LATENCY: XITE VERSUS PCI CREAMWARE CARD?

The Sonic Core XITE hardware platform for Scope

Moderators: valis, garyb

Post Reply
m.my91
Posts: 119
Joined: Sat May 10, 2014 12:44 am

LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by m.my91 »

Hello.
i notice that pci creamware card make a little latency when you plug many dev plugins in serial mode...
i know that creamware card have only one type of dsp.
the xite have two type of dsp.
6 old one like the creamware pci card and 4 or 12 new dsp.
in fact the signal pass through the 6 old dsp before being calculate by the new dsp...
if the xite give more dsp power i wonder if the 2 dsp calculations generate more latency than only one?
in another way is xite produce more latency than old creamware pci card?
thanks for reply
User avatar
Sounddesigner
Posts: 1063
Joined: Sat Jun 02, 2007 11:06 pm

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by Sounddesigner »

If you're hearing latency when chaining plugins in SCOPE then you need to use zero-latency plugins like dNa MasterComp or work at 96khz samplerate wich cuts latency more than in half for converters, plugins, Asio I/O, etc.

The XITE-1 generally has less latency than the older pci cards because the dsp's are bigger. There is a 2 samples delay when spanning from one dsp to another so if the Mixer or a plugin is using 2 dsp's for power there is a 2 samples delay involved. Because the old pci dsp's are so small hungry devices are more likely to span multiple dsp's, but XITE-1's dsp's are bigger so there is less need to span. XITE-1's latency is less because of the bigger dsp's. Usually the 2 samples delay is not a problem tho.

EDIT: XITE-1's 6 older dsp's are generally not used for mixers, effects and synths but are used for I/O modules, VDAT, and incompatible plugins. To my knowledge the older dsp's carry the same 2 samples spanning delay on XITE-1 as they do on older pci. XITE-1 differs from older pci in that if a problem does occur you can manually assign devices to wichever dsp you want them to draw power from.
Last edited by Sounddesigner on Sun Jun 21, 2015 2:29 pm, edited 1 time in total.
User avatar
garyb
Moderator
Posts: 23248
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by garyb »

even a chain of digital hardware delays, reverbs, etc will have the same several sample delay.
anything which requires calculation will take some time, that includes AD/DA converters.

it's still less delay than a fully wired major hardware studio, when using the patchbay. if there are gundreds of feet of wire, this generates delay. audio does not instantly flow across a cable, it moves at the speed of light. likewise across a pc board.
m.my91
Posts: 119
Joined: Sat May 10, 2014 12:44 am

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by m.my91 »

Sounddesigner wrote:If you're hearing latency when chaining plugins in SCOPE then you need to use zero-latency plugins like dNa MasterComp or work at 96khz samplerate wich cuts latency more than in half for converters, plugins, Asio I/O, etc.

The XITE-1 generally has less latency than the older pci cards because the dsp's are bigger. There is a 2 samples delay when spanning from one dsp to another so if the Mixer or a plugin is using 2 dsp's for power there is a 2 samples delay involved. Because the old pci dsp's are so small hungry devices are more likely to span multiple dsp's, but XITE-1's dsp's are bigger so there is less need to span. XITE-1's latency is less because of the bigger dsp's. Usually the 2 samples delay is not a problem tho.

EDIT: XITE-1's 6 older dsp's are generally not used for mixers, effects and synths but are used for I/O modules, VDAT, and incompatible plugins. To my knowledge the older dsp's carry the same 2 samples spanning delay on XITE-1 as they do on older pci. XITE-1 differs from older pci in that if a problem does occur you can manually assign devices to wichever dsp you want them to draw power from.
yes i understand that the dsp are bigger in xite .
but there's two type of dsp in xite (the old one like pci card and the knew one).
if the signal need to pass the two type of dsp i wonder how xite can be more speed than a old pci card?
ps :i work in 96 khz.
User avatar
garyb
Moderator
Posts: 23248
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by garyb »

m.my91 wrote: but there's two type of dsp in xite (the old one like pci card and the knew one).
if the signal need to pass the two type of dsp i wonder how xite can be more speed than a old pci card?
ps :i work in 96 khz.
that's not how it works...
User avatar
Sounddesigner
Posts: 1063
Joined: Sat Jun 02, 2007 11:06 pm

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by Sounddesigner »

m.my91 wrote:
Sounddesigner wrote:If you're hearing latency when chaining plugins in SCOPE then you need to use zero-latency plugins like dNa MasterComp or work at 96khz samplerate wich cuts latency more than in half for converters, plugins, Asio I/O, etc.

The XITE-1 generally has less latency than the older pci cards because the dsp's are bigger. There is a 2 samples delay when spanning from one dsp to another so if the Mixer or a plugin is using 2 dsp's for power there is a 2 samples delay involved. Because the old pci dsp's are so small hungry devices are more likely to span multiple dsp's, but XITE-1's dsp's are bigger so there is less need to span. XITE-1's latency is less because of the bigger dsp's. Usually the 2 samples delay is not a problem tho.

EDIT: XITE-1's 6 older dsp's are generally not used for mixers, effects and synths but are used for I/O modules, VDAT, and incompatible plugins. To my knowledge the older dsp's carry the same 2 samples spanning delay on XITE-1 as they do on older pci. XITE-1 differs from older pci in that if a problem does occur you can manually assign devices to wichever dsp you want them to draw power from.
yes i understand that the dsp are bigger in xite .
but there's two type of dsp in xite (the old one like pci card and the knew one).
if the signal need to pass the two type of dsp i wonder how xite can be more speed than a old pci card?
ps :i work in 96 khz.
Both the old and new dsp's in XITE-1 give the same 2 samples buffering why would this change just cause they are paired together? It's a common practice to mix two types of dsp's. The old Protools TDM farm cards had old and new dsp's and the new Protools HDX has both Texas Instrument dsp's for the plugins and FPGA for the Mixer but still has a Round-Trip-Latency of .7ms at 96khz. The two different dsp's in HDX doesn't appear to have changed anything and if so not of any real significance because the two are paired together. UAD has newer dsp's in their Octo cards and their Solo/duo/quads use older ones. There's nothing unique about XITE-1 using two types of dsp's it's a common practice. The dsp buffering with SCOPE has always been 2 samples and that 2 samples is what the new dsp's in XITE-1 have as well. It is illogical for their nature to magically change and extra latency pop out of nowhere cause they are paired together.. What are you basing your assumption on? The older pci cards smaller dsp's and constant spanning should be a bigger worry for you since that would add up to or exceed the problem that your theory on XITE-1 has. If you're monitoring threw a chain of effects on the old pci cards many of those effects may be on different dsp's and the mixer may span multiple dsp's to add to the I/O wich may be on a different dsp.

PS. If you are already working at 96khz on PCI card then your Round-Trip-Latency is at around .7ms without plugins, choose the right plugins and you should still be at .7ms or a tiny bit higher wich is about as good as it gets in the digital world. Unless you're going to pony up on a Fairlight system it's unlikely you'll find better monitoring RTL in the digital world then what you have. Choose the right plugins for Tracking.

EDITED
tgstgs
Posts: 526
Joined: Sun Jan 15, 2006 4:00 pm

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by tgstgs »

i fact you have different latency between dsps on xite;
old pci boards have 2 samples between each dsp where the 2 samples are samplerate dependant;
on xite you have
old to old;
new to old;
new to new on same slot;
new to new on different slot;
a nightmare for big devices__
in combination with the dsp optimizing algo > ramdom filtering_
only solution is to manually restrict each device to a specified dsp and watch your lines;

good vibes
User avatar
Sounddesigner
Posts: 1063
Joined: Sat Jun 02, 2007 11:06 pm

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by Sounddesigner »

Thanks for the response and info tgstgs . I was quoting the developer from DAS who stated on the http://www.xited.org forum long ago that there was a 2 samples delay with XITE-1 dsp's when he was pointing out possible phase problems IIRC. But judging by your info the DAS developer may have only been referring to same slot delay from different dsp's and not different slots, etc. I don't know, maybe I misread or misunderstood also but I'm fairly sure that was the case back then.

Do you have any figures as to the number of samples of the different dsp travel paths and how big of a difference the number of samples are for each? I mean if slot-to-slot samples number is different than different-slots how big is the difference and do you have a number? Is the same delay amount samplerate dependant like the old pci?
User avatar
dante
Posts: 5043
Joined: Sat Nov 24, 2001 4:00 pm
Location: Melbourne Australia
Contact:

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by dante »

Check out these tests from DAS, should give you an idea :

http://www.hitfoundry.com/issue_07/xitemast.htm

Specifically the tabs 'Test 1:' thru 'Test 4:'
User avatar
Sounddesigner
Posts: 1063
Joined: Sat Jun 02, 2007 11:06 pm

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by Sounddesigner »

Thanks for posting this link to information Dante. Will further check it out later.
User avatar
dante
Posts: 5043
Joined: Sat Nov 24, 2001 4:00 pm
Location: Melbourne Australia
Contact:

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by dante »

Good luck - with careful study of the animated gifs, it may give you enough info to reverse engineer the delay relationships between the 4 DSP farms.

At the moment, that is a secret unlocked and kept by one of our DAS friends.

Point is, you really need to know those delay relationships to effectively use manual assignment otherwise phase issues will most likely occur.
m.my91
Posts: 119
Joined: Sat May 10, 2014 12:44 am

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by m.my91 »

Sounddesigner wrote:
m.my91 wrote:
Sounddesigner wrote:If you're hearing latency when chaining plugins in SCOPE then you need to use zero-latency plugins like dNa MasterComp or work at 96khz samplerate wich cuts latency more than in half for converters, plugins, Asio I/O, etc.

The XITE-1 generally has less latency than the older pci cards because the dsp's are bigger. There is a 2 samples delay when spanning from one dsp to another so if the Mixer or a plugin is using 2 dsp's for power there is a 2 samples delay involved. Because the old pci dsp's are so small hungry devices are more likely to span multiple dsp's, but XITE-1's dsp's are bigger so there is less need to span. XITE-1's latency is less because of the bigger dsp's. Usually the 2 samples delay is not a problem tho.

EDIT: XITE-1's 6 older dsp's are generally not used for mixers, effects and synths but are used for I/O modules, VDAT, and incompatible plugins. To my knowledge the older dsp's carry the same 2 samples spanning delay on XITE-1 as they do on older pci. XITE-1 differs from older pci in that if a problem does occur you can manually assign devices to wichever dsp you want them to draw power from.
yes i understand that the dsp are bigger in xite .
but there's two type of dsp in xite (the old one like pci card and the knew one).
if the signal need to pass the two type of dsp i wonder how xite can be more speed than a old pci card?
ps :i work in 96 khz.
Both the old and new dsp's in XITE-1 give the same 2 samples buffering why would this change just cause they are paired together? It's a common practice to mix two types of dsp's. The old Protools TDM farm cards had old and new dsp's and the new Protools HDX has both Texas Instrument dsp's for the plugins and FPGA for the Mixer but still has a Round-Trip-Latency of .7ms at 96khz. The two different dsp's in HDX doesn't appear to have changed anything and if so not of any real significance because the two are paired together. UAD has newer dsp's in their Octo cards and their Solo/duo/quads use older ones. There's nothing unique about XITE-1 using two types of dsp's it's a common practice. The dsp buffering with SCOPE has always been 2 samples and that 2 samples is what the new dsp's in XITE-1 have as well. It is illogical for their nature to magically change and extra latency pop out of nowhere cause they are paired together.. What are you basing your assumption on? The older pci cards smaller dsp's and constant spanning should be a bigger worry for you since that would add up to or exceed the problem that your theory on XITE-1 has. If you're monitoring threw a chain of effects on the old pci cards many of those effects may be on different dsp's and the mixer may span multiple dsp's to add to the I/O wich may be on a different dsp.

PS. If you are already working at 96khz on PCI card then your Round-Trip-Latency is at around .7ms without plugins, choose the right plugins and you should still be at .7ms or a tiny bit higher wich is about as good as it gets in the digital world. Unless you're going to pony up on a Fairlight system it's unlikely you'll find better monitoring RTL in the digital world then what you have. Choose the right plugins for Tracking.

EDITED
thank you for your precise reply.
so PCI scope system made 0.7 ms at 96 khz without plugins ok but:
0.7 ms is AD DA time convertion (converter latency)?
is it the same if you only use scope in digital domain(adat smux spdif)?
do you knows the size of buffer used by dev scope plugins ,adat smux module,spdif module?
is there a "list" showing the buffer needed in sample for each dev scope plugins?
i'am using plugins like "VINCO" in insert on a protools hd 7.4 and it work very well for optimising the level recording.
i don't have a real latency problem but i wonder how many samples need dev plugins for put the right value in the insert HW delay of protools: (external effect compensation latency):
0.7 ms + vinco latency?
thanks
User avatar
Sounddesigner
Posts: 1063
Joined: Sat Jun 02, 2007 11:06 pm

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by Sounddesigner »

m.my91 wrote:
thank you for your precise reply.
so PCI scope system made 0.7 ms at 96 khz without plugins ok but:
0.7 ms is AD DA time convertion (converter latency)?
is it the same if you only use scope in digital domain(adat smux spdif)?
do you knows the size of buffer used by dev scope plugins ,adat smux module,spdif module?
is there a "list" showing the buffer needed in sample for each dev scope plugins?
i'am using plugins like "VINCO" in insert on a protools hd 7.4 and it work very well for optimising the level recording.
i don't have a real latency problem but i wonder how many samples need dev plugins for put the right value in the insert HW delay of protools: (external effect compensation latency):
0.7 ms + vinco latency?
thanks
i don't know the latency amount of plugins and different I/O Protocols, i've never been that concerned since I use SFP Mode at 96khz thus generally I have no latency problems. I told you in the other thread there is a latency checker plugin, why did'nt you inquire of it? Use SFP Mode and then If latency is bothering you at 96khz then digital is not for you. A few samples in SFP Mode is unnoticeable and of no concern to most people and if it is then that person should use analog hardware cause digital will never satisfy that person. i don't have latency numbers for plugins, and protocols but you can start a inquiry thread to get the latency checker device for SCOPE wich I think GOST created.

If you are picky about latency then use SFP mode not XTC mode otherwise you're pinching yourself and telling others it hurts :) . Don't use dsp plugins inside your DAW if you are concerned about latency. I'm picky about latency myself to an extent so I bought SCOPE and use SFP Mode, cause I learned that "If you see a problem solve the problem, don't go in circles over it". DSP plugins have a advantage over Native plugins with regards to latency BUT if you use those DSP plugins the wrong way (wich is inside the Native DAW as VST's) then Native plugins has the advantage in that regard. DSP plugins have more latency than Native plugins when you try to use them like Native plugins inside your DAW (not only will you have the plugins inherent latency but you will have an additionional round-trip Asio latency). If you care about low-latency then why create a bad situation and then complain about the natural results :) ? I wouldn't use XTC Mode cause it gives you disadvantages and greatly limits the SCOPE Platform.

There is a chainer plugin for SCOPE that reduces the latency of SCOPE plugins inside your DAW. If you insert the chainer plugin inside your DAW and then insert multiple SCOPE plugins inside the chainer you'll only get 1 roundtrip of latency added for the multiple plugins chained instead of a roundtrip for each chained plugin when not using the chainer.

PS. Altho I do not have the inherent amount of plugin latency and hardware I/O protocols latency I DO know that XTC/VST DSP plugins need to travel in and out of the Asio domain to process from the dsp card. So whatever your Native buffer-size is has to be added to your dsp plugin. If your Asio buffer is 256 then you will have 256 input and 256 output latency in addition to your dsp plugin latency. Using DSP plugins in XTC Mode inside your DAW will give you a significant disadvantage to Native plugins wich don't have the extra latency cause they are indigenous to that environment. The only way around this is to build the DAW around the dsp platform like Protools HDX then the dsp plugins will be indigenous.
m.my91
Posts: 119
Joined: Sat May 10, 2014 12:44 am

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by m.my91 »

Sounddesigner wrote:
m.my91 wrote:
thank you for your precise reply.
so PCI scope system made 0.7 ms at 96 khz without plugins ok but:
0.7 ms is AD DA time convertion (converter latency)?
is it the same if you only use scope in digital domain(adat smux spdif)?
do you knows the size of buffer used by dev scope plugins ,adat smux module,spdif module?
is there a "list" showing the buffer needed in sample for each dev scope plugins?
i'am using plugins like "VINCO" in insert on a protools hd 7.4 and it work very well for optimising the level recording.
i don't have a real latency problem but i wonder how many samples need dev plugins for put the right value in the insert HW delay of protools: (external effect compensation latency):
0.7 ms + vinco latency?
thanks
i don't know the latency amount of plugins and different I/O Protocols, i've never been that concerned since I use SFP Mode at 96khz thus generally I have no latency problems. I told you in the other thread there is a latency checker plugin, why did'nt you inquire of it? Use SFP Mode and then If latency is bothering you at 96khz then digital is not for you. A few samples in SFP Mode is unnoticeable and of no concern to most people and if it is then that person should use analog hardware cause digital will never satisfy that person. i don't have latency numbers for plugins, and protocols but you can start a inquiry thread to get the latency checker device for SCOPE wich I think GOST created.

If you are picky about latency then use SFP mode not XTC mode otherwise you're pinching yourself and telling others it hurts :) . Don't use dsp plugins inside your DAW if you are concerned about latency. I'm picky about latency myself to an extent so I bought SCOPE and use SFP Mode, cause I learned that "If you see a problem solve the problem, don't go in circles over it". DSP plugins have a advantage over Native plugins with regards to latency BUT if you use those DSP plugins the wrong way (wich is inside the Native DAW as VST's) then Native plugins has the advantage in that regard. DSP plugins have more latency than Native plugins when you try to use them like Native plugins inside your DAW (not only will you have the plugins inherent latency but you will have an additionional round-trip Asio latency). If you care about low-latency then why create a bad situation and then complain about the natural results :) ? I wouldn't use XTC Mode cause it gives you disadvantages and greatly limits the SCOPE Platform.

There is a chainer plugin for SCOPE that reduces the latency of SCOPE plugins inside your DAW. If you insert the chainer plugin inside your DAW and then insert multiple SCOPE plugins inside the chainer you'll only get 1 roundtrip of latency added for the multiple plugins chained instead of a roundtrip for each chained plugin when not using the chainer.

PS. Altho I do not have the inherent amount of plugin latency and hardware I/O protocols latency I DO know that XTC/VST DSP plugins need to travel in and out of the Asio domain to process from the dsp card. So whatever your Native buffer-size is has to be added to your dsp plugin. If your Asio buffer is 256 then you will have 256 input and 256 output latency in addition to your dsp plugin latency. Using DSP plugins in XTC Mode inside your DAW will give you a significant disadvantage to Native plugins wich don't have the extra latency cause they are indigenous to that environment. The only way around this is to build the DAW around the dsp platform like Protools HDX then the dsp plugins will be indigenous.
thanks for your reply.
no i don't use xtc mode i made insert in protools hd (with aes ebu io) i connects digi 192 io directly on the digital io of scope.
so i don't use computer cpu (only digidesign and scope dsp).so i don't have latency.
i have more latency than scope dsp plugins, when i insert a TDM or RTAS or asio (vsthost) VST.(with asio scope at 64 buffer).
using scope dsp on protools insert is like using digital hardware (tc finalizer,drawmer,zsys....) for optimised recording session.
can you tell me where i can find the "chainer plugins for scope 5.0"
the plugins chainer need only 1 sample! or 1 ms?
i don't see any "chainer plugins" nowhere.
do you have a link for download it?
thanks
User avatar
Sounddesigner
Posts: 1063
Joined: Sat Jun 02, 2007 11:06 pm

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by Sounddesigner »

Understood! I think the actual name of the chainer plugin is MultiFX. It should be located in your plugins folder for V5. It's a stock plugin.
m.my91
Posts: 119
Joined: Sat May 10, 2014 12:44 am

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by m.my91 »

Sounddesigner wrote:Understood! I think the actual name of the chainer plugin is MultiFX. It should be located in your plugins folder for V5. It's a stock plugin.
thanks a lot.
i will use multifx seen if it works when i chain plugins.
have a good day
m.my91
Posts: 119
Joined: Sat May 10, 2014 12:44 am

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by m.my91 »

Hello
i have tried the multifx module and insert eq and compressor.
The multi fx work better than chain the eq and compressor module without it.
multi fx module work better in term of latency.
and a good new!
I have found a vst rtas wrapper!
blue cat patchwork!
bluecat also reduce the latency
it work better than a rtas plugin!
so in term of latency the better is:
Scope plugins + multifx.
digidesign tdm.
Bluecat patch work (vst on protools).
rtas .
thanks
User avatar
ronnie
Posts: 788
Joined: Thu Jul 17, 2003 4:00 pm
Location: Varies Between 30Hz & 20KHz
Contact:

Re: LATENCY: XITE VERSUS PCI CREAMWARE CARD?

Post by ronnie »

Good info!
"I’ve come to the conclusion that synths are like potatoes, they’re no good raw—you’ve got to cook ‘em, and I cooked these sounds for months before I got them to the point where they sounded musical to me." Lyle Mays
Post Reply