Then this shouldn't happen. It should actually slightly increase the performance the higher the CPU generation, not the other way around.
So... several things before it's down to software testing again:
a) check the motherboard, update the BIOS if needed
b) set it up correctly
c) setup your OS correctly, drop all nunecessary ASUS or GIGABYTE drivers (whatever board you have)
d) reinstall your ADC and Cubase
If the issues still remain, then we're talking.
Things I hate about... - Cubase 7
-
- KVRAF
- Topic Starter
- 14739 posts since 19 Oct, 2003 from Berlin, Germany
-
- KVRer
- 5 posts since 5 Dec, 2013 from Germany
Hi,
Thanks for the suggestions. Before I get on to it, I just want to get out the way that I'm not squarely putting the blame on Cubase 7 here (though it's certainly possible!), this is why I did mention that I did in fact (moving from Cubase 5.5 to 7) experience a huge gain in performance with my old system. This is why I was alluding to the point that it's either the chipset/board or more likely this particular chipset/board in concert with Cubase 7 is what's not playing well with Cubase 7.
Ok, to the checks and problems:
1. I have already run DPC Latency checker as part of my tests, it gives me a steady current latency of 29 micro seconds with an absolute of 52 micro seconds.
I have CPU, system temperature monitored and the CPU hovers around 42C, doesn't really go past 47C even when I hammer it.
The system is an X79 Gigabyte UP4 I've gone from f4 BIOS to f5e, this has made no difference to audio performance, but did sort out some BIOS set up niggles.
RAM is 1866mhz G.skill populated Quad Channel 32Gb's.
Hard drives are all SSD's, boot drive running on its own 512Gb SSD with a further 4x 1TB SSD's for sample library streaming. All SATA 6Gbps.
CPU is a hexacore 4930k ivy bridge E (that'd be Extreme, not mobile as somebody mentioned!).
Cubase build is 7.0.2. I had 7.0.6 and that was much worse.
Audio card is an SSL MADIxtreme running with an AlphaLink Xlogic AX.
OS- Windows 7 Pro 64bit.
*bootnote- Quick comparison, my old i7 920 quad core would handle 15-20 vst's running and some fx, no problem.
My previous system i7 970 (Gigabyte X58 UD7) hexacore 56 vst's and several effects in tests, all with low buffer settings.
----
Kate.
Thanks for the suggestions. Before I get on to it, I just want to get out the way that I'm not squarely putting the blame on Cubase 7 here (though it's certainly possible!), this is why I did mention that I did in fact (moving from Cubase 5.5 to 7) experience a huge gain in performance with my old system. This is why I was alluding to the point that it's either the chipset/board or more likely this particular chipset/board in concert with Cubase 7 is what's not playing well with Cubase 7.
Ok, to the checks and problems:
1. I have already run DPC Latency checker as part of my tests, it gives me a steady current latency of 29 micro seconds with an absolute of 52 micro seconds.
I have CPU, system temperature monitored and the CPU hovers around 42C, doesn't really go past 47C even when I hammer it.
The system is an X79 Gigabyte UP4 I've gone from f4 BIOS to f5e, this has made no difference to audio performance, but did sort out some BIOS set up niggles.
RAM is 1866mhz G.skill populated Quad Channel 32Gb's.
Hard drives are all SSD's, boot drive running on its own 512Gb SSD with a further 4x 1TB SSD's for sample library streaming. All SATA 6Gbps.
CPU is a hexacore 4930k ivy bridge E (that'd be Extreme, not mobile as somebody mentioned!).
Cubase build is 7.0.2. I had 7.0.6 and that was much worse.
Audio card is an SSL MADIxtreme running with an AlphaLink Xlogic AX.
OS- Windows 7 Pro 64bit.
*bootnote- Quick comparison, my old i7 920 quad core would handle 15-20 vst's running and some fx, no problem.
My previous system i7 970 (Gigabyte X58 UD7) hexacore 56 vst's and several effects in tests, all with low buffer settings.
----
Kate.
egbert wrote:I would try to do some research and find out if your particular motherboard/cpu/OS combo is known to be a successful configuration for low latency audio. The Gigabyte X58 UD3/UD5 boards with i9X0 CPUs were good Win 7 systems but not all mobos were working well with setup.
I don't know how how experienced at configuring systems you are. Sometimes tweaks like turning off unneeded mobo features or updating a BIOS can make the difference. Eg it is not advisable to install all the utilities that come with your mobo on disk - for example some system monitor or EZY-overclock configuration tools are resource hogs. If your CPU is under-cooled or is running a Power setting that is racheting down power to some of your cores during heavy work you aren't going to get all the benefits of a fast CPU.
You don't mention your mobo or your audio hardware.
Run DPC Latency Checker if you are on Win 7 and see if anything is causing problems there. Any big peaks there will definitely mess with low buffer settings.
I'm sure plenty of others would like to know how you go and perhaps some others with similar systems can chime in. You might want to move this to its own thread in the hardware setup section. Before blaming Cubase I would see how performance is in other hosts. You could work in Studio One or Reaper (demos if necessary) to see how well they work on the same system.
Last edited by Katzy on Thu Dec 05, 2013 6:57 pm, edited 3 times in total.
-
- KVRer
- 5 posts since 5 Dec, 2013 from Germany
Done all this twice or more already.
Even did a fresh reinstall. No difference whatsoever.
I always remove the crap from my systems.
Even did a fresh reinstall. No difference whatsoever.
I always remove the crap from my systems.
Compyfox wrote:Then this shouldn't happen. It should actually slightly increase the performance the higher the CPU generation, not the other way around.
So... several things before it's down to software testing again:
a) check the motherboard, update the BIOS if needed
b) set it up correctly
c) setup your OS correctly, drop all nunecessary ASUS or GIGABYTE drivers (whatever board you have)
d) reinstall your ADC and Cubase
If the issues still remain, then we're talking.
-
- Banned
- 22457 posts since 5 Sep, 2001
[DELETED]
-
- Banned
- 22457 posts since 5 Sep, 2001
[DELETED]
-
- KVRAF
- 6097 posts since 5 Jul, 2001 from Just about .... there
That isn't entirely true, although in a round about way. Audio streaming requires very specific behavior through the MB. Not all chipsets behave the same in buss management scenarios. Just like the scheduling behaviors of OSs can cause cross platform problems in unexpected ways, routing logic even from point releases of the same chipset firmware can have screwy impacts on high priority tasks.TheoM wrote: Cubase *itself* can't have an incompatibility with certain intel chipsets. That doesn't make sense.
If you have to ask, you can't afford the answer
-
- KVRer
- 5 posts since 5 Dec, 2013 from Germany
Yeah to be honest I wasn't entirely sure if Cubase would be affected. I was just totally running out of ideas!
I disabled power saving features in the BIOS, but just tried the speedstep EIST(?), after disabling that, I have noticed the CPU temperature shot up after a few minutes, I then tried Cubase again and noticed this time the temperature going up quite a bit. So I think we're onto something here. If I'm guessing rightly based on what somebody suggested it may be dipping the core frequencies to keep that temperature down. I'm just doing a stress test now with all fans forced to max to check this out... fingers crossed.
Thanks again for suggestions...
--
Kate.
I disabled power saving features in the BIOS, but just tried the speedstep EIST(?), after disabling that, I have noticed the CPU temperature shot up after a few minutes, I then tried Cubase again and noticed this time the temperature going up quite a bit. So I think we're onto something here. If I'm guessing rightly based on what somebody suggested it may be dipping the core frequencies to keep that temperature down. I'm just doing a stress test now with all fans forced to max to check this out... fingers crossed.
Thanks again for suggestions...
--
Kate.
-
- Banned
- 22457 posts since 5 Sep, 2001
[DELETED]
-
- Banned
- 22457 posts since 5 Sep, 2001
[DELETED]
-
- KVRAF
- Topic Starter
- 14739 posts since 19 Oct, 2003 from Berlin, Germany
Okay, you seem to have the basic thing down - and I assume you created an image prior to further testing.
I run on a RME HDSPe connected to a Digiface (now officially DxD'd). And I tested several Cubase versions since this original issue showed up.
If you have a clean image, try to put C7.02 and C7.06 side by side again - rule out that "roll back" is the issue, have a clean rig for each test. Actually, for me it doesn't matter which C7 (prior to C7.5), the performance is equally bad on a PC. While C6.5 clearly is superior in low latency. And since I'm usually working in 256samples and about 6ms in/out each - that sure makes a difference
C7 runs somewhat stable at 1024 samples (about 20ms for me). But I do run into issues real quick in real work environsments.
The tests I did (and my fellows as well) with an (in)official Steinberg project showed, that the higher the CPU generation, the better the performance in Cubase 6. Cubase 7 however only showed an improvement of 5% absoluite maximum - spanned over several CPU generations (Bloomfield to Haswell).
So what you encounter, is the infamous ASIO Engine issue. Just from your descriptions alone.
Unfortunately, there is no real "fix" for all this. You can only do further tests, revert to higher latency, channel freezing, etc... and pray that things improve with future versions (C8 and up).
Or... stick with C6.5
I am on an OC'd i7 920 on ASUS P6TSE. With DPC I currently (with this setup) have a latency of 86microsections average, and 106microseconds max. Always in the green area.Katzy wrote:*bootnote- Quick comparison, my old i7 920 quad core would handle 15-20 vst's running and some fx, no problem.
My previous system i7 970 (Gigabyte X58 UD7) hexacore 56 vst's and several effects in tests, all with low buffer settings.
I run on a RME HDSPe connected to a Digiface (now officially DxD'd). And I tested several Cubase versions since this original issue showed up.
If you have a clean image, try to put C7.02 and C7.06 side by side again - rule out that "roll back" is the issue, have a clean rig for each test. Actually, for me it doesn't matter which C7 (prior to C7.5), the performance is equally bad on a PC. While C6.5 clearly is superior in low latency. And since I'm usually working in 256samples and about 6ms in/out each - that sure makes a difference
C7 runs somewhat stable at 1024 samples (about 20ms for me). But I do run into issues real quick in real work environsments.
The tests I did (and my fellows as well) with an (in)official Steinberg project showed, that the higher the CPU generation, the better the performance in Cubase 6. Cubase 7 however only showed an improvement of 5% absoluite maximum - spanned over several CPU generations (Bloomfield to Haswell).
So what you encounter, is the infamous ASIO Engine issue. Just from your descriptions alone.
Unfortunately, there is no real "fix" for all this. You can only do further tests, revert to higher latency, channel freezing, etc... and pray that things improve with future versions (C8 and up).
Or... stick with C6.5
-
- Banned
- 22457 posts since 5 Sep, 2001
[DELETED]
-
- KVRAF
- 6097 posts since 5 Jul, 2001 from Just about .... there
No not better ... Pro Audio Builders were using MBs with VIA chipsets the worked under some scenarios, but when certain buss mastering audio cards were used, that chipset couldn't handle it. It took them forever to figure out it was the chipset.TheoM wrote:Cubase doesn't have incompatibility with the latest sandy E chipsets. Better? Pro audio builders are building those machines all the time, working perfectly.
Audio interfaces can have compatibility problems.
That scenario is how I figured out that the "pro builder" MB was the problem I was having with my UAD setup.
If you have to ask, you can't afford the answer
-
- Banned
- 22457 posts since 5 Sep, 2001
[DELETED]
-
- KVRAF
- Topic Starter
- 14739 posts since 19 Oct, 2003 from Berlin, Germany
Still shouldn't happen, Theo. Either way.
I really don't know what's happening here - so I'm also swimming in water.
Gigabyte boards had issues with Bloomfield and Sandy Bridge, but that issue should be gone by now (especially with BIOS updates). So I don't know where to direct.
What Kate could try, would be:
a) Generic ASIO driver
b) ASIO4all
Just for sh*ts and giggles, if this improves things. If so, there are some ADC issues. If not, I don't know where to help.
DPC is unholy low, the CPU is a beast, there's plenty of RAM, the OS is optimized. Only Cubase is acting irradically.
And to be honest:
20-25 plugin instances on an i7 920 quad
55 instances on an i7 970 hexa
and with a hexa Ivy E, the performance is halfed again?
Even if the same hardware/software was used?
This looks to me like an incompatibilty issue at first, and then a core issue with the host at second.
Heck even the old ScanPro Audio benchmarks with Quad vs Hexa on 9x0 were not that drastic IIRC.
Lost for words, currently. So... if possible, a side-by-side try could reveal some interesting results.
I really don't know what's happening here - so I'm also swimming in water.
Gigabyte boards had issues with Bloomfield and Sandy Bridge, but that issue should be gone by now (especially with BIOS updates). So I don't know where to direct.
What Kate could try, would be:
a) Generic ASIO driver
b) ASIO4all
Just for sh*ts and giggles, if this improves things. If so, there are some ADC issues. If not, I don't know where to help.
DPC is unholy low, the CPU is a beast, there's plenty of RAM, the OS is optimized. Only Cubase is acting irradically.
And to be honest:
20-25 plugin instances on an i7 920 quad
55 instances on an i7 970 hexa
and with a hexa Ivy E, the performance is halfed again?
Even if the same hardware/software was used?
This looks to me like an incompatibilty issue at first, and then a core issue with the host at second.
Heck even the old ScanPro Audio benchmarks with Quad vs Hexa on 9x0 were not that drastic IIRC.
Lost for words, currently. So... if possible, a side-by-side try could reveal some interesting results.
-
- Banned
- 22457 posts since 5 Sep, 2001
[DELETED]