donw@zehntel (Don White) (03/17/89)
[I got quite a few messages saying that I am wrong about there being no ] [inherent conflict between Amigas and SCSI, so, I'll post my reply. ] From donw Wed Mar 15 10:56:36 1989 To: zehntel!sun!sugar.hackercorp.com!karl Subject: Re: Drive Performance Degredation This is NOT an Amiga problem. This is a 2090A problem. If you use a different drive controller the problem goes away. Commodore admitted (hearsay) that when the 2090A was designed, whoever did it didn't think that loading pictures would be to large a part of the users activity, so they used the same DMA channels which display an image to transfer it. Naturally this causes some difficulty. The job of loading an image must be multiplexed. This causes slowing of the loading process. THE PROBLEM IS THE 2090!!! NOT THE AMIGA!!! (Sorry if this is nit picking.) My understanding is that they don't intend to fix it. They intend to make the multiplexing more efficient. (i.e. kludge) Summary - use a different drive controller. One that uses different DMA channels (or whatever). Don White zehtnel!donw
gay%elde.epfl.ch@cunyvm.cuny.edu (David Gay) (03/22/89)
In article <798@zehntel.UUCP> donw@zehntel (Don White) writes: > >From donw Wed Mar 15 10:56:36 1989 >To: zehntel!sun!sugar.hackercorp.com!karl >Subject: Re: Drive Performance Degredation > > > This is NOT an Amiga problem. This is a 2090A problem. If you use a > different drive controller the problem goes away. Commodore admitted (hearsay) > that when the 2090A was designed, whoever did it didn't think that loading > pictures would be to large a part of the users activity, so they used the > same DMA channels which display an image to transfer it. Seeing that the display is already using all the cycles that are available to access chip memory, it's a bit difficult to avoid it ... Haven't you noticed the slowdown with such a display present ? The only time anything can access the chip memory, with a HIRES 16 colour screen, is during horizontal and vertical retrace. > My understanding is that they don't intend to fix it. They intend to make > the multiplexing more efficient. (i.e. kludge) I'm not sure what else they could do ? (or have you got a magic wand ?) > Don White > zehtnel!donw ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ David Gay 6 x 9 = 42 GAY@ELDE.EPFL.CH, or GAY%ELDE.EPFL.CH@CLSEPF51.bitnet ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~