Puget Systems print logo

https://www.pugetsystems.com

Read this article at https://www.pugetsystems.com/guides/1461
Article Thumbnail

Windows 10 Security Patch Slowed Intel Core i9 9900K in Pix4D, Metashape, & RealityCapture

Written on May 20, 2019 by William George
Share:

Introduction

Over the last couple of years, several new vulnerabilities have been found in modern microprocessor architectures. Many of these have been related to speculative execution techniques used across most of Intel's product line, and a patch was released for Windows 10 last week which was intended to address some of those weaknesses. Unfortunately, it has also reduced performance in some applications - which this article will demonstrate by comparing processing speed in three photogrammetry programs before and after the update.

For the record, this performance degradation was not something I was intentionally looking for. I only noticed it because the patch happened to be released in the middle of other testing I was doing on these applications, and when I compared two sets of results I was surprised to see a substantial increase in how long the Metashape benchmark I built took. Digging into what had changed, I discovered that a "Cumulative Update for Windows 10", described in KB4494441, had been installed between benchmark runs.

Windows Update History Showing Cumulative Update for Windows 10 KB4494441

System Details

The system I first discovered this on used an Intel Core i9 9900K processor, installed on a Gigabyte Z370 AORUS Gaming 5 motherboard with 64GB of DDR4 2666MHz memory and an NVIDIA GeForce RTX 2080 Ti Founders Edition graphics card.

I had two other systems running the same sets of tests, one with an Intel Core i9 9940X and the other an AMD Threadripper 2990WX. Neither of those had any statistically significant impact from the same security patch, though. AMD's processors have often escaped when Intel chips have been found to contain flaws, and from what I've read Intel's latest 9th Gen Core X chips (which the 9940X is part of) were not vulnerable to the latest exploits... so it makes sense that these platforms were unaffected. Because of that, however, my data is limited to just a single processor - but I suspect that all Intel chips affected by the recent discovery would exhibit similar performance loss in these applications. Other programs may or may not be impacted in the same way.

Unfortunately, I also cannot go back and duplicate the original results or try a different CPU in the same system. The update in question did not have any option to roll back, and the only Windows Restore Point on the system was one made automatically... after the update had been applied. At this point I have other testing that needs to be completed for future articles, so I am not in a position to go back and spend a lot of time trying to get to a pre-patch condition in order to dig further. I am simply publishing the information I do have as a PSA.

Results Overview

I think the easiest way to show the overall impact of this performance loss is a graph showing the added time taken to fully process each image set in our benchmark, on each of the three photogrammetry applications we currently test. The original performance of the system is shown in blue and is always 100% - with the performance after the patch shown in red. A result of 110%, for example, would mean that it now takes 10% longer to process that image set than it did before the security update. I am not going into detail about the image sets used in these tests, as that information will be available in the performance articles I am working on for each of these programs.

Detailed Results Tables

For those who want even more detail, I put together charts of the average processing times (in seconds) before and after the security patch, including each step in these benchmarks (rather than just the total times, which were used for the summary graphs above):

Conclusion

A single patch, to address a couple of CPU vulnerabilities, caused Intel's Core i9 9900K processor to take 2 to 11% longer to process image sets in these photogrammetry applications. This is not the first update that has addressed such weaknesses in modern CPUs, so I wouldn't be surprised if the overall impact from those cumulative patches is even bigger. Some applications might be affected even more heavily, while others could see little or no difference - and of course, different CPUs may be affected more or less with each discovered vulnerability and subsequent patch. I look forward to seeing if this performance drop on the 9900K (and related models) impacts what CPUs are the best for photogrammetry moving forward, and we will have articles covering that topic in the coming days.

Looking for a
Photogrammetry Workstation?

Puget Systems offers a range workstations tailored for Pix4D, Metashape, and RealityCapture. Even in the most demanding situations, our workstations and designed to minimize downtime and allow you to work as efficiently as possible.

Tags: photogrammetry, Pix4D, Metashape, RealityCapture, Intel, CPU, Processor, Windows 10, Security, Core i9, 9900K, Performance, Benchmark
Hwgeek

Lets wait and see if we gonna need 9900K in 4 days :-).

Posted on 2019-05-23 12:24:36

I presume you are referring to the launch of AMD's next generation of processors? I too am curious about their performance potential in these applications, but unfortunately they haven't sent us any high-end samples from Ryzen's 3rd generation. I am hoping such chips show up at some point, so that we can include them in future CPU comparison articles.

[edited to remove possible NDA information]

Posted on 2019-05-23 16:07:26
Hwgeek

Interesting, so it looks like AMD is keeping the cards close to the chest and only will give the Ryzen 7/9 after May 27th.
Maybe they send the low end parts so Reviewers can test and give feedback to AMD if there are Bios/Ram/Windows/Software issues so they can fix it before the Reviews go live?
Thanks for this article, it's very useful.

Posted on 2019-05-23 19:20:19

That is funny but that makes sense. That’s like going to a Maserati dealership to buy a preowned Ford.

Posted on 2019-05-24 14:30:42
Adriano

Whether you realize it or not you basically let the cat out of the bag about the relative performance, on a per core basis, of AMD's new processors versus the new offerings, long story short, barely faster in most workloads but makes a compelling case when number of cores per dollar is taken into account.

Posted on 2019-05-24 15:22:40

Huh? I haven't yet run or tested any systems with the new 3rd-Gen Ryzen chips, so I don't have any information (or cats) to let out of the bag :)

In past generations, though, what you have described is exactly the situation we've seen. Per-core clock speeds and IPC did not match Intel, so AMD's processors only made sense in applications which scaled very well with additional cores and cared less about per-core performance. The Threadripper line, in particular, has been very good in that respect - and is often the best option for a single-socket CPU rendering system (for example).

Now if I had to guess, looking at the things that have leaked online, I would suspect that AMD is doing their darnedest to close that per-core performance gap. Will they succeed? We'll have to wait till we have samples to test in order to answer that :)

Posted on 2019-05-24 15:43:34
Dsdfa Sefes

Ryzen CPUs IPC is only couple percents lower than Intel. Difference in performance is result of clock speed, memory latency and optimisation.

Posted on 2019-05-26 01:53:40
tguy

always test the single threaded performance? to see if it tops intel? clock for clock obviously.

Posted on 2019-05-25 08:34:04
Alejandro Bertinelli

So you have a 3600... Any info on core count/frequencies for that SKU?

Posted on 2019-05-25 09:23:06
celtiberian

This! Is it 6 or 8 cores?

Posted on 2019-05-25 12:38:52

I can't say anything about details like that, as this CPU hasn't been released yet, but just looking at the leaked specs on various other websites it seems like it wouldn't be a model we'd be interested in anyway. If the leaks are anything close to accurate, there are a few models above it that would be much more appealing.

Posted on 2019-05-25 15:28:33
Alejandro Bertinelli

Okay, next time don't say anything if you're not going to give information about it.

Posted on 2019-05-25 15:43:40

I didn't mean to disappoint anyone, or to spoil things for AMD. Perhaps I should not have mentioned it at all. The reason I did, though, was to set expectations appropriately: as much as I would like to, we won't have performance articles about Ryzen 3 at or shortly after launch. I suppose I could have just said that instead, though :)

Posted on 2019-05-25 15:49:30
Lucas Biron

Wait, so why wouldnt you be interested in testing a cpu that is 8c/16t with comparable per core performances to a 9900k (according to leaks) if you are interested in testing/retesting the 9900k? Unless I am missing something ? (sorry we are like a bunch of sharks trying to get info from you :( )

Posted on 2019-05-25 16:39:16
juanme555

Because if they're expensive they wouldn't be worth it , you get very close to that performance with current Ryzen 7's and they are much cheaper , why would you buy a new one for over 300$ when you can get an 8C/16T 2700 for 210$ that gets almost 90% of the performance of a 9900k??.
Zen 2's whole hype is about the 12 and 16 core cpus.

Posted on 2019-05-25 16:42:22
Hwgeek

Because AMD will continue with the current core count for Ryzen 2600-3600 as 6c and 2700-3700 as 8c, it would be stupid and confusing Idea to change that, IMO Ryzen 3800/X will be 12C and Ryzen 9 3900/X will be the 16c.

Posted on 2019-05-25 20:27:34

I was basing my assumption on leaks like the ones listed in this article:

https://www.pcgamer.com/amd...

According to that, the Ryzen 5 3600 would be 8 cores / 16 threads at 3.6 to 4.4GHz. Even if AMD pulls of a miracle and has the same instructions per clock (IPC) performance as Intel, that would still be substantially slower than the 9900K. The models above that are what would interest me more :)

Posted on 2019-05-25 22:16:00
celtiberian

Ask them for a Ryzen 9 3xxx and a Zen 2 Threadripper to test and review.

Posted on 2019-05-26 02:03:08
La Frite David Sauce Ketchup

3600 is bad ? :'(

Posted on 2019-05-25 10:51:41
mene toto

Thinking upgrading to one of those and OC it!

Posted on 2019-05-25 12:34:58
Luca Pupulin

Window 10 is a shame...even the Pro version....

Posted on 2019-05-23 13:25:20

I don't think I'd pin this on Windows or Microsoft, though I do think they've gone downhill since Windows 7 and there are a lot of legitimate things to complain about with Windows 10 (forced updates, for one). The problem is a vulnerability in Intel processors, and while I've read a lot about it and have a lot of questions regarding how much of a threat these weaknesses really are - especially in the context of a desktop operating system - I cannot blame Microsoft for wanting to patch their OS to mitigate such issues.

Posted on 2019-05-23 16:38:54
Luca Pupulin

Hi William,
you are perfectly right!
The problem is a series of vulnerabilities in Intel processors but these are more pertinent in a Server OS...

so these paches,in my opinion,shouldn't be automatically applied to a desktop/workstation OS;
Anyway I don't consider Windows 10 Pro a good OS for a Workstation....forced updates or updates wich cause severe issues,RAM and TDR (Timeout Detection and Recovery) managements and so on and so forth...

Linux,especially CentOS,gives me headaches at times,but at least I don't have an OS which calls programs "apps",or have functionalities related to games (Xbox) etc...

I really miss XP Pro x64 or Win 7 at least..

Posted on 2019-05-24 08:14:17
John Smith

If you want decent windows, use LTSB/C.
Forced updates suck, but that's kind of on you for not using the good versions of windows. Security updates should always be installed though. More light needs to be shone on this so intel stops cutting corners.

Posted on 2019-05-25 07:57:14
γιαννης

in theory you are correct BUT
unlike the rest of the holes intel has we already saw a lot of videos of people running scripts for zombieload as demo on youtube
that means scripts are probably already being sold around

Posted on 2019-05-25 10:44:28
yezhacker

Surely it would require too much work for you guys to redo all the benchmark tests for the Intel processors, but, at least in several applications it would be useful to know how much performance was lost due to this patches. Especially for the i9 9900k on Adobe CC.

Hope you can get soon a sample of the Ryzen 3700x and 3800x/3850x to test them

Posted on 2019-05-23 23:03:21

We're actually right in the midst of updating a lot of our video/photo editing benchmarks, so we probably won't be doing much until that is done. Working on After Effects, Premiere Pro, and Resolve benchmarks in particular.

Posted on 2019-05-23 23:10:21
yezhacker

I would like to say "that's great!" But bad news are never great. Anyway, I'll be expecting them!

Posted on 2019-05-23 23:19:19
A vW

Can you please redo the photoshop benchmark as well, seems like my 8700k running all cores at 5GHz has lost quite a bit of performance since having applied these patches starting with the spectre/meltdown patches till the most recent.

Posted on 2019-05-24 15:18:14

As always thanks for the benchmarks. One thing I incorrectly noticed when ‘quickly’ glancing at the benchmarks is that Intel processors gained speed after the patch: 100% scores with 105%, etc.

For some reason, in my mind the original performance should be at 100% just like you had but a slowdown in post-patch should be a value <100%. While a gain in performance would be a value >100%.

Just my 5c.

Posted on 2019-05-24 14:36:45
Adriano

I concur, this is the way I read it, especially since it's consistent with the way you post the relative performance of various systems when you guys do a review.

Posted on 2019-05-24 15:20:51

Thank you for the feedback! I was a bit divided myself on how best to convey the slowdown, but since these benchmarks measure performance in time taken - and shorter is better - I thought I should stick with that style of measurement. Maybe it would have helped if I had included (on the graph) text saying something like "shorter is better" or "lower is better"? I nearly did, but it felt crowded already.

Posted on 2019-05-24 15:46:53
Sid

I think the current method makes more sense.. Its easier to understand % change compared to older version. Change with respect to new system would be odd.

Posted on 2019-05-25 21:11:37
Hwgeek

Zombie Loading: Intel Patches Slow Down SSDs as AMD Gains Ground
https://www.tomshardware.co...
Maybe SSD slowdown also makes tasks to take longer to finish, IMO it's worth to test your fast storage configuration after this update.

P.S Gigabyte just teased PCIe Gen4 SSD with with speeds up to 5000 MB/s, I can imaging your build with X570 and those ultra fast SSD's :-).

Posted on 2019-05-24 17:01:08
Herbert Hoffman

Hi! My 9900K already is a hot tube. One months ago i built my own workstation copy the specs of Puget systems machines ( 9900k, z390 Designare, rtx4000...ect.)

I set the BIOS according to your gigabyte Designare BIOS preview link, but did't work.

Now i change the cooler ( from Noctua u12s to a u15s) but puget's temps still better.

in Bios:

Puget Custom PC CPU core voltage 1.164 + 24 Celsius
My Custom PC: 1,284 + 33 Celsius

This is because of the lower voltage?

Any suggestion?

Posted on 2019-05-31 13:35:18
Herbert Hoffman

i'm from Germany, here it isnt available Pugetsystems...:(

Posted on 2019-05-31 13:40:37

Hmm, I'm not sure which articles you are referring to with BIOS settings and stuff like that (we have a lot of folks here who write, and I don't have a chance to read everything that is published) - but in general, yes: higher core voltage settings will lead to higher temperatures.

Posted on 2019-05-31 16:18:50
leexgx

if your doing benchmarks ideally should use gpedit to set windows update to notify before download and install (you have to press check for updates or download to trigger update download and install) as it could impact benchmark results (in this case it was good that it did as you would of been scratching heads why the results was lower than normal at a later date when you rerun the tests on a new pc)

on the windows update side of things and system restore not been reliable by default need to make shadow copy snapshot to make it work,windows 10 onwards MS have disabled snapshot auto creation on windows 10 (witch is why previous file versions is empty on folders when you go into properties) also rolling back on windows updates is not reliable, i have mine set to make a snapshot every day and just left it at default 10GB space, Do note some windows 10 installs (clean or OEM install) for some dumb reason defaults to system restore off and 0gb space default but with the way windows feature updates are rolled out mite be better to set it to 20GB) wmic shadowcopy call create Volume=C:\ is the command to restore shadow copies or this site has very good instructions on how to do it (adblock recommended on ths site) https://www.repairwin.com/h... ( i have it set to once every day 10 minute after system is idle) use "shadow copy viewer" to verify its creating new shadow copy points when you run the task (you also see a "previous file versions" point shown on any folder)

Note if you Clone to a drive you must go into system system info > system protection and press remove the system protection from the missing partitions and re enable it on new partitions (it's silly MS does not do this on its own) or system restore is completely disabled

every computer i setup is using 10 Pro and i set it to 100 days delay on feature updates and 15 days delay on normal updates (so all home and normal pro customer can beta test a untested release update so when it breaks enough PCs they pull it before my customers get it) and import shadow copy snapshot task schedule task (as you can export and import schedule tasks) so if a customer loses a file they norm have upto a month to get it back or if windows update breaks the PC i can undo it

Posted on 2019-06-08 02:01:03
joms

Hi, just wondering if perhaps you received or will receive an AMD 3900x before the release date? I'm thinking of getting this or the intel 9700x or 9900k. I'm kind of leaning towards the AMD 3900x due to the numerous vulnerabilities of intel and its impact on its performance. (as well as the price savings). Note: I plan to use it on Adobe premiere and later on after effects.

Posted on 2019-06-11 22:26:55

We cannot reveal anything about what samples we may or may not get, nor can we publish any performance data, before manufacturers life their press embargo. Everything pre-launch is under NDA. However, I am sure that plenty of websites will have coverage of new CPU performance as soon as they are allowed to :)

Posted on 2019-06-11 22:29:35