Best Processor for Video Conversion at this time of July 29 2015

[johnharlin]

I do lots of video conversions and with a faster processor I would like to convert a lot of old rare video cassettes I have. The Skylake I7 6700 is coming out next week but there is also the I7 5820 with 6 cores that has been out for awhile. Which would be faster. I think they would be priced about the same. It seems like 6 cores would be faster since you would divide the work among 6 cores. AMD also has an an 8 core processor but it doesn't seem like the AMDs are as good.

I know that some of you have the latest hardware. What kind of conversion speed are you getting out of them? Like how long would it take to convert an hour long video say from avi to DVD or something like that? I suppose the quality makes a difference as well like it would probably take longer in converting HD quality videos.

It would be nice to have a conversion chart of times for different types of conversions with different processors because time is an important element in video work.

[LigH.de]

If your preference is strictly speed over price, go for modern intel Core i# processors. AMD produces only compromises at the moment.

[manono]

I do lots of video conversions and with a faster processor I would like to convert a lot of old rare video cassettes I have.

'Video cassettes', as in VHS tapes?

I suppose the quality makes a difference as well like it would probably take longer in converting HD quality videos.

What's Hi-Def about VHS tapes? If this question is really about VHS tapes and if quality figures at all in your calculations, much better would be to buy a decent capturing setup and learn how to capture and convert properly, instead of worrying about what will convert them the fastest. What's the point of speed if the output is garbage?

It would be nice to have a conversion chart of times for different types of conversions with different processors because time is an important element in video work.

Nonsense. The encoding speed of your computer is much less important than the time spent restoring the captures and preparing them for encoding.

[ezcapper]

x265 is the most intensive encoding, so this conversion chart should help. I would wait for the skylake to be released and see how much of an improvement it is over the 5820. I'm thinking you want something over 30 fps:
http://x265.ru/en/x265-hd-benchmark/

you can run the benchmark on your own computer, here are the results of mine, it says that the results are wih preset fast, but they fall more in line with preset ultrafast:

x265 HD BENCHMARK 0.1.4 RESULTS

Please do NOT compare it with older versions of the benchmark!
Please copy/paste everything below the line to to report your
data to http://x265.ru/x265-hd-benchmark/.html

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Results for x265.exe build 1.4+5
x265 Benchmark: 64-bit
==========================

CRF-20 preset-"fast"
--------------------
encoded 1128 frames in 322.06s (3.50 fps), 2266.23 kb/s
encoded 1128 frames in 314.58s (3.59 fps), 2266.23 kb/s
encoded 1128 frames in 312.72s (3.61 fps), 2266.23 kb/s
encoded 1128 frames in 312.97s (3.60 fps), 2266.23 kb/s


System Details
--------------
Name Intel Pentium 2020M
Codename Ivy Bridge
Specification Intel(R) Pentium(R) CPU 2020M @ 2.40GHz
Core Stepping E1/L1
Technology 22 nm
Stock frequency 2400 MHz
Core Speed 2394.6 MHz

Northbridge Intel Ivy Bridge rev. 09
Southbridge Intel HM70 rev. 04

CAS# latency (CL) 11.0
RAS# to CAS# delay (tRCD) 11
RAS# Precharge (tRP) 11
Cycle Time (tRAS) 28
Command Rate (CR) 1T
Memory Frequency 798.2 MHz (1:6)
Memory Type DDR3
Memory Size 3982 MBytes
Channels Single

Windows Version Microsoft Windows 8.1 (6.3) 64-bit (Build 9600)

Number of processors 1
Number of threads 2
Number of threads 2 (max 16)
L2 cache 2 x 256 KBytes, 8-way set associative, 64-byte line size
Instructions sets MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, EM64T, VT-x
Package (platform ID) Socket 988B rPGA (0x4)

Temperature 0 51°C (123°F) [0xCAA] (THRM)
Temperature 1 n.a. (CPU)

[jagabo]

There is no simple answer to this question. Different programs can use different encoders. Different encoders deliver different quality and run at different speeds. Even the same encoder using different settings can deliver different quality and take different amounts of time. For example, x264 encoding a 90 minute DVD source may take anywhere from 4 minutes to 400 minutes depending on the settings used. If and editor uses Intel Quick Sync it may encode that DVD in 2 minutes (if the decoder can feed it fast enough). But the quality (per bitrate) won't be as good as an x264 encoding at settings that takes 30 minutes. Are you going to be filtering while encoding? If that filtering is slow the speed of the encoder may not matter much.

AMD's 8 core CPUs can encode faster than more expensive 4 core (8 thread) Intel i7 CPUs under some situations -- like x264 CRF encoding. They also consume twice as much power requiring bigger, louder cooling solutions. But with other software the i7 will outperform the AMD CPU.

So if anybody gives you a table of CPU vs. encoding speed it will only be accurate for a specific encoder under specific conditions. If you're using a different encoder or different settings that table may be misleading.

[ezcapper]

Here are 2nd pass benchmark tests for x264 for different processors from anandtech, in fps. Looking at it, I still think you should wait and see how the skylake does on these tests:
http://www.anandtech.com/bench/CPU/54

I can't find good mpeg2 benchmark test results, I guess it is too old and considered outdated. Maybe someone else can. I think these two are really enough to figure out which processor will have the faster encoding speeds and about how much time it will save you, etc.

Tom's Hardware does h.264 benchmarks with handbrake, adobe's program, etc. but it is not comprehensive, they only have certain processors:
http://www.tomshardware.com/charts/cpu-charts-2013/benchmarks,140.html

[jagabo]

Skylake won't be appreciably faster than current Intel processors. Intel is concentrating on reducing power consumption now, not increasing performance.

Most MPEG 2 encoders are not well multithreaded so Intel CPUs with their better IPC are faster with MPEG 2 encoding. Unless you want to run multiple conversions at the same time.

[poisondeathray]

There are leaked benchmarks at various Taiwanese sites. You get about 5-10% increase across the board on various applications - which is the same every new generation of processors. There hasn't been a big jump in performance in a long time. Thanks to no competition.

[jagabo]

Skylake will have a hardware HEVC encoder. It will probably be faster than the software h.265 encoders and deliver lower quality.

[poisondeathray]

Skylake will have a hardware HEVC encoder. It will probably be faster than the software h.265 encoders and deliver lower quality.

Yes, I'm not expecting much in terms of quality in the 1st generation. It too them 3 generations to make QuickSync h.264 good enough to be usable

The 1st generation Nvidia NVENC HEVC encoder on the GTX960 is pretty poor too. Fast , but quality is very poor.

But valid reason to upgrade might be the HEVC HW QS decoder . CPU decoding of HEVC UHD right now eats up almost all the CPU . Or maybe just pick up a Nvidia card with VP7 (only GTX960 and 950 right now have the HEVC decoder)

[ezcapper]

<deleted, didn't see post above>

[SameSelf]

I do lots of video conversions and with a faster processor I would like to convert a lot of old rare video cassettes I have. The Skylake I7 6700 is coming out next week but there is also the I7 5820 with 6 cores that has been out for awhile. Which would be faster. I think they would be priced about the same. It seems like 6 cores would be faster since you would divide the work among 6 cores. AMD also has an an 8 core processor but it doesn't seem like the AMDs are as good.

I know that some of you have the latest hardware. What kind of conversion speed are you getting out of them? Like how long would it take to convert an hour long video say from avi to DVD or something like that? I suppose the quality makes a difference as well like it would probably take longer in converting HD quality videos.

It would be nice to have a conversion chart of times for different types of conversions with different processors because time is an important element in video work.

Skylake i7-6700, yawn. Only four cores. This is 2015. 4 cores is so 2000 and late. If you are looking to drop some coin on a new cpu that has more threads than you can shake a stick at, I would wait for a Skylake-E version which should be out next year. I am hoping for a 12 core variant that gives a whopping 24 threads of deliciousness.

And with that, you can start deinterlacing all the old rare video like a hot knife through butter.

[-Habanero-]

I do lots of video conversions and with a faster processor I would like to convert a lot of old rare video cassettes I have. The Skylake I7 6700 is coming out next week but there is also the I7 5820 with 6 cores that has been out for awhile. Which would be faster. I think they would be priced about the same. It seems like 6 cores would be faster since you would divide the work among 6 cores. AMD also has an an 8 core processor but it doesn't seem like the AMDs are as good.

I know that some of you have the latest hardware. What kind of conversion speed are you getting out of them? Like how long would it take to convert an hour long video say from avi to DVD or something like that? I suppose the quality makes a difference as well like it would probably take longer in converting HD quality videos.

It would be nice to have a conversion chart of times for different types of conversions with different processors because time is an important element in video work.

Skylake i7-6700, yawn. Only four cores. This is 2015. 4 cores is so 2000 and late. If you are looking to drop some coin on a new cpu that has more threads than you can shake a stick at, I would wait for a Skylake-E version which should be out next year. I am hoping for a 12 core variant that gives a whopping 24 threads of deliciousness.

And with that, you can start deinterlacing all the old rare video like a hot knife through butter.

Your assumption that everything can be perfectly parallelized is cute. I'm surprised you still post here after embarrassing yourself in that 10 year CPU thread.

[sophisticles]

And with that, you can start deinterlacing all the old rare video like a hot knife through butter.

Not even close, most deinterlace filters are not that well optimized for high core counts, a 12 core cpu would not speed up that task appreciably. Pure clock speed and IPC improvements are what you want.

This holds true for x265 as well, the company rep that goes buy x265 on various forums has said that x265 loves fast single thread performance and gobs of memory bandwidth.

[sophisticles]

There are leaked benchmarks at various Taiwanese sites. You get about 5-10% increase across the board on various applications - which is the same every new generation of processors. There hasn't been a big jump in performance in a long time. Thanks to no competition.

I have been reading this no competition theory for years now but to me it never made any sense. Consider, if Intel wants to maximize sales of the new processors they would want as many people to run out and upgrade, including people that bought Ivy Bridge and Haswell based cpu's. BUT, if Skylake only delivers an average 10% improvement over current cpu's then those consumers will not be in any hurry to upgrade, most likely even most Sandy Bridge owners will decide to wait until the next offering or AMD's 2016 cpu's.

From Intel's perspective thy would want to release a cpu with 50%+ greater performance because that would entice many Ivy Bridge and Haswell owners to upgrade again in a short time frame.

I think that we are seeing the limits of the x86 architecture, coupled with the limits of current manufacturing technology, added to the limits of electrical circuits.

Intel could offer a much faster cpu, but they would have to make it physically much bigger, to fit more circuitry on a wafer and it would consume way more power, but within the current size and power constraints I think they have extracted as much as they possibly can out of their current cpu's

It wouldn't surprise me to see Intel, in a generation or two, release a non-x86 cpu for the desktop, they tried years ago to move to a RISC based architecture with the Itanium and failed due to legacy software concerns.

With the advent of Win10, I think we're probably 2-3 years away from a 128bit RISC cpu.

[poisondeathray]

There are leaked benchmarks at various Taiwanese sites. You get about 5-10% increase across the board on various applications - which is the same every new generation of processors. There hasn't been a big jump in performance in a long time. Thanks to no competition.

I have been reading this no competition theory for years now but to me it never made any sense. Consider, if Intel wants to maximize sales of the new processors they would want as many people to run out and upgrade, including people that bought Ivy Bridge and Haswell based cpu's. BUT, if Skylake only delivers an average 10% improvement over current cpu's then those consumers will not be in any hurry to upgrade, most likely even most Sandy Bridge owners will decide to wait until the next offering or AMD's 2016 cpu's.

From Intel's perspective thy would want to release a cpu with 50%+ greater performance because that would entice many Ivy Bridge and Haswell owners to upgrade again in a short time frame.

That's certainly one way to look at it. But you have to factor in the cost of R&D schedule depreciated over time . They've invested many years ago, for skylake and future architectures. Those are previous capital expenditures that they have to make up for - all their financial statements that they release, all their investors and financial projections rely on it. This is the model of planned obsolecense - they need to maximize sales of each generation (ie "milk it") . It's not as if they are hurting financially doing what they are doing now - it's difficult to "hurt" when you're essentially a monopoly LOL. Jumping a few notches early (instead of "tick tock") at means 100's of billion down the drain, poor yields because of low maturity. They can afford to go slowly and test out each fab process on the desktop (desktop users are the guinea pigs, early silicon) , make refinements to the final golden silicon before selling to the real cash cow - the enterprise market. They can't afford problems with their real customers. That's why the Xeons and "E5, E7" series chips come later. However, if another company "X' were to introduce something remotely competitive, it would force their hand

[SameSelf]

Not even close, most deinterlace filters are not that well optimized for high core counts, a 12 core cpu would not speed up that task appreciably. Pure clock speed and IPC improvements are what you want.

This holds true for x265 as well, the company rep that goes buy x265 on various forums has said that x265 loves fast single thread performance and gobs of memory bandwidth.

I use QTGMC MT regularly and have optimized the settings to use all 8 threads at 98+% on my i7-3770K. Without the MT version, it crawls at 20% to 30% utilization on four cores with no hyperthreading. QTGMC is a hungry beast even on SD video at slow settings.

I should also add that x264/265 are the very last steps of my workflow after everything else is done from deinterlacing, denoising, color grading, building timelines, etc. IOW, the final encoding is the easiest part of the process. It's the other stuff that bogs down my workflow and cpu clocks are like liquid gold for me.

[hogger129]

Obviously the most powerful one on the market right now (whatever the top-notch i7 is, 5960x?). But also get what you can afford. When I built a new computer this past winter, I opted for the 4770k as it was powerful but also affordable. Obviously I'm not going to be able to use x265 at a reasonable speed with it, but hard drive space is super cheap right now too so I'm happy just doing BD9's or BD25's with x264 and BD Rebuilder.

[sophisticles]

Not even close, most deinterlace filters are not that well optimized for high core counts, a 12 core cpu would not speed up that task appreciably. Pure clock speed and IPC improvements are what you want.

This holds true for x265 as well, the company rep that goes buy x265 on various forums has said that x265 loves fast single thread performance and gobs of memory bandwidth.

I use QTGMC MT regularly and have optimized the settings to use all 8 threads at 98+% on my i7-3770K. Without the MT version, it crawls at 20% to 30% utilization on four cores with no hyperthreading. QTGMC is a hungry beast even on SD video at slow settings.

I should also add that x264/265 are the very last steps of my workflow after everything else is done from deinterlacing, denoising, color grading, building timelines, etc. IOW, the final encoding is the easiest part of the process. It's the other stuff that bogs down my workflow and cpu clocks are like liquid gold for me.

This contradicts what I said how exactly?

[-Habanero-]

Not even close, most deinterlace filters are not that well optimized for high core counts, a 12 core cpu would not speed up that task appreciably. Pure clock speed and IPC improvements are what you want.

This holds true for x265 as well, the company rep that goes buy x265 on various forums has said that x265 loves fast single thread performance and gobs of memory bandwidth.

I use QTGMC MT regularly and have optimized the settings to use all 8 threads at 98+% on my i7-3770K. Without the MT version, it crawls at 20% to 30% utilization on four cores with no hyperthreading. QTGMC is a hungry beast even on SD video at slow settings.

I should also add that x264/265 are the very last steps of my workflow after everything else is done from deinterlacing, denoising, color grading, building timelines, etc. IOW, the final encoding is the easiest part of the process. It's the other stuff that bogs down my workflow and cpu clocks are like liquid gold for me.

MSU quality evaluator takes 99% of my 8 threads too. You know what the awesome part is? It's 5 times faster when I run it on a single thread! I think people from the land of Steers and Queers shouldn't be allowed near any electronic device. They'll make companies rich by spending $10,000 on a system they think will last them 10 years. Oh wait...

And PDR is right, planned obsolescence is the name of the game. They released P4 at 1.3 GHz in 2001 and at 3.8 GHz in 2004. 20 of the exact same products with a different clock speed, because it was technologically impossible to clock them so high from the start, right? The computers you see commercially available today are garbage compared to what can be and was done a long time ago.

[SameSelf]

I can see I have confused some by my statements. Others are butt hurt. Let's KISS.

For a video editing machine, only an Extreme version of the i7 will suffice. Anything less is ghetto, like trying to mine BTC with nVidia. That means the 5820K, 5930K, and ideally the 5960X with its 8 cores. But, 2016 will hopefully see a refresh of the Extreme line up on Skylake with perhaps 12 cores. So it comes down to should you wait or not.

[glenpinn]

Skylake i7-6700, yawn. Only four cores. This is 2015. 4 cores is so 2000 and late. If you are looking to drop some coin on a new cpu that has more threads than you can shake a stick at, I would wait for a Skylake-E version which should be out next year. I am hoping for a 12 core variant that gives a whopping 24 threads of deliciousness.

LOL, here we go again, the speed demons are at it again.

I think Jagabo mentioned something earlier about this stuff not being all about speed or brute power, and he is right, and that is what really gripes me when people carry on about video editing and not understand the subject properly, and are simply after the sheer speed and horsepower from a CPU rather than focus on the software and the quality of your outputs.

Speed does not mean quality.

Those who want to use Intel Quick Sync encoding have no choice but to use a Quad cpu, because only those have the Intel HDxxxx or Iris Pro graphics chip on board, none of the bigger Chips have graphics built in, so you can only use software encoding.

Depending on your software and how well it uses Quick Sync, those 4 Core Chips will pump out a conversion faster than a 6 core, and i know because i have a friend who has a 6 core (Haswell) and we did some testing in Handbrake, outputting the same 1080/50p mp4 videos to 720/50p, he had to use software, i used QS, and my 3770 killed his 6 core.

Quick Sync was not very good in earlier times, i believe it has been much improved as far as quality goes, and i know many who are now using it, however i am from the old school where we still do our encoding using software.

I currently use VRD Pro with the unlocked x264 encoder, and for HD video editing and encoding, a 3770 Ivy Bridge quad is perfectly fine, but it is a bit of a dog with 4k editing.

Vrd has its own implementation of Quick Sync, and it is only about twice as fast as doing the same encode using the software, however using Handbrake to output the same encode using equivalent settings, it is about 2 times faster than the Quick Sync in VRD.

I will be moving to 4k soon, and because i don't use Quick Sync, i will be building myself a new 6 core micro editing computer that will handle 4k editing and i will have a 4k screen, as well as a 1080 screen attached.

I can see I have confused some by my statements. Others are butt hurt. Let's KISS.

For a video editing machine, only an Extreme version of the i7 will suffice. Anything less is ghetto, like trying to mine BTC with nVidia. That means the 5820K, 5930K, and ideally the 5960X with its 8 cores. But, 2016 will hopefully see a refresh of the Extreme line up on Skylake with perhaps 12 cores. So it comes down to should you wait or not.

Seriously dude, you have no idea what you are talking about, sounds like you might be a young teenager who has been sucked into all this crazy logic that brute power is everthing.

Bad news chum, for 1080p editing, an i5 and i7 quad core is perfectly fine, so you go out and grab you big mutha 12 core cpu that will cost you a small fortune, and you have a nice time editing with it, but most of what you have said is rubbish.

[SameSelf]

^ http://www.anandtech.com/show/8426/the-intel-haswell-e-cpu-review-core-i7-5960x-i7-593...5820k-tested/5

[glenpinn]

^ http://www.anandtech.com/show/8426/the-intel-haswell-e-cpu-review-core-i7-5960x-i7-593...5820k-tested/5

And your point is what ??????

[SameSelf]

Wht was urs?

[glenpinn]

Wht was urs?

I was asking you what the point was by posting up that link, and not say anything, so again, i am asking you what the link was all about ?

You are the one who said only an Extreme version of the i7 will suffice. Anything less is ghetto (as well as other stuff that i won't bother quoting) and that is just plain rediculous, and you just come over as one of these people who are infatuated by speed and power, but seemingly know very little, nor understand much about video editing or encoding.

You can throw up all the benchmark links you like, editing and encoding is not just about brute speed or power, if i want speed, i can just use my 3770 cpu with Quick Sync and get as good, or faster results than you will with some of these bigger cpu's that need to encode using software, and are more expensive to buy.

You can have your 12 core cpu, i choose to do my work on what i feel is right for me, and while i am still doing 1080p editing, my 3770 cpu is fine.

[SameSelf]

Whoa u need to chll dude tl;dr

[-Habanero-]

Wht was urs?

I was asking you what the point was by posting up that link, and not say anything, so again, i am asking you what the link was all about ?

You are the one who said only an Extreme version of the i7 will suffice. Anything less is ghetto (as well as other stuff that i won't bother quoting) and that is just plain rediculous, and you just come over as one of these people who are infatuated by speed and power, but seemingly know very little, nor understand much about video editing or encoding.

You can throw up all the benchmark links you like, editing and encoding is not just about brute speed or power, if i want speed, i can just use my 3770 cpu with Quick Sync and get as good, or faster results than you will with some of these bigger cpu's that need to encode using software, and are more expensive to buy.

You can have your 12 core cpu, i choose to do my work on what i feel is right for me, and while i am still doing 1080p editing, my 3770 cpu is fine.

For f*cks sake just ignore the trolling half-wit. He is right up there with newpball and isn't worth the intellectual effort.