Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!hplabs!hplabsz!dleigh From: dleigh@hplabsz.HPL.HP.COM (Darren Leigh) Newsgroups: comp.sys.amiga Subject: Re: Hard Drive & Chip Contention Summary: maybe fixable? A little? Message-ID: <3149@hplabsz.HPL.HP.COM> Date: 29 Mar 89 23:10:33 GMT References: <4573@xyzzy.UUCP> Reply-To: dleigh@hplabs.UUCP (Darren Leigh) Organization: Open Fly Systems Lines: 76 In article <4573@xyzzy.UUCP> poirier@dg-rtp.dg.com (Charles Poirier) writes: >In article <19314@srcsip.UUCP> carpent@coltrane.SRC.Honeywell.COM (Todd Carpenter) writes: >>In article <8903232240.AA05683@yorkville.csri.toronto.edu> drz@csri.toronto.edu (Jerry Zarycky) comments on hd slowness while displaying graphics screens: >>Bull! Excuse me, but I've been viewing *lots* of 640x400 pictures, overscan, >>and others. > >You haven't said if your pictures are 4 bitplanes or not. This is an essential >ingredient. I'll grant, though, they probably are 4 planes. > >> NO WAY does this cause slowdown on the order of minutes. You have >>some screwy software, I think. Yes, I get slowdown. Probably about a factor >>of 2 to 5, at MAX. I never exactly timed it but, I get a load of a high >>res screen in less than one second. If there is already a high res picture up >>there, and I load in another, it is maybe 3 - 5 seconds. > >I timed it. A 27Kbyte, 640 x 400 x 4 image loads into DPaintII in 4.5 seconds >with Workbench in front, or in 45 seconds with DPaintII in front. Extrapolate >to a more complex (less compressed) image and that easily enters the "minutes" >realm. My system, for reference: B2000, 2090A, 1.3, Quantum Pro 80s I just did some timings with Butcher. I consistently load a 640x400x4 image (while displaying one) in about twelve seconds. Writing one takes substantially longer (about 90 seconds). The file is 102078 bytes long. What's really weird is that things didn't used to go this fast. I had some hard disk problems a while ago and ended up reformatting my entire drive and changing some things. Before then, image loads used to take forever so I was quite pleased when they went faster after the reformat (etc.). If anyone's really interested in exactly what I changed, send me mail. Diskperfa w/ workbench in front: File create/delete: create 16 files/sec, delete 52 files/sec Directory scan: 102 entries/sec Seek/read test: 119 seek/reads per second r/w speed: buf 512 bytes, rd 84562 byte/sec, wr 29127 byte/sec r/w speed: buf 4096 bytes, rd 238312 byte/sec, wr 163840 byte/sec r/w speed: buf 8192 bytes, rd 327680 byte/sec, wr 218453 byte/sec r/w speed: buf 32768 bytes, rd 524288 byte/sec, wr 291271 byte/sec Diskperfa w/ 640x400x4 image in front: File create/delete: 1 create files/sec, 6 delete files/sec Directory scan: 44 entries/sec Seek/read test: 28 seek/reads per second r/w speed: buf 512 bytes, rd 18460 byte/sec, wr 9637 byte/sec r/w speed: buf 4096 bytes, rd 22995 byte/sec, wr 12542 byte/sec r/w speed: buf 8192 bytes, rd 2449 byte/sec, wr 1105 byte/sec r/w speed: buf 32768 bytes, rd 2085 byte/sec, wr 2007 byte/sec So we do take a substantial hit when displaying hi-res 4 bit deep images, especially with large buffer sizes. Since the buffer size is so critical, apparent disk speeds will be application dependent. >>I have a *fast* disk. Other disks may not be able to handle the rate. But I >>would think that would make the problem *less* apparent. > >Not so. The problem is that the disk overruns the 2090's tiny little buffer. >The faster the disk, the more likely the overrun. Case in point, my older >Miniscribe (68 ms) drive had no noticeable problem, but my Quantum Q280 >(28 ms) has the abovementioned severe problem. BTW, I am using a 2090 not >a 2090A, but I was assured the buffer sizes are equal. Given that some people >with fast drives are having little trouble, it seems that the amount of >slowdown depends on the detailed timing of the drive and not just its >"average" speed. The 80s is a fast disk (19ms access, even better) and still doesn't take very long to load images. ======== Darren Leigh Internet: dleigh@hplabs.hp.com UUCP: hplabs!dleigh