|
|
|
|
| Welcome, Guest | Home | Search | Login | Register | |
| Author | MiniBench Scores (Read 101319 times) | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
dpaanlka
1024 MB  Posts: 1646 
|
on: October 04, 2006, 05:24
I finally finished up the System 7 Today benchmark utility. Optimized for both 68k and PowerPC, it is a simple quick test to gauge CPU power of your 68k or PowerPC Mac. See the MiniBench page for more information and to download. If you have a spare minute, run the test on your classic Mac and post the scores you get below! Scores posted will be added to the list of scores on that page. Be sure to read the instructions, and boot with Extensions Off to run the test. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
RacerX
|
32 MB Posts: 36
Reply #1 on: October 04, 2006, 15:58
|
The image doesn't work on any of my systems... old Mac OS disk images (.img files) keep part of the required image info in their resource fork... that is stripped off when shared in any way other than an AppleTalk/AppleShare connection. Neither HTTP nor FTP preserve Mac OS resource forks. If you want to distribute apps as disk images, I can convert them to Self Mounting Images (.smi files) which can be transfered via the net and don't require that the person have Disk Copy installed on their systems.
|
dpaanlka
|
1024 MB Posts: 1646
Reply #2 on: October 04, 2006, 18:35
|
Yeah I forgot to hqx it.
|
dpaanlka
|
1024 MB Posts: 1646
Reply #3 on: October 04, 2006, 18:47
|
All fixed now!
|
madmax_2069
|
16 MB Posts: 28
Reply #4 on: October 05, 2006, 02:58
|
ok i got a 160.5 score my system is a upgraded Power Macintosh (Beige G3 AIO) CPU : G3 466mhz Memory : 448mb Storage : 120gb hdd Optical : Pioneer DVR-110D video : ATI Radeon 7000 Mac Edition 32mb vram PCI/ onboard ATI Rage Pro 6mb vram Operating System : OS 9.2.2/ Tiger OS X 10.4.8 the program worked flawlessy. no problems opening it up or running it . didnt take to long to run the benchmark.
|
dpaanlka
|
1024 MB Posts: 1646
Reply #5 on: October 05, 2006, 04:04
|
So that means your G3 finishes the calculations on average 2.3x faster than a 250mhz Power Mac 8600 (the test runs the calculations 10 times, hence the progress bar). Interesting, I would have expected a better improvement than that. I wonder why that is?
|
RacerX
|
32 MB Posts: 36
Reply #6 on: October 05, 2006, 04:45
|
79.7 System: Power Macintosh 8600 Processor: PowerPC 604e at 300 MHz, 1 MB L2 Memory: 416 MB of RAM Video: 4 MB of VRAM on logic board (monitor being used) ATI Rage 128 with 16 MB of VRAM (monitor not being used) Operating System: Rhapsody 5.6 (Blue Box: Mac OS 8.6) 154.2 System: Power Macintosh 8600 Processor: G3 at 450 MHz, 1 MB L2 Memory: 256 MB of RAM Video: ixMicro Ultimate Rez (TwinTurbo 128M-3D) with 8 MB VRAM (monitor being used) ixMicro Pro Rez (TwinTurbo 128M-3D) with 8 MB VRAM (monitor not being used) Operating System: Mac OS 8.6
|
RacerX
|
32 MB Posts: 36
Reply #7 on: October 05, 2006, 05:13
|
60.0 System: PowerBook 3400c Processor: PowerPC 603e at 200 MHz, 265k L2 Memory: 80 MB of RAM Video: 2 MB of VRAM on logic board Operating System: Mac OS 8.6 28.7 System: PowerBook Duo 2300c Processor: PowerPC 603e at 100 MHz Memory: 56 MB of RAM Video: 512k of VRAM Operating System: Mac OS 8.6 158.3 System: Power Macintosh 8100av Processor: G3 at 500 MHz, 1 MB of L2 Memory: 208 MB of RAM Video: 2 MB of VRAM Operating System: Mac OS 8.6
|
RacerX
|
32 MB Posts: 36
Reply #8 on: October 05, 2006, 05:32
|
Is it me or does the 604e not perform that well on this test? I can't imagine how my PowerBook 3400c (603e at 200 MHz) can score 14% slower than your 8600 (604e at 250 MHz) when the 3400c is 20% slower in clock speed. In most of the tests I've seen the 604e is 15% to 25% faster than the 603e at the same clock speed. I had put together a graphic of relative speeds of some processors so I would have a reference... as I recall (as it is an old graphic) these are SPEC scores. (http://www.shawcomputing.net/racerx/processors_rhap.gif)
|
dpaanlka
|
1024 MB Posts: 1646
Reply #9 on: October 05, 2006, 05:55
|
On thinking about it more, the benchmark I suppose isn't pure calculation. It's hard to explain but to make what I thought would be a more real world type of thing, I have RealBasic perform a set of routines ten times, but each seat consists of multiplying a number by 3, then dividing it by 2.99, then multiplying it by 3 again, the entire time moving it back and forth between two storage locations in memory. It does this 2 million times per set, and it performs ten sets, then takes the average. Each set looks like this... Code: [Select] index = 1So I would imagine this benchmark is not entirely processor dependent, and perhaps I should change the wording on my page and in the app. So this probably reflects other things besides CPU power, such as motherboard and memory speed, etc... Would that make sense? Or am I missing something? Unless of course there is some error in the way I generate the scores, which are generated by dividing 1 billion by the average of the ten sets of scores. I did this to reverse the effect of faster machines having lower scores, so that the faster the machine is, the higher it's score is. So on a PowerBook 1400c, 1 billion divided by the average comes out to a nice 40.3. So the question is, will this scoring system be out of ratio, with each equal jump in performance producing smaller and smaller score additions? I don't think so, but I'm really bad at conceptualizing math.
|
RacerX
|
32 MB Posts: 36
Reply #10 on: October 05, 2006, 06:35
|
Well, I can see where something like L2 cache would help a system in this case... which may have helped the 3400c. What size L2 does your 8600 have? And actually what you are doing right now is the best way to see how the test works out... trying it on as many processors as possible and seeing how they pan out compared to what you know of each. And in this set of test, throw out the Blue Box score... I found out here that there is some performance penalty from Blue Box. So it may be a good idea to get a few more results and then plot them to see where they land compared to where you would think (from experience) that they would land. Once you have an idea how this test is working, it'll be easier to tweak it to get a more consistent set of results. Trying to make changes right now would be working in the dark.
|
dpaanlka
|
1024 MB Posts: 1646
Reply #11 on: October 05, 2006, 06:39
|
Quote from: "RacerX" What size L2 does your 8600 have? That particular 8600 was not actually mine, that was a test that another user performed and sent to me. Same with the 2400c. The only test I ran was the 1400
|
RacerX
|
32 MB Posts: 36
Reply #12 on: October 05, 2006, 07:12
|
Oh... okay. I was thinking that the 8600/250 was different than the 8600/300, but it isn't. Both have 1 MB of L2 on a special 100 MHz bus direct to the processor. The 8600/200 was the one that came with 256k standard on the same bus as the rest of the memory (at 50 MHz). I'd give it another day or so to collect scores and then we can see just how the scores are skewed. If we see a curve along a family of processors that should have a pretty straight progression, then it may just end up being the last calculation to make the faster scores larger. ... but for now it is still too early to tell. Um... this isn't homework, is it? If you have a deadline for this as a project, I'll pull more systems out of storage to get some more results.
|
dpaanlka
|
1024 MB Posts: 1646
Reply #13 on: October 05, 2006, 07:34
|
No, it isn't homework, this was just something I did on my own time. I did a simplified version of the final calculation though, which was 1000 / 20 which equals 50 1000 / 40 which equals 25 1000 / 10 which equals 100 If a computer took 20 microseconds to complete the operation, it's score was half that of a computer which only took 10 microseconds. A computer that takes 40 microseconds to complete the operation's score is half that of the computer that takes 20 microseconds. So I wouldn't think the score calculation would mess up the score.
|
madmax_2069
|
16 MB Posts: 28
Reply #14 on: October 05, 2006, 18:45
|
Im thinking the memory bus being a bit faster in my system than the other ones could be the limiting factor in the score. don't benchmark have to deal with the memory some how . so a faster system bus and memory bus would result in a higher score even on a slower CPU clock speed. just like the result of that 8100AV with the G3 500mhz CPU upgrade scored a 158.3 it also has a 50mhz system bus and mine running at 466mhz scored a 160.5 and has a 66mhz system bus. like you said i think the memory and system bus has something to do with the outcome of the score which it should cause all of it play's a part in CPU speed in general. any way its a great benchmarking program. i myself would not know any other way to get results then how you did it ( AKA by allot of math).
|
|
Pages: [1] 2 3 4
|
| |||||||||||||||
|
© 2021 System7Today.com. |
