If ATA/33 and ATA/66 on the same channel, what speeds do they run at?

[AuraEdge]

Ive heard many conflicting answers on this one..
Say you have an ATA/33 device, and a ATA/66 devices, on the same ATA/100 channel, with a 80 pin ribbon cable

Does the ATA/66 go down to ATA/33 because the ATA/100 channel goes the speed of the slowest device on the channel? or does the ATA/66 devices go at ATA/66?

I know on SCSI devices, a slow drive will drag down the rest of the chain, but does this happen on IDE too?

Most of the ones I hear about the IDE devices slowing down to the slowest device on the chain are in reviews in such, such as this one - http://www.cdrlabs.com/review.php3?reviewid=38&page=Features - the text next to the UDMA 66 Icon.

From personal experience, it seems it isn't the case - Ive had a PIO4 CD drive and a ATA/66 Harddrive Drive on the same channel, and speeds were still high. Also, Windows 2K showed the CD drive as PIO only for the first devices but Ultra ATA for the Harddrive.

Other evidence is that I have an ATA/33 DVD drive now and the same ATA/66 harddrive as its slave in my system. The harddrive benches similar to those of other ATA/66 drives in Sandra.

I really would like to know, as to know how to best configure IDE devices in the future when setting up peoples computers.

[krusty the klown]

Hmmm... I've been told that mixing 33&66 on a 66/100 ribbon cable will pull the entire channel down to 33...

BUT: (and this makes as much sense as Sandra 2001 benchmarks on a celery 600 http://www.sysopt.com/forum/wink.gif)

My m8 bought a new ATA100 IBM drive and was running it on a Soyo board with ATA66. He was using an ATA66 ribbon and was slaving his old Fujitsu ATA33 drive. So, of course I say that you won't be getting the best performance from the IBM and it'll be running at ATA33....

Well, I benched it as he'd set it up using HD Tach and much to my amazement, it was running at ATA66 speed! Then I disconnected his old ATA33 drive and the remaining IBM dropped back to ATA33. Much weirdness!! (and yes, the remaining IBM drive was on the correct connector on the ribbon cable!)

You could actually hear the difference in speeds during the benchmark!

You can d/l a shareware version of HD Tach: http://www.tweakfiles.com/drives/hdtach.html

It will not run the write test, or run under Win2000 without a licence.

Guess ya just bench & see

[AuraEdge]

I just ran HD TACH on my WD 20.5 GB ATA/66 Drive that is a slave to my Pioneer 105s ATA/33 CD/DVD drive on a ATA/100 channel (686B primary channel). This is the second time I ran HD Tach it through on this drive, and the harddrive has not been written to since last defrag. Its about half full of data.

<IMG SRC="http://a.postmypic.com/a.nsf/Z/HD+Tach+-+ATA66+w+ATA33WTHO/$file/HD+Tach+-+ATA66+w+ATA33.jpg" border=0>

Can ATA/33 do that?

I guess I'll try unplugging the DVD drive sometime and benching again...
Sometime soon if my curiously isnt satisfied.

[krusty the klown]

That looks like ATA66 to me. I remember that my m8s IBM drive was giving around 55MB/s burst speed.

I just benched my old Fujitsu 6.4G ATA33 under Win98 (dual boot/2K) and it was turning in 23MB/s - about what I'd expect for it.
http://b.postmypic.com/b.nsf/Z/HDDCQHU/$file/HDD.jpg

[AuraEdge]

Just tryed it

WD 20.5GB ATA/66 slave to no master on 686B Primary channel.

<IMG SRC="http://b.postmypic.com/b.nsf/Z/HD+Tach+-+ATA+66+wo+ATA33VALH/$file/HD+Tach+-+ATA+66+wo+ATA33.jpg" border=0>
Looks about the same...
Is ther anything I missed in my experimentation? If not, guess IDE doesn't share the same 'slow down to slowest device on chain' effect as SCSI does.



[This message has been edited by AuraEdge (edited 04-11-2001).]

[Jeff7]

I have an ATA/100 primary master and an ATA/66 primary slave. Max burst speed of the ATA/100 drive is 77MB/sec. Max burst of the ATA/66 drive is 46.7MB/sec. Doesn't seem to have much effect there.

[AuraEdge]

Hmm thats wierd my RAID0 of ATA/100's can only reach about 80MB/s (masters on seperate channels, no slaves)
I figure it should be higher shouldnt it?

[dosmastr]

i think newer IDE can handle it.

but i know that when i hooked an OLD DMA 8 drive up to a DMA32 channel everything fell down to a snails pace (imagine, clicking start, going to the bathroom and comming back just in time to see it pop up UG!)
i would think newer SCSI could get around it to, the slow drive does slow everything down, but its "disconnected" when not in use or when it doesn't have any relevant info to transmit...right??

anyway, you can get one of those promise IDE cards, independant timing for all 4 devices.

[dosmastr]

auraEdge
that's about what my RAID 0 benched too, fastest IDE drive i know of is the 60GXP by IBM 40mbps sustained, run two together, you should get about 80MBps sustained. (on paper anyway)
the best i ever got with my dual 75GXP's (the 60GXP's older brother running at 37mbps sustained) was 75mbps with an average of only 60mbps. so i guess they max out there, but those are only dual drive configs....
is that a registered copy of hd tach?
i have the free version and it only did the first 8GB of my 60GB RAID 0.