keith@beamfoundation.

I have been keeping a log of performance data using the same test patch (originaly done by Dave

Stanford). I include comparisons to other Max versions back to 3.6.2. All tests were on a 1G

tiBook, with a 512 sample I/O and signal vector size. Overdrive off.

Would someone please run this test on the new Intel hardware and post results.

I have been keeping a log of performance data using the same test patch (originaly done by Dave
Stanford). I include comparisons to other Max versions back to 3.6.2. All tests were on a 1G
tiBook, with a 512 sample I/O and signal vector size. Overdrive off.

Would someone please run this test on the new Intel hardware and post results. 

  Max 4.6.0b4 (OS 10.4.7 PPC)
  buttons: 92 - 139
  triggers: 162 - 243
  bangbangs: 104 - 150

  Max 4.5.7 (OS 10.4.7)
  buttons: 58
  triggers: 92 - 104
  bangbangs: 58

  buttons: 91
  triggers: 149
  bangbangs: 103

  buttons: 3249
  triggers: 178
  bangbangs: 131

  buttons: 393
  triggers: 108
  bangbangs: 39

  buttons: 232
  triggers: 46
  bangbangs: no result

Keith McMillen 
BEAM Foundation 
http://www.beamfoundation.org/ 
510.502.5310


max-4-6-0b4-performance-measurements

Quote: keith@beamfoundation. wrote on Fri, 14 July 2006 15:21

----------------------------------------------------

> I have been keeping a log of performance data using the same test patch (originaly done by Dave

> Stanford). I include comparisons to other Max versions back to 3.6.2. All tests were on a 1G

> tiBook, with a 512 sample I/O and signal vector size. Overdrive off.

> Would someone please run this test on the new Intel hardware and post results. 

Quote: keith@beamfoundation. wrote on Fri, 14 July 2006 15:21
----------------------------------------------------
> Hi All,
>
> I have been keeping a log of performance data using the same test patch (originaly done by Dave
> Stanford). I include comparisons to other Max versions back to 3.6.2. All tests were on a 1G
> tiBook, with a 512 sample I/O and signal vector size. Overdrive off.
>
> Would someone please run this test on the new Intel hardware and post results. 
>
> Thanks,
>
> Keith McMillen

buttons: 11, sometimes 0
triggers: 23 or 34
bangbangs: 11 or 0

>
>   Max 4.6.0b4 (OS 10.4.7 PPC)
>   buttons: 92 - 139
>   triggers: 162 - 243
>   bangbangs: 104 - 150
>
>   Max 4.5.7 (OS 10.4.7)
>   buttons: 58
>   triggers: 92 - 104
>   bangbangs: 58
>
>   Max 4.5 (OS 10.2.8)
>
>   buttons: 91
>   triggers: 149
>   bangbangs: 103
>
>   Max 4.3 (OS 10.2.8)
>
>   buttons: 3249
>   triggers: 178
>   bangbangs: 131
>
>   Max 4.1 (OS 9.2.2)
>
>   buttons: 393
>   triggers: 108
>   bangbangs: 39
>
>   Max 3.6.2/MSP 1.7 (OS 9.2.2)
>
>   buttons: 232
>   triggers: 46
>   bangbangs: no result


On 15-Jul-2006, at 0:21, Keith McMillen wrote:

> Would someone please run this test on the new Intel hardware and 

Max 4.6b11, OS X 10.4.7, iMac 1.83 GHz Intel Duo, 1GB

While we're here: I've done numerous comparisons with various Litter 

objects, using the patch below to test MSP performance. Compared to 

Rosetta I'm seeing improvements of anything from 10% to ~3times as fast.

(Requires lp.stacey from the Litter Starter Pack)

Peter Castine +--> Litter Power & Litter Bundle for Jitter

Interface Building for |home | chez nous|

Max/MSP Extremely cool |bei uns | i nostri|

> Would someone please run this test on the new Intel hardware and  
> post results.

   Max 4.6b11, OS X 10.4.7, iMac 1.83 GHz Intel Duo, 1GB

   buttons: 21
   bangbangs: 21
   triggers: 33

While we're here: I've done numerous comparisons with various Litter  
objects, using the patch below to test MSP performance. Compared to  
Rosetta I'm seeing improvements of anything from 10% to ~3times as fast.

(Requires lp.stacey from the Litter Starter Pack)


--------------    http://www.bek.no/~pcastine/Litter/    -------------
Peter Castine             +--> Litter Power & Litter Bundle for Jitter
                                Universal Binaries on the way
iCE:  Sequencing,  Recording &
       Interface  Building  for                   |home    | chez nous|
       Max/MSP   Extremely cool                   |bei uns |  i nostri|
       http://www.dspaudio.com/                   http://www.castine.de

What is the improvement from OS 9 to the new intel build?

> On 15-Jul-2006, at 0:21, Keith McMillen wrote:

>> Would someone please run this test on the new Intel hardware and post 

> Max 4.6b11, OS X 10.4.7, iMac 1.83 GHz Intel Duo, 1GB

> While we're here: I've done numerous comparisons with various Litter 

> objects, using the patch below to test MSP performance. Compared to 

> Rosetta I'm seeing improvements of anything from 10% to ~3times as fast.

> (Requires lp.stacey from the Litter Starter Pack)

> #P hidden newex 308 115 69 196617 speedlim 600;

> #P hidden newex 25 114 69 196617 speedlim 600;

> #P hidden newex 115 26 69 196617 speedlim 600;

> #P hidden newex 203 91 99 196617 unpack 0. 0. 0. 0. 0.;

> #P hidden newex 99 91 99 196617 unpack 0. 0. 0. 0. 0.;

> #P comment 0 327 257 196617 Freeware , enjoy!;

> #P number 224 142 35 9 0 0 160 3 0 0 0 221 221 221 222 222 222 0 0 0;

> #P flonum 224 206 35 9 0 0 160 3 0 0 0 221 221 221 222 222 222 0 0 0;

> #P number 224 185 35 9 0 0 160 3 0 0 0 221 221 221 222 222 222 0 0 0;

> #P number 224 164 35 9 0 0 160 3 0 0 0 221 221 221 222 222 222 0 0 0;

> #P comment 215 114 90 196617 Last x seconds (max. 10);

> #Q window flags nogrow close zoom nofloat;

> #P hidden newobj 261 328 59 196617 thispatcher;

> #P hidden newex 386 2 44 196617 onecopy;

barry threw :: sound | (if you would see the stars clearly,

 | look hard at the surrounding darkness)

-- 
barry threw :: sound      |   (if you would see the stars clearly,
http://www.barrythrew.com |   look hard at the surrounding darkness)
bthrew(at)gmail(dot)com   |                       -Ooka Makoto
857-544-3967              |


On 16-Jul-2006, at 4:57, Barry Threw wrote:

> What is the improvement from OS 9 to the new intel build?

With so many factors effecting performance, I'm not sure how 

meaningful any answer is. The new Intel machines have so much more 

raw horse power at their disposal than machines that would typically 

be running OS 9 (or 8.6), that, well, of course you get more fps and 

lower DSP-CPU readings and faster [Uzi 1000000]'s. OS X eats up some 

of that HP (as discussed ad naseum when Max.MSP 4.2 came out), but 

it's a pittance compared to basic VRRRoooom.

The short answer is: compared to a G3/G4 running a Classic Mac OS, 

On 16-Jul-2006, at 4:57, Barry Threw wrote:
> What is the improvement from OS 9 to the new intel build?

With so many factors effecting performance, I'm not sure how  
meaningful any answer is. The new Intel machines have so much more  
raw horse power at their disposal than machines that would typically  
be running OS 9 (or 8.6), that, well, of course you get more fps and  
lower DSP-CPU readings and faster [Uzi 1000000]'s. OS X eats up some  
of that HP (as discussed ad naseum when Max.MSP 4.2 came out), but  
it's a pittance compared to basic VRRRoooom.

The short answer is: compared to a G3/G4 running a Classic Mac OS,  
the Intel Macs are a lot faster.

On 17-jul-2006, at 18:06, Mihalic Alexander wrote:

>> Would someone please run this test on the new Intel hardware and 

MacBook Pro 15" 2.16 2.0 gb ram
osx 10.4.7 max 4.6b11

buttons : 20 (sometimes 10)
bangbang : 21 (sometimes 33)
triggers : 33 (sometimes 44)

>> Would someone please run this test on the new Intel hardware and  
>> post results.

mobile.: +31 6  16614770
home...: +31 10 4112603
fax....: +31 84 7559757
iceland: +354   6626132

dlf@telcosystems.net
www.telcosystems.net

Stadhoudersweg 49B
3038 EE Rotterdam
the Netherlands

Pc (Intel MB with Intel Pentium4 3.20 ghz, Sapphire radeon 9200SE),

Windows Xp SP2, 512 MB Ram, Max/Msp 4.57.

> Would someone please run this test on the new Intel hardware and post results.

Pc (Intel MB with Intel Pentium4 3.20 ghz, Sapphire radeon 9200SE),
Windows Xp SP2, 512 MB Ram, Max/Msp 4.57.

buttons: 11or 0
triggers: 11 or 0
bangbangs: 11 or 0

2006/7/15, Keith McMillen :
> Hi All,
>
> I have been keeping a log of performance data using the same test patch (originaly done by Dave
> Stanford). I include comparisons to other Max versions back to 3.6.2. All tests were on a 1G
> tiBook, with a 512 sample I/O and signal vector size. Overdrive off.
>
> Would someone please run this test on the new Intel hardware and post results.


Can someone please explain what these figures mean? High = good? Low 

In short: is there a performance improvement or hit when I switch to 

Op 18-jul-2006, om 16:43 heeft Alessandro Fogar het volgende geschreven:

> Pc (Intel MB with Intel Pentium4 3.20 ghz, Sapphire radeon 9200SE),

> Windows Xp SP2, 512 MB Ram, Max/Msp 4.57.

>> I have been keeping a log of performance data using the same test 

>> Stanford). I include comparisons to other Max versions back to 

>> tiBook, with a 512 sample I/O and signal vector size. Overdrive off.

Can someone please explain what these figures mean? High = good? Low  
= good?

In short: is there a performance improvement or hit when I switch to  
vs 4.6 on my PPC(g4 & g5)  puters?

Zip Boterbloem
Media Mechanics
Zwaluwstraat 54
2025 VR  Haarlem
The Netherlands
+31627014758
zip@knoware.nl

> Hi,
>
> Pc (Intel MB with Intel Pentium4 3.20 ghz, Sapphire radeon 9200SE),
> Windows Xp SP2, 512 MB Ram, Max/Msp 4.57.
>
> buttons: 11or 0
> triggers: 11 or 0
> bangbangs: 11 or 0
>
> All the best
>
> Alessandro Fogar
>
> 2006/7/15, Keith McMillen :
>> Hi All,
>>
>> I have been keeping a log of performance data using the same test  
>> patch (originaly done by Dave
>> Stanford). I include comparisons to other Max versions back to  
>> 3.6.2. All tests were on a 1G
>> tiBook, with a 512 sample I/O and signal vector size. Overdrive off.
>>
>> Would someone please run this test on the new Intel hardware and  
>> post results.


I don't think these tests are a good benchmark at all. The patch is clearly

not capable of using multiple processors, which make the results of the quad

and intel macs rather useless. My 1 year old pentium M 1.86 is getting

better results than a quad 2.5 or a macbook pro. (0 sometimes 11 on all

tests). The results from these tests tell you something about your L2 cache

I don't think these tests are a good benchmark at all. The patch is clearly
not capable of using multiple processors, which make the results of the quad
and intel macs rather useless. My 1 year old pentium M 1.86  is getting
better results than a quad 2.5 or a macbook pro. (0 sometimes 11 on all
tests). The results from these tests tell you something about your L2 cache
more than anything else.

My understanding is that Max/MSP (unlike Jitter) is not multi-processor capable. Has this changed? If not, these tests do indicate relative scheduler/event performance. If so, could you look to see how the processes are balanced among the cores/CPUs ?

I don't think these tests are a good benchmark at all. The patch is clearly not capable of using multiple processors, which make the results of the quad and intel macs rather useless. My 1 year old pentium M 1.86 is getting better results than a quad 2.5 or a macbook pro. (0 sometimes 11 on all tests). The results from these tests tell you something about your L2 cache more than anything else.

My understanding is that Max/MSP (unlike Jitter)  is not multi-processor capable. Has this changed? If not, these tests do indicate relative scheduler/event performance. If so, could you look to see how the processes are balanced among the cores/CPUs ?

Thijs Koerselman  wrote:
  I don't think these tests are a good benchmark at all. The patch is clearly not capable of using multiple processors, which make the results of the quad and intel macs rather useless. My 1 year old pentium M 1.86  is getting better results than a quad 2.5 or a macbook pro. (0 sometimes 11 on all tests). The results from these tests tell you something about your L2 cache more than anything else.

> My understanding is that Max/MSP (unlike Jitter) is not multi-processor

> capable. Has this changed? If not, these tests do indicate relative

> scheduler/event performance. If so, could you look to see how the processes

I think you're right about max/msp not supporting mp. I have no other

computer to compare my test results with, but I still think it's a strange

that my laptop would outperform a macbook pro running max/msp. I think its a

good idea to put together a more sophisticated benchmark patch. Maybe I have

some time later this week to build one. I am interested in this as well,

because I'm about to buy a new computer soon....and it could finally be a

On 7/18/06, Keith McMillen  wrote:
>
> My understanding is that Max/MSP (unlike Jitter)  is not multi-processor
> capable. Has this changed? If not, these tests do indicate relative
> scheduler/event performance. If so, could you look to see how the processes
> are balanced among the cores/CPUs ?
>

I think you're right about max/msp not supporting mp. I have no other
computer to compare my test results with, but I still think it's a strange
that my laptop would outperform a macbook pro running max/msp. I think its a
good idea to put together a more sophisticated benchmark patch. Maybe I have
some time later this week to build one. I am interested in this as well,
because I'm about to buy a new computer soon....and it could finally be a
mac this time;-)

There's a test by Ali Momeni that's gotten quite a bit of mileage 

http://cnmat.cnmat.berkeley.edu/~ali/zShare/max/

Really for MSP (dsp), rather than Max (scheduler), but maybe you'll 

Also - Thank you, but I shouldn't be given credit for the test being 

used in this thread. It came out of another discussion:

https://cycling74.com/forums/index.php?t=msg&goto=28500

I just modified the patch to work with OS9.

On Jul 18, 2006, at 1:49 PM, Thijs Koerselman wrote:

> I think its a good idea to put together a more sophisticated 

> benchmark patch. Maybe I have some time later this week to build one.

There's a test by Ali Momeni that's gotten quite a bit of mileage  
that's somewhere around:

Really for MSP (dsp), rather than Max (scheduler), but maybe you'll  
still find it useful.

Also - Thank you, but I shouldn't be given credit for the test being  
used in this thread. It came out of another discussion:

> I think its a good idea to put together a more sophisticated  
> benchmark patch. Maybe I have some time later this week to build one.
>


Sorry to be the breaker of bad news, but the original post to this thread from July 14th provided 2 versions of the same patch originally created by David Stanford. If you look at the text files, the second version is missing one line that connects one of the bangs in the first test chain, thus giving very low results. I happend to run both versions simultaneously and had to track this down. I wondered why there was such a variety of results posted for the tests.

Weird how some times cut and passed into the internet can drop information. I've noticed this before in a few other patches posted to the list.

Sorry to be the breaker of bad news, but the original post to this thread from July 14th provided 2 versions of the same patch originally created by David Stanford.  If you look at the text files, the second version is missing one line that connects one of the bangs in the first test chain, thus giving very low results.  I happend to run both versions simultaneously and had to track this down.  I wondered why there was such a variety of results posted for the tests.

Weird how some times cut and passed into the internet can drop information.  I've noticed this before in a few other patches posted to the list.


Max 4.6.0b4 performance measurements