Topic: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

Hi,
currently I am testing a Lenovo P1 Gen2, Win 10 1809, MADIface USB.

I can't set ASIO buffersize lower than 1024, otherwise recording a full MADI stream will cause some rare ASIO drop outs (Sequoia 15). That's quite disappointing to me, as the P1 is quite an expensive piece of hardware. I did some tweaking, driver updates, energy options, all that stuff.

Any experience here? Recommendations for other modern laptops? What about HP Zbook?

Best,
Ulrich

2

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

Out of interest: what does Latency Mon say on this machine?

Regards
Matthias Carstens
RME

3 (edited by Ulrich 2020-05-28 11:03:25)

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

MC wrote:

Out of interest: what does Latency Mon say on this machine?

"Your system seems to be having difficulties".

The bars still look green.

Wdf01000.sys, 640 µs
ACPI.sys, 1251 µs
Hard pagefault count: 754

When being in Playback mode in Sequoia, or in "Standard" recording mode (that means: with Playback engine running) I can monitor the Sequoia status bar. Even while playing an empty project (64 tracks) sometimes the "DSP Max" value suddenly jumps over 100%, and in that instant a lost ASIO buffer occurs.

Using MADIface USB and 1024 buffersize seems to be the safest option.
MADIface XT (via USB) still produces (only a few) lost buffers with the same settings.

I have the impression the hardware is not perfectly suitable. Do you have any recommendations for the choice of a laptop?

Thanks for caring,
Ulrich

4

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

No longer. Dell ruined my XPS7590 to the same, ACPI.sys, 1251 µs. Waiting long enough peaks of up to 6 ms show up. There is no tweak that can change this, only a new BIOS could do.

You remember the other thread where FFrank mentioned his Lenovo X Extreme gen2 also failing? No good times for us audio users....

Regards
Matthias Carstens
RME

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

Oh dear...

Should we try to connect the MADIface XT via thunderbolt? What kind of hardware adapters to we need?

6

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

If that computer has TB then it's worth a try, but most probably shows issues too. On these newer machines the Apple TB3 to 2 adapter usually works.

Regards
Matthias Carstens
RME

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

Hi Ulrich

Sorry to hear of your P1 troubles.  I was planning on replacing my venerable T430 with one soon.  Is the P1 running with the Nvidia card?  Do you have the option in UEFI to disable the Nvidia and use only the built-in Intel graphics?  As a fellow XT/Sequoia user I would try that if it's available.

With regard to getting the XT connected through TB, this below may work.  Note, I use the XT with Matrox external PCIe adapters very successfully, but directly into PCIe, not with Thunderbolt.

1. https://www.sonnettech.com/product/echo … rbolt.html
2. https://www.bhphotovideo.com/c/product/ … d_for.html
3. https://www.bhphotovideo.com/c/product/ … _MXO2.html  OR  https://www.bhphotovideo.com/c/product/ … _MXO2.html

Failing all that, you could look for a workstation notebook with an expresscard slot (probably an HP Zbook or Lenovo P51), and omit #1 from above.

Simon

Madiface XT, Madiface, 3x Micstasy, ADI8QS
Sequoia 15,  W10 x64, W7 x64

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

Actually, reading a bit more, the ACPI.sys thing seems like a real blocker.  Oh dear, I've been relying on Lenovo's for years without this kind of issue that has plagued Dell's.

Madiface XT, Madiface, 3x Micstasy, ADI8QS
Sequoia 15,  W10 x64, W7 x64

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

with regard to ACPI.sys, see also:
https://www.sweetwater.com/sweetcare/ar … cy-issues/

Madiface XT, Madiface, 3x Micstasy, ADI8QS
Sequoia 15,  W10 x64, W7 x64

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

This P1 has a so-called hybrid graphics solution with Intel and NVidia. It turned out switching off the Intel graphics and working only with NVidia gave the best results in Latency Mon. (Of course we also tried vice versa.) However, real-world audio capability didn't improve vastly. With high buffer settings the issues are quite rare but not completely absent. The ACPI issue doesn't disappear completely.

Keep the T430...!

11 (edited by Potscrubber 2020-06-03 22:27:55)

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

It would be worth posting about this in detail and explaining the issue/needs on the official Lenovo forums. I have seen staff respond there, they may fix it in a bios update.

T430 with an i7 still works very well as a bit acquisition bucket, but it is getting pretty creaky.  It’s frustrating we are still grappling with these issues (if not more so) in 2020.

Madiface XT, Madiface, 3x Micstasy, ADI8QS
Sequoia 15,  W10 x64, W7 x64

12 (edited by Potscrubber 2020-06-03 22:34:56)

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

Some more recent experiences here:
https://www.gearslutz.com/board/music-c … tency.html

Madiface XT, Madiface, 3x Micstasy, ADI8QS
Sequoia 15,  W10 x64, W7 x64

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

A bit more on this as I think it's an important issue for a lot of typical RME users.

Notebook Check reviews often include LatencyMon results, but not always and sometimes they are only run for 10 seconds or so.  As we know 5~10 minutes is a meaningful test, I've written to the editors to request they test LM consistently.

Some laptops that show up as having favorable DPC results include HP Zbooks in general (look for models without the U designation - U means a mobile CPU). 

Also the Asus Zenbook Flip 14 UX463FA   https://www.notebookcheck.net/Asus-ZenB … 996.0.html

Really is tough times for notebooks though.  Really sad about the P1.

Madiface XT, Madiface, 3x Micstasy, ADI8QS
Sequoia 15,  W10 x64, W7 x64

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

A little bit of success here.

With "everything" turned off in the BIOS, no Hyper threading, Intel graphics turned off (device manager), starting again from a fresh recovery image and much OS optimization by the dealer, all network disconnected, but Windows Firewall and Defender active, behaviour is now reliable with buffer size 1024 and Sequoia recording 64 or 128 tracks (MADIface USB or MADIface XT, mode: recording independant from playback). That's encouraging.

With buffer size 512 there are rare lost buffers (MADIface XT always a bit more than MADIface USB), buffer size 256 doesn't work at all for this kind of recording.

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

Are you sure your usb3 chipset is supported by the madiface xt? There is a compatibility list somewhere in the manual.
Georg

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

georgob wrote:

Are you sure your usb3 chipset is supported by the madiface xt? There is a compatibility list somewhere in the manual.
Georg

RME says: "Fully compatible to Intel's USB 3 implementation, which - on current Windows and Mac computers - is part of the chipset. USB 3 sockets that are connected via an internal cable (not directly soldered onto the motherboard) can cause transmission errors. These are shown in the Settings dialog."

It's Intel AFAIK, and I don't experience transmission errors.

Re: Lenovo P1 Gen2 needs buffersize 1024 for MADIface USB

For what it's worth i have a Lenovo X1 Extreme (Gen1) which is more or less the same as a P1 with a GTX instead of a Quadro - and i've had a lot of DPC issues as well (with a FireFace 400 UC).

It is my understanding, that Lenovo took a short-cut, and instead of implementing a MUX to switch graphics cards, they are copying data from GPU to GPU instead. The internal display (4k in my case) is hardwired to the Intel GPU and the HDMI port is hardwired to the NVidia GPU. So - this means a lot of additional PCIe traffic depending on your setup.

I've managed to get my laptop fairly stable through a lot of low-level tweaking, but i can't really point to any single fix that will do the trick - except running on the Intel GPU (without external displays) results in much more stable ASIO performance for me. At one point i had my mouse in one of the thunderbolt ports using a USB-C > USB 3 dongle, and this caused alot of low-level spiking - which went away as soon as i used the USB-A ports on the right side. This kind of BS really shouldn't be happening in 2020 ..

Rune Borup @ FishCorp
Producer / Engineer / Composer
RayDAT > 2 x ADI-8 QS | AES+SPDIF > ADI-2 Pro