Puget Systems print logo

https://www.pugetsystems.com

Read this article at https://www.pugetsystems.com/guides/969
Article Thumbnail

Premiere Pro CC 2017.1.2 CPU Comparison: Skylake-X, Kaby Lake-X, Broadwell-E, Kaby Lake, Ryzen 7

Written on June 26, 2017 by Matt Bach
Share:

Introduction

AMD's launch of Ryzen in March marked the first time in years that Intel was seriously challenged in the enthusiast CPU market. Now, with the Skylake-X and Kaby Lake-X CPUs officially launched, we get our first taste of Intel's response. It is really just a taste, however, since Intel has only released their 4, 6, 8, and 10 core CPUs so far. According to Intel's E3 press release, there will be a 12 core CPU available in August along with 14, 16, and 18 core CPUs in October. Because of this, we won't be able to look at the entire Skylake-X line-up today, but what we can do is see how the CPUs that are currently available compare to Intel's previous generation CPUs as well as AMD's Ryzen 7 CPUs.

There are a wide variety of tasks we could test in Premiere Pro to see how these new CPUs perform, but in this article we will specifically be looking at:

  1. Rendering previews
  2. Exporting
  3. Performing a Warp Stabilization
  4. Live playback performance

Our testing includes test footage with resolutions of 4K, 6K, and 8K using six different codecs (more information in the test setup section). In total, we ran nearly 90 unique tests on 10 different CPUs resulting in almost 900 data points. If you would rather skip over our analysis of the individual benchmarks, feel free to jump right to the conclusion section.

Test Setup

Since the new processors include two different CPU architectures, our testing platform is less straightforward than normal. Most of the new CPUs support DDR4-2666 RAM, although the Intel Core i7 7740X (Kaby Lake-X) CPU only supports four sticks of RAM rather than eight. In addition, the Intel Core i7 7800X (Skylake-X) only supports DDR4-2400 RAM for some odd reason. We opted to test with the highest officially supported RAM configuration for each CPU, so components used in our X299 test platform requires an entire table just for itself:

To act as a comparison to the new CPUs, we will also be testing AMD's Ryzen 7 1700X/1800X CPUs as well as the Intel Core i7 7700K (Kaby Lake) and Intel Core i7 6850K/6900K/6950X (Broadwell-E) CPUs. Again, we will be testing with the best officially supported RAM configuration for each CPU.

Inconsistent RAM aside, one thing we do want to point out is how much more affordable the Skylake-X CPUs are compared to the previous generation Broadwell-E CPUs. Where a six core CPU from Intel used to cost over $600, you can now purchase an eight core CPU for roughly the same cost. In fact, the i7 7800X is even a hair less expensive than AMD's Ryzen 7 CPUs. This means that regardless of any performance gains we may see, Skylake-X is at the very least a big deal in terms of cost.

Before getting into our testing, we also want to point out is that while our test platforms are using a single hard drive, that is not actually what we would typically recommend to our customers. We have found in our testing that using at least a two drive configuration with the media cache and scratch files on a secondary drive can make a big impact when it comes to importing footage and tasks like conforming audio. However, since we will not be testing any of these actions in this article we opted to use a single drive simply to cut down on the number of variables.

Most of the media we will be using is available from the Sample R3D Files and were transcoded to the various codecs we wanted to test.

4K

EPIC DRAGON
3840x2160
23.976 FPS

6K

WEAPON 6K
6144x3077
23.976 FPS

8K

WEAPON 8K S35
8192 x 3456
50 FPS

To test exporting and rendering previews we used a moderately complex timeline involving multiple clips, Lumetri Color, multicam footage, and some other effects like a logo overlay, Gaussian Blur and Cross Dissolves. If you want a more in-depth look at what our timelines look like, we recorded a short video explaining our test process:

Our 4K VR testing was performed using the "Sample 1 - Ring road motorbike ride" footage from the Autopano Video Benchmarking page. We tested using both some built-in Premiere Pro effects (Lumetri Color, text overlay, and cross dissolve) as well as using the Mettle Skybox 360 VR Tools and Skybox 360/VR Transitions plug-ins to apply a number of effects such as Denoise, Rotate Sphere, Sharpen, and Iris Wipe. We typically try to avoid using plug-ins in our testing, but since Premiere Pro only has basic support for VR at the moment we felt it made sense to also look at the popular Mettle plug-in for VR projects. In addition, Adobe recently acquired Mettle Skybox which means all Creative Cloud customers will be able to use this plugin for free by the end of the year.

Rendering Previews

Rendering previews is something you never want to have to do since it interrupts your workflow, but if you do complex editing it is sometimes unavoidable. Because of this, being able to render previews as quickly as possible is often an important part of a Premiere Pro workstation.

Premiere Pro Skylake-X 7900X 7820X 7800X Render Previews Benchmark

[+] Show Raw Results

Since we are comparing 10 different CPUs across 16 different projects that include a range of resolutions and source codecs, it would take us a long time to go through the results one by one. Because of this, we decided to compile all the results into an overall average for each CPU compared to the Intel Core i7 6850K. This CPU has long been our entry recommendation for most video editing workstations which should make it a great point to compare the other CPUs against. If you wish to examine the raw results yourself, you can do so by clicking on the "Show Raw Results" link under the chart.

Even with the results averaged, there is a lot to discuss. Starting at the bottom of the chart with the quad core CPUs, the Intel Core i7 7740X ended up being about 5% faster than the Core i7 7770K. This is more than we expected given that they have the same maximum Turbo Boost frequency, but the higher base clock and support for DDR4-2666 RAM made a bigger difference than we expected. However, for only another $50 the Core i7 7800X is about 13% faster than the Core i7 7740X. In fact, it is even a bit faster than AMD's Ryzen 7 1700X for a few dollars less, and only a few percent slower than the Ryzen 7 1800X.

Moving up to the Core i7 7820X, we saw a really nice 17% jump in performance over the Core i7 7800X. This makes it both cheaper and 20% faster than the similarly priced Core i7 6850K and fares well compared to the AMD Ryzen 7 1800X from price-to-performance viewpoint.

At the (current) top of Intel's new CPU stack, the Core i9 7900X was a bit faster than the Core i7 7820X but only by about 5%. For a $400 price premium that isn't all that great, but compared to Intel's previous generation processors it is about 6% better than you could previously get at almost half the cost.

Exporting

Exporting is one of the biggest time sinks for a Premiere Pro user and is often the go-to metric for measuring performance. For this test, we looked at 35 different combinations of source footage and export settings. This includes 4K, 6K, and 8K resolutions along with H.264, DNxHR HQ, ProRes 422 HQ, ProRes 4444, RED, and H.265 codecs. In addition, we will also be looking at VR projects using both built-in effects as well as with the popular Mettle plugin.

Premiere Pro Skylake-X 7900X 7820X 7800X Exporting Benchmark

[+] Show Raw Results

Just like in the previous section, since we are comparing 10 different CPUs across 35 different projects that use a range of resolutions, source codecs, and export settings, we decided to compile all the results into an overall average for each CPU compared to the Intel Core i7 6850K. If you have the time and will, feel free to examine the raw results yourself by clicking on the "Show Raw Results" link below the chart.

The results for exporting are not much different than what we saw when rendering previews, but again starting with the quad core CPUs we saw about a 6% performance gain with the Core i7 7740X compared to the Core i7 7770K. For a small price increase, however, the Core i7 7800X is even faster - coming in at about 14% faster than the Core i7 7740X. This also makes it a hair faster than AMD's Ryzen 7 1700X for a few dollars less and only 3% slower than the Ryzen 7 1800X.

Moving up to the Core i7 7820X, we saw a terrific 20% performance gain over the Core i7 7800X which makes it both cheaper and significantly faster than any of the previous generation Core i7 6XXX CPUs. Even compared to the i7 6950X the Core i7 7820X should be noticeably faster when exporting projects while being less than a third the cost!

The performance gain isn't quite as large with the Core i9 7900X, but it was still about 6% faster than the Core i7 7820X on average. If you dig into our raw results, you will notice that this performance gain was mostly from the projects that used either RED or VR footage. If your work involves either of these types of footage, you will likely notice a large improvement in performance but otherwise there is not a large difference in export speeds compared to the i7 7820X.

Warp Stabilize

While exporting and rendering previews may be the easiest and most common thing to benchmark in Premiere Pro, we have received a lot of feedback that performing a warp stabilize is another task where high performance is important. We have found that the time it takes to complete a warp stabilize analysis does not vary much on the source codec, but the resolution does make a big difference so we will be testing with a 4K H.264 clip as well as a 8K ProRes 4444 clip.

Since warp stabilize is not well threaded (meaning that it does not take great advantage of multiple CPU cores), we also opted to split our test clip into multiple parts and analyze all of them at the same time in order to force Premiere into making more effective use of all the CPU cores. This is a trick some people use to speed up the time it takes to analyze a single clip, but it is also a great indicator of performance when you have multiple clips that need to be stabilized. With this in mind, we not only timed how long it takes to apply a warp stabilization effect to a single 10 second clip, but also how long it takes if the clip is split into 2, 4, 8, and 16 "sub-clips" that are all analyzed at the same time.

Premiere Pro Skylake-X 7900X 7820X 7800X Warp Stabilize Benchmark

[+] Show Raw Results

While we were able to get away with an overall average in the last two sections, the results for warp stabilize depends so heavily on the number of clips you are analyzing that we decided to show both the average result with a single clip and with 16 clips. Feel free to examine the raw results if you would like to see the results for 2, 4, and 8 clips, but the results for those fell between the single and 16 clip results exactly as you would expect so we opted to not muddy the chart with even more data.

Starting with a single clip, all of the 6-10 core CPUs (both Broadwell-E and the new Skylake-X) performed roughly on par with each other. There was really no advantage to using a Broadwell-E CPU or a Skylake-X one. However, due to the higher single-threaded performance of the Core i7 7740X and 7770K, both of those CPUs gave excellent results coming in at 20-27% faster than the higher core count Intel CPUs. Between those two CPUs, the Kaby Lake-X Core i7 7740X was about 7% faster than the Core i7 7770K.

Upping the clip count to 16, however, gives us vastly different results. With this many clips, the higher core count CPUs are able to really take advantage of all their cores. Starting with the Core i7 7800X, we found that it performed roughly on par with the previous generation Core i7 6850K even though it is only about 60% the cost. Moving up to the Core i7 7820X, we saw about a 19% performance gain over the i7 7800X. The Core i9 7900X was a further 8% faster than the i7 7820X, outperforming or matching the more expensive i7 6900K and 6950X.

Live Playback

Live playback performance is a challenge for us to accurately test since whether you can play a timeline at full, half, quarter, etc. resolution is highly dependent not only on your source footage but also what effects you have applied to the timeline. To try to keep things universally applicable, we opted to test 10 different projects using 4K, 6K, and 8K footage with multiple codecs across three relatively simple timelines. What we wanted to see was if we would be able to play the timeline at either full or half resolution without dropping any frames - even if it was just one or two at the very start of playback.

Basic

  • 4 clips in series
  • No effects
  • No transitions

Lumetri Color

  • 4 clips via multicam sequence
  • Lumetri Color Correction
  • No transitions

Lumetri & Cross Dissolve

  • 4 clips via multicam sequence
  • Lumetri Color Correction
  • Cross dissolve

Premiere Pro Skylake-X 7900X 7820X 7800X Live Playback Benchmark

[+] Show Raw Results

If there is a codec you typically use that we happened to use it in our testing, we recommend taking a look at the raw results to see how each CPU performed for that specific codec. For a more general overview, we created a scoring& system based on how many timelines each CPU was able to run at either full or half resolution. Every successful playback without dropping any frames at full resolution is 1 point while every playback at half resolution is .5 points. This isn't a perfect system by any means, and if you have any recommendation on a better method we would love to hear it in the comments!

Using our score-based system, you can get a pretty decent idea of how each CPU might affect live playback performance. Honestly, the results were a lot closer than we expected. The Core i7 7800X and 7820X both had a score of 20 while the i9 7900X had a core of 21.5 since it was able to play an additional three timelines at half resolution - the 6K RED Basic and Lumetri timelines along with the 8K ProRes 4444 Basic timeline. This means that the i9 7900X is slightly better than the previous generation CPUs for live playback, but the i7 7800X and i7 7820X are really about the same.

Conclusion

New CPUs are always exciting to test, but the new Intel Skylake-X and Kaby Lake-X CPUs are a bit odd since only about half the CPUs are currently available. At the moment, we have four, six, eight, and ten core options with a 12 core CPU due to arrive in August. After that, there are another three CPUs coming in October with 14, 16, and 18 cores.

This means that we really only have the low and mid-range CPUs available right now which were only moderately faster than the CPUs they are replacing. What is different, however, is the pricing of these CPUs. Looking at the overall performance including rendering previews, exporting, warp stabilize, and live playback, we saw the following average performance from the CPUs we tested:

Premiere Pro Skylake-X 7900X 7820X 7800X Overall Benchmark Results

If you simply compare the new CPUs to the ones from the previous generation that have similar core counts, the average performance gains are only around 5%. However, there were a few times not shown in the average where Skylake-X did have a significant nice bump in performance - namely if you work with RED or VR footage. In these cases, we found that the Core i9 7900X was about 20% faster than the Core i7 6950X when exporting and about 10% faster when rendering previews. The i9 7900X was also better at playing back 6K RED footage live - it was the only CPU tested that was able to play two of our three 6K RED timelines at half resolution without dropping any frames.

Pricing may not be as exciting as raw performance, but cost is really the bigger story here. In most cases you should be able to get about 5% higher performance for just a bit more than half the cost! This also affects the ever on-going "AMD vs Intel" debate. When we looked at the Premiere Pro performance of AMD's Ryzen CPUs back in March, the AMD CPUs were able to hold their own pretty well against the Intel CPUs. While the AMD Ryzen CPUs have seen performance gains since then (due to driver/BIOS and software improvements), the lower price of Intel's Skylake-X CPUs puts them firmly in the lead at the moment.

If you examine the individual results from each section, you will find that the Intel Core i7 7800X is a hair cheaper than the AMD Ryzen 7 1700X while being slightly faster for exporting/rendering previews and about 24% faster for warp stabilize. Averaged out, this makes the i7 7800X about 10% faster overall for $10 less. One step up from that, the Intel Core i7 7820X is a bit more expensive than the Ryzen 7 1800X, but in exchange it is a solid 17% faster at exporting, 15% faster at rendering previews, 32% faster at warp stabilize, and better at live playback. On average, this makes the i7 7820X about 18% faster than the AMD 7 1800X for $100 more. X299 motherboards do tend to be around $120 more than the lowest end AMD motherboard which will negate most of this pricing advantage, but keep in mind that the more affordable AMD motherboards are often lacking in features like WiFi or USB 3.1 and often use lower quality components to achieve the low price points. In addition, the AMD Ryzen CPUs only support 64GB of RAM and 24 PCI-E lanes while the Intel 6-10 cores support up to 128GB officially (or up to 512GB on X299 boards that support Registered RAM) and have either 28 or 44 PCI-E lanes.

The bottom line is that these new Intel CPUs are attractive more due to their pricing than their raw performance. They showed nice performance gains with RED and VR footage, but overall they are only about 5% faster than the previous generation Broadwell-E CPUs. This may change when the higher core count CPUs are launched later this year, but for now there is little reason to upgrade to Skylake-X if you have a Broadwell-E CPU. On the other hand, if you are already in the market for a new workstation the large drops in price means that you should be able to get much higher performance for your dollar than you could with the previous generation Intel CPUs.

Tags: Premiere Pro, Skylake-X, Ryzen 7, Broadwell-E, Kaby Lake-X
Tuchulu

"the Intel Core i7 7800X is a hair cheaper than the AMD Ryzen 7 1700X"

That's objectively true, but...
x299 motherboards are pricer than X370 and B350 motherboards.
Factoring in motherboard cost would've made everything much more complicated, so I understand the decision to leave them out of the discussion. However the added cost of x299 is real and it should at least be acknowledged.
e.g. It would make no sense to recommend Kaby Lake X over its equivalent Kaby Lake CPU for a gaming build, even if the CPU prices are roughly the same.

Ryzen 7 1700 can easily be overclocked to match the 1700X and it costs two thirds of the price.
I realize overclocking is not the same for a manufacturer, since it's much harder to guarantee stability of aggressively overclocked systems, so that's probably a moot point.

I also realize that power consumption and thermals are beyond the scope of this article so I won't bring it up.

Posted on 2017-06-27 02:01:01
Hugo Scali

Also the fact that the earrings memories of 2400mhz with the ryzen, if they used memories of 3200mhz like the gskill flare x would have a 20% of difference in favor of the R7 although in intel the increase of speed of ram does not vary in but of 2% increase in premiere.

Pd: I have it trying on my youtube channel although in SPANISH

Posted on 2017-06-28 17:19:31
Tuchulu

Puget Systems is more concerned with stability and being able to provide the best support in the industry. Overclocking the ram over the guaranteed specs is a bad idea for them. As a home user you can do whatever you want. This is the same problem I pointed out about overclocking the Ryzen 7 1700. If you do it at your own risk, that's fine, but if Puget does it then they also have to support it.

Also, training their Tech Support for Ryzen would take time and money, they need a really compelling reason to do so.
Maybe Thread Ripper will be that reason, but we won't know until we test it.

Posted on 2017-06-28 18:06:14
Chris Markes

Well, R7 1700 cost is now down to $310 from newegg (and you get a cooler). It takes about 20 seconds to go into the bios and adjust it to 3.7ghz and 1.215 volts. Save & exit and you're done. Mine has been stable this way for over a month of rendering 4k, GTAV, on the stock fan (saved $$ there), and this was my very first PC build. ZERO issues, even on hot days. I agree Pugets' target audience is the higher end crowd that has money to burn, but I looked at it and well, now I can get 85% of the performance, but I have a 1080ti + Samsung 960 pro, and a beautiful Phanteks Etho Lux (it's just unbelievable!!! flush mounted SSD's, big enough to fit any board size/video card size) case for less than I would have paid with a puget system.

With everything being prettymuch plug n play, the home user/hobbyist should really build their own. I can see the professional users wanting the tax writeoff / support, but I just could not justify the additional costs. Much respect for Puget for being the best in the business and responding to my myriad of emails though. If you're going to have a machine built, but I would encourage those that are "afraid" to go for it, it's honestly easier than working on most cars and it's pretty difficult to break things, especially with a cmos reset on the board itself. If you can use a philips screwdriver and not rush through things, and read a manual, it's very simple.

I would like to point out that Puget should consider the heat issues with the new Intel chips, which I would think would affect long term reliability. If I were an Intel fan, I would actually skip this round and wait: The new ars technica article just released today is very telling: https://arstechnica.com/gad...

Posted on 2017-07-04 18:23:09
Tuchulu

Puget Systems Premiere Pro Workstations are ordered either consumers that don't mind spending the money to buy the best of the best (top dog premium, I think it's called) or companies that require stability and reliability. Saving a few hundred dollars today but having 20% reduced performance for the next three years makes no sense for a company. Also, every minute that PC isn't working at full capacity that means reduced productivity, and that's not good.
As an enthusiast, if I needed a new PC I would just build it myself, because my income isn't depending on it, so clearly the people and companies that buy from Puget have different needs than I do.
It is probably not a worthwhile investment for Puget Systems to fully support another platform and CPU lineup unless it has a significant performance advantage. When Thread Ripper comes out, if the performance is there, I have no doubt it will be available and supported.

Posted on 2017-07-04 18:53:00
AlbertS

Chris - I don't get your point. Many of us "non-pros" see our next upgrade for 4K as a significant investment and don't believe "cheaper is always better" or have money to burn or get a tax break.. Puget has tested Ryzen but Intel's new i9s at lower prices (thanks to AMD) and better performance than Ryzen is where my money will go as I've never had an Intel CPU fail and nor have my pals. Puget builds a range of quality systems using quality components at reasonable prices and assists users in sizing the solution to customer needs of a wide range of HEDT apps based on actual testing and benchmarks. AMD fell so far behind in both CPUs and GPUs and won't get my money for at least another year while their products prove themselves. In spite of the demand for Ryzen, they are still running at a loss and may not even survive. Intel's first quarter profit was greater than AMDs total turnover. You didn't skimp on the rest of your components so I don't see why you would want to save a few dollars on your CPU, cooling, and m/board but I wish you luck with your DIY investment. Like my next car, my next PC will come from a reputable manufacturer. I'm also looking for a 4K software/hardware editing solution and have found no better source of info than Puget. Both hardware and software are improving so I'll wait to see the results later this year.

Posted on 2017-07-05 06:36:05

I agree a review service should base it's reviews on a standard. Not an overclock. However, I hope they will consider returning to this test now that BIOS 1.0.0.6 Agesa has released to most motherboards which brings support for 3200MHz with compatible ram. I updated my bios. rebooted and boom, natively running 3200MHz

Posted on 2017-07-12 15:44:25
lin lin

Thanks for doing this review. X299 motherboard will cost at least 220USD + tax or and higher, but a good overclockable B350 will only cost 80USD + tax. or 51USD + tax for non overclockers.
In my quick tests. GPU only lightly taxed with one luma curve. the r7 1700 @ 3.9ghz with 16gb 2666 ram GTX960 its actually quite a bit quicker at exporting than my 5820K @ 4.4Ghz 64gb 2400 ram. GTX1060 and about the same as my 5960k @ 4.2 with 32GB 2666 ram. GTX1060 Maybe, just Maybe, the RAM speed might affect the rendering speed when rendering with GPU. just like how ryzen is more affected by RAM speed in programs and able to push more frames in video games when faster GPU is paired with. and how can your 7740K be 5% faster than 7700K when most other benchmark website show no improvement in speed between 7740k and 7700k when paired with 3200rams. 2666 is really making 7740k that much faster than 2400?

Posted on 2017-06-27 03:20:13
BArt

will you do a similar comparison for Photoshop and Lightroom?

Posted on 2017-06-27 11:17:24

Yes, we will have articles on both those (as well as After Effects) hopefully in the next few days.

Posted on 2017-06-27 16:52:50
CASEYV

+1 After Effects. Can't wait!

Posted on 2017-06-27 21:13:29
MarkN

Would be very interested in CPU comparisons in Illustrator (as would a large number of other Illustrator users, I suspect)! Also GPU comparisons with and without GPU support enabled in preferences, if only to prove GPU power isn't much of a factor in Illustrator.

Posted on 2017-12-31 21:50:39
hernan

start using 3.200 mhz memory with ryzen https://www.youtube.com/wat...

Posted on 2017-06-28 17:22:43

Hey hernan, at the moment we don't plan on testing with >2400 MHz RAM since we value reliability and stability so highly. Being able to render a bit faster doesn't really mean anything if the machine is unstable and causes you to lose your work.

We actually have wanted to do some testing comparing different speeds of RAM with Ryzen simply to settle these debates, but in our testing we have run into stability issues with even DDR4-2666 RAM (which from what I've read is still a very common issue with that platform). I can't imagine getting DDR4-3200 stable in a professional environment - especially not at CL14 like they did in the video you linked! Keep in mind that not even DDR4-2400 is officially supported by AMD when using 4 sticks. Their officially-supported configuration is just DDR4-1866 if using 4 stick in dual channel mode or DDR4-2133 with 4 sticks in single channel mode: https://community.amd.com/c...

I'm sure higher speed RAM would improve performance a bit - although I don't believe it will be as much as that video shows in most situations. Our testing is first and foremost to help us determine what hardware to offer to our customers and since going that far outside of spec is something we usually heavily avoid (unless there is an extremely compelling reason to do so) something like higher frequency RAM is pretty low on our list of things to test. If you build your own computer and are willing to take on whatever risks come with being out of spec, that is 100% your choice. But the majority of our customers come to us because they are not the type of people who want to tinker or even think about their computer. They just want it to work so they can get their work done - hence why reliability is much more important to us than raw performance in most cases.

Posted on 2017-06-28 17:37:43
johndoe

Did you mean to write "rank" instead of "channel"?

Posted on 2017-06-29 08:54:07

Your right, that should have been dual and single rank, not channel. I got dual rank and dual channel mixed up in my head, but the rest of my comment still stands

Posted on 2017-06-29 17:13:05

Your stability concerns could be with the Prime Pro you're using. With the BIOS 1.0.0.6 Agesa update my X370 Taichi and G.Skill 32GTZ runs out the gate at 3200MHz CL15 and I've experienced no issues. Ran Prime95 and Memtest with no errors. My main point i keep making on these review sites is, while we're all excited for Ryzen, it's brand new. and there a kinks to work out in the system for sure. Stress test environments are going to change drastically over the first 6 months of a new launch. I hope Pudget will follow up with updated bench builds as that happens. within reason of course.

Posted on 2017-07-12 15:59:13
pillybilly

You are pretty boring... No one who works seriously, overclock a new cpu. They are barely stable now with default values...

Posted on 2017-08-29 21:19:54
AlbertS

"No one who works seriously, overclock a new cpu"
If you mean Intel, all K and X series are unlocked specifically for overclocking. I've never had one fail. 7700K is stable close to 5Ghz water cooled. I've never tried Ryzen but have seen latest Threadrippers running all cores overclocked to 4Ghz. Temps are good and systems stable.

Posted on 2017-08-30 07:16:17
Andrew Pealock

Only thing about this article, it is still using old Ryzen pricing. The 1700x in particular has dropped at least 50-60 dollars since then. I got mine for $330 shipped. The 1800x is also around $400 now, so about $90-100 dollars cheaper. So the price to performance has still got AMD pretty far ahead.

Posted on 2017-06-30 04:47:29
CASEYV

Hi Matt, Can you advise what CPU cooling you were using for these benchmarks? And also if you experienced any super high CPU temperatures whilst running the tests?

Posted on 2017-06-30 23:31:28

Cooler was a Noctua NH-U12DX i4, but I didn't log CPU temperatures, but I know they weren't anything to high because the CPU fan wasn't really ramping up much. If I had to guess, I was probably hitting around 70C under load but with fan ramping and all that I couldn't really tell you if it was running hotter than the previous CPUs. I also do the majority of my testing on a test bed (to make swapping hardware easier) so cooling is better than it would be in a chassis.

Posted on 2017-07-01 03:29:47
CASEYV

I'm surprised at the low temps on the i9-7900X on an air cooler! All the talk at the moment is that it's generally impossible. Open air test bench obviously helps. But thanks for the info! Makes me slightly more confident in my AIO cooler. Looking forward to more benchmarks on the i9-7900x!

Posted on 2017-07-01 03:47:51
AlbertS

Matt - Thanks for another great test. Even though the i9s are based on existing Xeons with ECC removed in response to Ryzen it's great to see improved performance at lower prices (thanks AMD) from Intel. In your previous Ryzen benchmarks the i7-770K did well against Ryzen but here the new 7740K under-performed and the 7740K has disabled the Intel Iris Graphics iGPU for some strange reason. I know PrPro has Intel iGPU Quick-sync Video support (QSV) but I'm not sure if the 7700K iGPU was enabled in the Ryzen benchmarks for H.264 or if it affected the 7740K results above with no iGPU.
I'm not an AMD fan and my experience is "cheaper is not always best" but competition benefits us all. I'm trying to decide on a 4k H.264 edit workstation and H.265 is here on the GH5 and DJI cameras. Software timeline decoding of these highly compressed codecs is CPU intensive with limited multi-threading for most NLEs so it would seem that that hardware decoding using Intel Iris iGPU for AVC and HEVC would reduce the CPU load. Edius Pro and MAGIX have both added Iris iGPU decoding support so I'm wondering if it's worth waiting for the 6 core Coffee Lake with iGPU coming later this year??
http://www.anandtech.com/sh...

Posted on 2017-07-03 08:19:18

I'm not following how the i7 7740X under-performed in our benchmarks - it was consistently ~5% faster than the i7 7700K. As far as I can tell in my testing, Premiere doesn't make any use of the Intel Graphics Hardware Decoding - or at least, it doesn't if you have a discrete video card installed (which you really, really should have). We did actually leave the onboard graphics enabled on the i7 7700K, so if Premiere Pro did support the CPU hardware decoding, I would expect the i7 7700K to be faster than the i7 7740X when in reality it was ~5% slower.

Posted on 2017-07-03 17:39:51
AlbertS

Apologies - I should have read more closely - mis-read averages and didn't see raw results dropdown. Was the Kaby Lake iGPU utilised at all?
re Intel iGPU decoding - I was referring to a post on a user forum which referred to ""Enable Intel H.264 acceleration decoding" so I assumed it is an option in Premiere Pro.
https://forums.creativecow....

Posted on 2017-07-04 09:04:28

I'm honestly not 100% sure how all the hardware-level H.264 decoding works and whether it is even being used if you have a discrete video card. My impression is that the "Intel H.364 accelerated decoding" is really only important with low-end systems that have no discrete GPU. It might make a small different with the types of configurations we tested, but I believe any performance gain is completely outweighed by the performance of the GPU. Honestly, I wouldn't worry about it too much. We had the integrated graphics enabled in our testing, so if it did make a difference it is included in our results. Since the i7 7740X was faster than the i7 7700K, that just tells me that either it isn't doing anything (since we have a discrete GPU) or it is such a small difference that the faster clock speed of the i7 7740X is more than enough to compensate

Posted on 2017-07-05 18:44:02
editor0400

I love these tests.
Looks like the latest gen of Intel processors isn't a huge leap for performance, however the cost is much more reasonable and there are more options that lead to more PCI lanes. This is important when incorporating multiple GPUs and M2 storage in your build!

Posted on 2017-07-04 21:09:05
AlbertS

"According to Intel's E3 press release, there will be a 12 core CPU available in August along with 14, 16, and 18 core CPUs in October. Because of this, we won't be able to look at the entire Skylake-X line-up today, but what we can do is see how the CPUs that are currently available compare to Intel's previous generation CPUs as well as AMD's Ryzen 7 CPUs."
Thanks again for your comprehensive testing. I am amazed at the effort you put into it.. I understand individual workflow and source camera formats will determine best user CPU/GPU and averages can be misleading - but there is a declining value return on more cores vs CPU clock speed due to parallelism limitations in a single app. In your Overall Average it's interesting to note that the i7-7800X 6 core @ $389 is 102% vs the i7-7900X 10 core @ $999 is 113% is only 16% more performance for 257% more money. That 16% may also be due, in part, to the fact that the 7800X is missing Intel's Turbo Boost 3 which is present only in the 8 & 10 core CPUs. Likewise, the 7820X 8 core vs 7900X 10 core is shows only 5% more performance for 167% more money. 12-18 core Xeons (SkylakeX is based on existing Xeons with ECC removed) always performed best as server CPUs as HEDT apps can't use the extra cores. Do you expect to see much, if any, further performance gains going from 10 cores at $999 to 18 cores at $2000+?
I do see an advantage of new Intel 6 vs 4 cores at a similar cost but it's clear that the 4 core 7740X 4.3-4.5Ghz @ $339 matches the the 8 core Ryzen 1800X 3.6-4Ghz @ $499 performance at 95% where faster clock speed offsets the advantage of double the number of cores.
For my next 4K edit PC, I would like to see the new 8th gen 6 core Coffee Lake i7-8700K (due soon) tested as Intel now claims it's 30% faster than Kaby Lake at a similar price and retains socket 1151 so motherboard cost will be lower than X99 and use something like the Asus Prime Z270A used in your Kaby Lake benchmark. You may be able to offer a consumer priced HEDT for 4K editing which would appeal to a larger market.
http://www.trustedreviews.c...

Posted on 2017-07-20 08:27:25

The issue with CPUs like the i7 7700K and (presumably) the new Coffe Lake CPUs is that they are limited to 64GB of RAM. For people doing basic work in Premiere Pro or hobbyists that is probably just fine but for anyone doing 4K that is really the minimum amount of RAM we would recommend. If you use Photoshop, After Effects, or even have a lot of internet browser tabs open at the same time, you should probably upgrade to 128GB of RAM since 64GB isn't quite enough in most cases.

Also, I doubt Intel's claim of a 30% performance improvement will apply to everything. That will probably be true for heavily threaded applications like 3D rendering, but probably not Premiere Pro or similar applications. If they have to reduce the frequency too much to compensate for the higher core count, it is entirely possible that the 6 core will actually be slower in Premiere Pro than the quad core. That probably won't be the case, but I would expect a closer to 10-15% performance gain in Premiere. That's just a complete guess, however, since there is no real way to know until we can run actual benchmarks.

Posted on 2017-07-20 17:24:18
AlbertS

Thanks Matt - I didn't know PrPro used so much memory and that may be limit for some users so I will investigate further. Would a new M.2 fast SSD help on a 1151 system for faster memory paging if demand exceeds 64GB? What are your feelings about multi-threading performance/parallelism performance with more than 10 cores as I mentioned above? One professional user I know claims he can edit 4K on a laptop using the PrPro Cineform implementation as an intermediate edit codec for smooth timeline decoding playback and reduced hardware demands. I'm less concerned about render speed than timeline performance.

Posted on 2017-07-21 06:11:29

M.2 drives are certainly better than a standard SSD for memory overflow, but even if you have a pair of Samsung 960 Pro NVMe drives in RAID0 they won't be as fast as system RAM. This is actually something I've wanted to test for a while but haven't had the time to do so. I'd love to be able to say "you will lose X% performance if you run out of RAM" and measure how much the speed of your cache drive changes that number.

Performance with more than 10 cores is really a complete unknown right now since there are no 12 core CPUs available with decent clock speeds. There are some Xeon E5 V4 CPUs, but they have relatively low clock speeds. We've done testing with dual CPU setups with 12+ cores but there is a lot of overhead associated with dual CPU systems that in the case of Premiere Pro actually causes worse performance in most cases.

If the new Intel Core i9 CPUs that are coming out soon are close in frequency to the i9 7900X they may give a small ~5& performance gain over the 7900X but I think it is equally likely they will have to sacrifice frequency too much and will be the same (or worse) performance compared to the i9 7900X. No way to really make a guess until Intel publishes the full specs, however, and no way to know for sure until we can actually benchmark it though.

Edit: Sorry, I forgot about your last question. There are a couple ways you could make it so timeline performance is great no matter what your workstation hardware (within reason of course). The first is to do what you mentioned and transcode all your footage to something that is easy on the system. Cineform is a good one, ProRes 422 is another that is popular with Mac users, DNxHD is another I'm hearing some about as well. Cineform is probably your best bet though. The problem with this is that you are reducing the video quality every time to transcode to one of these codecs. Another thing you could do is to adopt a proxy workflow. Same idea (replace your footage with a lighter weight format), but at the end you switch back to the original footage for the final export. https://helpx.adobe.com/pre...

I think either method is perfectly acceptable, it just depends on the quality you need for the final product.

Posted on 2017-07-23 20:51:47
AlbertS

Thanks Matt - Appreciate all the info. I saw a benchmark using a dual Xeon $50K workstation and it was slower than a single Xeon, Look forward to you future benchmarks. Thanks again for a great informative website. Also hoping to see tests with Resolve 14 once it's stabilized and released.

Posted on 2017-07-24 08:07:56

Thanks! I hope to get Resolve 14 testing done as well. I had everything going on Resolve 13.5 and was partway through testing when they announced Resolve 14. Hopefully they'll have it released before NAB New York in October - although maybe they are waiting so they can launch it fully at that event.

Posted on 2017-07-24 17:39:45
El Perro

Incredible work. Congratulations.
What is most involved with the live playback?
According with the test the 7900x has the best LPB but not much better than lesser cored cpus.

Posted on 2017-07-28 14:07:48

Live playback is really a mix of everything (CPU, GPU, Storage, RAM, etc.) and which component is more important depends on the codec and resolution. We actually expected the CPU to make a bigger difference than it did so this testing was very eye opening for us as well.

It is really hard to say what exactly is holding us back on the higher resolution footage, but I suspect it is actually the GPU. We are testing with the fastest GPU currently available for Premiere Pro, however, so at the moment there isn't much to do about that. We've testing dual GPU in the past and it didn't make much of a difference but after seeing these results with the newer CPUs it might be time for us to revisit that.

Posted on 2017-08-04 16:49:54
Lars Passic

"In fact, the i7 7820X is even a hair less expensive than AMD's Ryzen 7 CPUs."

It appears that you have mistakenly listed a $599 Intel CPU as cheaper than AMD's most expensive $499 CPU in all of your Adobe articles.

Posted on 2017-08-02 16:59:36

Thanks for pointing out that typo, we've corrected it to say "7800X" instead of "7820X" and updated the other articles as well.

Posted on 2017-08-04 16:14:22
El Perro

First threadripper benchmarks are appearing and while Amd win in multithread syntetics it seems not to be ideal in less cores tasks

Posted on 2017-08-06 15:16:05
Carles Bargues

Considering that it will be used both in Premiere and Resolve, and having a limited budget, would you choose 7800X + GTX1080 or 1700 (would OC so = 1800X) + GTX1080Ti? (up to 5K footage)

Posted on 2017-08-13 19:29:02
AlbertS

I'm going to wait another week for the new 6 core Coffee Lake.

Posted on 2017-08-14 06:03:35
soksok

remember 7800x have more pci lanes and more connectivity options and all so quad channel (128gb ram max).for a more render heavy usage and for a long run investment the 7800x is more powerfull,thats my 2 cents

Posted on 2017-08-18 04:29:30
El Perro

I would prefer a 7700k as the 7800x is marginally faster only in rendering. 7700k is much faster in after effects and photoshop. Why would you need those extra pci lanes (double GPU does not improve much performance)? Just asking

Posted on 2017-08-18 08:27:13
soksok

In My upcoming build , i need to connect 4 ssd, 4 hdd, and 2 nvme, x299 in some mobo configs as in asus x299 mark 1 can use all those lanes for connectivity with a 28 pci lanes cpu as in the 7800x, i use only 1 gtx 1070,this why ryzen or even upcoming coffee lake cant compete in connectivity terms.

7700k will be faster then 7800x on some cases but, for a system for the next 4 years i choose 7800x(in the 400 doller range choices i have)
Comming by the way from a 6 year old 2600k who i plan to retire

Posted on 2017-08-18 11:39:28
El Perro

If you read the articule, with the 1800X you get almost the worst results at a similar price of a 7820x (which is much faster in most of the tasks)

Posted on 2017-08-18 08:30:51
Carles Bargues

Well, at a similar price... not in Spain. Ryzen 1700 (really similar to 1800x when OC'ed) with included cooler = 308€. 7820X = 650€ plus a needed high quality cooler and more expensive mobo = 800€. 500€ more compared to Ryzen.

Posted on 2017-08-19 23:11:10
Johnnie Karafyllakis

Threadripper? When will we see some benchmarks from threadripper and Vega

Posted on 2017-08-22 07:42:40

Threadripper is coming - we will likely have an article up early next week but there is a small chance of it getting done late this week. Vega we don't really have a timeline on quite yet. With all the CPU launches from Intel and AMD we are a bit hard pressed just getting CPU articles out in a timely manner. It is definitely on our list though, so Vega testing is something we will hopefully get done in the near future.

Posted on 2017-08-22 17:08:26
El Perro

Are you planning to do a threadripper in photoshop and after effects too?

Posted on 2017-08-22 19:03:44

Neither application will likely be very good on Threadripper since they are only lightly threaded, but we do plan on at least testing After Effects since that is often used alongside Premiere Pro. Photoshop we might do since the testing is relatively quick but since it is really one of the worst applications for high core count CPUs it isn't a high priority for us. We might be able to sneak it in between other tests, however.

Posted on 2017-08-22 19:08:32
Johnnie Karafyllakis

Awesome thanks for the quick response, cant wait.
Looking to build a new editing rig and would love to see the performance.

Posted on 2017-08-22 22:36:15

Hey Johnnie, I remembered you asking about Threadripper so I wanted to make sure you saw our Threadripper vs Skylake-X Premiere Pro article: https://www.pugetsystems.co... . Threadripper does pretty well when it comes to exporting and rendering previews, but overall the live playback performance falls behind Intel. It's pretty close though and AMD is definitely competitive which is a really good thing to see.

Posted on 2017-08-31 16:54:51
Johnnie Karafyllakis

Will definitely check it out now, thanks.

Posted on 2017-08-31 22:31:23
El Perro

Thanks for the answer. Which one of your actual builds or which cpu woukd be best for a heavy photoshop usage with a medium or light premiere pro and marginal after effects use?
7820x sounds good but I would lose much performance in Ps I'm afraid

Posted on 2017-08-22 23:56:18

If you haven't seen them, we actually have identical articles to this one for Photoshop https://www.pugetsystems.co... and After Effects https://www.pugetsystems.co... if you want to see how these CPUs compare in all the applications you listed.

Personally, I would probably go with either the i7 7820X or the i7 7700K. The i7 7820X is the better all-around CPU for Photoshop and Premiere Pro I would say, and the fact that it can use more than 64GB of RAM is good if you start to do more heavy work in After Effects. The i7 7700K is technically a tiny bit better for Photoshop, but actually only by a few percent in general tasks so you probably would never notice in the real world. The much better performance in Premiere Pro with the i7 7820X is probably worth that small loss in Photoshop in my opinion.

Posted on 2017-08-23 17:53:50
El Perro

Thanks Matt
Yeah I already saw them and was thinking about those options too

Posted on 2017-08-23 19:22:35
Adriano

In the UHD videoediting with a GTX 1080Ti videocard and Edius 9, in general is it better i9-7940X or Threadripper1950x? Thanks for some info

Posted on 2018-01-17 16:49:31
AlbertS

Edius 9 uses the Intel Graphics GPU and QuickSync (QSV) for accelerated 4K timeline decoding of H.264 &H.265. i9s dont have the GPU. Check with Edius users forum but I think the i7-8700K 6 core with Intel 630 GPU might be your best bet (you may not need the 1080TI). :)

Posted on 2018-01-17 18:40:08
Adriano

I thank you but .... I'm sorry, I asked another different question. It seemas you answered to another guy. ;)

Posted on 2018-01-17 19:26:08