Return to

A little teaser of what is to come :)


Yes, but look at the requirements, the application must be signed with a certificate specifically for this purpose.


If you have to email a sales guy from Microsoft, you aint getting it. haha


When will this be open for public testing. Im really excited for this to become a reality.


Go up and read, this has been answered many times now.


I think they’ve increased the 75W power budget and that that’s the reason why some boards have an additional motherboard power connector closer at the bottom of the board where the slots are.

I wasn’t aware that a separate controller was needed, and was under the impression that it’s any PCIe slot/card that’s expected to support this functionality. Various places on the internet talk about 2 pins on opposite sides of the card connector having to be short circuited and constructed longer/shorter to detect hot plugging…

@marcan42 on Twitter might know (he’s been running PCIe over serial and FPGAs for his porting Linux to PS4 efforts at some point), I’ll ask.


Doesn’t mean it conforms, it isn’t about what connects first so much as the inrush current. If you have some capacitors on there, the inrush can be 100’s of amps for a split second to charge them, which is enough to disrupt things in the system.


Didn’t I read a story recently about DRM on DRM and the game suffering :). It is pointless when its hacked in days and the pirates get better gaming.


Though if I’m not mistaken, some of that can be mitigated with some cheating in the bios with aspm (should the board manufacturer actually patch it in) and power states to limit the possible current, though definitely not as good as a hardware solution for sure. Thank god for pcie gen4 actually standardizing the power design.


Exciting news!!! I just figured out how to get the sync even closer, this is nuts!

By adopting a hybrid method of frame sync with the guest (calculated sleep delay + polling with wait on event if polling takes too long) and a call to a method normally avoided in OpenGL apps (glFinish) I am able to both ensure that the frame is delivered to the GPU on time, retain low CPU usage and obtain a maximum of 200us latency WITH vsync.

That is not a mistake, MICROSECOND latency!

Btw, figured out a way to calculate the latency :smiley:

Edit: With vsync disabled this drops to a maximum of 40us.

EDIT3: I think I was a little premature in this announcement… there is certainly an improvement but there might be something off in this new code


I think I broke my jaw it hit the floor so hard…:rofl:


Now I feel sorry for going off on a tangent on this thread…, maybe a mod can be more effective in forking the thread into a separate one about PCIe hotplugging.

I think that’s what PRSNT#1 and PRSNT#2 are meant to help with, that is to help electrically isolate the card until it’s actually sitting in the slot, but you’re right, should work != does work. And the fact there’s ways to do slow start, doesn’t mean cards or motherboards actually have all the circuitry on them.

I wonder if saphire or msi can confirm/deny doing any even if minimal testing of these scenarios, regardless of whether it is or isn’t a supported use case for the products in an actual production device, just out of curiosity.

Also, pcie slots provide both 12V and 3v3 on various pins, 12V is a lot in a box full of semiconductors, even if you could switch those powerlines on/off with software and that part is working, or if you had a physical clicky power switch that would disconnect those pins, it doesn’t mean that you will always be able to do so, and doesn’t mean that something won’t die if you try to unplug / plug back in a card in a live system… it’s definitely not a typical use case, lots can go wrong.


And here’s the reply:

So I’m guessing there’s at least some boards where this works.


Just run with auto login. You are probably only using it for gaming anyway. nothing important


I don’t know if that has been asked, but will this work with linux guests too?


I already asked him about it

TL:DR - Windows is the primary focus of the project, but it is possible for someone else to add Linux support.


macOS support will be the real challenge because of OSX’s gatekeeper security. And that would really help people still using Adobe apps on OSX (cause that’s the ONLY proper way to render ProRes)


I thought Clover completely bypasses the whole gatekeeper security.

I actually would love to see this support a MacOS guest, but I feel like someone else would have to implement this.


there are legal implications for this too. the only legal way to run a mac os VM in on mac hardware that runs mac os


You are right, running a Hackintosh is in the gray area. However, I don’t think that would stop someone from creating an addon for macOS.

The worst I have seen from Apple in terms of lawsuit, was the Apple V. Psystar case. They, understandably, don’t like it when people sell hardware that runs macOS on non-apple hardware.

However, I haven’t seen Apple try to shut down Clover or Chameleon. Apple has sent cease and desist to various websites before (Source 1|Source 2) but I haven’t seen them do anything recently.

So if someone were to actually implement a driver for macOS, I doubt they would actually get in trouble with Apple.


Any OpenGL developers here willing to have a crack at this problem?