 minutes-- with DMA turned
|>on. All the while I was online with OE open & running.
|
| I had no apps running during the test.
|
|>I can't look up any
|>specs-- I don't know precisely what that thing is! I recall seeing a
|>Philips in the box-- but can't recall where I may have written its
|>model number. And I hesitate to turn that DMA off again for another
|>test! Here is about all my Compaq Diagnostics says about it...
|>
|>IDE-CD R/RW 4x4x24 CD-RW
|> Firmware Revision . . . . C12a
|
| This discussion ...
|
| http://discussions.virtualdr.com/archive/index.php/t-101564.html
|
| ... suggests that you have a "CompaQ CDD4401/71 CD-RW, which is
| basically a Philips PCRW404K".
|
| I found these Philips support URLs:
| http://www.p4c.philips.com/files/p/pcrw404k_00/
| http://www.p4c.philips.com/files/p/
|
| ... and this discussion ...
|
| http://www.broadbandreports.com/forum/remark,16350636
|
| ... which suggests that the following C1.8 firmware upgrade is for
| your CDD4401 drive:
| ftp://ftp.compaq.com/pub/softpaq/sp16501-17000/sp16704.exe
| ftp://ftp.compaq.com/pub/softpaq/sp16501-17000/sp16704.txt
|
| This is the Identify Packet Device data I extracted from the C18
| firmware:
| http://www.users.on.net/~fzabkar/DVD/C18.txt
|
| It indicates that your drive supports MW DMA mode 2 and PIO mode 4,
| but it does not support UDMA. It powers up in MW DMA mode 2. The "DMA
| supported" bit is set.
Thanks for all of that. I took the download. That SP16704 wasn't among
the many I once downloaded for this Compaq 7470. It wasn't listed for
this computer, but I see now not all of them came with a Philips inside.
I wish my Compaq Diagnostics would precisely spell out which Philips
I've got-- in case more than one is a possibility. I only know I've got
a Philips at all because I vaguely recall seeing it in the box.
|>| I tried this again. I used the Debug method (in real DOS) to
|>| configure the drive for multi-word DMA mode 2 and then typed "win".
|>| This time the machine didn't hang but booted into Windows (the DMA
|>| box was ticked beforehand). The same test now took 13m 16s. <shrug>
|>|
|>| The above transfer rate works out as 0.72 MB/s. The transfer rate
|>| for my DVD 10x drive is 2.0 MB/s. The latter is configured for
|>| UDMA-2.
|>
|>Mine comes to about 0.625 MB/s. But it seemed quick enough for me.
|>Part of that was "preparing to copy". Also, the device seemed to
|>power down & up again at least once in the midst of the copy.
|
| My FSB is set to 75MHz, so that means my PCI bus is running at
| 37.5MHz. Yours is probably set to 33MHz. Therefore I would expect a
| 13% speed advantage for the same transfer mode.
I'm not sure of my figure. In the BIOS, IDE Devices requestor, I see
that "Ultra 33/66" is enabled. My processor speed is written: 533/97
MHz. I did kept avast! running & it scanned all the executables during
the copy process.
Neither of us got 16.6 MB/s for MW DMA 2-- but I'm not really sure when
they start/stop counting! Are they speaking of the device already has
everything set & is about to move its heads? And as soon as the heads
stop moving they are done counting? We are going through "preparing to
copy", watching the CD-ROM spin down & up again midst the copy, & even
waiting for Explorer to update its display after the copy is done.
Nevertheless, it all seemed quick enough for me. Possibly the highest
figures will only apply to hard drives, I'm thinking.
|>| After all my experimentation, I finally found this Ricoh FAQ:
|>| http://www.ricoh.com/drive/asia/support/faq/allmodels/dma_d.html#04
|>|
|>| DMA transfer mode supported by each model:
|>|
|>| * MP6200A 11.11 MB/sec. (Max.) PIO mode 3
|>| MP7040A 16.7 MB/sec. (Max.) Multiword DMA mode 2, PIO mode 4
|>| ....
|>| MP5240A / MP5125A / MP5120A 33 MB/sec. (Max.) Ultra DMA mode 2
|>|
|>| Now why doesn't the user manual have this information, and why does
|>| the drive allow M-W DMA modes 0,1, and 2 to be set via the Set
|>| Features command if it doesn't support them?
|>
|>Can it be they had early plans to support them but later aborted
|>without undoing the prep work?
|
| Maybe. Or maybe the programmers used the same basic code for several
| models and just disabled the modules they didn't need.
Now I'm thinking the drive software is just not looking at the time the
Set Features command puts info into the PCI Register. It could put
anything in there, probably. When Windows & the drive are ready to use
it to determine what mode to use, its up to them what to do if/when an
invalid value is found. Apparently, neither will correct the value, but
possibly they will ignore it & use a default that is valid-- who knows?
|>| And why does the DMA
|>| checkbox remain ticked in Device Manager if the drive is in PIO
|>| mode?
|>
|>Maybe there is something non-standard about one of the PIO modes that
|>makes it look a lot like one of the DMA modes.
|
| - Franc Zabkar
| --
| Please remove one 'i' from my address when replying by email.
--
Thanks or Good Luck,
There may be humor in this post, and,
Naturally, you will not sue,
Should things get worse after this,
PCR
pcrrcp@netzero.net
Re: A hack, anyone, to turn on dma ? SOLVED ???
Franc Zabkar wrote:
| On Sun, 6 Jul 2008 18:20:51 -0400, "PCR" <pcrrcp@netzero.net> put
| finger to keyboard and composed:
|
|>Franc Zabkar wrote:
|>| On Sun, 06 Jul 2008 07:26:37 +1000, Franc Zabkar
|>| <fzabkar@iinternode.on.net> put finger to keyboard and composed:
|>|
|>|>On Sat, 05 Jul 2008 10:16:44 -0300, Shadow <sh@dow> put finger to
|>|>keyboard and composed:
|>|
|>|>> Does it write any faster ?
|>|>
|>|>The maximum write speed (2x) is limited by design, so whether the
|>|>drive is in PIO mode or DMA mode would make no difference. One would
|>|>expect an improvement in read performance, though.
|>
|>Remember that article...?...
|>http://support.microsoft.com/?kbid=159560
|>DMA Check Box Does Not Remain Checked
|>.......Quote......................
|>MORE INFORMATION
|>
|>DMA (also referred to as bus mastering) reduces CPU overhead by
|>providing a mechanism for data transfers that do not require
|>monitoring by the CPU. The transfer rate for a particular data
|>transfer event will not noticeably increase. However, overall CPU
|>overhead should be reduced using DMA mode.
|>.......EOQ.........................
|
| Thanks, that makes sense, although PIO mode 3 is spec'ed for 11.1 MB/s
| and MW DMA 2 for 16.6 MB/s.
|
|
http://en.wikipedia.org/wiki/Atapi#ATA_standards_versions.2C_transfer_rates.2C_and_features
Alright. That article does state there is some variance in what speed
may actually be attained for any of the modes, due to clock cycle
considerations & bus congestion. But those were probably much the same
for your PIO 3 vrs. MW DMA 2 test. As far as whether the CPU was busy
(which the MS article implies gives DMA the edge because the IDE device
will do the transfer)... you do say you had no apps running. (Naturally,
the OS itself is always running, but CPU usage from that alone can often
show up as zero in System Monitor.)
Nevertheless -- because these various DMA modes have different speed
ratings (although that could be because devices got quicker themselves
by the time the UDMA modes came out)-- I'm beginning to wonder whether a
busy CPU is all that should make DMA quicker. I'm tempted again to turn
mine off & see what happens-- but I see PIO 4 & MW DMA 2 have the same
16.6 MB/s rating. So, that won't help!
Anyhow, let me start Process Explorer. That has made CPU Usage jump to
the 70's! With DMA still on, let me copy that 300 MB's again to see
whether it still takes only about 8 minutes....
YIKES, it took 19 minutes that time!!! Why???
(Well, I normally won't have a CPU that busy during CD-ROM usage,
anyhow.)
|>... I copied about 300 MB from
|>my "IDE-CD R/RW 4x4x24" to HDD in about 8 minutes-- with DMA turned
|>on. All the while I was online with OE open & running.
|
| I had no apps running during the test.
|
|>I can't look up any
|>specs-- I don't know precisely what that thing is! I recall seeing a
|>Philips in the box-- but can't recall where I may have written its
|>model number. And I hesitate to turn that DMA off again for another
|>test! Here is about all my Compaq Diagnostics says about it...
|>
|>IDE-CD R/RW 4x4x24 CD-RW
|> Firmware Revision . . . . C12a
|
| This discussion ...
|
| http://discussions.virtualdr.com/archive/index.php/t-101564.html
|
| ... suggests that you have a "CompaQ CDD4401/71 CD-RW, which is
| basically a Philips PCRW404K".
|
| I found these Philips support URLs:
| http://www.p4c.philips.com/files/p/pcrw404k_00/
| http://www.p4c.philips.com/files/p/
|
| ... and this discussion ...
|
| http://www.broadbandreports.com/forum/remark,16350636
|
| ... which suggests that the following C1.8 firmware upgrade is for
| your CDD4401 drive:
| ftp://ftp.compaq.com/pub/softpaq/sp16501-17000/sp16704.exe
| ftp://ftp.compaq.com/pub/softpaq/sp16501-17000/sp16704.txt
|
| This is the Identify Packet Device data I extracted from the C18
| firmware:
| http://www.users.on.net/~fzabkar/DVD/C18.txt
|
| It indicates that your drive supports MW DMA mode 2 and PIO mode 4,
| but it does not support UDMA. It powers up in MW DMA mode 2. The "DMA
| supported" bit is set.
Thanks for all of that. I took the download. That SP16704 wasn't among
the many I once downloaded for this Compaq 7470. It wasn't listed for
this computer, but I see now not all of them came with a Philips inside.
I wish my Compaq Diagnostics would precisely spell out which Philips
I've got-- in case more than one is a possibility. I only know I've got
a Philips at all because I vaguely recall seeing it in the box.
|>| I tried this again. I used the Debug method (in real DOS) to
|>| configure the drive for multi-word DMA mode 2 and then typed "win".
|>| This time the machine didn't hang but booted into Windows (the DMA
|>| box was ticked beforehand). The same test now took 13m 16s. <shrug>
|>|
|>| The above transfer rate works out as 0.72 MB/s. The transfer rate
|>| for my DVD 10x drive is 2.0 MB/s. The latter is configured for
|>| UDMA-2.
|>
|>Mine comes to about 0.625 MB/s. But it seemed quick enough for me.
|>Part of that was "preparing to copy". Also, the device seemed to
|>power down & up again at least once in the midst of the copy.
|
| My FSB is set to 75MHz, so that means my PCI bus is running at
| 37.5MHz. Yours is probably set to 33MHz. Therefore I would expect a
| 13% speed advantage for the same transfer mode.
I'm not sure of my figure. In the BIOS, IDE Devices requestor, I see
that "Ultra 33/66" is enabled. My processor speed is written: 533/97
MHz. I did kept avast! running & it scanned all the executables during
the copy process.
Neither of us got 16.6 MB/s for MW DMA 2-- but I'm not really sure when
they start/stop counting! Are they speaking of the device already has
everything set & is about to move its heads? And as soon as the heads
stop moving they are done counting? We are going through "preparing to
copy", watching the CD-ROM spin down & up again midst the copy, & even
waiting for Explorer to update its display after the copy is done.
Nevertheless, it all seemed quick enough for me. Possibly the highest
figures will only apply to hard drives, I'm thinking.
|>| After all my experimentation, I finally found this Ricoh FAQ:
|>| http://www.ricoh.com/drive/asia/support/faq/allmodels/dma_d.html#04
|>|
|>| DMA transfer mode supported by each model:
|>|
|>| * MP6200A 11.11 MB/sec. (Max.) PIO mode 3
|>| MP7040A 16.7 MB/sec. (Max.) Multiword DMA mode 2, PIO mode 4
|>| ....
|>| MP5240A / MP5125A / MP5120A 33 MB/sec. (Max.) Ultra DMA mode 2
|>|
|>| Now why doesn't the user manual have this information, and why does
|>| the drive allow M-W DMA modes 0,1, and 2 to be set via the Set
|>| Features command if it doesn't support them?
|>
|>Can it be they had early plans to support them but later aborted
|>without undoing the prep work?
|
| Maybe. Or maybe the programmers used the same basic code for several
| models and just disabled the modules they didn't need.
Now I'm thinking the drive software is just not looking at the time the
Set Features command puts info into the PCI Register. It could put
anything in there, probably. When Windows & the drive are ready to use
it to determine what mode to use, its up to them what to do if/when an
invalid value is found. Apparently, neither will correct the value, but
possibly they will ignore it & use a default that is valid-- who knows?
|>| And why does the DMA
|>| checkbox remain ticked in Device Manager if the drive is in PIO
|>| mode?
|>
|>Maybe there is something non-standard about one of the PIO modes that
|>makes it look a lot like one of the DMA modes.
|
| - Franc Zabkar
| --
| Please remove one 'i' from my address when replying by email.
--
Thanks or Good Luck,
There may be humor in this post, and,
Naturally, you will not sue,
Should things get worse after this,
PCR
pcrrcp@netzero.net
.] Seriously, I know you