[Info-vax] MicroVAX/VMS newbie, couple of questions

[John E. Malmberg]

Jordi Guillaumes i Pons wrote:
> En/na John E. Malmberg ha escrit:
>> Jordi Guillaumes i Pons wrote:
>>
>>> En/na John E. Malmberg ha escrit:
>>>
>>>>
>>>> I am not sure how big of a x86 system running Linux is needed to be 
>>>> faster than my VAX 4000 Model 500.
>>>
>>>
>>> If it can be of help, yesterday I compiled OpenSSL both in one 
>>> 4000-200 and SIMH on a MacBookPro with an Intel 2 Duo at 2.4 GHz. The 
>>> compilation time in the mac was about 1/3 than in the 4000/200. I 
>>> don't know the difference between the -200 and the -500, but perhaps 
>>> you can make yourself an idea.
>>
>>
>> VAX 4000 Model 200 is  5 VUPs.
>> VAX 4000 Model 500 is 24 VUPs.
>>
>> Do you mean SimH on the Mac is 3 times faster than your VAX?
> 
> Well, doing THAT workload, which is very CPU-bound, yes. BUT the mac was 
> not idle, I was using it at the same time (anyway, SIMH eat almost an 
> entire 2.4 core).

A note to the original poster, your MicroVAX 3400 is rated at 2 VUPs.

So SimH on the MacBookPro is at least 7 times faster than the MicroVAX 
3400 hardware by that measurement.

>> That would put your MacBookPro SimH at 15 VUPs, and means that I would 
>> need a 3.8 Ghz processor to retire my VAX 4000 Model 500.
> 
>> Of course the disk drives in use will also affect your compile speed.
> 
> I was using the internal drive of the MacBook (SATA, 160 GB), and the 
> VAX was working with a RF72 (DSSI) as system disk, but the compilation 
> was done over a RZ26 (SCSI). OTOH the 4000-200 has just 40 MB of main 
> memory, while the SIMH is configured with 128MB.
> 
> 
> This is the accounting record for the compile in the 4000-200:
> Processor time:              0 15:15:03.27
> 
> And this is the one for the SIMH on mac:
> Processor time:              0 10:15:50.57
> 
> As you can see, I did the compilation in batch on the VAX, but did it 
> interactively on the mac. So both records are not comparable "per se" (I 
> did a lot more things on my interactive session, hence the difference in 
> the number of images executed). If you want, I can't repeat the compile 
> in batch in the mac so you'll have a set of comparable numbers.

Is there anyway to do the compile on an RF72?

If you are using a KZQSA, it is limited to about 1 MB/sec throughput, 
although disk caching may compensate for that.

A third party Q-BUS SCSI adapter can run at about 3.5 MB/sec throughput.

I assume that the RF72s are on the internal DSSI bus as that is a bit 
faster than a KFQSA DSSI bus.

Which also means that on a 100 Mbps Eithernet link, an emulated VMS 
system should be able serve files at close to 5 MB/sec, which is faster 
than the speed of IDE drives on a DS10/DS10L, which conveniently has a 
second ethernet adapter that can be dedicated to accessing the cluster 
storage.

-John
wb8tyw at qsl.network
Personal Opinion Only