It’s fair to dream. I’ve no doubt the CPU core count will increase somewhat in the next iteration. I wouldn’t be surprised to see more fixed function hardware in their next (theoretical) SoC.

I’m not sure if we’d ever really see that much gfx resource in a handheld, at least for now. I agree it would be very cool but vendors need to strike a very fine balance as far as power is concerned. Could go for the ‘dock to unlock’ approach though I genuinely appreciate that the steam deck’s performance characteristics are 1:1 plugged in and on battery. Besides that, area is expensive, and the steam deck came in at an extremely attractive price in 2022 relative to other x86_64 handhelds on the market. I would hope price remains a focus to get Linux gaming and desktop experiences into more hands.

As for halo in particular, significant improvements have been made at a packaging level to minimise idle draw with the mcm design (and I think that is somewhat reflected in current OEM offerings) but it’s still not quite where you’d want it to be in a handheld system. That’s not to say it won’t get there eventually.

The two gfx IPs were in direct competition for 2020, if anything, the two platforms should be on par in terms of base level capabilities. Interesting to see them go with a fullHD panel, though. 1080p@120hz bodes pretty well for general performance expectations. Curious what the battery life will be like.

strix halo in any gaming handheld would be bizarre, much as I’d love to see it. I get the feeling valve would hold out just a little while longer before pulling the trigger on a second gen.

aaaaaaaaaaaaa!

Absolutely no player moderation infrastructure whatsoever. It’s as if they never made a halo game before.

I recall some streamer lady getting relentlessly harrassed via lobby voice chat at the beginning of infinite’s life cycle. Zero recourse; there’s no in-game reporting function (the game directs you to the halo waypoint website, it’s a fully manual process, you supply the offending player’s name, you’re even expected to manually capture infractions via the (still) broken in-game theatre mode).

As for betrayal booting, I have a kind of roundabout theory. 343 in their infinite wisdom decided to disable player collision and friendly fire in the sandbox by default.

On one hand, this is behaviour in-line with contemporary shooters to prevent griefing. On the other hand, it’s entirely detrimental to sandbox immersion and can lead to bad habits when it comes to player positioning.

I suspect this change lead to the oversight of any player booting mechanism, though it’s still possible to team-kill via vehicle collisions.

There’s so much wasted potential in infinite, yet the core mechanics are pretty nailed down.

I’m astounded they released this multiplayer focused game with no betrayal booting or any form of in-game player reporting.

I don’t believe FPO was expanded to >2 displays, though having identical panels would eliminate VBI complexities. Will be interested in understanding how this behaves with rdna 4.

Huh interesting. Are these displays all the exact same model?

There’s a relatively new feature in the amd gfx display abstraction layer called freesync power optimisation. This feature leverages panel VRR to help mclk idle low at the desktop. This was introduced for single display use with RDNA 2, and expanded to dual display along with RDNA3s MALL advancements. I’m not sure if this is expanded further with RDNA 4 but I can try to find out.

This isn’t true for Vega 10 and 20 due to their use of HBM2. In general what you’re describing is a comparative weakness of GDDR as a technology. I don’t think there’s anything to suggest there’s an inherent issue with idle power on older gen asics at >60 Hz save from the typical limitations with VBI compatibility in an array of panels or display bandwidth thresholds. In the case of VBI compact issues, modifying EDIDs can indeed help.

That’s kind of curious. I don’t think 3D_FULLSCREEN should inherently determine idle mclk behaviour in and of itself. Is this just a single 1440p display at 120Hz? Is VRR enabled?

huh, I generally expect Vega10 based GPUs to idle at ~3W TGP (not inclusive of other board power losses like VRM). Can you tell us what you display setup is? Can you get a reading of your idle mclk using something like CoreCtrl?

video playback will likely kick the asic out of idle. What’s your power use at true desktop idle?

Generally speaking, vega 10 (56, 64) and 20 (radeon vii) are able to achieve decently low desktop idle with varied display configs due to the memory technology they employ.

Can you tell us which distro and display config this is with?

I’m also in 3D_FULLSCREEN on NV21XT but my idle mclk is 96 MHz. TGP (not to be confused with TBP which is only communicated on RDNA 3 and up) is 6 watts at idle with my browsers open. GNOME 47 + Wayland, 2560 x 1440 @ 180Hz + 1920 x 1080 @ 60Hz, VRR enabled on both displays. This is with Fedora 41, kernel 6.13.6-200.fc41.x86_64

For context, the GPU index (0,1,2 etc) will depend on the number of video adapters you have in the system. If you have a CPU with integrated graphics, there’s a chance this will be registered as 0000 whereas the dGPU will be 0001.

Oh I see, good to know

Radv is also a part of mesa, along with anv, nvk, turnip etc

I mean, it runs everything I need. But what is mainstream gaming to everyone else? Is it fortnite? Call of duty? Destiny 2? Pubg? Valorant? GTA? Battlefield? (weirdly a lot of shooters), Apex? Siege?

May not matter to people like us but they each command something to the effect of hundreds of thousands of concurrent players. Capable as Linux distros are for gaming (truly the best way to experience classic games) the anticheat situation is no less dire.

right, good to know. I was under the impression this was included with Mesa. Admittedly I’ve never provisioned a system on Linux without amd gfx from the get go. Do users switching from nvidia (for example) to amd gfx need to manually install radv on their systems?

Is this analogous to radv? If so, does it need to be manually installed?

Was this the one backed by a streamer?

Idle power is determined by your display setup. Is your friend a comparable arrangement to yours? Do you run a couple high res high refresh rate displays?

I’m a little surprised you’re reaching the same idle draw as vega10 given its use of HBM2. Also worth noting that everything prior to RDNA 3 showed TGP instead of TBP. This was particularly annoying as it didn’t account for other board power losses (like VRM), and didn’t give monitoring software an accurate read :/

For perf deficits in rtrt with heavier effects (reflections, GI, AO), you may find some luck in leveraging AMDVLK instead of RADV. I don’t doubt the latter will catch up in good time.