home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!zephyr.ens.tek.com!uw-beaver!uw-coco!nwnexus!kanefsky
- From: kanefsky@halcyon.com (Steve Kanefsky)
- Newsgroups: comp.sys.mac.hardware
- Subject: Re: Daystar Powercache (need info...)
- Message-ID: <1992Sep13.030822.21282@nwnexus.WA.COM>
- Date: 13 Sep 92 03:08:22 GMT
- References: <15990.2AAE8B94@zeus.ieee.org> <1992Sep10.165937.22578@nwnexus.WA.COM>
- Sender: sso@nwnexus.WA.COM (System Security Officer)
- Organization: The 23:00 News and Mail Service
- Lines: 121
-
- In article <1992Sep10.165937.22578@nwnexus.WA.COM> I wrote: writes:
- >
- >As soon as I've got my PowerCache installed, I'll test it out and fill in
- >the net. Naturally I'll post those all-important Speedometer numbers :-)
-
- OK, I've had the PowerCache for about 24 hours now (it took an extra day
- because MacWarehouse apparently calls back to verify credit-card
- purchases of this amount. Since I ordered late on Wednesday they didn't
- call back until Thursday and I didn't get the package 'till Friday.
- Keep that in mind if you're ordering something expensive from
- MacWarehouse).
-
- Not to keep anybody in suspense, the PowerCache improves my LC II's
- performance incredibly! Most things are literally three times faster.
- The only area of performance which isn't increased substantially is disk
- throughput (and I have a relatively slow Quantum 80MB LPS drive).
- Nevertheless, even disk intensive tasks like launching a big application
- with a big data file are much faster (launching Excel 4.0, for example is
- at least twice as fast -- even moreso when it's launched with a big set
- of complex linked spreadsheets.)
-
- Programs that are intended to run at a constant speed (mostly games) are
- running a whole lot more smoothly now. Solarian, for example -- which
- used to get "bogged down" in the later rounds -- doesn't slow down at all
- now no matter how much action is occuring on-screen. Spectre is also
- much easier to play now. It used to be difficult to aim on the higher
- levels (level 20 or so) when there were a lot of tanks firing at you
- because one tap of the key would move you 10 degrees or so. Now you can
- aim much more precisely and fire more rapidly and everything seems much
- smoother (I could even swear that the rendering of the 3D scenes was
- more realistic). I haven't decided yet whether this makes games easier
- or harder to play. It certainly makes them more fun to play. I just
- tend to get more reckless (just as one tends to get more reckless when
- driving a high-performance car than an old jalopy).
-
- CPU intensive programs really fly now! I use DeHQX a lot to translate
- ..hqx files to Mac files, and it used to be annoyingly slow. Now it
- really flies! Compression programs are much much faster now too.
-
- You may recall from my last article that I decided not to get the FPU
- with my 50mhz PowerCache, although I intend to upgrade to the 33mhz
- 68040 PowerCache with built-in FPU when it comes out. The only regret I
- have so far is that I'd like to see how well my LC II + PowerCache would
- perform on number-crunching tasks like ray-tracing. I tried to run
- Persistence of Vision and got an error dialog saying that the program
- wouldn't run without an FPU, after which the machine bombed. That's the
- only bomb I've gotten so far with the PowerCache, and I've got lots of
- extensions (all the Now Utilities plus 6 or 7 other ones). I've also run
- lots of programs, even some old games from '88 or so, and everything that
- ran before still runs with the PowerCache. I'm not sure if the POV crash
- would have happened on my LC II without the PowerCache or not. After
- the POV crash, I got a copy of SoftwareFPU and tried to do some
- rendering with that. Even the 50mhz '030 with Software FPU is a far cry
- from a 16mhz '030 with 16mhz FPU. I had the Daystar PowerMath LC before
- I got the PowerCache, and POV was a lot faster with the real FPU.
-
- I wish there was a way in software to completely disable the PowerCache,
- so I could easily switch back and forth and give lots of concrete
- numbers, but the only thing you can do with software is disable the 32K
- RAM cache. I did, however, have Speedometer 3.1 results from my plain
- LC II and of course I have results with the PowerCache installed. My
- setup is a Mac LC II with 4MB RAM, 80MB Quantum LPS hard disk, 512K
- VRAM, 13" Apple Hi-Res RGB monitor (set to 1-bit video before running the
- tests). I rebooted with all extensions, startup items, and virtual
- memory off in each case, except that I needed the DayStar PowerCentral
- extension to enable the cache for the tests with the PowerCache.
- I've included the raw results here for each setup as well as the
- percentage improvement in each test ((B-A)/A * 100, where B is with
- PowerCache and A is without).
-
- I'd really appreciate it if someone could post or email to me some
- Speedometer readings on IIfx's and/or Quadra 700's. I'd be interested
- to see how an accelerated LC II compares to those machines.
-
- Remember that my these benchmarks are with no FPU. Steve Tuttle from
- DayStar, in a note posted on sumex.stanford.edu
- (info-mac/report/daystar-powercache.txt), listed the following Speedometer
- figures for an LC with 50mhz PowerCache w/ FPU:
-
- P.R. Rating: 11.88
- Bench Avg: 20.82
- FPU Avg: 4.05
- Color Avg: 2.30
-
-
- Here are my figures for the LC II with and without 50mhz PowerCache (no
- FPU in either case) Keep in mind that 0% faster would mean no
- improvement, 100% faster means twice as fast, 200% faster means three
- times as fast, etc:
-
- Benchmark w/o PC w/ PC %faster
- -----------------------------------------
- CPU 4.03 13.11 225%
- Graf 4.80 10.52 119%
- Disk 1.65 1.95 18%
- Math 5.63 17.35 208%
- P.R. Rating 3.95 10.52 166%
-
- KWhet 6.67 18.55 178%
- Dhry 2.64 10.44 295%
- Towers 2.57 8.32 224%
- Quicksort 3.26 10.73 229%
- Bubble Sort 4.66 14.21 205%
- Queens 4.36 11.74 169%
- Puzzle 4.70 16.56 252%
- Permute 3.69 7.43 101%
- Fast Fourier 4.53 13.73 203%
- F.P. Matrix 4.43 12.85 190%
- Int. Matrix 3.94 15.12 284%
- Sieve 4.87 15.07 209%
- Bench. Ave. 4.19 12.90 208%
-
- Monochrome 1.15 2.63 129%
- 2 Bit 1.23 2.49 102%
- 4 Bit 1.28 2.17 70%
- 8 Bit 1.28 1.76 38%
- Color Ave. 1.23 2.26 84%
-
-
- --
- Steve Kanefsky
-