Is this true? Logic and hyperthreading

Getting back to the topic of the thread, I ran the "benchmark" test from the gearslutz poster;

http://www.gearslutz.com/board/music-computers/371545-logic-pro-multicore-benchm arktest-22.html

I got 68 tracks. I turned OFF the hyperthreading and got 74. Huh?

Anyway, apparently 74 is quite a few, considering that every track has spacedesigner, so I guess I shouldn't complain. According to that test, my old machine should have only got around 25.

It's possible that your session is set up in such a way that it's not balancing the load well between the cores. If one instrument maxes out a core, that load can't be shared with another core. HS is definitely a library that has the possibility of doing that.

Do you have each instrument on a separate instance of the plugin? And are you using the "lite" HS patches as opposed to the really huge ones?

Yes and yes. Thank you again for your time, Mike

When I ran the project again without HT it ran better and the machine went ok. I wonder if the new HS 16 bit is available at a discount to buyers of the original? It would probably be fine for most of what I do.

Another update.

Re the original post, on my new 8 core westmere 2.4, logic can use 8 cores, but runs better with hyperthreading off, so I guess that means that HT is not implemented well or at all for 8 cores.

As per my previous post, running the logic "benchmark" test at

http://www.gearslutz.com/board/music-computers/371545-logic-pro-multicore-benchm arktest-22.html

I got a track count of 68 with HT, and 74 without.

Now, here's the chilling statistic. I ran my trusty 2007 MPB 2.2ghz dual core with the same test, and I got 17 tracks. That's 17 tracks, 2 x 1st gen Intel cores at 2.2 ghz

With the new 4th gen Intel westmere 8 core at 2.4ghz, I get 68 out of the box, 74 if I disable HT.

***? 17 x 4 = 68. This seems to suggest that all the c**p I was told at the mac store about not being able to compare the clock speeds of the old gen vs the new gen chips is c**p.

I'm not a computer guru, but if 2 cores on my old MBP at 2.2 can do as much as 2 of the 8 cores on the new machine AND RUN THE SYSTEMS, my new whizzbang machine suddenly looks slow, and in fact is choking with ONE instance of Hollywood Strings. CPU speed is CPU speed, I would say, and the 2.4gig westmere is a comparative dud.

Or am I wrong with my maths?

QtheMusic wrote:
With the new 4th gen Intel westmere 8 core at 2.4ghz, I get 68 out of the box, 74 if I disable HT.

***? 17 x 4 = 68. This seems to suggest that all the c**p I was told at the mac store about not being able to compare the clock speeds of the old gen vs the new gen chips is c**p.

I'm not a computer guru, but if 2 cores on my old MBP at 2.2 can do as much as 2 of the 8 cores on the new machine AND RUN THE SYSTEMS, my new whizzbang machine suddenly looks slow, and in fact is choking with ONE instance of Hollywood Strings. CPU speed is CPU speed, I would say, and the 2.4gig westmere is a comparative dud.

For VI's clockspeed is the most important factor as a VI runs as a single thread. And... in case you haven't noticed. Apple generally doesn't know diddly about Logic or how it works in regards to resources, nor much about how Logic runs under various configurations.

pancenter-

Is there a way to keep HT turned off?

Bump?

QtheMusic wrote:
I wonder if the new HS 16 bit is available at a discount to buyers of the original? It would probably be fine for most of what I do.

All owners of the full version of HS get a free second install. Now that the lite version is announced, that second license can be for the 16 bit version and get the disks for the cost of shipping.


You can't directly compare machines by multiplying by the number of cores - no 8 core machine is going to get 8 times the speed of a one core machine, that's just the nature of how multithreading works. If you want a more direct comparison per core, use the same tool you used to turn off HT and run the test with only 2 cores enabled on your MP (and do it twice, with HT on and HT off).


And your system choking with ONE instrument of Hollywood Strings isn't a problem with your machine, it's a problem with HS. Plenty of users with the fastest machines are struggling to get the library working without issues, that's the whole reason they released a 16 bit version so fast (and gave all users a second license, they realize that many people will need to split it over two machines to really use it).

djanthonyw wrote:
Is there a way to keep HT turned off?

None that I am aware of. It's a pain, but as far as I know you have to open the control panel after every time you reboot the machine.

I don't want to start a riot here, but I thought I would let you know, Mike, that East West have allowed me to have an extra license for HS Diamond (for a second computer in the future) AND a license for HS Gold, for only the shipping cost of the disks. Wow! Problem solved... kinda. They are nice guys after all!

Mike Connelly wrote:
And your system choking with ONE instrument of Hollywood Strings isn't a problem with your machine, it's a problem with HS. Plenty of users with the fastest machines are struggling to get the library working without issues,

What did they develop and test on anyway?

pancenter-

Who knows. It does sound like their stuff generally runs better on PC than mac overall, so maybe if they catch their mac development up someday...

You can try adding the kernel flag cpus=1 to your com.apple.boot.plist - this file can be found in Library/Preferences/SystemConfiguration
Edit at your own risk though!

Won't that disable one of the cores, not the hyper-threading?

I am not sure about it. I cannot test it, I have no HT on my machine. Maybe you should try other (Apple) forums for a better answer.
You have 2 cores with hyperthreading? cpus=2 might do it... I assume the HT-cores will be 'turned off' first by this kernel flag, so you won't get a machine only running on HT cores...