home *** CD-ROM | disk | FTP | other *** search
- Path: grapool30.rz.uni-frankfurt.de!news
- From: raftopou@stud.uni-frankfurt.de (Joerg Raftopoulos)
- Newsgroups: comp.sys.amiga.hardware
- Subject: Re: Proper mask, transfer rate, etc.
- Date: 15 Apr 96 22:54:02 +0100
- Organization: J. W. Goethe-Universitaet Frankfurt/Main
- Message-ID: <3184.6679T1374T130@stud.uni-frankfurt.de>
- References: <4k1740$ddo@hobbes.compusult.nf.ca> <4k1lkb$a3j@freenet-news.carleton.ca>
- <2365.6671T6T1540@stud.uni-frankfurt.de> <dalec.06j5@zorro.amitrix.com>
- NNTP-Posting-Host: dialin139.rz.uni-frankfurt.de
- X-Newsreader: THOR 2.22 (Amiga;TCP/IP) *UNREGISTERED*
-
-
- DC> Actually, the 3000 users manual has no such reference. You must be talki
- DC> about the section in the 2.04 Workbench manuals, which pertains to all
- DC> machines, not just the 3000 (BTW, it's section 6-87 in the 2.1 manual).
-
- alas! i always should have my personal notes handy before i write nonsense.
-
- the correct mask value for the internal amiga 3000 series scsi hostadapter is
- indeed 0x7ffffffe. (as stated in earlier postings). mask should be 0xffffff
-
- i promise next time i`ll be more carefull before posting.
-
- DC> However, the default value in HDToolBox has always been 0x7ffffffe, up to
- DC> and including 3.1, AFAIK. That is just a couple of bytes shy of 2 Gigs,
- DC> which is the limit of the filesystems addressing. Anything larger is
- DC> invalid.
-
- as far as i understand the mask parameter it determines address space for dma
- operations. a3k memory map says every address >= $8000 0000 is reserved (or z3
- autoconfig from $ff00 0000 to $ff01 0000) and it`s really a good idea not to
- access an area where today nothing is located and nobody knows what in future
- amigas will be put in.
-
- DC> As a matter of interest, I once ran some DiskSpeed tests with both a "c"
- DC> and an "e" mask. The differences were so small as to be within the
- DC> variation in numbers between successive tests with no changes. However,
- DC> usual, your mileage may vary, I'll stick with the default, it works.
-
- at least the "c" it won`t crash your amiga or corrupt harddrive data. but it`s
- slower on word aligned buffers:
-
-
- MKSoft DiskSpeed 4.2 Copyright ⌐ 1989-92 MKSoft Development
- ------------------------------------------------------------
- CPU: 68030 AmigaOS Version: 40.68 Normal Video DMA
- Device: bin0: Buffers: 120
- Comments: DiskSpeed 4.2
-
- { Amiga 3000, 68030@25 Mhz, Kickstart 40.68, Wb 40.42, WD 33C93A-00-08PL }
-
- CPU Speed Rating: 1320
-
- Testing with a 512 byte, MEMF_FAST, WORD-aligned buffer.
-
- 7ffffffe mask || 7ffffffc mask
- Create file: 41216 bytes/s | CPU : 50% || 78400 bytes/s | CPU : 26%
- Write to file: 43072 bytes/s | CPU : 51% || 38464 bytes/s | CPU : 36%
- Read from file: 48512 bytes/s | CPU : 44% || 83968 bytes/s | CPU : 13%
-
-
- *** but ***
-
- Testing with a 4096 byte, MEMF_FAST, WORD-aligned buffer.
-
- 7ffffffe mask || 7ffffffc mask
- Create file: 244736 bytes/s | CPU : 27% || 54254 bytes/s | CPU : 52%
- Write to file: 276992 bytes/s | CPU : 23% || 28823 bytes/s | CPU : 57%
- Read from file: 250880 bytes/s | CPU : 23% || 47261 bytes/s | CPU : 52%
-
-
- Testing with a 32768 byte, MEMF_FAST, WORD-aligned buffer.
-
- 7ffffffe mask || 7ffffffc mask
- Create file: 391259 bytes/s | CPU : 22% || 57420 bytes/s | CPU : 55%
- Write to file: 437179 bytes/s | CPU : 20% || 27702 bytes/s | CPU : 59%
- Read from file: 402485 bytes/s | CPU : 14% || 44281 bytes/s | CPU : 56%
-
-
-
-
-
- SOL LUCET OMNIBUS
-
- Joerg Raftopoulos, Amiga User since 1987
- 3 x A1000, 1 x A500, 2 x A2000, 1 x A1200, 2 x A3000 Owner
- Student at J. W. Goethe Universitaet, Frankfurt am Main, West Germany, Sol III
-
-