Title: RE: [t13] INTRQ
Jeff
 
    I think you are making many assumptions about the application...  In my case, it was BIOS.  BIOS is by nature single threaded, CPU utilization is not a factor.  BIOS does however benefit from UDMA since some drives can now sustain up to 40MB/s.  So, regarding your points:
 
1. Yes it is an SFF-8038 host
2. Incompete easily in transfer rate, runs faster than anything windows has to offer, and CPU utilization is 100%.  Keep in mind that polling is faster than interrupting.  If all you want to do is stream data, you poll
3. Cpu utilization is directly proportional to the polling rate...
4. If you were using an ADMA based host (project 1510d) you would not have any of these issues.
 
 

-----------------------
Curtis E. Stevens
Pacific Digital Corp.
2052 Alton Parkway
Irvine, CA 92606

Phone (949) 477-5713
Fax (949) 252-9397

E-Mail: [EMAIL PROTECTED]
WEB: www.PacificDigital.com

The face of a child can say it all, especially the mouth part of the face...

-----Original Message-----
From: Wolford, Jeff [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, January 22, 2002 12:04 PM
To: [EMAIL PROTECTED]
Subject: [t13] UDMA without INTR

Curtis,
 
I chopped all the other stuff out
 
 OK Jeff, this E-Mail is way too long to read.  However, just looking @the first point, I did UDMA without using the interrupt.  The interrupt is truely not a requirement for UDMA.
[JWW>] 
1) Please describe in high level and was it using a SFF-8038 type host ?
 
2) Can a non-INTR driven UDMA compete with one that does in either:
a) Transfer Rate
b) CPU Utilization
 
3) If it involves ANY polling are you not either
a) Consuming the CPU that you are trying to off load with UDMA
b) Have a large gap after the data transfer is completed to when you
    checked for status ?
 
Jeff

Reply via email to