XITE-1 PC CONFIGURATION REFERENCE

The Sonic Core XITE hardware platform for Scope

Moderators: valis, garyb

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

Re: XITE-1 PC CONFIGURATION REFERENCE

Post by Sounddesigner »

valis wrote: Mon Dec 18, 2023 7:08 pm You can try the HPET to RTC trick in your EFI/BIOS, but most likely another driver is hogging too much time in the background
I never knew of this trick. I'll keep it in mind in the future. I did disable HPET in the Device Manager.

I think the problem of runnig my Projects at the lowest latency possible is solved. I called tech support at Magic Micro to get a few ideas to try and the guy told me that some of their Audio Production customers have problems with 'Turbo Boost'. Well 'that's a blast from the past' i thought. I never considered this cause it was a problem i had with the very first generation of i7 processors and i thought Intel worked out all the kinks by now but i was wrong!

I turned off everything in the BIOS that i use to turn off in the past just now, wich was Turbo boost, C-State, Eist, etc and the problem seems to have disappeared. My last i7 4790k did not have this problem wich is part of the reason i thought it was just a first-generation problem. The more things change the more they stay the same!

Even tho i wanted the Turbo Boost wich went to 4.9 GHz, i've never been a big fan of substitute-power. Substitutes such as turbo-boost, hyperthreading, and some instruction sets are not better than the real deal wich is just a high base speed per core. I believe in just making sure the CPU has a high base speed around 4 GHz and one will be better served. Substitute-power often brings problems like i just experienced.

I don't know if i truly needed to turn off all that i did in the Bios and later i may turn some of it back on to deduce the culprits and the innocent. I just went off memory and turned off everthing i turned off with the first generation i7 920 and i'll figure out everything else later. So far so good and i am definitely doing MUCH better at 1 ms buffer now. Ultra low latency at high samplerates is a great way to test a computer system, if somethings not right you will know it quickly.

I still need further testing of this computer for a final conclusion, but so far everything is working fine. This is definitely a PC Configuration other SCOPERs should consider. Combined with XITE-1 you can easily handle the most ambitious projects. Plus it's generally better to get a system that someone else already played guinea pig and tested :) ...
User avatar
valis
Posts: 7318
Joined: Sun Sep 23, 2001 4:00 pm
Location: West Coast USA
Contact:

Re: XITE-1 PC CONFIGURATION REFERENCE

Post by valis »

Disabling HPET in BIOS/EFI is the correct way, that would change the driver in Driver Manager as it switches everything to RTC (realtime clock) which is less interrupts/lower precision. Note this fix works simply reduces the number of calls that badly coded drivers make, and mostly applies to hardware from 10+ years ago. I've had no issues with Scope & HPET on my 2015 era PC, which is in contrast to the later eras of Core2 and early models of i7 (which you say you used), where HPET often caused problems with hardware from that era.

In regards to Turbo, the issue here is that you want all of your cores in the *highest* possible state as much of the time as possible. There's a few reasons why, but it generally comes down to multi-threading in audio applications loading cores in a manner that favors the faster cores. A few things can happen here:

You're loading up mostly 1 core with the high cpu usage (native/vst) plugins when you start working, and it hits a very high cpu turbo state as a result (cores are limited by TDP, the more cores in action, the lower the turbo states available are). You add more plugins and suddenly things start getting wonky, necessitating reshuffling around the process load. In this era that's usually done simply by stopping and starting the playback or recording again, but in some DAWs you may find that certain plugin chains will not be happy no matter what you do as they won't be distributed across multiple cores. This is the MOST common complaint from people (once my project get going, things fell down).

You get used to working on 1-2 cores with a specific master chain, and then late into a project load up something semi-heavy, and reduce the CPU to the lowest state for all cores turbo (instead of the 2 cores being turbo), this is most common on a 4-6 core CPU in terms of issues.

Hyperthreading is a different story, but again things have changed drastically since the early days. The main thing to understand about all current implementations is that it's a way of keeping the pipeline as active as possible so as few CPU cycles are wasted as possible. DIsabling it entirely with any machine after 2014 really shouldn't be necessary.

Disabling core parking and the low core sleep EIST states can still help Scope PCI based systems, but I don't even have to do that with my Scope box unless I'm doing a lot of complex projects. I found that simply changing power modes to high performance in Win10/11 does enough that I don't bother with BIOS fixes. I switch back to balanced when done, so that the cpu isn't at 100% (even with scope open to a default/idle project) as this reduces my power load from the wall by about 40%.

I'm not sure how much--if any--of this is necessary with Scope XITE as it's PCIe based and seems to play much better with modern systems.
User avatar
astroman
Posts: 8410
Joined: Fri Feb 08, 2002 4:00 pm
Location: Germany

Re: XITE-1 PC CONFIGURATION REFERENCE

Post by astroman »

Even if I‘m not planning to leave my old P4mobile, this is valuable information. 8)
Thank you, Valis.
User avatar
Sounddesigner
Posts: 1063
Joined: Sat Jun 02, 2007 11:06 pm

Re: XITE-1 PC CONFIGURATION REFERENCE

Post by Sounddesigner »

Very indepth and informative Valis, especially on Turbo Boost!

My philosophy with Services on your computer is 'turn them off if their not truly needed'. Anything running takes up resources and may increase dpc-latency. At high Samplerates ones machine needs to be lean and mean.

I do leave hyperthreading activated cause i've seen significant benifits from it in terms of larger projects with many Effects and Instruments, but with a single instrument or Effect that is EXTREMELY hungry this does not benifit you, only high clock speeds can save you.

The main problem i had with the first-generation i7 was that it just was not fast enough. The first i7 ran at 2.66 GHz and its speed was cut in half since i ran at 96khz. There was many demmanding Synths and Effects that my computer just could'nt handle cause they required more clock speed of a single core. When it comes to core-killers such as synths and effects that require a bigger and more powerfull core no substitute-power will remedy this. When i came accross core-killers nothing was there to help, not Turbo boost, not more cores, not more virtual-cores/hyperthreading. My only solution i had was to upgrade to a faster i7 wich ran at a greater speed of 4 GHz.

More cores and virtual-cores do help in regards to larger projects with many plugins so i do welcome them, but for single threaded software VSTs that are VERY demmanding only raw juggernaught clock speed will help, especially in my situation of high samplerates and low-latency.

I know your aware of all this already, i'm just trying to explain my personal prefences and why i'm so fixated on high clock speeds.

Since the 12th generation of Core CPUs intel has been using a hybrid Architecture of P-Cores and E-Cores. While i welcome this i still would prefer that all the Cores on my new 10-Core ALL ran at 3.7 GHz. Instead only the 6 P-Cores run at the full 3.7 GHz, the 4 E-Cores only run at 2.8 GHz. I understand that E-Cores are efficient cores that handle background tasks and lite duty processing to free up Performance-Cores to focus soley on the most demmanding tasks making them essentially faster. Do we really need 4 E-cores? Why not just 2? Must E-Cores run only at 2.7ghz, why not at 3.7ghz like the P-Cores? I prefer ALL of my Cores to run at 3.7 GHz. This smells like another Substitute wich usually are'nt as good as the real deal.

I'm no fool tho, i know we must get the most of whatever tools are available whether flawed or not just like the Audio Engineers of the past. Ultimately i welcome whatever power intel can get to us no matter if substitute or genuine, i just don't care for too much substitutes as they reduce CPU Clock speeds in favore of more cores and other substitutes. Especially since many of us have had insufficient clock speed problems.

I mostly have praises for intel and think overall they did a great job. I also blame mainly Microsoft for many of the problems i had.

I do agree with you hyperthreading is overall very usefull and IMO one of the better substitutes, i'm just saying there are situations i ran into that only higher clock speeds per core could solve.


EDITED
Last edited by Sounddesigner on Tue Dec 26, 2023 1:27 pm, edited 1 time in total.
User avatar
valis
Posts: 7318
Joined: Sun Sep 23, 2001 4:00 pm
Location: West Coast USA
Contact:

Re: XITE-1 PC CONFIGURATION REFERENCE

Post by valis »

Yes, for people who do mixing of mostly pre-recorded audio and thus a lot of (less cpu intensive than VSTi) insert effects, more cores may be better than faster cores. For people who prefer high end virtual instruments that do advanced synthesis (uHe for instance) algorithms and lots of circuit modelling, high speed cores is best. In the consumer realm the market is laid out so that as you go up in cost you go up in cores, so there's not much to consider there, but in the 'workstation' and 'HEDT' market where you have Xeons and Threadrippers, this means that a lesser core count SKU with higher clockspeed may be better for some workflows, and worse for others.

Looking at Cinebench and Gaming benchmarks in reviews tells audio users nothing. :wink:
circlingkailas
Posts: 16
Joined: Sun Nov 27, 2011 4:09 pm

Re: XITE-1 PC CONFIGURATION REFERENCE

Post by circlingkailas »

Hi everyone - Checking in 3 years after my last post. I am still using my original PowerPulsar on a XP Carillon PC, and my XITE is still sitting in its box. Looks like everyone is still mainly doing IT rather than making music. Does anyone out there have a single go-to PC or laptop that is a turn-key solution running the XITE and Scope? Or another way of asking, is anyone in the market for a brand new, carefully packaged XITE?

I would be using it if there was a straightforward answer to what PC desktop or laptop to match it with. Including, ideally, a way to link up the XITE with a Mac.

Thanks!
User avatar
valis
Posts: 7318
Joined: Sun Sep 23, 2001 4:00 pm
Location: West Coast USA
Contact:

Re: XITE-1 PC CONFIGURATION REFERENCE

Post by valis »

I just built an XITE box. We also have a for sale forum, and I started a thread to encourage PCI users to consider migrating to XITE.

To connect XITE to a Mac, you'll want to link the XITE and Mac via ADAT, analog i/o and MIDI as with the PCI cards. I prefer RME soundcards still for latency+stability++matrix routing options, but you have lots of options on Mac to choose from. For the XITE's native software, you can install RTPMidi (free for Windows users, connects to AudioMIDI setup's network driver on Mac) to handle MIDI, and route through the XITE as needed. Note the XITE has the same limitations for 64bit ASIO support as the PCI cards, so I use a legacy version of Bidule (0.667 if I recall--I can look that up if neede) to host native software.

Options abound!
User avatar
garyb
Moderator
Posts: 23246
Joined: Sun Apr 15, 2001 4:00 pm
Location: ghetto by the sea

Re: XITE-1 PC CONFIGURATION REFERENCE

Post by garyb »

the XITE is not as sensitive to the motherboard as XP days PCI cards. there are an extremely few number of PCs that will make trouble. aslmost everything works, so there is no definitive list because it isn't needed.
Post Reply