1 |
I've tried re-routing, and even switching out the cable with a |
2 |
braided/shielded one. No effect. I've also tried switching the drive's |
3 |
mode to mdma2 (via hdparm), but that locks up my system. |
4 |
|
5 |
Maybe the driver-optimization is the key. I don't have a different |
6 |
manufacturers card to try at the moment (only a few promise ultra66s). |
7 |
It's kind of coincidental though... I tried the same thing with a |
8 |
couple of different SCSI drives (plugged into an Adaptec 2940), and |
9 |
received similar throughput (9M). Maybe the Adaptec driver isn't |
10 |
optimized either? |
11 |
|
12 |
|
13 |
|
14 |
David S. Miller wrote: |
15 |
|
16 |
>From: Derek Pizzagoni <pizz@×××××××.net> |
17 |
>Date: Wed, 31 Aug 2005 16:09:49 -0700 |
18 |
> |
19 |
> |
20 |
> |
21 |
>>As you can see, the drives are identical. The first one comes up in |
22 |
>>mdma2 mode, and the second one comes up in udma4 mode. This makes |
23 |
>>sense, since the motherboard doesn't support the faster interface (only |
24 |
>>up to 33, right?). |
25 |
>> |
26 |
>> |
27 |
> |
28 |
>The limit is mdma2 mode for the onboard IDE controller. |
29 |
> |
30 |
> |
31 |
> |
32 |
>>I've run them multiple times, and it always comes out similar to above. |
33 |
>>This is on a relatively dormant system. Can anyone tell me why a drive |
34 |
>>plugged into the motherboard's IDE interface (running in mdma2 mode) |
35 |
>>would outperform the same drive plugged into an Ultra66 card (running in |
36 |
>>udma4 mode)? With these results, I'd be better off putting the second |
37 |
>>drive back on the chain with the CDROM. |
38 |
>> |
39 |
>> |
40 |
> |
41 |
>If there is any noise on the cable, UDMA performance can suffer |
42 |
>dramatically, because commands will be retried when parity errors |
43 |
>are detected on the bus. |
44 |
> |
45 |
>It's nice that UDMA handles parity errors nearly transparently |
46 |
>like this, but it can be a silent performance killer. |
47 |
> |
48 |
>Another possibility is that the promise driver isn't optimized |
49 |
>for sparc64 so well. This kind of thing has been discovered |
50 |
>before. |
51 |
> |
52 |
> |
53 |
|
54 |
-- |
55 |
gentoo-sparc@g.o mailing list |