Just wondering if there was a Best Video Encoding CPU for theMoney?

[BenjaminCachimbear]

Trying to find a "Best Video Encoding CPU for the Money' article. Anybody got any leads? If there isn't one, there should be! I know I can compare the charts, but it's not the same.

 

[Leaps-from-Shadows]

Intel's Sandy Bridge CPUs have Quick Sync technology, which is by far the best video encode/decode/transcode method. Quick Sync is faster at it than any discrete graphics card without any loss of quality. The 2500K and 2600K Sandy Bridge CPUs have the best version of the integrated graphics and Quick Sync.

Note that you must use either an H67 mainboard (available now) or an X68 mainboard (which hasn't been released yet, but is coming soon) to enable Quick Sync. P67 mainboards disable the integrated graphics and Quick Sync.

 

[Guest]

I think the i5 2500k, and the Phenom II x6 1090T at ~ $225 and $200 or so, are the best video encoding cpu's for the money.

 

[MU_Engineer]

Trying to find a "Best Video Encoding CPU for the Money' article. Anybody got any leads? If there isn't one, there should be! I know I can compare the charts, but it's not the same.

It depends on how much you want to spend. I'd nominate the Athlon II X4 640 ($100), Phenom II X6 1090T (~$190), the Core i5-2500 ($200-something), two Opteron 4180s ($380), and four Opteron 6128s ($1070) in their respective price ranges. The latter two are especially interesting as two Opteron 4180s are frequently faster than a Core i7-990X but cost a lot less, and four Opteron 6128s cost about what that i7-990X does but are MUCH faster.

Not the phenom it lacks SSE4 instructions unfortunately.

There is no "SSE4" as there wasn't a consensus on what constituted SSE4 and each vendor has some instruction they call "SSE4-something" and don't support the other vendor's "SSE4-something" instructions. The Phenoms have SSE4a and Intel has their SSE4.1 and 4.2.

Op it will depend on the application your using and if you do video editing as a hobby. If you use Photoshop in your sparetime the 2600k would be the best all round performer.

Photoshop does video? I thought the super-expensive Adobe video editing software was Premiere Pro.

If you do it professionally then the I7 9xx models with a 700usd Pci-e raid add in card.

You would be much better with a multiprocessor setup if you're going to spend that kind of money than a desktop single-processor i7-900 setup.

Remember video editing a balance system is more important than a super fast cpu. Nvidia cards are also advised as its widely supported and hardware mpe give you big boost as to software mpe.

You are also very limited to what codecs, resolutions, bitrates, and encode options you an use. Also, quality can suffer in a lot of GPU implementations compared to CPU-based encoders.

So the 2600k with the Gtx 560ti, 460, 470 or 480 gpus would be your best bet. The quadro 400 that came supposedly outperforms the Gtx 580 by 5 fold thanks to its driver support that's what was reported but we all know what nvidia can do so ill leave that one to the driver modders to bring out the truth

I'd suggest dual Opterons or Xeons and doing the encode with the CPU to get the best possible product, but it again depends on how much video editing you want to do and how much you care about exactly what the final product looks like.

 

[jaguarskx]

If I were to upgrade my current Q9450 for video encoding, I would probably go for the i7 2600k since video encoding can take advantage of HyperThreading. Not to mention the overclocking capabilities of the "k" series.

The difference in price between the i5 2500k and the i7 2600k was $90 the last time I checked on Newegg.com.

HyperThreading does nothing for games.

 

[ElMoIsEviL]

Intel's Sandy Bridge CPUs have Quick Sync technology, which is by far the best video encode/decode/transcode method. Quick Sync is faster at it than any discrete graphics card without any loss of quality. The 2500K and 2600K Sandy Bridge CPUs have the best version of the integrated graphics and Quick Sync.

Note that you must use either an H67 mainboard (available now) or an X68 mainboard (which hasn't been released yet, but is coming soon) to enable Quick Sync. P67 mainboards disable the integrated graphics and Quick Sync.

Hmm,

According to the Tests I've seen Quick Sync is actually lower quality output than AMDs solution overall (http://www.anandtech.com/show/4083/the-sandy-bridge-review-intel-core-i7-2600k-i5-2500k-core-i3-2100-tested/9) but quicker. nVIDIAs CUDA technology is all about performance and as such quality suffers (like anything else nVIDIA does). That being said they are all of lower quality than using a pure x86 path with SSE4 support and filters (it is still top dog). You can use various programs including XMPEG+ for x86.

I encode a ton of video and nothing comes close to using the CPU without any other gimmicks. Sony Vegas does a really good job as well but it is damn slow to encode.

But yeah... Carry on

 

[MU_Engineer]

Photoshop = CS5 that's what I meant.

DISK I/O test:
The overriding factor is disk speed here.  The test uses many small reads and a large sequential write (nearly 13 GB). Number of cores makes no real difference (it is not well multithreaded), but clock speed does.
MPEG2 DVD test:
The two overriding factors here are amount of memory and number of cores. More is better here. Additionally the location and speed of the pagefile can be important especially if you have a small amount of RAM.
H.264 test:
Here the speed of CPU/RAM communicationis king. Number of cores, clock speed and the amount of CPU cache are very important. Dual processor systems are hampered by the 2 chip communication.
CPU / GPU Test Result:
This is almost solely based on the video card and whether hardware or software MPE is used.
MPE Gain:
This shows how much faster hardware MPErendering is than software only rendering. The minimum score is of course 1, since if there is no hardware MPE available, there is no performance gain.

Then

Our Hardware Design Conclusions
*.Absolutely most important is the CPU, right now that is the Intel i7 processor or the dual processor Xeon 56xx series. Specifically for MPEG encoding the amount of memory is the most important facor to influence performance. As you can see from the performance data the speed ofthe processor is also very significant. AMD processors lack the full SSE set of instructions and are not as effective as the Intel processors.

*.A dual processor setup profits from the extra cores during MPEG encoding, but is hampered during H.264 encoding due to latencies between the 2 chips, a single i7 profits from more cache.
*.With CS5 4 or 6 GB of RAM is minimum, 8or 12 GB is suggested for most users. For heavy multitasking or dynamic linking, 16 or 24 GB is optimal

The Adobe minimum basic disk system isan absolute minimum of two 7200 rpm disk drives.  Our personal preference is for a 10,000 rpm drive for the Operating System & Applications disk and a RAID array for the project files. Specific functionslike a separate dedicated drive for writing Output files or Preview files are of less value as it may just slow things down compared to a high performance RAID.

.A CUDA/MPE card makes a huge difference in performance and improves quality of the output over software MPE.

No way a dual opteron will outperform a Xeon nor the 980x in video editing. The overhead of the cpu communication smacks it into the ground. Here's proof
http://ppbm5.com/Benchmark5.html

Opteron getting smashed by everything of intel in all those encodings
http://ppbm5.com/Test.html

That's ONE benchmark with a program that is very well known to be highly optimized for Intel CPUs. (The quip about AMD CPUs "not supporting the full set of SSE instructions" makes that one apparent.) It is also not very well multithreaded based on their results. The benchmark website is also poorly done with somewhat broken English and little information backing up their list of assertions.

CPU intercommunication is as good or better with Opterons than with Xeons. All of the current Opterons run their coherent HyperTransport links at 16 bits @ 6.40 GT/sec. The only Xeons that are allowed to run their QPI links at 16 bits effective (20 bits, but at 8/10b encoding) @ 6.40 GT/sec are the X-series units. The rest run theirs at 5.86 GT/sec or 4.80 GT/sec. The Opteron 6100s also have three 16-bit coherent HT links compared to one for the Xeon 5500/5600 CPUs, which gives it at least three times the interprocessor communication bandwidth.

 

[ElMoIsEviL]

That's ONE benchmark with a program that is very well known to be highly optimized for Intel CPUs. (The quip about AMD CPUs "not supporting the full set of SSE instructions" makes that one apparent.) It is also not very well multithreaded based on their results. The benchmark website is also poorly done with somewhat broken English and little information backing up their list of assertions.

CPU intercommunication is as good or better with Opterons than with Xeons. All of the current Opterons run their coherent HyperTransport links at 16 bits @ 6.40 GT/sec. The only Xeons that are allowed to run their QPI links at 16 bits effective (20 bits, but at 8/10b encoding) @ 6.40 GT/sec are the X-series units. The rest run theirs at 5.86 GT/sec or 4.80 GT/sec. The Opteron 6100s also have three 16-bit coherent HT links compared to one for the Xeon 5500/5600 CPUs, which gives it at least three times the interprocessor communication bandwidth.

I would have to question what I placed in bold. Although that Opteron model may have more HT Links... in a Dual Processor setup (as is being discussed here) only one of those links is utilized for communication between both processing units.

Now if you mean Internal communication between cores, Intel's is superior by virtue of a much more efficient usage of caching technology allowing for far superior communication between various cores through the caching mechanism. This is a remnant of the Core 2 days when an FSB was still being utilized and remains present in Nehalem and Sandy Bridge. This explains why Intel still retains superior Core to Core communication bandwidth and far lower latency than AMD (a. AMD has inferior caching technology and B. HT offers less bandwidth and higher latency than cache for core to core communication).

As an overall platform... it is a bit of a mixed bag with Xeons taking the overall lead in terms of computational performance but things getting quite close when it comes to point to point interconnect throughput and latency.

Worth mentioning that a single Nehalem based Core i7 will outpeform a Dual Opteron (Dual Quad Phenom II generation) in x86 encoding tasks. So he would not be best served with a Dual Processor setup. Don't believe me?? Try it out yourself... my friend Obi Wan (his pseudo name but he goes under the name "Gold Leader" over at 2CPU.com) is a big Opteron fan and we compared results... I slaughtered him in XMPEG, Sony Vegas Pro, Premiere Pro and Cyberlink Power Director in terms of encode times using the same files and settings. When I mean slaughtered I mean it wasn't even a fair fight. In fact there isn't a single benchmark where he bested me outside of synthetics.

This was when I had a Core i7 920 Overclocked to 4GHz. He has two Quad Core Opterons (I forget which models he has though but he trounced me in terms of the amount of RAM he has (16GB vs. 6GB at the time)).

 

[MU_Engineer]

I would have to question what I placed in bold. Although that Opteron model may have more HT Links... in a Dual Processor setup (as is being discussed here) only one of those links is utilized for communication between both processing units.

The Opteron 6100s have two dies per CPU. There are multiple HT links used between the two sockets to make sure that every die can communicate directly with every other die in a two-socket setup. When a third or fourth CPUs are added, there are some dies that are two hops away.

Now if you mean Internal communication between cores, Intel's is superior by virtue of a much more efficient usage of caching technology allowing for far superior communication between various cores through the caching mechanism. This is a remnant of the Core 2 days when an FSB was still being utilized and remains present in Nehalem and Sandy Bridge. This explains why Intel still retains superior Core to Core communication bandwidth and far lower latency than AMD (a. AMD has inferior caching technology and B. HT offers less bandwidth and higher latency than cache for core to core communication).

HyperTransport is not used for inter-core communication. AMD uses a crossbar for this purpose. The Core 2s use the L2 cache in Core 2 Duos and the L2 cache and FSB in Core 2 Quads. The L3 is used in Nehalem/Gulftowns and the ring bus is used in Becktons and Sandy Bridge. I won't argue too much on caches as exclusive caches don't really make much sense any more with transistor budgets in the hundreds of millions to billions per chip nowdays. But even then, AMD's L3 is inclusive and the reason its latency is high is mostly lower clock speed.

As an overall platform... it is a bit of a mixed bag with Xeons taking the overall lead in terms of computational performance but things getting quite close when it comes to point to point interconnect throughput and latency.

Agreed.

Worth mentioning that a single Nehalem based Core i7 will outpeform a Dual Opteron (Dual Quad Phenom II generation) in x86 encoding tasks. So he would not be best served with a Dual Processor setup. Don't believe me?? Try it out yourself... my friend Obi Wan (his pseudo name but he goes under the name "Gold Leader" over at 2CPU.com) is a big Opteron fan and we compared results... I slaughtered him in XMPEG, Sony Vegas Pro, Premiere Pro and Cyberlink Power Director in terms of encode times using the same files and settings. When I mean slaughtered I mean it wasn't even a fair fight. In fact there isn't a single benchmark where he bested me outside of synthetics.

This was when I had a Core i7 920 Overclocked to 4GHz. He has two Quad Core Opterons (I forget which models he has though but he trounced me in terms of the amount of RAM he has (16GB vs. 6GB at the time)).

I know Gold Leader, he has 2384s IIRC. Many encoders are not really prepared to take advantage of more than 8-12 cores at the moment, which is probably why the highly-overclocked i7 may beat two stock-clocked quad-core Opterons. It was actually funny to see my 16-core Opteron setup score basically the same in Phoronix's x264 benchmark as a 48-core Opteron 6172 machine. However, I usually have more than one video to encode at a time, so queuing up several to run at the same time and seeing no loss of performance when doing so is very handy.

 

[oby20]

Trying to find a "Best Video Encoding CPU for the Money' article. Anybody got any leads? If there isn't one, there should be! I know I can compare the charts, but it's not the same.

more core, more memory and a good gpu is the best setup for a video encoding.
sandy bridge is not the best in video editing, encoding. it is by far the fastest but not in programs that use more memory.
sandy bridge is good for an average user, what do an average user do, games, browse net, email, edit photos and so on.
its speed wont do much in a heavily threaded program such as adobe photoshop, video editing, and animations.
it is just a turbo version of 1156 series, good but not great.
i7 970 and i7 980x 6 core 12 threads together with a good gpu and large memory will be the top choice.
but with a budget i would say the i7950 is the second best choice for video editing.

 

[amdfangirl]

Hmm,

According to the Tests I've seen Quick Sync is actually lower quality output than AMDs solution overall (http://www.anandtech.com/show/4083/the-sandy-bridge-review-intel-core-i7-2600k-i5-2500k-core-i3-2100-tested/9) but quicker. nVIDIAs CUDA technology is all about performance and as such quality suffers (like anything else nVIDIA does). That being said they are all of lower quality than using a pure x86 path with SSE4 support and filters (it is still top dog). You can use various programs including XMPEG+ for x86.

I encode a ton of video and nothing comes close to using the CPU without any other gimmicks. Sony Vegas does a really good job as well but it is damn slow to encode.

But yeah... Carry on

Quick Sync does have a lower quality output, I don't find it acceptable, the limited set of encoding options and file formats.

You might want to wait for CS5.5, which adds 64bit encoding and better multi-core support for Adobe Media Encoder. I don't honestly know anything about Sony because after all, it's not my field of work (Photoshop & Illustrator).

Oh and one last thing. Photoshop does do video, you can import it as frames and edit it frame by frame as layer by layer or basic animation tools inside extended. However, they are quite surprisingly useless, you can get so much better by I dunno, using an actual video editing software .