Dhrystone for 68k Macintosh

[Patater]

Howdy comrades,

I was asked to build Dhrystone for 68k Macintosh, just for fun. I felt I should share it with you all. This ancient benchmark will measure the integer performance of your Macintosh. Let us compare results, if feel so moved.

Dhrystone for 68k Macintosh

Results:

Macintosh SE/30

40000 Runs

Dhrystones per Second: 2666.7

 

[bbraun]

Neat! Thanks for doing this, and thanks for sharing!

 

[Gorgonops]

And just for grins here's Dhrystone 2.1 on the somewhat elderly Opteron workstation I'm in front of (Ubuntu 10.4):

dhry$ ./a.out

Dhrystone Benchmark, Version 2.1 (Language: C)

Program compiled without 'register' attribute

Please give the number of runs through the benchmark: 1000000000

Execution starts, 1000000000 runs through Dhrystone

Execution ends

Final values of the variables used in the benchmark:

(snip)

Microseconds for one run through Dhrystone: 0.1

Dhrystones per Second: 14285714.0

While the 2.53 Ghz Core i5 MacBook Pro running Snow Leopard racks up 24390244.0. Yay?

 

[techfury90]

An IBM PS/2 Model 56 gets about 5900 it seems, under DOS 6.22 with the code built using Open Watcom. Not bad for a 20 MHz 386SX built under license by IBM with added on-die cache...

 

[johnklos]

Results:Macintosh SE/30

40000 Runs

Dhrystones per Second: 2666.7

I'm guessing that the division which reports 2666.7 (or 500 ms) doesn't have proper accuracy, which is why it's coming up as such an exact number. Have you figured that out?

It's interesting to see that Dhrystones measured in Speedometer 4 and in NetBSD on the same machine come pretty close. I'll have to download this program and try it, too, for comparison.

Dhrystones in Mac OS 8.1 on Q610 (50 MHz m68040) in Speedometer 4:

35888

Same machine in NetBSD 5:

36764.7

For comparison, it seems that the m68030 in the SE/30 beats out the processor in the VAXstation 4000/30 (VLC), which is the slowest computer I own which runs NetBSD. It only gets 2164 Dhrystones a second.

 

[Patater]

I sold my Macintosh IIfx to a cool dude about two years back. It was missing a rubber foot or two. I told him that I thought I had seen one around and that I would send it to him, but I didn't find that foot until last month. He was happy to finally receive it.

I also asked him to run Dhrystone for me. Here are the results from his wicked fast Macintosh IIfx, with 32 MiB memory, running System 7.5.5.

100000 runs:

Microseconds for one run through Dhrystone: 130.0

Dhrystones per Second: 7692.3

200000 runs:

Microseconds for one run through Dhrystone: 125.0

Dhrystones per Second: 8000.0

500000 runs:

Microseconds for one run through Dhrystone: 128.0

Dhrystones per Second: 7812.5

Unfortunately, it isn't over 9000.

 

[Gorgonops]

An IBM PS/2 Model 56 gets about 5900 it seems

Is it a 386SLC or a 486SLC model? The 486SLC(2) was actually a pretty potent CPU, at least on benchmarks. Integer-based 'marks small enough to fit into cache scored almost in the same ballpark as a real 80486. Granted that's the most useless kind of benchmark for determining real system performance, but... that's what Dhrystone is all about!

(The *stupid* thing about that CPU was of course its 24 bit address bus. The 386sx sort of made sense in 1988, not so much in 1992.)

Speaking of useless benchmarks, I screenshotted the OP's program results alongside Speedometer 4.0's running in BasiliskII on my Opteron:



They come out in the same ballpark although the actual result varies quite a lot from run to run, undoubtedly a consequence of the JIT compiler rather than anything to do with the program. It is mildly impressive to some degree that the emulated performance in some runs is close to 1/10th that of Dhrystone running natively, I guess.

(This is a recently built-from-source Basilisk II binary which seems to be both wickedly fast and wickedly unstable. The 2.8Ghz Opteron is scoring over twice what a 2.0 Ghz Mac Pro scored in Speedo-4.0 using an older binary dating from when BasiliskII was in the Ubuntu package collection, and I *know* the 2.8Ghz Opt isn't that much faster than the M'Pro.)

 

[MidnightCommando]

Fun little benchmark.

My PowerBook scored 5714.3 Dhrystones per second.

Apparently, my PB180c is about as fast as an IBM PS/2 Model 56.

 

[Patater]

If you are interested in benchmarking with your cache disabled, I wrote a tool to help you disable the cache. Disabling the cache should make a big difference on Dhrystone results.

 

[techfury90]

An IBM PS/2 Model 56 gets about 5900 it seems

Is it a 386SLC or a 486SLC model? The 486SLC(2) was actually a pretty potent CPU, at least on benchmarks. Integer-based 'marks small enough to fit into cache scored almost in the same ballpark as a real 80486. Granted that's the most useless kind of benchmark for determining real system performance, but... that's what Dhrystone is all about!

386SLC 20 MHz. The SCSI controller performance on that thing blows just about every 68k out of the water too... it's kind of scary that it can reliably pull 3-4 MB/s.

 

[techfury90]

Here's another fun result:

Microseconds for one run through Dhrystone: 0.1Dhrystones per Second: 7448789.5

HP Integrity rx2600, 1.3 GHz/3MB L3 Itanium2 "Madison", OpenVMS 8.4. The compiler is:

$ cc/versionHP C V7.3-020 on OpenVMS IA64 V8.4

$

IIRC it's a rebadged Intel compiler along with the C++ and FORTRAN ones. I know Pascal and COBOL are modified versions of the DEC Alpha compilers.