Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!zaphod.mps.ohio-state.edu!caen!umich!sharkey!nstar!jwt!john From: john@jwt.UUCP (John Temples) Newsgroups: comp.sys.ibm.pc.hardware Subject: IDE Transfer Rates (was: Re: TOSHIBA IDE HARDDRIVE INFO) Keywords: HD Message-ID: <1991Feb26.033111.7347@jwt.UUCP> Date: 26 Feb 91 03:31:11 GMT References: <4663@cocoa46.UUCP> Distribution: comp Organization: Private System -- Orlando, FL Lines: 29 In article <4663@cocoa46.UUCP> reichert@motcid.UUCP (Chuck KD9JQ) writes: >The current 1 to 1 interleave gives 920 - 970kbs transfer rate. Using what -- CoreTest? I just ran CoreTest on a Conner CP3204, and got a very impressive 1100 kb/sec transfer rate with no software disk caching enabled. I suspect that this figure is almost totally meaningless. Since the Conner's on board cache is at least as large as the maximum block size can CoreTest use, I think all I'm seeing is the speed at which I can read from the drive's cache. I doubt that CoreTest is able to force a physical read from the drive. After using the drive for the while, it certainly didn't feel like I was getting the impressive performance I expected. I ran a simple C program I use to benchmark disk performance which writes 1024 records of 1024 bytes each, then reads them back in sequence. The write rate was an appalling 70 kb/sec, with the read rate not much better at 140 kb/sec. Results were similar under both DOS and UNIX. Suspecting the non-cached 20 MHz 386 wasn't fast enough to keep up with 1:1 interleave at 12 Mbits/sec, I tried the drive on a cached 386/33, with the same results. Now that I find the Conner to be significantly slower than a 1:1 ST-506/MFM drive in "application level" transfer rates, I wonder if I've set something up incorrectly, or is a 386/33 just not fast enough to run 1:1 at 12 Mbits? Should the drive be reformatted to 2:1, or should I get a 486/33? Will I get better performance from ESDI at 10-20 Mbits/sec? -- John W. Temples -- john@jwt.UUCP (uunet!jwt!john)