Ist it possible to get more CPU-usage
Moderator: Ken Berry
Ist it possible to get more CPU-usage
Hi,
My T5250 is not the fastest CPU out there, but when i start encoding my VSS11+ only uses my CPU up to about 60% and so VSS+ is is very slow compared to other NLE's which i only had tested as Demos.
Is there a way to force VSS11+ to more CPU-usage?
Best Regards
MC
My T5250 is not the fastest CPU out there, but when i start encoding my VSS11+ only uses my CPU up to about 60% and so VSS+ is is very slow compared to other NLE's which i only had tested as Demos.
Is there a way to force VSS11+ to more CPU-usage?
Best Regards
MC
-
vze22jf9
I second this request. In my case, the average cpu usage hovers around 40% across the cpus (2 AMD 270s).
Is there something I am missing (settings) or VS11+ simply won't maximize cpu utilization (perhaps by design to allow some responsiveness in the application)? Has anyone achieve high cpu utilization with VS11?
I would appreciate some suggestions on improving cpu utilization or any comment that may shed light on why the seemingly low cpu utilization.
Thanks.
PS. My typical project (instr. videos) involves using the Movie wizard to import a HDV video file, add a title screen, and outputing the video to H.264 (ipod) format.
Is there something I am missing (settings) or VS11+ simply won't maximize cpu utilization (perhaps by design to allow some responsiveness in the application)? Has anyone achieve high cpu utilization with VS11?
I would appreciate some suggestions on improving cpu utilization or any comment that may shed light on why the seemingly low cpu utilization.
Thanks.
PS. My typical project (instr. videos) involves using the Movie wizard to import a HDV video file, add a title screen, and outputing the video to H.264 (ipod) format.
On a single-core system, I would expect nearly 100%.
I believe the MPEG-2 encoder can take advantage of 2 CPUs. So, if you encode to MPEG-2, from AVI/DV, on a dual-core machine, you will get more than 50% utilization. However, some of the other encoders and decoders may not, so it depends on what format you are decoding from, and what format you are encoding to. (DivX and the other MPEG-4 variations are notoriously slow to decode, and I believe these decoders only use a single core/CPU.)
And, there can be other bottlenecks in a system... With most real-world applications, the CPU is often "idle" while waiting for data. The faster your CPU is, and the more processors you have, the more likely you are to have a CPU that's strarved for data!
I believe the MPEG-2 encoder can take advantage of 2 CPUs. So, if you encode to MPEG-2, from AVI/DV, on a dual-core machine, you will get more than 50% utilization. However, some of the other encoders and decoders may not, so it depends on what format you are decoding from, and what format you are encoding to. (DivX and the other MPEG-4 variations are notoriously slow to decode, and I believe these decoders only use a single core/CPU.)
That's a quad-core machine, right? I'm not sure if Video Studio can take advantage of more than 2 CPUs.I second this request. In my case, the average cpu usage hovers around 40% across the cpus (2 AMD 270s).
And, there can be other bottlenecks in a system... With most real-world applications, the CPU is often "idle" while waiting for data. The faster your CPU is, and the more processors you have, the more likely you are to have a CPU that's strarved for data!
Only the Corel programmers can do that (or the CODEC programmers who may not work for Corel).. The programmer has to break-up the program into "threads". And, it can be very tricky to make 2 or more equal threads. For example, if you make one thread for encoding, and another for decoding, it is unlikely that these processes will require equal time.Is there a way to force VSS11+ to more CPU-usage?
[size=92][i]Head over heels,
No time to think.
It's like the whole world's
Out of... sync.[/i]
- Head Over Heels, The Go-Gos.[/size]
No time to think.
It's like the whole world's
Out of... sync.[/i]
- Head Over Heels, The Go-Gos.[/size]
Comparing to another video editing program you need to compare the full encoding times of both programs and the quality of the encodes.
Just because one program is using all 4 cores at a high cpu usage doesn't mean it's better. So you have to compare the encoding times & quality from one to another.
I have one program that uses all 4 cores at 80% cpu usage.
VS11+ uses all 4 cores at 45% cpu usage.
The encoding times are almost identical, so it's in the code.
For Intel quad processors VS11+ definitely uses all 4 cores converting from mpeg2 to AVCHD, the cpu usage is around 60% to 70%, MoiveFactory uses all 4 cores and the cpu useage is 80% or higher.
But like Doug says, it does depend on the source & target formats.
Not sure about AMD processors.
Just because one program is using all 4 cores at a high cpu usage doesn't mean it's better. So you have to compare the encoding times & quality from one to another.
I have one program that uses all 4 cores at 80% cpu usage.
VS11+ uses all 4 cores at 45% cpu usage.
The encoding times are almost identical, so it's in the code.
For Intel quad processors VS11+ definitely uses all 4 cores converting from mpeg2 to AVCHD, the cpu usage is around 60% to 70%, MoiveFactory uses all 4 cores and the cpu useage is 80% or higher.
But like Doug says, it does depend on the source & target formats.
Not sure about AMD processors.
Re: Ist it possible to get more CPU-usage
One thing you can do is, just before you start encoding, hit ALT-CTL-DEL and this will pop up the Windows Task Manager. Click on the processes tab and find VS. Right click and change the priorty from Normal to AboveNormal. (Don't go any higher than that as you would be denying Windows, itself, process time). Of course it always helps if you have minimal things running in the background.MC76 wrote:Hi,
My T5250 is not the fastest CPU out there, but when i start encoding my VSS11+ only uses my CPU up to about 60% and so VSS+ is is very slow compared to other NLE's which i only had tested as Demos.
Is there a way to force VSS11+ to more CPU-usage?
Best Regards
MC
I don't have a Vista machine, but, in XP, you can turn off all the bells and whistles of the OS and this does help, i.e., turning off the fancy look and feel.
Re: Ist it possible to get more CPU-usage
I've always been a firm believer in setting up a dedicated video editing hardware profile in windows, which maximises resources available to VS, particularly for the encoding stage. I wholeheartedly agree with using the "Windows Classic" appearance and turning off the "Telly Tubbie" visual effects.MrA wrote:I don't have a Vista machine, but, in XP, you can turn off all the bells and whistles of the OS and this does help, i.e., turning off the fancy look and feel.
Recently, however, having bought 2GB of RAM for my C2D laptop, I re-ran some rendering tests on the C2D laptop, my old P4 2.8c (hyperthreading) desktop, a Celeron M laptop and a Sempron 3400 laptop, all running XP, either Home or Pro.
To my surprise, it turned out that there was no measurable improvement in rendering speed for the C2D or the P4 2.8c with HT enabled when using my video editing profile, which has only 13 Windows processes running.
There was a benefit for the P4 2.8c when HT was disabled in bios, but encoding was slower than with HT enabled. The other two single core, single threaded laptops both showed a smallish but measurable improvement in rendering speed when using a video editing profile.
Mind you, my normal profiles only use 25 or so processes, compared with over 50 I've seen running on other peoples pc's.
Although I still believe it's a good idea to use a dedicated video editing profile, in order to avoid conflicts with other software and so on, I don't believe there's much of a speed benefit to be had for dual core, dual threaded or even quad core machines.
And of course when considering dual core performance, even if you were to see both cores working at only 40%, they would still probably be encoding the video faster than an equivalent single core working at 100%.
My own experience with my C2d laptop, which uses a 1.83GHz T2400 is that both cores hover around the 70% mark when rendering DV type 1 avi to mpeg-2.
JVC GR-DV3000u Panasonic FZ8 VS 7SE Basic - X2
-
sjj1805
- Posts: 14383
- Joined: Wed Jan 26, 2005 7:20 am
- System_Drive: C
- 32bit or 64bit: 32 Bit
- motherboard: Equium P200-178
- processor: Intel Pentium Dual-Core Processor T2080
- ram: 2 GB
- Video Card: Intel 945 Express
- sound_card: Intel GMA 950
- Hard_Drive_Capacity: 1160 GB
- Location: Birmingham UK
You can try this - if you do please post the results here for the benefit of other forum members.
1. With VideoStudio running press the [Ctrl] [Alt] [Del] keys at the same time to bring up the Windows task manager.
2. Go to the [Processes] tab
3. Find vstudio.exe and click once to select it
4. Right click your mouse to bring up a list of options
5. Look for [set priority] and change it from 'normal' to 'High'
You cannot create a [url=Vhttp://phpbb.ulead.com.tw/EN/viewtopic.php?t=13950]Video Editing Profile[/url] on a Vista Computer - this is something that has upset a large number of computer users across a wide spectrum of computing tasks especially games players. Microsoft have removed the ability to set up hardware profiles with Vista which is the main core involved in creating these special profiles.
EDIT.
Just noticed that MrA has already beaten me to this suggestion!!
1. With VideoStudio running press the [Ctrl] [Alt] [Del] keys at the same time to bring up the Windows task manager.
2. Go to the [Processes] tab
3. Find vstudio.exe and click once to select it
4. Right click your mouse to bring up a list of options
5. Look for [set priority] and change it from 'normal' to 'High'
You cannot create a [url=Vhttp://phpbb.ulead.com.tw/EN/viewtopic.php?t=13950]Video Editing Profile[/url] on a Vista Computer - this is something that has upset a large number of computer users across a wide spectrum of computing tasks especially games players. Microsoft have removed the ability to set up hardware profiles with Vista which is the main core involved in creating these special profiles.
EDIT.
Just noticed that MrA has already beaten me to this suggestion!!
I tried the to give VSS a higher priority but the CPU-usage won't go higher.
When i compared other NLE's I did always the same test, a 2:35min 1080 50i HDV m2t clip from a HV20 rendering to Pal widescreen DVD file, my Thinkpa
the
The rendering time was on XP:
VS11+ ~7min
Vegas7 ~4min
Liquid 7.2 ~9min
Premiere Pro 2 ~10
There are some slower than VS, but they too don't use the Cores more than 70%. The only one which user more thean 90% on both cores ist Vegas 7, so it beomes incredibly fast compared to the others, i don't know if the small version is as fast as Vegas7.
MC
When i compared other NLE's I did always the same test, a 2:35min 1080 50i HDV m2t clip from a HV20 rendering to Pal widescreen DVD file, my Thinkpa
the
The rendering time was on XP:
VS11+ ~7min
Vegas7 ~4min
Liquid 7.2 ~9min
Premiere Pro 2 ~10
There are some slower than VS, but they too don't use the Cores more than 70%. The only one which user more thean 90% on both cores ist Vegas 7, so it beomes incredibly fast compared to the others, i don't know if the small version is as fast as Vegas7.
MC
-
vze22jf9
Thanks for the replies. Many suggestions are sound and useful while others led to some discoveries (for me anyway). I will address them, not in any particular order, below.
In the example you cited, it seems clear that VS11 employs a more efficient codec, but wouldn't VS11 be much faster if it also utilized 80% cpu, like its competitor, instead of merely 45%?
=====
I did some testing based on the comments. I hope the results might be of some use to someone. Comments are welcomed.
However, I tried both and neither resulted any change in performance in my testing. However, I didn't (and usally don't) run any other extraneous processes when I am converting/encoding videos. I'd use another computer if I have to.
If interested, please see discussion on encoding/decoding further below for more details.
On the other hand, putting the source file on a RAID 0 SATA array and the output file on a different hard drive did not result in any noticeable further improvement when compared to having the source file on a single 10K rpm drive.
In all tests, the source file is a HDV (mpeg2?) video at 25Mbps CBR. I am still relatively new to VS11 so I only used the default video output types that's available (SHARE TAB/Create Video File).
1. HDV -> MPEG4 (IPOD 640x480) ----- Average cpu usage was 26%. However, it's a steady 100% on 1 cpu only and virtually 0% on the other 3 cores. Also, 100% utilization alternates on cpu 1 and 3, but never cpu 2 and 4. This suggests this decoding/encoding process is probably single threaded and not smp, dual/multi-core aware.
2. HDV -> MPEG4 (ipod H.264) ----- Average cpu usage was around 40%. The usage were on cpus 1 and 3 only with large spikes up and down. This result seems to suggest this particular encoding/decoding process is multi-threaded and smp aware, but does not utilize dual/multi-core cpus. Furthermore, the large spikes up and down on cpu suggest the process alternates between decoding and encoding stages (as opposed to dedicated threads for decoding and encoding running concurrently). The spikes can also suggest data starvation, but I don't think that's case here.
3. HDV -> AVCHD (1440x1080) ----- Average cpu usage was around 40%. The usage was steady and across all 4 cpus. This would suggest this process is multi-threaded, smp & multi-core aware.
=====
Granted this is a pretty limited test, but doesn't it seems like there is room for improvement?
I wonder if the mixed results on cpu utilization might be due to VS11 using a mix of Ulead and non-Ulead codecs... Does anyone know for sure?
This comment goes to heart of my request. I wasn't suggesting that VS11 is slow, but was wondering why it doesn't make better use of the cpus to reach even better performance.I have one program that uses all 4 cores at 80% cpu usage.
VS11+ uses all 4 cores at 45% cpu usage.
The encoding times are almost identical, so it's in the code.
In the example you cited, it seems clear that VS11 employs a more efficient codec, but wouldn't VS11 be much faster if it also utilized 80% cpu, like its competitor, instead of merely 45%?
=====
I did some testing based on the comments. I hope the results might be of some use to someone. Comments are welcomed.
5. Look for [set priority] and change it from 'normal' to 'High'
Minimizing number of processes and running programs when possible, and giving higher process priority when not is sound advice.Click on the processes tab and find VS. Right click and change the priorty from Normal to AboveNormal
However, I tried both and neither resulted any change in performance in my testing. However, I didn't (and usally don't) run any other extraneous processes when I am converting/encoding videos. I'd use another computer if I have to.
Not exactly. It is a total of 4 cores, but it's 2 physical cpus, each with 2 cores. I make this distinction because in my testing I found that depending on the encoding process (within VS11) chosen, there was differences in cpu utilization when it comes to physical cpu and multi-core cpus.That's a quad-core machine, right? I'm not sure if Video Studio can take advantage of more than 2 CPUs.
If interested, please see discussion on encoding/decoding further below for more details.
I found one such bottleneck my testing, which was obvious in hindsight. If both the source and output files were on the same hard drive (a single 10K rpm drive), the cpu utilization dropped when compared to having the files on different physical drives. (30% vs 40%, HDV-AVCHD)And, there can be other bottlenecks in a system... With most real-world applications, the CPU is often "idle" while waiting for data. The faster your CPU is, and the more processors you have, the more likely you are to have a CPU that's strarved for data!
On the other hand, putting the source file on a RAID 0 SATA array and the output file on a different hard drive did not result in any noticeable further improvement when compared to having the source file on a single 10K rpm drive.
In my testing, which is similar to yours, HDV (which is a form of mpeg2, no?) ---> AVCHD, the cpu utilization on my AMD is 40% even distribution across all 4 cpus.For Intel quad processors VS11+ definitely uses all 4 cores converting from mpeg2 to AVCHD, the cpu usage is around 60% to 70%, MoiveFactory uses all 4 cores and the cpu useage is 80% or higher...
... Not sure about AMD processors.
However, some of the other encoders and decoders may not, so it depends on what format you are decoding from, and what format you are encoding to. (DivX and the other MPEG-4 variations are notoriously slow to decode, and I believe these decoders only use a single core/CPU.)
This suggestion turned out to provide the most insight to cpu utilization for me. Some interesting results/observations detailed below.But like Doug says, it does depend on the source & target formats.
In all tests, the source file is a HDV (mpeg2?) video at 25Mbps CBR. I am still relatively new to VS11 so I only used the default video output types that's available (SHARE TAB/Create Video File).
1. HDV -> MPEG4 (IPOD 640x480) ----- Average cpu usage was 26%. However, it's a steady 100% on 1 cpu only and virtually 0% on the other 3 cores. Also, 100% utilization alternates on cpu 1 and 3, but never cpu 2 and 4. This suggests this decoding/encoding process is probably single threaded and not smp, dual/multi-core aware.
2. HDV -> MPEG4 (ipod H.264) ----- Average cpu usage was around 40%. The usage were on cpus 1 and 3 only with large spikes up and down. This result seems to suggest this particular encoding/decoding process is multi-threaded and smp aware, but does not utilize dual/multi-core cpus. Furthermore, the large spikes up and down on cpu suggest the process alternates between decoding and encoding stages (as opposed to dedicated threads for decoding and encoding running concurrently). The spikes can also suggest data starvation, but I don't think that's case here.
3. HDV -> AVCHD (1440x1080) ----- Average cpu usage was around 40%. The usage was steady and across all 4 cpus. This would suggest this process is multi-threaded, smp & multi-core aware.
=====
Granted this is a pretty limited test, but doesn't it seems like there is room for improvement?
I wonder if the mixed results on cpu utilization might be due to VS11 using a mix of Ulead and non-Ulead codecs... Does anyone know for sure?
-
skier-hughes
- Microsoft MVP
- Posts: 2659
- Joined: Thu Jul 21, 2005 10:09 am
- System_Drive: C
- 32bit or 64bit: 64 Bit
- motherboard: gigabyte
- processor: Intel core 2 6420 2.13GHz
- ram: 4GB
- Video Card: NVidia GForce 8500GT
- sound_card: onboard
- Hard_Drive_Capacity: 36GB 2TB
- Location: UK
I'm not an IT guy, but I've just been doing some reading on Vista and in that it said, if I remember correctly, that for optimum pc useage the cpu shouldn't run above 70% for prolonged periods as this could affect stability, producing too much heat etc.
If I can find out the piece again I'll quote it in more detail, not that I think it was terribly in depth!!!!!
If I can find out the piece again I'll quote it in more detail, not that I think it was terribly in depth!!!!!
