Messages in this thread | | | Date | Thu, 6 Aug 1998 12:53:34 -0400 (EDT) | From | "Mike A. Harris" <> | Subject | Re: IDE drive data transfer rates, kernel optimization? |
| |
On Wed, 5 Aug 1998, Aaron Tiensivu wrote:
> > I am concerned about the advertised data rates of IDE hard disks, > > more specifically the "true maximum throughput" of said drives. > > No present drive seems to approach the 33 spec..
I'm not surprised. If they did, it would be a sure sign that they were actually spending money in engineering instead of marketing. What would be nice is to see some Finnish computer science student develop his own hard disk technology that is technically perfect, but only useable by hackers because you have to put it together yourself, and then offer the drives at cost and put them under GPL.
Hey Linus, do you have a younger brother in Univ? ;o)
> > I have a Quantum Fireball SE 4.3G drive that claims UDMA > > capability. I am getting 7.8Mb/s out of this drive. This is > > I get ~9.5MB on an "old" IBM Deskstar 6.4GB UDMA on a non-UDMA (!) motherboard. > Once my UDMA-able motherboard comes, I hope the rates will rise but I have my > doubts. I'll be driving the drive at a 'sane' mhz instead of the current 41mhz > bus I'm driving it at.
I finally took the plunge last night and went from 66Mhz to 75Mhz. My K6 got a 20% speed boost. My chipset keeps PCI at 33Mhz though from what I understand from the docs... Actually I think my BIOS lets me set the PCI clock...
> > As you can see, it says "BuffSize=80kB". That is NOT 128k for > > sure. > > It's probably dedicated part of the buffer to firmware is my guess on the > matter. Also, some drives report incorrectly (for some odd reason) the amount > of cache they really do have.
Yes, that is basically what others have said too. Interesting. I guess, as internal cache size gets larger, the firmware usage will stay the same, so it will become negligible. I noticed that my CD writer (HP7200i) claims 768k buffer, however reports 3.9Mb buffer from hdparm. I am guessing that the manual is old (it mentions the older model drives here and there... and that they've increased the buffer in this drive.
I wonder how much a true IDE caching controller would help? Can anyone recommend one? IDE that is...
> > This is NOT the rated 33megs per second that the sheet claims for > > UDMA operation. The rate for PIO/DMA is 16.6Mb/s. I get 7.8Mb/s > > or roughly half of spec. Does this mean that if I use UDMA, I > > can expect around double, or 15Mb/s? If so, that is far from > > 33Mb/s, and still not quite even the calculated 19.75Mb/s that > > the internal buffer supports. > > Currently, the 33 is a marketing gimmick.. ATA66 is coming but I don't see > drives hitting that rate for a long while.
Yeah, I saw ATA66 on Quantum's website. Just another crock. The spec probably says that the drive cables can't be longer than 4 inches or so... impossible...
> > /dev/hda: > > multcount = 16 (on) > > If your drive can go higher, it is usually beneficial to push it above that. > More data per interrupt.
That is maxed. I maxed out everything according to the hdparm -i output.
> > unmaskirq = 0 (off) > > You can usually get a performance boost by enabling it. I've only seen > DTC VLB I/O cards munge this up.
Hmm... can anyone else recommend on this setting? TX-Pro chipset, Quantum fireball SE.... > > using_dma = 0 (off) > > This should be enabled.
Can't. No support in Linux for my chipset yet. > > Now back to your regularly scheduled programming. (pun intended) > > You can munge with the various -X settings but usually, drives default to their > optimal level..
Hmmm.. I've tried to change that stuff, but it made no difference, so you're probably right.
Thanks for your reply. TTYL
-- Mike A. Harris - Computer Consultant - Linux advocate
Escape from the confines of Microsoft's operating systems and push your PC to it's limits with LINUX - a real OS. http://www.redhat.com
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html
| |