Turns out the M1 pro gpu (16 core) is faster than my rtx 3070 laptop when using a metal-optimized game

[hefeglass]

I recently noticed that one of the FPV drone simulator games I play (Liftoff) has the option to use "metal rendering". I wanted to see how the m1 pro 16 core GPU in the 16" base model macbook pro compares to my gaming laptop, a Zephyrus G15 with RTX 3070 and ryzen 9 5900hs. I was suprised to see that the m1 pro consistently had a higher FPS throughout the run (hard to get a direct comparison since the runs arent identical..but its decent). The zephyrus was on Turbo mode and plugged in, you can hear the fans going pretty hard in the short video. Both machines had the same game settings except for 2560x1440 resolution on the g15 and 2560x1600 resolution on the m1 pro macbook.
Too bad there arent more metal-enabled games to compare..

(edit: I just realized I was in low-power mode on the mac when I did this test..apparently there is a separate low-power mode setting for battery and power adapter. I will have to test again and report back)
(edit #2 : its a HUGE difference..I have to record a quick video, at least 10fps faster now, maybe more)

Another game that has metal support on MacOS is Dirt Rally (still not native and runs through rosetta)
This time the m1 pro embarassed the 3070, even though the 3070 is pulling almost 90w by itself.

RTX 3070 - 76 fps average
M1 Pro - 96 fps average

 

[bsamcash]

Exciting time to be a Mac user.

 

[leman]

I would love someone do a side-by side comparison using the new Baldur's Gates 3 M1 patch (just out today).

 

[turbineseaplane]

"...when using a metal-optimized game"

That's sort of a big caveat unfortunately

 

[Pressure]

I would love someone do a side-by side comparison using the new Baldur's Gates 3 M1 patch (just out today).

You have been able to expose Metal rendering in the config file for some time, although it were in beta.

 

[Kpjoslee]

Well, even M1 pro GPU should pack many more transistors than mobile RTX 3070 ( 33B with majority dedicated to GPU vs 17.4B with 5/6 of shaders enabled) so technically it should hang with mobile 3070 at least on pure rasterization performance.

 

[howlingsun]

I would love someone do a side-by side comparison using the new Baldur's Gates 3 M1 patch (just out today).

I am also really interested in how the new Pro and Max runs the game. I am sure they would do a good job.

It feels like the M1 patch made the performance go down, this patch added some graphics improvements. Its no longer playable on my M1 Mini 16GB. I get 25 fps at medium and 1080p at witch point the game looks terrible. To get it playable (30+) I need to set it to low and then it looks god awful.

 

[aeronatis]

I would love someone do a side-by side comparison using the new Baldur's Gates 3 M1 patch (just out today).

I tried it on MSI Creator Z16 (i7-11800H, 32 GB, RTX3060 65-watt) and also on my M1 Max MacBook Pro; however, on the Mac, game only launched via the external launcher on the local files of the game directory, thus, showing me no FPS count. When I try to open directly on Steam, the game crashes ?

Creator Z16 runs the game at around 40-44 FPS when graphics set to High and 33-35 FPS when set to ultra @ 1920x1200.

I am quite sure M1 Max is running at much higher than that even though I couldn't measure.

 

[leman]

You have been able to expose Metal rendering in the config file for some time, although it were in beta.

BG3 was always using Metal (as are the majority of modern games), but their Apple Silicon version for the latest update was not ready until today. What is special about BG3 however is that it is optimized for Apple Silicon, making it pretty much the first AAA game to do so.

Well, even M1 pro GPU should pack many more transistors than mobile RTX 3070 ( 33B with majority dedicated to GPU vs 17.4B with 5/6 of shaders enabled) so technically it should hang with mobile 3070 at least on pure rasterization performance.

I don’t think that number of transistors is a good predictor of rasterization performance.

 

[JMacHack]

Well, even M1 pro GPU should pack many more transistors than mobile RTX 3070 ( 33B with majority dedicated to GPU vs 17.4B with 5/6 of shaders enabled) so technically it should hang with mobile 3070 at least on pure rasterization performance.

I was waiting for the “performance per transistor” argument. Never disappoint me Macrumors.

 

[Kpjoslee]

I don’t think that number of transistors is a good predictor of rasterization performance.

Not very scientific I know. But not too far off method in terms of predicting where it should fall in terms of performance against AMD/Nvidia counterparts with clockspeed considered.

 

[Kpjoslee]

I was waiting for the “performance per transistor” argument. Never disappoint me Macrumors.

I wasn't really downplaying M1 Pro's performance so why you sound so triggered?
If you look at the transistor count * clockspeed vs competing Nvidia parts, their performance falls within the expected range vs similar RTX parts if properly supported with Metal drivers.

 

[dgdosen]

wrong thread... deleted

 

[hefeglass]

so..I just realized my macbook was in low power mode for this test. Apparently there is a separate low-power mode setting for battery and power adapter (not sure why youd want to use low power mode on power adapter, but they made it an option)

edit: its a HUGE difference..at least 10fps faster now, maybe more

 

[pi=e=3]

so..I just realized my macbook was in low power mode for this test. Apparently there is a separate low-power mode setting for battery and power adapter (not sure why youd want to use low power mode on power adapter, but they made it an option)

edit: its a HUGE difference..at least 10fps faster now, maybe more

If I had to guess, you'd use low power mode on a power adapter if you only had access to a lesser adapter, like a 30W charger.

That way your draw wouldn't exceed its capability.

 

[Pressure]

BG3 was always using Metal (as are the majority of modern games), but their Apple Silicon version for the latest update was not ready until today. What is special about BG3 however is that it is optimized for Apple Silicon, making it pretty much the first AAA game to do so.

I was thinking of League of Legends for some strange reason. You can enable metal there by changing the config file.

 

[Analog Kid]

If I had to guess, you'd use low power mode on a power adapter if you only had access to a lesser adapter, like a 30W charger.

That way your draw wouldn't exceed its capability.

Or if you're super sensitive to fan noise-- maybe for audio work...

(BTW, I found myself super triggered by your user name... ? Well done.)

 

[mi7chy]

Something wrong with that laptop or user error. I'm getting ~90 fps average, 110 fps high and 70 fps low with 70W 3060 at 1440p fantastic quality and no vsync.

 

[ikir]

Something wrong with that laptop or user error. I'm getting ~90 fps average, 110 fps high and 70 fps low with 70W 3060 at 1440p fantastic quality and no vsync.

70W is already double the power draw for entire MBP, just to say. Maybe the test was run with v-sync, resolution on MacBook Pro is higher and it was on low power mode. just to be clear.

 

[crazy dave]

so..I just realized my macbook was in low power mode for this test. Apparently there is a separate low-power mode setting for battery and power adapter (not sure why youd want to use low power mode on power adapter, but they made it an option)

edit: its a HUGE difference..at least 10fps faster now, maybe more

So I don’t have one to test but how I saw it explained in a review was that: yes, if you plugged in an M1 laptop with low power mode on, then low power mode would still be “on” but effectively ignored. That way when you came back on battery, low power mode would still be the default but it wouldn’t affect performance when it wasn’t needed. This would contradict that.

 

[ikir]

I recently noticed that one of the FPV drone simulator games I play (Liftoff) has the option to use "metal rendering". I wanted to see how the m1 pro 16 core GPU in the 16" base model macbook pro compares to my gaming laptop, a Zephyrus G15 with RTX 3070 and ryzen 9 5900hs. I was suprised to see that the m1 pro consistently had a higher FPS throughout the run (hard to get a direct comparison since the runs arent identical..but its decent). The zephyrus was on Turbo mode and plugged in, you can hear the fans going pretty hard in the short video. Both machines had the same game settings except for 2560x1440 resolution on the g15 and 2560x1600 resolution on the m1 pro macbook.
Too bad there arent more metal-enabled games to compare..

(edit: I just realized I was in low-power mode on the mac when I did this test..apparently there is a separate low-power mode setting for battery and power adapter. I will have to test again and report back)
(edit #2 : its a HUGE difference..I have to record a quick video, at least 10fps faster now, maybe more)

I suppose you need to update your signature

 

[hefeglass]

So I don’t have one to test but how I saw it explained in a review was that: yes, if you plugged in an M1 laptop with low power mode on, then low power mode would still be “on” but effectively ignored. That way when you came back on battery, low power mode would still be the default but it wouldn’t affect performance when it wasn’t needed. This would contradict that.

there is a separate low power mode check box on both the battery tab and the power adapter tab though
I guess its just so the fans dont spin up when you need complete silence (now that I actually read what it says under the checkbox ..yes its for quieter use and does reduce energy usage even when plugged in)

 

[hefeglass]

Im not sure how you can possibly test this ..unless you are using the same input type (transmitter) and flying with the same camera angle and speed of the drone there is no way to compare framerates (even I said its not a direct comparison and im doing everything as close to the same as possible). And as you can see in my video, when there is little to no movement (start and end) the 3070 goes well over 100 fps, but as soon as you start to fly it drops significantly.

On top of everything, the m1 pro actually performs better than the video showed because low power mode was enabled.

Please post a video of you flying through this level with the FPS displayed. I dont doubt your claim, id like to see how you tested the game though. Maybe you fly fpv too and have all the same stuff I do, in which case its a great comparison.

 

[AppleFanBoy888]

Amazing. Just wish more games would support Metal. Sigh. I guess gaming on macOS isn't that popular enough for game developers to justify. I was using Boot Camp to game before my M1 Max and I'm missing it. Tried Parallels but it is still not as responsive as Boot Camp. Bought a Dell XPS 8940 just for gaming.

I suppose you need to update your signature

Haha. ? That's the first thing I noticed actually. I was gonna PM OP about it but never mind.

 

[mi7chy]

Typical response when misleading info is exposed. I've also exposed fake Blender tests where they ended up editing out the Blender comparisons but my posts still have the screen shots for evidence. When I get my 16" M1 Max I'll perform my own comparisons and not have to rely on fakes.

https://forums.macrumors.com/threads/3d-rendering-on-apple-silicon-cpu-gpu.2269416/post-30540840