Reason 8.1 & above CPU Stress Test (songfile included)!

[jam-s]

Given the poor performance of Reason on MacOS-X I suggest to also tell what OS you've used for your test.

[Troyvasanth]

Updated

[swamptooth]

Toshiba S855-S5378 laptop. 16gb ram.
Core™ i7-3630QM Processor (6M Cache, up to 3.40 GHz)
Turbo Boost disabled
M-Audio Fast Track Ultra (yes, it IS 8 years old) 4096 sample buffer
craps out around 12 seconds (bar 6.4) every time
Turbo enabled stalls at 20-21 secs

[Jivethompson]

New custom PC:

I7 6700k s1151 4Ghz (not overclocked)
MSI Z170A SLi plus s1151 (motherboard)
Corsair Vengeance LPX 2 x 8GB 2400 mh (16gb RAM)

Focusrite 2i4 (forgot to note buffersize, but the stress test was done at minimum latency; around 3 or 4 ms)

- Song stops at 38 seconds with internet connection
- Songs stops at 53 seconds when pc is offline (!)

Pretty average result i guess for a new pc
But the increase of performance in offline mode is remarkable!!!

[Kategra]

New custom PC:

I7 6700k s1151 4Ghz (not overclocked)
MSI Z170A SLi plus s1151 (motherboard)
Corsair Vengeance LPX 2 x 8GB 2400 mh (16gb RAM)

Focusrite 2i4 (forgot to note buffersize, but the stress test was done at minimum latency; around 3 or 4 ms)

- Song stops at 38 seconds with internet connection
- Songs stops at 53 seconds when pc is offline (!)

Pretty average result i guess for a new pc
But the increase of performance in offline mode is remarkable!!!

Interesting find.
I think if you increase the buffer to something above 12 ms, it will be less of a difference between Offline / Internet connection results.
Though an impact of CPU performance should be considered if there is actually traffic going trough the network interfaces, the more traffic, the bigger the impact.

[elMisse]

Interesting find.
I think if you increase the buffer to something above 12 ms, it will be less of a difference between Offline / Internet connection results.
Though an impact of CPU performance should be considered if there is actually traffic going trough the network interfaces, the more traffic, the bigger the impact.

Interesting indeed. In my Reason 9 / 8-core Xeon system, the difference is over 20secs No matter what buffersize I choose (128-1024), when LAN-adapter disabled, it goes through the whole test without problems. And with LAN on, worst case stop at 54 secs.

[Jivethompson]

New custom PC:

I7 6700k s1151 4Ghz (not overclocked)
MSI Z170A SLi plus s1151 (motherboard)
Corsair Vengeance LPX 2 x 8GB 2400 mh (16gb RAM)

Focusrite 2i4 (forgot to note buffersize, but the stress test was done at minimum latency; around 3 or 4 ms)

- Song stops at 38 seconds with internet connection
- Songs stops at 53 seconds when pc is offline (!)

Pretty average result i guess for a new pc
But the increase of performance in offline mode is remarkable!!!

Interesting find.
I think if you increase the buffer to something above 12 ms, it will be less of a difference between Offline / Internet connection results.
Though an impact of CPU performance should be considered if there is actually traffic going trough the network interfaces, the more traffic, the bigger the impact.

Ok will try that

Btw, how can we judge the performances of our systems based on this stress test? For example, is 53 seconds more than enough to be not limited by the DSP load in more 'realistic' projects? I know this depends on how much processing one uses and with which RE's, but it would be nice to see this stress test in some perspective if you know what i mean.

[Jivethompson]

Interesting find.
I think if you increase the buffer to something above 12 ms, it will be less of a difference between Offline / Internet connection results.
Though an impact of CPU performance should be considered if there is actually traffic going trough the network interfaces, the more traffic, the bigger the impact.

Interesting indeed. In my Reason 9 / 8-core Xeon system, the difference is over 20secs No matter what buffersize I choose (128-1024), when LAN-adapter disabled, it goes through the whole test without problems. And with LAN on, worst case stop at 54 secs.

Wow! That is a huge difference!!

[elMisse]

Wow! That is a huge difference!!

I cannot figure the reason for the difference being this huge.... Hopefully others will post their testreports on this too, so maybe we will get some answers someday

[Kategra]

New custom PC:

I7 6700k s1151 4Ghz (not overclocked)
MSI Z170A SLi plus s1151 (motherboard)
Corsair Vengeance LPX 2 x 8GB 2400 mh (16gb RAM)

Focusrite 2i4 (forgot to note buffersize, but the stress test was done at minimum latency; around 3 or 4 ms)

- Song stops at 38 seconds with int+ ernet connection
- Songs stops at 53 seconds when pc is offline (!)

Pretty average result i guess for a new pc
But the increase of performance in offline mode is remarkable!!!

Interesting find.
I think if you increase the buffer to something above 12 ms, it will be less of a difference between Offline / Internet connection results.
Though an impact of CPU performance should be considered if there is actually traffic going trough the network interfaces, the more traffic, the bigger the impact.

Ok will try that

Btw, how can we judge the performances of our systems based on this stress test? For example, is 53 seconds more than enough to be not limited by the DSP load in more 'realistic' projects? I know this depends on how much processing one uses and with which RE's, but it would be nice to see this stress test in some perspective if you know what i mean.

My opinion comes down to these:

- If your PC can handle your projects routinely, then it's powerful enough even if it only plays 20 seconds into the "benchmark".

- If you can't finish your songs because it reaches the DSP limit and playback stops or begins to crackle, even though you optimize your processing (don't use 3-4 same FX as inserts to corect the correction, bounce tracks and remove the devices that generate it when you don't need to change it anymore, use more native devices instead of REs when you know you can use the native EQ\ reverb\compressor to easily get the same result as the RE), well - if the PC/MAC can't handle it, then you should be thinking get a better one.

PS1. My PC (i5 3570K 4.2 Ghz) currently plays around 30+ seconds into the benchmark.
I have a song that I can't finish because it has so many tracks (half of them bounced) and so many FX that it crackles a bit even if I mute half them in the sequencer. It's a mess, I haven;t touched it in 2 years, don't know what to cut form it to make it more efficient, and I still have to track the vocals for the verses which will need even more DSP.

So when I'll upgrade later next year (I want a 6 core i7) I will be able to sort that song out and stay out of trouble with new projects and not go overboard with heavy combinators and processing in general, because now I have a bit more experience on how to make something sound good with less then the # of devices that I used 3 years ago.

PS2.Realistically speaking, you can't expect any desktop PC/MAC to be powerful enough to record produce and mix a full commercial song (with high track number, synths, polished vocals) in one go without heavy use of bouncing the tracks to free up DSP.
Maybe an dual XEON workstation with a price tag > $6000 could do that with no problems.

[Jivethompson]

New custom PC:

I7 6700k s1151 4Ghz (not overclocked)
MSI Z170A SLi plus s1151 (motherboard)
Corsair Vengeance LPX 2 x 8GB 2400 mh (16gb RAM)

Focusrite 2i4 (forgot to note buffersize, but the stress test was done at minimum latency; around 3 or 4 ms)

- Song stops at 38 seconds with int+ ernet connection
- Songs stops at 53 seconds when pc is offline (!)

Pretty average result i guess for a new pc
But the increase of performance in offline mode is remarkable!!!

Interesting find.
I think if you increase the buffer to something above 12 ms, it will be less of a difference between Offline / Internet connection results.
Though an impact of CPU performance should be considered if there is actually traffic going trough the network interfaces, the more traffic, the bigger the impact.

Ok will try that

Btw, how can we judge the performances of our systems based on this stress test? For example, is 53 seconds more than enough to be not limited by the DSP load in more 'realistic' projects? I know this depends on how much processing one uses and with which RE's, but it would be nice to see this stress test in some perspective if you know what i mean.

My opinion comes down to these:

- If your PC can handle your projects routinely, then it's powerful enough even if it only plays 20 seconds into the "benchmark".

- If you can't finish your songs because it reaches the DSP limit and playback stops or begins to crackle, even though you optimize your processing (don't use 3-4 same FX as inserts to corect the correction, bounce tracks and remove the devices that generate it when you don't need to change it anymore, use less native devices instead of REs when you know you can use the native EQ\ reverb\compressor to easily get the same result as the RE), well - if the PC/MAC can't handle it, then you should be thinking get a better one.

Mmmm...now i'm getting nervous haha

This is the exact reason why i got a new pc, hoping i will never get the message "your computer is too sloooooowwww" again! I'm used to reason 5 but after upgrading to 8 my computer was not up the task. The difference between reason 5 and 8 is immense btw; i think the SSL mixer takes a big chunk of the workload. I never had any dsp problems with reason 5 on my old computer and boy, there were some heavy duty projects in that one.

I guess i have to wait and see then if my PC holds up to my demands. Perhaps i should have gotten a 6-core though i thought that would be overkill for what i truly needed.

Time will tell!

[Carly(Poohbear)]

New custom PC:

I7 6700k s1151 4Ghz (not overclocked)
MSI Z170A SLi plus s1151 (motherboard)
Corsair Vengeance LPX 2 x 8GB 2400 mh (16gb RAM)

Focusrite 2i4 (forgot to note buffersize, but the stress test was done at minimum latency; around 3 or 4 ms)

- Song stops at 38 seconds with internet connection
- Songs stops at 53 seconds when pc is offline (!)

Pretty average result i guess for a new pc
But the increase of performance in offline mode is remarkable!!!

Interesting find.
I think if you increase the buffer to something above 12 ms, it will be less of a difference between Offline / Internet connection results.
Though an impact of CPU performance should be considered if there is actually traffic going trough the network interfaces, the more traffic, the bigger the impact.

Got all excited for a minute, but my online and offline results are the same for me.

Did a lot of benchmarks between R8 and R9 Beta and R9 release.

The R9 beta RC's (non logging of course) were slower, not by much.
R8 and R9 release I get more or less the same results however every 5-6 days I suddenly see a big performance drop in R9 (stress test stops at 23 seconds) which I have not really tried to track down (done basic tracking through Task manager etc nothing pops out) as it happens at times when I don't have spare time and reboot sorts it, so it could be well to do with my system, also note Reason runs 24 hours on my system, can't really compare this to R8 as I did not run R8 except for stress test results (came from R7 which I did not see the problem with)

Asus G751JT Laptop
i7 4750HQ@2.0 Ghz, 16GB
Windows 10 64bit
Audio Interface: Multiple, (DX) RealTek\Nvidia\Yeti > ASIO4ALL

Slight different results depending on interface.
ASIO4ALL avg 40 Seconds
Realtek avg 39.6 seconds
Nvidia avg 39.2 seconds
Yeti avg 38.7 seconds

Note Yeti is an USB device, also I got the largest range of spurious results from that interface (which TBH I would expect as it's sharing a bus with quite a few other devices), I found the ASIO4ALL to be the most consistent with the results, Note the ASIO4ALL at the time of testing was setup to use the Realtek.
Also note real spurious results I excluded when doing my avg. (seamed to get one every 20 to 30 runs).

Would be interested to know if anyone else has tested via ASIO4ALL and then using DX through to their audio interface if they see different results?

[svenh]

With Reason 9, the song stops at 59 seconds on my computer:

CPU i7 6700K @ 4,0 Ghz (not overclocked)
RAM 32 GB @ 3,0 GHz
Focusrite Scarlett 6i6
Sample rate 44100 Hz, buffer size 884 samples (max)

[devilfish]

My i7-2700k @ 4x 4,2GHz got 33 seconds

Windows 10 / RME Hammerfall DSP 9632 (@128 Samples)


-----

Today I ordered a new Intel Xeon E5-2683 V3 (Engineer Sample from China, 233€ on eBay)

14 Cores / 28 Threads @ 2GHz
(2,3GHz All Core Turbo and 3GHz two Core Turbo)

Hope the new machine will fully complete the Benchmark-Song

[Jivethompson]

My i7-2700k @ 4x 4,2GHz got 33 seconds

Windows 10 / RME Hammerfall DSP 9632 (@128 Samples)


-----

Today I ordered a new Intel Xeon E5-2683 V3 (Engineer Sample from China, 233€ on eBay)

14 Cores / 28 Threads @ 2GHz
(2,3GHz All Core Turbo and 3GHz two Core Turbo)

Hope the new machine will fully complete the Benchmark-Song

14 cores??? You going to the moon?

[mcatalao]

PS2.Realistically speaking, you can't expect any desktop PC/MAC to be powerful enough to record produce and mix a full commercial song (with high track number, synths, polished vocals) in one go without heavy use of bouncing the tracks to free up DSP.
Maybe an dual XEON workstation with a price tag > $6000 could do that with no problems.

Kategra, we already talked about this in the puf.
Realistically speaking, I don't know what you're doing to get to this conclusion... i was making arrangements, and recording full complete songs with a Q6600. I'm currently recording songs and making arrangements for other artists with a i7-4790k (as matter of fact, i even started mixing a song today in a Microsoft surface 4 pro, which is quite under powered when you compare it to the desktop i7 ).

The big difference at this point is that with the Q6600 i barely use Rack Extensions, whereas now, i use more than 30 rack extensions in a single project, considering instruments and effects. I still do what i always did. I always separated different production stages, specially recording and sequencing from mixing from mastering. Meanwhile, with Bounce in place in R9, can't you take the DSP burden from that project a little more?

[mcatalao]

My i7-2700k @ 4x 4,2GHz got 33 seconds

Windows 10 / RME Hammerfall DSP 9632 (@128 Samples)


-----

Today I ordered a new Intel Xeon E5-2683 V3 (Engineer Sample from China, 233€ on eBay)

14 Cores / 28 Threads @ 2GHz
(2,3GHz All Core Turbo and 3GHz two Core Turbo)

Hope the new machine will fully complete the Benchmark-Song

That's going to be interesting!

[mcatalao]

Btw, how can we judge the performances of our systems based on this stress test? For example, is 53 seconds more than enough to be not limited by the DSP load in more 'realistic' projects? I know this depends on how much processing one uses and with which RE's, but it would be nice to see this stress test in some perspective if you know what i mean.

IMHO, this test is good to compare peoples systems, and it's also a good tool to measure the performance of new versions of Reason when they get out. It is al

You judge your system against you're own projects.
When you start having the computer too slow messages, you start tweaking project performance.
Separate production stages, arranging/Recording from mix from mastering, etc.

When you start having the computer too slow on a single stage, you bounce, and bounce, and bounce. At some point the workflow gets too cluttered, and that's time to get a new machine.

[Kategra]

PS2.Realistically speaking, you can't expect any desktop PC/MAC to be powerful enough to record produce and mix a full commercial song (with high track number, synths, polished vocals) in one go without heavy use of bouncing the tracks to free up DSP.
Maybe an dual XEON workstation with a price tag > $6000 could do that with no problems.

Kategra, we already talked about this in the puf.
Realistically speaking, I don't know what you're doing to get to this conclusion... i was making arrangements, and recording full complete songs with a Q6600. I'm currently recording songs and making arrangements for other artists with a i7-4790k (as matter of fact, i even started mixing a song today in a Microsoft surface 4 pro, which is quite under powered when you compare it to the desktop i7 ).

The big difference at this point is that with the Q6600 i barely use Rack Extensions, whereas now, i use more than 30 rack extensions in a single project, considering instruments and effects. I still do what i always did. I always separated different production stages, specially recording and sequencing from mixing from mastering. Meanwhile, with Bounce in place in R9, can't you take the DSP burden from that project a little more?

I've watched Pensado's Place where professional mixers talked about Pro Tools projects (TOP commercial songs) that come to them directly in wav - no FX no synths, samplers with a track count higher higher than 100. I doubt that any desktop PC/MAC would be able to run synths and the EXTRA plugins added in the mix stage in live mod without bouncing. [that was my statement above ->>" full commercial song (with high track number, synths, polished vocals) in one go without heavy use of bouncing the tracks to free up DSP]"

So I think it depends a lot on the type of song you make, how good you know your tools and experience in general, how much time you have to finish the project etc AND of course how powerful for audio the desktop/workstation is.
It's great that you have the ability to separate all stages of making a song! I'm a hobbyist, I finish 1 or 2 per songs year and don't have that much inspiration or experience to envision the finished song - so I change a lot of things all the time, no clear stages.

[devilfish]

My i7-2700k @ 4x 4,2GHz got 33 seconds

Windows 10 / RME Hammerfall DSP 9632 (@128 Samples)


-----

Today I ordered a new Intel Xeon E5-2683 V3 (Engineer Sample from China, 233€ on eBay)

14 Cores / 28 Threads @ 2GHz
(2,3GHz All Core Turbo and 3GHz two Core Turbo)

Hope the new machine will fully complete the Benchmark-Song

14 cores??? You going to the moon?

No, I want to go outside our galaxy

I hope the CPU will work fine. It is an Engineering Sample (Es) QEY7 with C1 Stepping. The Production CPU has 200Mhz higher all Core Turbo clock, ES=2,3Ghz vs. 2,5GHz production CPU... and hopefully no more changes. And yes, I know this CPU is not legal, but very very cheap xD

[mcatalao]

I've watched Pensado's Place where professional mixers talked about Pro Tools projects (TOP commercial songs) that come to them directly in wav - no FX no synths, samplers with a track count higher higher than 100. I doubt that any desktop PC/MAC would be able to run synths and the EXTRA plugins added in the mix stage in live mod without bouncing. [that was my statement above ->>" full commercial song (with high track number, synths, polished vocals) in one go without heavy use of bouncing the tracks to free up DSP]"

So I think it depends a lot on the type of song you make, how good you know your tools and experience in general, how much time you have to finish the project etc AND of course how powerful for audio the desktop/workstation is.
It's great that you have the ability to separate all stages of making a song! I'm a hobbyist, I finish 1 or 2 per songs year and don't have that much inspiration or experience to envision the finished song - so I change a lot of things all the time, no clear stages.

I'm sorry, missunderstood you. You do refer one go for the whole process. I really don't think it is the best way to do it, and at some point it will always make you buy another stronger PC.

Here's 2 ideas, if you keep a version of the song at different stages and really need to go.bakc to a line after bouncing you could just copy the original track between projects. I almost sure that works, but even if its not, you can always use midi and audio export in reason.

Second idea is, if you bounce synth stuff the muted track keeps tolling the Cpu. To really cope with that we usually need to delete it. But an interesting way to make it would be to have a blank or dummy device based in a nn-XT. Just drag and drop the dummy over the old synth track and store the info of the original track inside the with remark or Jp stuff.

I know this stuff ends up being somewhat a workflow delay, but imho, its always possible to solve.

[Marco Raaphorst]

• Your CPU model (Intel i7 3770k, AMD FX 8350, etc) and clock speed - as well as whether or not you are overclocking
2,3 GHz Intel Core i7 - MacBook Pro Retina RUNNING FROM FULL BATTERY
• The amount of RAM you have
16 GB 1600 MHz DDR3
• Your audiocard Brand and Model
Build-in soundcard
• Your audiocard driver settings (sample rate 44100 Hz recommended and max buffer settings)
4.096 samples
• CPU Useage Limit must be set to 95% in Reason's settings
• When playback stops / Computer too slow warning appears ----> (bars & time)
39 seconds

Audio was stuttering and glitching before it stopped. I would not be able to work on this track and simply render things to audio.

[devilfish]

My new computer is ready and it runs through the whole song
Attached are 2 photos. It is an Intel Xeon 2683v3 (Engineering Sample) CPU
Stepcode QFQK (Qualitiy Sample CPU).
My first CPU from China has unfortunately remained stuck at customs in Germany. It was destroyed
The current I bought from the United States via Ebay GPS. There were no problems with the delivery and it took me nearly 300 Dollars. An original CPU cost 1800-2000 Dollars !!!

The benchmark song had to be extended. It seems to be there is much more ASIO/CPU power available

[AttenuationHz]

My new computer is ready and it runs through the whole song
Attached are 2 photos. It is an Intel Xeon 2683v3 (Engineering Sample) CPU
Stepcode QFQK (Qualitiy Sample CPU).
My first CPU from China has unfortunately remained stuck at customs in Germany. It was destroyed
The current I bought from the United States via Ebay GPS. There were no problems with the delivery and it took me nearly 300 Dollars. An original CPU cost 1800-2000 Dollars !!!

The benchmark song had to be extended. It seems to be there is much more ASIO/CPU power available

Damn son! How many instances of the benchmark playing simultaneously maxes it?

[devilfish]

Two Songs simultaneously will stop at 25 seconds...
My old PC Stops at 32 seconds with just one Song *lol*

I really Love my new Baby