M1 Macs can now run Windows apps and games through Crossover 20

[handymanshandle]

Lmao at the Sydney bot joining dude's TF2 match

[Bombastinator]

20 hours ago, hishnash said:

Note quite, 

 

wine is not emulation:
*  it provides dynamic libs that exposed the windows apis to these apps the map these the unix system apis.
Rosetta2 is not emulation:
* it is a translation tools that convert x86-64 binaries into ARM64 binaries. 

Isnt wine an acronym that actually stands for 

Wine

Is 

Not 

Emulation?

I dunno. Maybe it’s just a joke.

[Spindel]

1 hour ago, Bombastinator said:

Isnt wine an acronym that actually stands for 

Wine

Is 

Not 

Emulation?

I dunno. Maybe it’s just a joke.

Emulation: Pretends to be a specific type of hardware in software.

Translation: Translates instructions from one hardware to an eqvivalent in another.

 

In theory:

- Emulation same output as on original hardware, but at a performance penalty because some resources has to be alocated to run the ”virtual hardware”.

- Translation almost the same output as running natively with no performance penalty.
The ”almost” is the important part because there might be cases where one type of instruction simply can’t be replaced by an equivalent, in those cases you might actually switch to emulation but only for those instructions, with the same rules as full eumulation but only on parts of the code.

 

[RedRound2]

20 hours ago, Arika S said:

people keep saying this like it's something amazing. even the "cheapest" model is still the price of a premium ultrabook, or a fairly decent gaming laptop (at least here in Australia)

Okay bud, but does those laptops have the same build quality, reliability, screen, support, etc? 

 

By cheapest in this context means the worst Apple Silicon processor in its worst configuration - a fanless design. An Intel fanless processor wouldn't stand a chance.

 

The cheapest Mac is the Mac mini for $699. For which you're getting a very capable machine able to run games, edit 4K HDR video with breeze with barely any fans ramping up and quite capable to run even Windows apps through multiple translation layers. And the same 15W chip is competing with much higher end chip, pretty much wrecking anything in the same class

 

I get that you guys just want to downplay anything postive or innovative Apple does. But for a community that gets hyped up about 10 FPS increase in new gen processor with exact same everything, it's a bit ironic

[RedRound2]

20 hours ago, Sauron said:

I don't think it's nearly as impressive as you're making it out to be. Of course it's usable, that was a given as soon as we found out that rosetta2 itself was "usable".

So does running all other layers minus Rosetta have a 100 percent performance mapping? And there's Zero CPU overhead when doing multiple translations? Is that what you're implying, because you're just saying that the whole thing only depended on Rosetta 2.

 

I would like to see a fanless Intel/AMD laptop running a game such as TF2 through Wine on Linux. Let's just forget about the 32 to 64, Rosetta 2, and Vulkan to Metal translations. If it performs as well as it does running natively on Windows, then alright maybe it wasn't a big deal

[Dabombinable]

20 minutes ago, RedRound2 said:

So does running all other layers minus Rosetta have a 100 percent performance mapping? And there's Zero CPU overhead when doing multiple translations? Is that what you're implying, because you're just saying that the whole thing only depended on Rosetta 2.

 

I would like to see a fanless Intel/AMD laptop running a game such as TF2 through Wine on Linux. Let's just forget about the 32 to 64, Rosetta 2, and Vulkan to Metal translations. If it performs as well as it does running natively on Windows, then alright maybe it wasn't a big deal

Problem is, AMD's 4000 series APU (which destroy Intel's price equivalents in performance and efficiency) are in short supply, which means that we may never actually see a comparison like that. In theory AMD would be on par however.

[Sauron]

1 hour ago, RedRound2 said:

So does running all other layers minus Rosetta have a 100 percent performance mapping? And there's Zero CPU overhead when doing multiple translations? Is that what you're implying, because you're just saying that the whole thing only depended on Rosetta 2.

Let me run you through it.

 

On an x86 CPU crossover offers a certain performance level with a given program. This is often close to 100% because wine translation isn't very costly, it's usually more a matter of whether the program runs properly rather than quickly - but let's say it's 90% in any given application just for simplicity's sake.

 

That means we expect 90% of the performance we'd expect in a macos native program running through rosetta2. rosetta2 itself provides about 80% of native arm performance in native macos applications. So we expect 90% of 80% of the performance, making it 72% of native performance. And you're saying that you don't expect a cpu to run a relatively lightweight program just fine at 72% of its theoretical maximum performance when we know it can do so at 80%?

[AndreiArgeanu]

That's cool and all, but TF2 isn't that hard to run. Frankly enough I think I could get it to run better on the cheap windows tablet I have with an Intel atom and 2gb of ram. We'll have to wait and see if support improves for x86 applications.

[suicidalfranco]

1 hour ago, RedRound2 said:

I would like to see a fanless Intel/AMD laptop running a game such as TF2 through Wine on Linux. Let's just forget about the 32 to 64, Rosetta 2, and Vulkan to Metal translations. If it performs as well as it does running natively on Windows, then alright maybe it wasn't a big deal

It's not under linux and Wine, but pretty much what AMD has been showcasing with renoir

[RedRound2]

1 hour ago, Sauron said:

Let me run you through it.

 

On an x86 CPU crossover offers a certain performance level with a given program. This is often close to 100% because wine translation isn't very costly, it's usually more a matter of whether the program runs properly rather than quickly - but let's say it's 90% in any given application just for simplicity's sake.

 

That means we expect 90% of the performance we'd expect in a macos native program running through rosetta2. rosetta2 itself provides about 80% of native arm performance in native macos applications. So we expect 90% of 80% of the performance, making it 72% of native performance. And you're saying that you don't expect a cpu to run a relatively lightweight program just fine at 72% of its theoretical maximum performance when we know it can do so at 80%?

But remember this is a game. Not your typical word or excel. A game requires second to second real time processing. Wine isn't very expensive and games do run on wine, I agree. But there's multiple things going on here, and the game is still playable. And how is a game relatively lightweight? Running Office or something would be relatively lightweight. 

 

The developer even said that he couldn't get his co worker to stop playing Witcher 3 on it.

https://www.codeweavers.com/blog/jwhite/2020/11/18/okay-im-on-the-bandwagon-apple-silicon-is-officially-cool

So it is cool, how the worst chip in worst config in Apple's lineup has such a high minimum base, no matter how much you'd like to keep saying this isn't a big deal

1 hour ago, AndreiArgeanu said:

That's cool and all, but TF2 isn't that hard to run. Frankly enough I think I could get it to run better on the cheap windows tablet I have with an Intel atom and 2gb of ram. We'll have to wait and see if support improves for x86 applications.

Mind you, this is running through multiple translations, not native. Plus the devs were also able to get Witcher 3 working on the M1 Mac.

[RedRound2]

1 hour ago, suicidalfranco said:

It's not under linux and Wine, but pretty much what AMD has been showcasing with renoir

That's pretty cool. Hopefully it actually makes it's way into the public and we can see a head to head comparison in the real world

[leadeater]

1 hour ago, suicidalfranco said:

It's not under linux and Wine, but pretty much what AMD has been showcasing with renoir

That's not just fanless, that's bare die? Hurts to watch lol

[leadeater]

6 minutes ago, RedRound2 said:

That's pretty cool. Hopefully it actually makes it's way into the public and we can see a head to head comparison in the real world

They do exist, Renior is Ryzen Mobile 4000 series. You do realize that is an actual laptop someone has taken apart to show you this?

 

Edit:

Also the extra part about WINE and Linux, that's irrelevant. WINE and Linux running Windows applications and games has been around for more than a decade, makes no difference to the CPU at all. If it doesn't work it's an issue with WINE not a CPU performance problem.

[Sauron]

12 minutes ago, RedRound2 said:

But remember this is a game. Not your typical word or excel. A game requires second to second real time processing. Wine isn't very expensive and games do run on wine, I agree. But there's multiple things going on here, and the game is still playable. And how is a game relatively lightweight? Running Office or something would be relatively lightweight. 

For the CPU, running a game is relatively light weight. The load is more on the GPU which isn't going though any binary translation.

[RedRound2]

11 minutes ago, leadeater said:

They do exist, Renior is Ryzen Mobile 400 series. You do realize that is an actual laptop someone has taken apart to show you this?

 

Edit:

Also the extra part about WINE and Linux, that's irrelevant. WINE and Linux running Windows applications and games has been around for more than a decade, makes no difference to the CPU at all. If it doesn't work it's an issue with WINE not a CPU performance problem.

Oh I assumed it was the upcoming 5000 series. Then we already have the numbers

https://www.anandtech.com/show/16252/mac-mini-apple-m1-tested/5

Also mind you, in the video, that die is exposed to air - which is definitely not representative of a real world usage where the CPU die is usually suffocated with stagnant air in a finless design

 

And I saw this on another thread, quite impressive. Not really sure what people want more here. People who actually have the new Macs (apart from Linus - since he can't accept that he was wrong) are getting blown away by the performance

 

[leadeater]

2 minutes ago, RedRound2 said:

Also mind you, in the video, that die is exposed to air - which is definitely not representative of a real world usage where the CPU die is usually suffocated with stagnant air in a finless design

Bare die is much much worse than having any kind of heatsink at all, regardless of lack of airflow. Bare die triggers on die Tjmax sensors in time frames milliseconds or less. The whole point of the video is to show how good the thermal control on the CPU is, usually such a test is CPU suicide, never do this, ever. It's actually dangerous and high chance of hardware failure.