Tested: AMD RYZEN vs INTEL SKYLAKE CPU's for BlueIris

[Lukas Beskow]

Good afternoon everyone,

I'm looking to set up my camera security system in a hostel I own in Pomerode - Brazil,
I bought a ryzen AMD CPU and become a bit worried because I readed some posts indicating that INTEL should perform much better for BI.
I ran a simple test both in my Intel computer at home and in the AMD computer at the hostel,

I think the test can give us a good comparison because both CPU's have similar performance at benchmarks. The results are surprising and I think it's a good idea to share it with you,

The test contenders:

Intel Skylake i5-6600 @ 3,3 GHz (base clock) - 4 cores 'n 4 threads
16 GB 2133 MHz RAM DDR4 dual Chanel (4x 4GB dimms)
Geforce GTX 1060 3GB
OS and BI installed on SSD's

AMD Ryzen 5 1400 @3,2 GHz (base clock) - 4 cores 'n 8 threads
8 GB 2666 MHz RAM DDR4 single channel (1x 8GB dimm)
Geforce GT 1030 2 GB
OS and BI installed on SSD's

For the test, two cameras were used:
Webcam: Microsoft Lifecam Studio (1080p capable sensor)
Ipcam: Foscam FI9816P (720p capable camera)

Attached you will find CPU-Z benchmarks for both systems. I ran these benchmarks to see the real result for my very system. You may see that the intel has a 17% faster single core speed, whereas AMD has a 15,5% multicore advantage (it has multi threading),

The test consisted in two steps
first running only the webcam in the following resolutions:
320x240;
640x480;
1280x720;
1920x1080;
Each resolution at 5fps, 15 fps and 25 fps,

The second test: running both the webcam and the ip camera in 1280x720; at 5 fps, 15 fps and 25 fps;

What was measured: CPU load (min, average and maximum in a 1 minute time lapse) and total memory allocated,

The Blue Iris Configuration was:
Direct to disk write: Disabled
H.264 decode: default
Rotate: No
De-interlace: No
Delay: 0 msec
Area of interest: unchecked
Anamorphic: unchecked
Flip: No


Some of the results:






the complete results follows attached,

Conclusion: the purpose of this test was to evaluate if my system will suport the security system. A further test could be done with more cameras. In just one scenario (memory allocated for both cameras), the intel CPU performed better, on all other the results went surprisingly well for AMD Ryzen, having less variable CPU load, which I suspect could translate for instability on bigger systems. The results are more impressive considering the 1400 being something like 40% cheaper than the 6600 was.

Hope you enjoy it, what do you think about that?

Best regards, Lukas,

 

[Gymratz]

I like that you ran through this - a very good start!
I think the actual tests, if the goal is a real world Blue Iris load, is done with many incorrect assumptions as to settings that should be used.
I would suggest connecting a full load to both systems, optimizing per recommendations on the forum (e.g. direct to disk), and also removing the video card from the Intel box.

 

[fenderman]

This test is completely useless as you didnt utilize the intel HA, you didnt utilize direct to disk...the AMD is an inferior processors...and here in the US, systems with the intel processors are cheaper than AMD...

 

[fenderman]

So i got curios about the Ryzen vs Intel thing and tried my own test. Core i5-6600K 3.5 Ghz (w/ HW accel enabled) compared to a AMD Ryzen 5 1600 3.6 Ghz (no HW accel). I exported my reg file for my 9x 3Mp setup with Direct to disk, motion detection, alert pre-buffer, the whole 9 yards.

I validated intel Quick sync was working on the i5, as I'd see a 20% increase in CPU load when I'd disable it.

On average the the intel sat at 25-30% CPU usage and the Ryzen sat at 9-10 % usage. and the Ryzen burned 10 less Watts too as measured by a killawatt.

I imagine even with HW accelerated advantage, the Intel's 4 threads (no HT on that model) still lost to the Ryzen's 12 threads.

Both CPUs are from last year and are still sold. The i5 is $20 more expensive than the Ryzen.

So is it really still valid to have all the hate for Ryzen if it (in software) can do what the intel (in hardware) can but faster for the same price? And if so what am I missing that invalidates these results?

yes you are missing the fact that you compared apples to oranges...you should have compared to an intel i5 eighth gen or an i7 7th/6th gen...also I suspect you didnt properly test - you dont state of the console was open..you dont provide complete system details...the 8th gen intel is even more powerful - and its a matter of months before those machines are available cheap.
you are also testing an intel K processor which uses more power...
intel 6th generation i5-6500 systems are available for 300...i7-6700 for about 400- that is a complete system with 3 year warranty...intel is way cheaper...
There is no point in building your own system...even if you want to build your own, you need to factor in the cost of the video card for the ryzen...

 

[fenderman]

So I'm comparing a 2016 $200 AMD with a 2015 $220 Intel that are both still on sale. The 2017 i5 8600k sells for $260, $60 more than the AMD. And Ryzen is set to release the 2018 Ryzen second gen in a few months. So I don't see how it's more fair to compare a 2016 processor to a 2017 processor but not a 2015 processor to a 2016 processor. I assume when the second gen Ryzen comes out it won't be fair to compare to the 8th gen intel anymore by that logic?

Though, dollar for doughnuts I compared two processors at the closest pricepoint, which is kinda my point to begin with.

Both tested as identically as possible. same .reg file in blue iris. Yes console open on both during sampling. 9 3MP cameras at 15fps all with motion detection and alert pre-buffer recording. Both systems have SSD for OS and 5400 RPM disk drives for recording.

And we aren't talking a few percentage points here, the Intel CPU usage was 2.5x the Ryzen.

Again you are comparing prices for the actual cpu, only a fool would pay 200 for a cpu when an ENTIRE system can be had for 300 (i5) or 400 (i7).
I dont believe for a second that your number are correct..you fail to even provide BASIC info about the systems, power supplies, number of discs, video cards and most importantly ACTUAL power consumption in watts???....did you enter the licence key when testing?

 

[fenderman]

Sorry I build my own systems so I know I'm getting decent hardware for the motherboard and other devices. And I'm only here asking if the whole "never by AMD because of lack of quick sync" is really valid, not because of possible deals on questionable sale rigs.

Both 500W PS, both have 2 disks, the one SSD and the one 5400 RPM disk. Both have graphic cards. Nvida GTX 1050 on the Intel and and AMD FirePro W2100 on the Ryzen. Key active for both. When BI running, power for the i5 was 75W and the Ryzen was 90W. So yes Rzyen burned a little hotter, but at my rates 24-7 usage that's only an annual increase of $18. Longer than a year to make up the price difference of the CPU.

I think the bottom line is a Ryzen build will work fine with BI, and may even give you more CPU time to do other things like host Minecraft servers and run Plex at the same time.

Both have graphics cards - mistake number 1...you need to measure power consumption in the intel without the card. It doesnt need it. On top of that have two different cards so you cant directly compare power consumption. Mistake number 2.
Now for some reason, the intel uses 15w less, not 10w more. What happened?
FYI Here is TRUE power consumption data for an older 4th gen i5-4590 - an i5-6500 will do better (no need for a k processor). 25w at 30% and 36w at 50%
PC NVR Power Consumption Sample
Sooo, bottom line is you are using about 60w+ more for the ryzen...and overpaying when you buy it...

Thinking that building your own systems will get you "decent" hardware.... I have over 100 dell/hp optiplex/elitedesk system running...over 20 running BI 24/7, no problems...dont fall for the myth of building your own. You should dedicate a pc to blue iris or any other vms...running other software and tasks is asking for problems.

 

[fenderman]

LOL. I measured power with a killawatt so i collected true power consumption. no onboard video w/ the ryzen so cant remove it. I baslined the intel build at 50W no load so the 75W at 25% load matches your chart (25w increase). I also baselined the ryzen at 75W at no load (15w increase for the 10% load of BI). So by baslining the systems ive accounted for differences in the builds like video cards.

So you've kinda answered my question by resorting to making a strawman out of DIY build vs. system bargain sales. For those of us that want to build DIY boxes, not only is Ryzen fine to run BI its actually a better value in regards to performance, even without HA.

Yes, the ryzen numbers were true...the intel were not as you included an unnecessary card...
Buying a dell or hp business system is not a "strawman" argument, its reasonable logic...why kind of fool would pay 2-300 more for the luxury of building their own system?
Again, EVEN if you take your foolish building of a system into account, the Ryzel system would cost you about 80 dollars PER YEAR more to run, in perpetuity....how is that about a better value? math isnt your strong point.

 

[fenderman]

Further your entire post makes no sense..the i5-4590 uses 25W for the ENTIRE system at 30 percent load.

 

[fenderman]

K well i measured actual data and asked if i missed something to explain the Ryzen hate by BI people. i got called a liar and insulted instead. So i guess i got my answer that it's just a fanboy thing.

Ok so your system went up 15W at 30% over idle. mine went up 25w. Maybe the K accounts for the 10w difference. Mine is a different CPU and build than yours as you pointed out so we have different HW. I have the GPU and you dont. I have the HDD and you unplugged it. How does me getting a higher baseline not make sense?

I didnt call you a liar, I said you were wrong, and I was quite right...now with the additional data you provided you have shown that you were in fact lying. Yeah, you did miss something...60w!!
Turns out you have no idea how much each cpu burns as you are using two different cards and systems.
Your PURPOSELY hid the fact that the ryzen burns more power by comparing idle to load vs load to load. You only admit it when directly questioned for hard numbers.
What we do know, is that your ryzen system uses 90w of power (ill take your word for it but at this point your data is suspect) vs a properly configured intel system that would use about 30w under the same load. 60w difference...

 

[fenderman]

Ok so you want to force the comparison to be between your optimized intel and my non-optimized Ryzen? I don't get to unplug the HDD like you did and i can't remove the GPU like you did?

Yep no fanboy bias at all....

My intel IS NOT OPTIMIZED. its a standard system that I recommend here all the time. IT DOES NOT NEED A CARD- THAT IS THE WHOLE DAMN POINT. It is more efficient naturally and MORE SO because you dont need discreet graphics. Oh boy you got me, adding a hdd to the ssd will add 4w....so now your system is only wasting an extra 56w....please....you are a dumber amd fanboy than I originally thought...
Folks like you who lack basic math and cognitive reasoning skills is why I am forced to emphasize this point over and over...AMD makes ZERO sense, it will cost (in your example) hundreds more over 5 years....even more if you are foolish enough to build your own.

 

[DouglasteR]

Lucas,

Refaça os testes com as configurações que todos usamos . Só assim o comparativo vai ser útil, visto que queremos saber o custo beneficio.

Mas eu duvido que bata o Intel com aceleracao.

 

[Adam Hahler]

I don't think the point that OP was trying to make was that Ryzen is better than Intel. Just that Ryzen is perfectly capable of running BI. Which everyone on here has said that it isn't. So I'm glad that he did this test. Yes, I know you can buy dirt cheap intel systems for running BI. But some of us like to build our own systems for doing more than running BI and we prefer to use AMD processors. I'm glad to know that we have the choice! Especially with the new Ryzen APU's out where we don't even have to purchase a high priced graphics card and it will run most games at 1080p too! Thank you OP for doing this test!

 

[fenderman]

I don't think the point that OP was trying to make was that Ryzen is better than Intel. Just that Ryzen is perfectly capable of running BI. Which everyone on here has said that it isn't. So I'm glad that he did this test. Yes, I know you can buy dirt cheap intel systems for running BI. But some of us like to build our own systems for doing more than running BI and we prefer to use AMD processors. I'm glad to know that we have the choice! Especially with the new Ryzen APU's out where we don't even have to purchase a high priced graphics card and it will run most games at 1080p too! Thank you OP for doing this test!

No one ever said it can't run bi...it's a freaking processor...it's just inefficient... building systems is a waste of money a big waste.... Using your system for Blue Iris and other purposes or software is a common mistake newbies make then bitch about issues... You can be foolish and go with AMD my point is to warn others that it's a dumb idea...

 

[fenderman]

No problem! Well, i was one of the posters doing real comparison tests if not the OP.

My one computer I'm running 28MP Blue Iris with full motion detection, Windows domain server, File server (w/ drive pooling software and monthly full platter scans), Plex media server (w/ HD OTA recording), client backup server, cloud sync software, syslog daemon for 3 networks, network config deployment server, and a Modded Minecraft server. Works great. I'm glad i questioned the repeated sentiment I found on this forum. And I'm repeatedly blown away at how much a performance boost the Ryzen offers over similarly priced Intels. And I had been exclusively Intel for 8 ish years now.

Complete lie...Intel will always be the cheapest and with hardware acceleration more powerful...see Dell and hp refurbes discussed here ...i cannot stress enough how dumb it is to run AMD for bi...your bi load is medium and can be run on a 100 dollar machine...but if you want to pay more initially and also perpetually to your power company go ahead but don't make false statements to mislead others...furthermore running bi together with a Plex server is about the dumbest thing you can do...it will bite you in the ass....

 

[jmg]

why kind of fool would pay 2-300 more for the luxury of building their own system?

To be fair, I build all my own systems... used to build big honk'n systems with multiple rads and custom loops, and only used 316 stainless steel tubes (started with copper, but can't stand how that looks), recently been into building small htpcs packed with insane hardware, with custom water loops and remote rads to keep them silent and cool... my current htpc is a 7700k w/ dual titans, the fans never crack idle, and the cards never creep past low 50's.

Dells are nice, and I completely understand paying lowest possible dollar for adequate hardware .. but man, they're just so vanilla and boring!!! As for paying 2- $300 more.. I'd say more likely it cost me a few thousand more than a dell.. so I guess there's at least one fool on the board!!!

completely on board with you on the ryzen train. I built one AMD system a while ago, and it's the last one I'll ever build. I'll take an Intel any day