The No.1 Website for Pro Audio
i7 Nehalem Hyperthreading in DAW Performance
Old 2nd September 2009
  #1
Lives for gear
 
TAFKAT's Avatar
 
🎧 15 years
Cool i7 Nehalem Hyperthreading in DAW Performance

Note : 9 November 2009 : This thread originally started out focused on Cubase, but has now expanded to include a host of other DAW applications, including Logic, Protools, etc.

Hey All,

I initially posted this at the Cubendo Forums specifically in relation to i7 Hyperthreading Performance with Cubendo , and Steinbergs official cross platform stance which is causing more than a little confusion.

I'll open up the discussion here as well to those using other DAW applications on these systems across both Windows and OSX , so those that have direct experience please drop a quick note with some detail on the System, O.S, , Host DAW Application used , and whether you experience better performance with HT ON or Off .

Information specific to Cubendo - Cross Platform below:

Quote:
So how does Hyper-Threading affect the performance in our applications?

We have found that - depending on the system environment - having Hyper-Threading (HT) enabled can lead to performance issues and spikes in the VST performance meter. At very low latencies, even dropouts may occur.

* If you are running a single Pentium 4 CPU system, enabling Hyper-Threading results in an increase of processing power.

* If you are running a Xeon HT multi-processor system (not all Xeon models support HT) or a Pentium Extreme Edition dual-core system, you should disable HT in the BIOS.

* The same applies to the new Intel i7 ("Nehalem") processors that re-introduced Hyper-Threading. We are currently investigating using an Intel i7 with HT enabled in conjunction with our sequencers. As long as our tests are still ongoing, we recommend turning off Hyper-Threading if any performance issue appear.
From Steinbergs New Knowledge Base Article Here ( Now requires a MySteinberg Login In to access )

Quote:
Originally Posted by Ekhard Doll
No, Hyper-Threading and real time audio just don't get along too well as it is now and thus there won't be much improvement on Hyper-Threading.

Nevertheless, we will continue our research on this topic in cooperation with Intel but based on the results we have, our recommendation to disable Hyper-Threading will still be valid with upcoming versions of Cubase and Nuendo.....

..... there's no specific plug-in that has issues with HT. It's rather a general issue that HT almost all the time has a negative effect on the overall performance when it comes to realtime audio. Especially if you what to run on very low latencies around 64 samples, turning off HT might do the trick.


.......It is not prohibited to use HT and there's no death penalty either. If you think that HT works for you, then leave it on. Cubase works with Hyper-Threading turned on.

However, our tests which were made by development showed that HT decreases the performance in most scenarios (meaning real life projects).
_________________
Support Manager
Steinberg Media Technologies
From Cubase.net Here:

Time to put this to bed.

With Steinbergs continued stance that Hyperthreading on the current i7's needs to be disabled , which is in direct contradiction to the experiences reported by i7 users on this forum, as well as several Professional DAW builders , not to mention other DAW Host Developers , I have decided to take the challenge up to Steinberg to give us quantifiable evidence why they are maintaining this line.

Hyperthreading on the current I7's has absolutely no correlation what so ever to the earlier Netburst architecture , so if as Steinberg claim they have quantifiable and qualified evidence , in collaboration with Intel ( or so they claim ) that HT on the current systems is in any way detrimental to real time audio application, lets see the fine detailed evidence.

Mr Doll has clearly stated they have some tests provided by development, why not make them publicly available , easy, we can then collectively increase the data base on the real state of play with HT, obviously they are seeing something that the vast majority of end users and DAW builders are not .

I have volumes of quantifiable and qualified test results to show that HT on the current i7 architecture scales incredibly well using Cubendo on Windows , on OSX its still a mixed bag , but that does in no way mean that Steinberg can maintain a blanket stance on the cross platform performance.

Detailed Cubendo / i7 & Nehalem Xeon Performance reports Here and a summary Here

What is it that they are still needing to investigate, i7's systems have been in the wild since November 2008 ?

They need to amend the information in greater detail to reflect a more accurate state of play in regards to the cross platform performance , or provide the evidence to support their current stance on HT performance on Windows at least.



V:
Old 3rd September 2009
  #2
Lives for gear
 
🎧 15 years
bump
Old 4th September 2009 | Show parent
  #3
Gear Addict
 
🎧 20 years
Quote:
Originally Posted by jcschild ➑️
bump
But I'm assuming you've been benchmarking the i5 750 without HT, anyway. Just the NDA's got your tongue, right?
Old 4th September 2009 | Show parent
  #4
Lives for gear
 
TAFKAT's Avatar
 
🎧 15 years
Cool

Quote:
Originally Posted by Gone Fission ➑️
But I'm assuming you've been benchmarking the i5 750 without HT, anyway.
Not sure what your point is ?

i5/750 has no HT , so is not relevant to the focus of the discussion.

Old 4th September 2009 | Show parent
  #5
Lives for gear
 
Immersion's Avatar
 
🎧 10 years
From most sources I have heard HT is recommended for the i7 platform. I have never tried to disable HT, but would be interesting to see if the DPC Latency will change.. I assume in that case it would decrease...I still have some problems when using 32 samples...a few click sounds from time to time... not sure if it could be HT related...

It might be so that you have prioritize , better multitasking (HT ON) or better ASIO latency (HT OFF).
Old 4th September 2009 | Show parent
  #6
Lives for gear
 
Jovas's Avatar
 
🎧 10 years
It doesn't matter wheter I enable or disable it in Cubase 4 or 5 with a Core i7 920 processor, I get the same performance.

I'm not sure wheter that's good or bad..
Old 4th September 2009 | Show parent
  #7
Lives for gear
 
TAFKAT's Avatar
 
🎧 15 years
Cool

Quote:
Originally Posted by Jovas ➑️
It doesn't matter wheter I enable or disable it in Cubase 4 or 5 with a Core i7 920 processor, I get the same performance.

I'm not sure wheter that's good or bad..
Hey Jovas,

How are you gauging that the performance is the same ?

Old 4th September 2009 | Show parent
  #8
Lives for gear
 
Jovas's Avatar
 
🎧 10 years
Quote:
Originally Posted by TAFKAT ➑️
Hey Jovas,

How are you gauging that the performance is the same ?

I watch my ASIO and processor usage(taskmanager) while playing back a track(s) with lots of plugin's and stuff going on(verbs, compressors, eq's etc)

Cubase 5.1 and 4.5.2, Focusrite Saffire LE, input latency: 5.6ms, output 9.5ms. Latency's in Windows 7 are lower, but still not impressive. It does the trick for me, I'm just a humble home techno producer

I have other problems with Cubase, like skips during the recording of a vinyl dj mix..
Old 4th September 2009 | Show parent
  #9
Lives for gear
 
🎧 15 years
I have an 8 core 2.26 Nehalem Mac and w/ OS 10.6 and Logic 8.02 I get about 40% better performance with HT turned off.
Old 4th September 2009 | Show parent
  #10
Lives for gear
 
🎧 15 years
yes we all know Apple OS sucks for HT
Snow leopard is not showing to be much improved either...
not enough time in front yet to give a complete comment.

Scott
ADK
Old 4th September 2009 | Show parent
  #11
Lives for gear
 
TAFKAT's Avatar
 
🎧 15 years
Cool

Quote:
Originally Posted by Jovas ➑️
I watch my ASIO and processor usage(taskmanager) while playing back a track(s) with lots of plugin's and stuff going on(verbs, compressors, eq's etc)

Cubase 5.1 and 4.5.2, Focusrite Saffire LE, input latency: 5.6ms, output 9.5ms. Latency's in Windows 7 are lower, but still not impressive. It does the trick for me, I'm just a humble home techno producer

I have other problems with Cubase, like skips during the recording of a vinyl dj mix..

Hey Jovas,

Thanks for your time and energy .. :-)

Trying to quantify scaling via the ASIO meter and TM is not very accurate, as there are many variables involved and in a lot of cases the 2 do not correlate, especially at lower to moderate session loadings. You could easily get the same ASIO / TM visual feed back between the same session running on a 4 physical vs an 8 physical core system I have found.

Its easy to quantify the variance using incremental DAWbenching sessions such as my DAWbench DSP or Blofelds DSP I/II Test sessions , but the argument will be raised how they correlate to Real World sessions. Of course what a RW session is to one, it isn't for another , so its not so easily quantified.

Where the the benefits of the higher core count ( physical or virtual) in RW sessions will become apparent is if you have a very heavy session running on the system with HT ON, and then switch HT off, you will find that the session will not run at all, and this has been a given in Cubendo on Windows , while OSX unfortunately has a few more minefields to navigate..


Which leads me to...

SFTPH,

Thanks for the feedback with Logic 8.0 , much appreciated.

With Cubendo we have known that MP scaling performance on OSX has not been comparable for many years, the Intel Macs even before the new Nehalems would not scale accordingly either, and thru some investigation we ( we meaning myself and some high end Cubendo cross platform users ) found that a lot of the scaling issues are caused by the inability of OSX to be able to arbitrate the EIST ( Intel Speed Stepping ) correctly.

OSX does not give any indication that the Core Clock speed is constantly ramping up and down even under heavy loads, which is causing significant scaling issues, of course not openly obvious to those without the kernel level applets to monitor what is really going on under the bonnet. There is a second variable involved called C1 Halt state, which is another low level thermal/clocking protocol that is used by the Intel chips , which also throttles clockspeed independent to the EIST.

Both of these protocols can be easily disabled at BIOS level on a PC , but Apples EFI ( their version of a BIOS ) is simplistic and far less accessible , and cannot control any of these protocols at that level. In Windows EIST is arbitrated correctly, so even if the EIST and C1 Halt state are active the systems still scale quite well. This is evidenced when comparing cross platform scalability on a new Nehalem Mac using an incremental benchmarking session.

Some results below:



There is also another issue with OSX Mutithreading across any more than 6 physical/virtual cores , something again evidenced and reported on the Logic benchmarking thread here , where Logic is favouring on 6 cores. Apple obviously know where the bodies are buried with the low level API's, and are trying to navigate around them, its interesting that you have reported that Logic is also suffering scaling issues with the Nehalem HT , so obviously there is some more work to do there. Hopefully SL improves things there, as I have read reports at C.net that the scaling is substantially better on SL with Cubendo , but still nowhere near on par with Windows.

Old 8th September 2009 | Show parent
  #12
Lives for gear
 
TAFKAT's Avatar
 
🎧 15 years
Cool

Some More fuel for the fire,

First off some detail on the test session, Cubendo DSP is a new test I developed purely for cross platform shootouts with Cubedo , using Steinbergs own VST3 native plugins. The Multiband Compressor in this instance . This will ensure the most consistent cross platform performance for Cubendo across Windows and OSX . Its essentially the audio data from DAWbench DSP Universal and the plugins from Blofelds II DSP.

The Systems tested :

System 1: The i7 920 test system is my tried and true development system which has been used for all previous i7 testing, but I have also configured an OSx86 10.5.8 install. Unlike a regular MAC, this system allows control of EIST and C1 Halt at BIOS level, which will eliminate those variables from the equation, leaving only OSX and Cubendo. All Windows testing has been on XPSP3 on this run, Win7 64 coming later.

O.K, the numbers..

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

System 1 :

i7 920 - 2.66 GHZ : X58 : 3GB DDR3 :OSx86 - 10.5.8 : Cubase 5.01 : RME AIO : Driver : 2.67 :

032 : HT OFF - 10 MBC / HT ON - 15 ( Hard to get a results as it was spiking and clipping really early on , haven't seen that behavior since pre Core 2 on windows )

064 : HT OFF - 40 MBC / HT ON - 48 MBC

128: HT OFF - 57 MBC / HT ON - 82 MBC

256 : HT OFF - 70 MBC / HT ON - 95 MBC


i7 920 - 2.66 GHZ : X58 : 3GB DDR3 : XPSP3 : Cubase 5.01 : RME AIO : Driver : 3.081 :

032 : HT OFF - 96 MBC / HT ON - 113 MBC

064 : HT OFF - 108 MBC / HT ON - 128 MBC

128: HT OFF - 114 MBC / HT ON - 140 MBC

256 : HT OFF - 124 MBC / HT ON - 153 MBC

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

System 2 :

2009 MAC PRO : Dual Nehalem Xeon E5520 : 2.26 GHz : 6GB RAM : OSX 10.5.8 : Nuendo 4.3 : RME MADIe : Driver 2.67

032 : HT OFF - 12 MBC / HT ON - 10 MBC

064 : HT OFF - 26 MBC / HT ON - 22 MBC

128: HT OFF - 61 MBC / HT ON - 64 MBC

256: HT OFF - 78 MBC / HT ON - 119 MBC


2009 MAC PRO : Dual Nehalem Xeon E5520 : 2.26 GHz : 6GB RAM : OSX 10.6.0 : Nuendo 4.3 : RME MADIe : Driver 2.67

032 : HT OFF - 38 MBC / HT ON - 42 MBC

064 : HT OFF - 87 MBC / HT ON - 120 MBC

128 : HT OFF - 107 MBC / HT ON - 152 MBC

256 : HT OFF - 126 MBC / HT ON - 160 MBC




Some interesting points,

Its obvious from the i7 results that the performance of C5 is not even on the same page between Windows and OSX , there is something seriously amiss on OSX , even with the EIST/C1Halt state variable removed. At first I was questioning whether my OSx86 configuration was not performing to spec, until I compared my OSx86 results with the MAC Pro results, which then placed them in perspective.

In a nutshell, the i7 920 in OSX 10.5.8 is in a lot of cases outperforming the MAC Pro Dual Nehalem, the EIST and C1 Halt control is a major factor there. So to be clear, even with 1/2 the cores/threads, the i7 is outperforming the official hardware, which is being crippled by the EFI implementation not handling the Intel thermal / clocking routines well at all.

Once we get past that , we then have OSX's MP task scheduling and threading in 10.5.8 tripping over itself, with HT On or Off , and the great white hope of Snow Leopard, although substantially better as displayed, IMO opinion simply polishing a turd.

To top it off Cubendo's ASIO efficiency on OSX sucks compared to Windows, so add it all up , and we have a pretty embarrassing state of affairs.

We still have some more test sessions to run thru that will cover a wider variable in regards to plugins, etc , but I really can't see any significant change to the cross platform results.

In short Steinbergs cross platform stance on Multiprocessor Performance & Hyperthreading is highly inaccurate , and needs to be clarified.

Logic test results coming soon, I am in the process of navigating the testing methodology to Logic8/9.

To Download the Test Session Files :

Original DAWbench DSP session with Audio Files : Here

Cubendo DSP Session File : Here

Both downloads are needed , as Cubendo DSP uses the audio files of DAWbench DSP , simply drop the Cubendo DSP session file into the DAWbench project folder and launch the session.

Full instructions on the methodology used to run the tests in the DAWbench DSP download file.

Old 8th September 2009 | Show parent
  #13
Lives for gear
 
🎧 15 years
"To top it off Cubendo's ASIO efficiency on OSX sucks compared to Windows"

Who's fault is this - Steinberg or Apple?
Old 8th September 2009 | Show parent
  #14
Lives for gear
 
🎧 15 years
Why don't Steinberg use core audio like everyone else on OSX?
Old 8th September 2009 | Show parent
  #15
Lives for gear
 
TAFKAT's Avatar
 
🎧 15 years
Cool

Quote:
Originally Posted by Gordon ➑️
"To top it off Cubendo's ASIO efficiency on OSX sucks compared to Windows"

Who's fault is this - Steinberg or Apple?
I suspect that fingers will be pointed at each other there.

RE Core Audio / AU over ASIO/VST performance , I don't have a good grounding in that area just yet to be honest , but I am hoping to do a lot more OSX based testing in future, and I will be focussing on Logic and DP as well, so hopefully I can get a clearer understanding on how the formats compare .

We can also throw RTAS into the mix here, but I think we all know the answer there already...


V:
Old 8th September 2009 | Show parent
  #16
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by TAFKAT ➑️

We can also throw RTAS into the mix here, but I think we all know the answer there already...


V:
Please enlighten those of us who are less informed.
Old 9th September 2009 | Show parent
  #17
Lives for gear
 
TAFKAT's Avatar
 
🎧 15 years
RTAS is known not to scale as efficiently as VST/AU.
Old 9th September 2009 | Show parent
  #18
Lives for gear
 
nikki-k's Avatar
 
🎧 15 years
RE: RTAS
Digidesign has officially and publicly acknowledged that issues with the RTAS aspect of Pro Tools do exist. They have stated it is going to take some time to resolve, as the code has been around for quite awhile, and with advances in OS, hardware, etc.. well, they said that the short answer is: the problem is deep, and will require examination first to determine what sections of code are affecting these issues, the best way to resolve those pieces, and how to provide for the future while "fixing" the code. I heard rumors of one year minimum to expect any headway... and that was months ago. VI's are the most *obviously* affected.

Remember HTDM? Hehehehe...

Pro Tools is a very interesting piece of kit. One way to see it is as a "hybrid" of sorts. It bridges "stand alone DAW" solutions (such as, say, RADAR), and completely native solutions (such as SAW). PARIS was *similar* IMO. "Consumer OS controlled expandable hardware DSP DAW system" might be one way to think of it. And then sequencer apps (Cubase, Logic, etc) added audio (DAW) capabilities. So, DIgi answered with sequencer/MIDI capabilities. And then offered native solutions to provide (initially) for "pros" to "bring their work home." AFAIK, no fully native coding for PT was ever produced; it was an adaptation, and resources to recode from ground up to provide for this.. well.. Thus, the native side has been a constant game of examine, perform surgery, recover, repeat.

Sorry- I find all this fascinating, and will babble on and on....
Old 9th September 2009 | Show parent
  #19
Lives for gear
 
nikki-k's Avatar
 
🎧 15 years
I will admit I am pretty ignorant of the testing being used here. That said..

I think it would be cool to have a utility that could emulate certain Dawquencer usage scenarios.. select an emulation type, execute, and it produces a report, inclusive of hardware, OS, drivers, etc. Once armed with this data, one could then fire up their fav DAWquencer, and utilize various testing scenarios with as many plug-in constants as possible (VI's available in all formats, buffer settings, etc) and compare.

For example: with the "Generic DAWquencer Utility," one could set it to emulate what, say, an orchestral composer might utilize while working on a pretty hardcore session, IF they were afforded a DAWquencer app (and plug-ins/VI's) that was written really well/*properly*. Thus, a solid baseline per hardware configuration could be established. Or, set the emulation for a hardcore mixing scenario, with lots of automation, dense edits, etc. Once optimum hardware/OS configs are discovered...

Armed with that "report," one would then load up that same "scenario".. say, the aforementioned hardcore orchestral composing session.. in their DAWquencer of choice (Cubendo, Logic, etc), and produce a report for that DAWquencer for that particular scenario. With both reports in hand, couldn't the shortcomings of hardware, OS, app, etc then be determined? Or have I over-medicated once again, and am now simply spouting nonsense in an incoherent manner? (unfortunately, this does happen with me.. honestly, it does; and I apologize. But I am just trying to figure out why apps/plug-ins in the audio world- in general- seem to be.. "broken"? Is it app devs? Plug-in devs? The hardware? Use of a "generalized use consumer" OS? Something else? All of these things?)
Old 28th September 2009 | Show parent
  #20
Lives for gear
 
Timur Born's Avatar
 
🎧 10 years
Quote:
Originally Posted by TAFKAT ➑️
OSX does not give any indication that the Core Clock speed is constantly ramping up and down even under heavy loads, which is causing significant scaling issues, of course not openly obvious to those without the kernel level applets to monitor what is really going on under the bonnet. There is a second variable involved called C1 Halt state, which is another low level thermal/clocking protocol that is used by the Intel chips , which also throttles clockspeed independent to the EIST.
There are workaround for this that enable you to get rid of throttling on OS X.

1. A combination of Coolbook + removing the OS X system kext that is responsible for Speedstep works quite good and without the need to create artificial load.

2. Creating artificial load (using the "Load" slider in Coolbook is a good tool for that) is the most performant way, but obviously creates load/heat. So in most cases 1. is preferable.

Different applications also behave differently to thermal clocking. Logic is far less affected than Ableton Live for example.

Last but not least Windows applications can just as much be affected as OS X applications. It depends on the app and circumstances.
Old 28th September 2009 | Show parent
  #21
Lives for gear
 
🎧 10 years
the reason steinberg recommend not using hyper threading is not a performance issue but a stability/quality issue. hyperthreading loads the cpu as if it has double the resources available to it. this means the scheduler can keep requesting data to be worked on in some circumstances and it improves performance by keeping the pipeline full.

some instruction/data dependencies work well with hyper threading and some do not. my guess is steinberg is seeing something wrong with test data run through the software under hyperthreading and is working on a patch to correct what they see as an issue. the performance of an i7 with hyperthreading off is still very good and the difference is like 80% at most with something that can fully utilize hyperthreading and -80% if it can't do so very well.

mountains out of molehills here really. the performance is there but something is coming out wrong in comparison to not having hyperthreading on.
Old 28th September 2009 | Show parent
  #22
Lives for gear
 
strauss's Avatar
 
🎧 15 years
Quote:
Originally Posted by newrigel ➑️
. The Mac will be the first consumer computer to go completely 64 bit and a FULLY functional 64 bit because they will release LOGIC 64 with an audio engine that "just works" he he...
actually 64 bit is one thing OSX is really behind on...just look at the facts.
I for one really appreciate all the research people like Tafkat does as real-time audio perfomance usually isn't that high on the priority list for system developers and there's soooo much ignorance and bullshit spread on the net.
Old 28th September 2009 | Show parent
  #23
Gear Nut
 
newrigel's Avatar
 
🎧 15 years
Quote:
Originally Posted by TAFKAT ➑️
Some More fuel for the fire,


In a nutshell, the i7 920 in OSX 10.5.8 is in a lot of cases outperforming the MAC Pro Dual Nehalem, the EIST and C1 Halt control is a major factor there. So to be clear, even with 1/2 the cores/threads, the i7 is outperforming the official hardware, which is being crippled by the EFI implementation not handling the Intel thermal / clocking routines well at all.
Your using a HACKED version of OS X... tutttutttutt but really, the MP stays right there with it (as far as I can tell by your color coding?) and with a LOT less headaches to boot and I get SUPPORT!!!
I'll stay all Apple thanks! WAY less headaches to get just 1-2% more performance. I'm a musician and if the computer starts to get in the way of my progress I'll just go completely analog thank you!

Your preaching to the choir! Man, SL JUST came out... how long has Vista been out and now W7 is going to save the world... The Mac will be the first consumer computer to go completely 64 bit and a FULLY functional 64 bit because they will release LOGIC 64 with an audio engine that "just works" he he... Then, all the other DAW applications will follow to keep market share.
And then, you'll all start saying "we need more power" then it all starts over again...
Just put your energies into that next great composition or performance as far as music is concerned. Having hyper threading doesn't make me a better musician or running 500 plugins @ 192 k don't make for better records. The more you get the more you want... it's funny but back in the day, artist produced epic albums on 4 tracks man! So I think anything over this is pure overkill!
I haven't seen the DSP overload YET with my new MP... not even ONCE! These projects locked my old MP up!
Old 28th September 2009 | Show parent
  #24
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by Timur ➑️
Last but not least Windows applications can just as much be affected as OS X applications. It depends on the app and circumstances.
But on a PC you can turn off the EIST and C1E halt state in the BIOS, where as the MAC PRO EFI does not allow this.


Not talking about Timur here but there is a lot of miss information going on in this thread, the performance difference is quite large without HT and has anyone here or on any other thread that TAFKAT started around the web found a real world problem with HT? Steinberg don't have a magic test what they have is an OSX machine causing problems and a face they want to save.

And newrigel, fully 64bit DAWs have been around for years on Windows so I think Logic will have a hard time being first, the 64bit issue isn't the DAW it's all the 3rd party plugins and no the great Logic can't solve that! Also just because you work in one way doesn't mean all people do, work on Film and TV you can use all the power available, it's not about being overkill. A lot of the stuff i do I can work with a 5 year old computer wouldn't matter but that doesn't hold true for all, I know some people who NEED 24GB+ of RAM and dual Quad CPUs with the HT turned on.
Old 28th September 2009 | Show parent
  #25
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by newrigel ➑️

. . . how long has Vista been out and now W7 is going to save the world... The Mac will be the first consumer computer to go completely 64 bit and a FULLY functional 64 bit because they will release LOGIC 64 with an audio engine that "just works" he he...
SONAR's 64-bit Audio Engine has "just worked" for years and the 8.5 version "Just works" awesome right now. It's the plugin makers who need get their act together though the $20 Jbridge application seems to work with everything.
Old 28th September 2009 | Show parent
  #26
Lives for gear
 
🎧 15 years
Quote:
Originally Posted by newrigel ➑️
I'll stay all Apple thanks! how long has Vista been out and now W7 is going to save the world... The Mac will be the first consumer computer to go completely 64 bit and a FULLY functional 64 bit because they will release LOGIC 64 with an audio engine that "just works" he he...
yeah so hows thats working out so far??
Snow still has major issues!
Logic 64 when did that release? oh right it hasnt yet...

as mentioned Vista 64 has been fine for awhile now. its the 3rd party apps that are an issue.

look the point of this thread is about core(s) Hyperthreading load balancing. not 64 bit software.

so lets stay on track eh?
and the point of using the hackin.... was to show the issues there are evident within a mac.

Scott
ADK
Old 28th September 2009 | Show parent
  #27
Lives for gear
 
Timur Born's Avatar
 
🎧 10 years
Quote:
Originally Posted by warmer ➑️
But on a PC you can turn off the EIST and C1E halt state in the BIOS, where as the MAC PRO EFI does not allow this.
Problem is that with some boards/BIOS turning off EIST leads to the CPU being fixed to its lowest multiplier, not its highest (same happens if you just remove the OS X Speedstep kext without using Coolbook).

And some of us use Windows on Apple hardware, which turns out to be problematic in some cases where dynamic CPU clocking kicks in.

Concerning Hyperthreading I wouldn't care too much for the theoretical benefits and drawbacks. Just try it out. My understanding is as follows:

One of the possible problems is that DAW applications expect every core to deliver the same performance which hyperthreaded "virtual" cores cannot always deliver. That means if a DAW load a real core it expects the second core to be free as long as it doesn't make use of it (not taking background processes into account).

With HT loading one core can also load the second "virtual" core, which is why you don't get 200% out of Hyperthreading but only 30+%. But if you are running 2 tracks full of identical effects than you might 200% aka 2 "real" cores.

Other applications than DAWs usually can live with "as fast as possible while dynamically changing", but for DAW users it makes finding the breaking point harder because it kind of behaves likes a CPU that has a changing maximum clock-rate/performance.

So if the breaking point (where audio output keeps stable and free of dropouts) is higher for you with HT on then turn it on, else turn it off. That breaking point may vary with different projects in different apps though.
Old 28th September 2009 | Show parent
  #28
MCJ
Here for the gear
 
🎧 10 years
Quote:
Originally Posted by newrigel ➑️

Your preaching to the choir!

No, he's not. Long time Mac user and I'm reading TAFKAT's tests with huge interest. Thanks TAFKAT.

Quote:
Originally Posted by newrigel ➑️
Man, SL JUST came out... how long has Vista been out and now W7 is going to save the world...
Nah, it's 10.5 and SL vs. old school XP:

Quote:
Originally Posted by TAFKAT

"All Windows testing has been on XPSP3 on this run, Win7 64 coming later."
Quote:
Originally Posted by newrigel ➑️

And then, you'll all start saying "we need more power" then it all starts over again...
Just put your energies into that next great composition or performance as far as music is concerned. Having hyper threading doesn't make me a better musician or running 500 plugins @ 192 k don't make for better records. The more you get the more you want... it's funny but back in the day, artist produced epic albums on 4 tracks man! So I think anything over this is pure overkill!
I haven't seen the DSP overload YET with my new MP... not even ONCE!
Cool.

Quote:
Originally Posted by newrigel ➑️

These projects locked my old MP up!

Zacktly... sometimes you need a faster machine.


TAFKAT, looking forward to your future tests.
Thanks,
MCJ.
Old 30th September 2009 | Show parent
  #29
Gear Head
 
🎧 15 years
Awesome awesome stuff guys...I love anandtech and tomshardware's reviews, but they don't really have any serious tests for pro audio apps. These comparisons are amazing and looking forward to your future tests.

also to newrigel.....mac fanboy rage much?
Old 30th September 2009 | Show parent
  #30
Lives for gear
 
mykhal c's Avatar
 
🎧 10 years
also...the question becomes...at what point does an app that is NOT efficiently coded to take advantage of the multiple/logical cores gain more of an advantage on sheer clock speed and a wide open BCLK. and if that is the case then one will/would be able to gain more of an OC with HT turned off (along with other settings changes too) to be able to achieve this. like Timur said, you just gotta try it/test it
πŸ“ Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 4622 views: 645230
Avatar for Gam_e_oveR
Gam_e_oveR 2 weeks ago
replies: 15929 views: 1592430
Avatar for Ragan
Ragan 11th January 2019
replies: 443 views: 58399
Avatar for Neil Martin
Neil Martin 14th February 2021
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump