Logo Logo
Adobe Premiere Pro CS5 und Hyperthreading - ist es sinnvoll, oder nicht?

Adobe Premiere Pro CS5 and Hyperthreading - it makes sense or not?

Frage von makiHD:
Januar 2011

After extensive search on the net I've found that it really does not call that exactly with hyperthreading employed in DV editing.

The following sources, I can make that say something at least to some extent:


-.> "How can such a rendering software to divide the work into two threads and assign each thread a different part of the picture In a multiprocessor system or a Hyper-Threading CPU for example, could a CPU to calculate the upper half of the image and independent of other the bottom of it. "

b) http://videoreisetagebuch.wordpress.com/2008/11/19/hintergrundwissen-intel-core-i7-and-powerdirector-7/

c) http://forum.egosoft.com/viewtopic.php?t=283750

Of course it depends on whether you have a true quad core or dual core with HT has.

The latter is the case for me, it tinkers with an Intel Core i5 560M 2.67 GHz in the notebook.

This turbo that comes on the 4 "cores" on a permanent 2.93 Ghz.

I would be interested to now how this CPU would cut vs. a "real" quad-core CPU.

Of course I know that a true quad core is faster, but the question is how much faster?

Here are three sample times CPUs:


The i7 is the Cinebench Multi nearly twice as fast.

The question is whether the transfer can be synonymous to Premiere.

I'm on this question because of the fact became interested because the rendering speed when changing from the Intel Core 2 Duo T9500 with 2 x 2.6 Ghz 6 MB L2 cache on the Intel Core i5 560M 2.67GHz HT MB L3 cache has multiplied 3 about the 4-fold.

This suggests that HT in PP but very, very much does.

Would love to test this, the question is: how?

s.Hat anyone have a notebook with a really fast dual-core CPU, such as a T9500 and could test the times?

b) Do you think you can, the simulated cores by the i5 - msconfig.exe>, turn off the number of CPUs to 2 temporary?

The question is dependent on whether the simulated order to test both CPUs are in use or a physical and a logical processor. It might at worst, only one physical core with a virtual core remain just that I want but do not test well ...

c) yes Programs benefit of HyperThreading, if they are massively parallel, so have many threads. In Task Manager, I could, the column "threads" can show. A high "thread" value would be a good evidence of suitability for HT, right?

d) Ideal would be a standard - AVCHD file that need to transcode some users at all times. I would file this one of a Panasonic HDC-SD707 out-of-cam "to deploy.

e) What are other proposals to illustrate the benefits of HT CPUs?

Namely, the reason is simple:

i5 is a core as dual core + HT significantly cheaper and more energy efficient (and thus cooler!) as a true quad-core.

If that is so extreme PP of HT may actually benefit (which is my guess), then it would almost make more sense to take one i5. Just cheaper, cooler, quieter, long life energy-and money-saving, but just barely slower - IF PP just massive benefits of HT.

Looking forward to an interesting thread that as a starting point in the network for the Comparison vs Dual Core & HT. could be Quad-Core:)

Antwort von tommyb:

A true quad-core is on a similar architecture and clock rate twice as fast. But it is on the software, how they can deal with multiple cores. Programs use two words a few core to the full. But there are four cores available, these are indeed used, but possibly only 80%.

Hyper Threading brings in optimized applications as well, although the performance gain is quite small (at most 10%). Together with HT increases in the rest of synonymous but the power consumption and the temperature (i7 920 then tinkers with 5 ° C more in the peak).

You should disable Hyper Threading in the BIOS the way.

Antwort von rudi:

"MakiHD" wrote:
After extensive search on the net I've found that it really does not call that exactly with hyperthreading employed in DV editing.

Our new Sandy Bridge System is already ordered and that is one of the key questions we want to clarify as soon as possible themselves. So something to hold out ...


Antwort von PeterB:

Interesting topic to me synonymous busy. For me, IT WORKS but felt the question whether i5 i7 2500k or 2600K and whether the value added to make.
@ Rudi: when is likely to you that 1 week, 2 or more?


Antwort von makiHD:

You are, that is obviously of interest:)

I have again found benchmark values from the old T9500 and compared with the i5 560m. The i5 renders consistently about 3 to 3.5 x faster than a T9500.

For about 30 seconds into the AVCHD Full HD 1920x1080 AVCHD video standard profile in PowerDirector (Trial on the fly) of the i5 560m takes about 80Sekunden.

My T9500 needs for such a thing about 240s.

But there is still a very interesting test with respect to the T9500:


I will try to recreate as many of the test scenarios and then to post comparisons, particularly in the Comparison synonymous when I "limiting" with me via msconfig.exe only 2 cores.

Antwort von Britta Leuchner:

"Peterb" wrote:
Interesting topic to me synonymous busy. For me, IT WORKS but felt the question whether i5 i7 2500k or 2600K and whether the value added to make.
@ Rudi: when is likely to you that 1 week, 2 or more?


Hoffentl I will. soon be able to tell ;-)

Antwort von PeterB:

"Britta Leuchner" wrote:
"Peterb" wrote:
Interesting topic to me synonymous busy. For me, IT WORKS but felt the question whether i5 i7 2500k or 2600K and whether the value added to make.
@ Rudi: when is likely to you that 1 week, 2 or more?


Hoffentl I will. soon be able to tell ;-)

What do you take? I would especially the comparison with and without HT among cs5 important as the 2500k and 2600K little else except maybe give a little more oc potential in 2600.
Thanks for the info;)

Antwort von makiHD:

So, I once benched bissl:)


Rendered, a FULL HD AVCHD file directly with the Adobe Media Encoder with quality setting 1080i25 HDTV - High Quality.

1 core: 99 seconds, 3.2 Ghz, 100% CPU load @ 1, Threads: 37
2 cores: 85 seconds, 3.2 Ghz, 60 to 75% load @ Core 1 and 2, Threads: 40
3 cores: 50 seconds, 2.93 Ghz, 60 to 75% load @ Core 1,2,3 Threads: 45
4 cores: 47 seconds 2.93 GHz, 75 to 95% load @ Core 1,2,3,4 Threads: 35

Observation: The scaling is quite visible. The rendering of 3 scores needed time almost halved, while all the cores can hardly push the value. Similarly s.besten the number of threads from 3 cores. This suggests that if only 3 cores are enabled HyperThreading s.besten can work because of the 4th Core (HT = 2-core) 1 HT-core does not work "snatches". MMN This is confirmed by the fact that the AME at 4 cores even fewer threads than 1 core port.

Now the "odd" cores are loaded alternately:

Core 0 & 1: 92s striking: CPU is almost constantly at 3.2 Ghz
Core 0 & 2: 61s
Core 1 & 2: 58s striking: almost permanently CPU to 2.93 Ghz
Core 1 & 3: 81's eye-catching: CPU is almost constantly at 3.2 Ghz

Observation here is that the burden of Core 0 and 1 whole 7s is faster than when running Core 0th This indicates a very strong parallel (to see the nearly identical number s.Threads). This fits with the fact that the fact that only one of the physical core is busy, the CPU can run one physical core to 3.2 ghz all day. The Virtual Core help with, so it comes down to 92s of 99s.

This represents a gain of about 7% by HT-core.

But what is noted in any case where the test was not a reeler. I just wanted to show the scale. Much more realistic (and synonymous interesting) is to insert the file into PP Project and the Project DIRECT with load file in the AME. When rendering a project directly in the AME in fact I have to load all 4 cores 100% - this is probably better Threaded.

Now, the test with the same video file in PP-Project:

1 Core: 94s, 3.2 Ghz, up to 100% CPU load @ 0, Threads: 38
2 cores: 78s, 3.06 Ghz, between 85 and 100% CPU load @ 0.1, Threads: 37
3 cores: 53s, 2.93 Ghz, between 85 and 99% CPU load @ 0,1,2 Threads: 33
4 cores: 48s, 2.93 Ghz, between 85 and 99% CPU load @ 0,1,2,3, Threads: 33

Most obvious change is that the respective cores are better utilized, often close to 100% s.der brand in contrast to the direct rendering of a video file in the AME. Thus, the load can be distributed better, the performance increases on average between 5 and 10%.

Conclusion 1: videos, the synonymous to be easily converted only render accessible via the project file.

It is striking that in the legal brief video sequence of about 17's never all cores at full capacity. So far it has rendered all of my projects so that all 4 cores at 100% Permanent full load were, as is the Scan Rate at 2.93 GHz.

Hyperthreading brings so few advantages. Adobe confirmed that CS5 work well with HT CPUs. The link is required for AE, Premiere CS5 is mentioned but synonymous.
Interesting fact: It is expressly pointed out that CS4 can not use virtual cores. Say, HT has no sense to CS4.

Think that is synonymous a good part to the increased performance of CS5 vs. CS4 contributed. Here's the link:

http://help.adobe.com/de_DE/after effects/cs/using/WS9F936D13-E76A-41e4-BF8F-577132AB4723a.html

Here is a benchmark of elsewhere on Multithreading:

Single-threaded Cinebench rendering vs. Multithreaded rendering.


It looks very good as the Athlon X6-1090T is catching up with multithreading well.

Finally, a smaller Comparison of CPUs:

-> Core 2 Duo T9500: 2.6 GHz with powerful dual-core
-> Core i5 560m Dual Core with HT, 4 x 2.67 Ghz to 3.2 Ghz
-> Core i7 740QM: true quad C

Antwort von DTEurope:

To date I have been used on two HexCore ADOBE events both a dual mode in HP. So a total of 24 cores. Whether this is really worth while I can see personally, but I think that the people of ADOBE there already see a meaning.
In the presentations it was certainly a pleasure to PP and AE with so watching a performance.
In the Task Manager to see 24 CPUs but does not look bad synonymous ...

Antworten zu ähnlichen Fragen:
avchd in premiere pro cs5
"Open Timeline" Premiere CS5 Question!
Premiere CS5 supported Grakas?
Premiere Pro CS5 and ASIO Drivers - A NO-GO?
Useful plug-ins for Premiere Pro CS5?
Adobe Premiere CS5 small program window
Premiere CS5 extremely slowly out of the blue
CS5 Premiere - Where is the plug Denoiser of Neat Magic or B.
People Videodateil airbrushed out (Premiere CS5)
Premiere Pro CS5 Full Screen Preview jerky - solution?
Loss of quality in Premiere CS5?
Textbook - Adobe Premiere Pro CS5
Projekt sichern mit Adobe Premiere Pro CS5.5
Adobe Engineering Hours: Was wünscht Du dir für Premiere Pro? Klartext mit Adobe reden und CC-Abo gewinnen
Import MP4 in Premiere CS5
Adobe - neue April Updates für Premiere Pro und Premiere Rush
Mac Pro: Adobe Premiere Pro demnächst mit ProRes Beschleunigung per Afterburner
Messevideo: Neue Funktionen Adobe Premiere Pro u. bestes, verstecktes Pro-Feature // NAB 2018
Adobe Premiere Pro
adobe premiere pro 1.5
Adobe Premiere Pro Streaming
Fragen zu Adobe Premiere pro 7
Adobe Premiere Pro Blitzeffekt
Problem mit Adobe Premiere 1.5 pro
Top 5 New Features in Adobe Premiere Pro CC
Adobe Premiere Pro 7 - Input 352x576

weitere Themen:

Adobe After Effects
Adobe Premiere Pro
After effects
Apple Final Cut Pro
Avid Media Composer
Final Cut
MAGIX video deLuxe
Pinnacle Studio