It would be great if we could get a steamdeck that runs one of these arm chips.
I wonder if valve is already experimenting with something like this. Maybe another they will have something like box64 too
this post was submitted on 24 Jun 2024
162 points (98.2% liked)
Linux
48074 readers
792 users here now
From Wikipedia, the free encyclopedia
Linux is a family of open source Unix-like operating systems based on the Linux kernel, an operating system kernel first released on September 17, 1991 by Linus Torvalds. Linux is typically packaged in a Linux distribution (or distro for short).
Distributions include the Linux kernel and supporting system software and libraries, many of which are provided by the GNU Project. Many Linux distributions use the word "Linux" in their name, but the Free Software Foundation uses the name GNU/Linux to emphasize the importance of GNU software, causing some controversy.
Rules
- Posts must be relevant to operating systems running the Linux kernel. GNU/Linux or otherwise.
- No misinformation
- No NSFW content
- No hate speech, bigotry, etc
Related Communities
Community icon by Alpár-Etele Méder, licensed under CC BY 3.0
founded 5 years ago
MODERATORS
Not sure why you'd want an ARM-based handheld to play PC games at this point in time. Pretty much all PC games are available in x86 only, and any efficiency gains these fancy new ARM chips supposedly have will be lost when translating x86 to ARM.
and any efficiency gains these fancy new ARM chips supposedly have will be lost when translating x86 to ARM.
Not a given. Translating can still be more efficient.
If both AMD/Intel and Qualcomm do a good job with their core design and the same process node is used, I don't see how a translation layer can be any faster than a CPU natively supporting the architecture. Any efficiency advantages ARM supposedly has over x86 architecturally will vanish in such a scenario.
I actually think the efficiency of these new Snapdragon chips is a bit overhyped, especially under sustained load scenarios (like gaming). Efficiency cores won't do much for gaming, and their iGPU doesn't seem like anything special.
We need a lot more testing with proper test setups. Currently, reviewers mostly test these chips and compare them against other chips in completely different devices with a different thermal solution and at different levels of power draw (TDP won't help you much as it basically never matches actual power draw). Keep in mind the Snapdragon X Elite can be configured for up to "80W TDP".
Burst performance from a Cinebench run doesn't tell the real story and comparing runtimes for watching YouTube videos on supposedly similar laptops doesn't even come close to representing battery life in a gaming scenario.
Give it a few years/generations and then maybe, but currently I'm pretty sure the 7840U comfortably stomps the X Elite in gaming scenarios with both being configured to a similar level of actual power draw. And the 7840U/8840U is AMD's outgoing generation, their new (horribly named) chips should improve performance/watt by quite a bit.
Not what i am saying. I said that it is not a given, that translation means less performance.
In theory you can achieve similar or even higher performance, all depending on how well or how bad the original machine code is. Especially when you can optimize it for a specific architecture or even a specific CPU.
And yes ARM has shown to be more power efficient then x86 CPUs even on higher load (not just low powered embedded stuff).
Wine/Proton on Linux occasionally beats Windows on the same hardware in gaming, because there's inefficiencies in the original environment which isn't getting replicated unnecessarily.
It's not quite the same with CPU instruction translation, but the main efficiency gain from ARM is being designed to idle everything it can idle while this hasn't been a design goal of x86 for ages. A substantial factor to efficiency is figuring out what you don't have to do, and ARM is better suited for that.
while this hasn't been a design goal of x86 for ages.
It has been since P4
As you said yourself, it's not the same thing. Proton can occasionally beat Windows because Vulkan might be more efficient doing certain things compared to DirectX (same with other APIs getting translated to other API calls). This is all way more abstract compared to CPU instruction sets.
If Qualcomm actually managed to somehow accurately (!) run x86 code faster on their ARM hardware compared to native x86 CPUs on the same process node and around the same release date, it would mean they are insanely far ahead (or, depending on how you look at it, Intel/AMD insanely far behind).
And as I said, any efficiency gains in idle won't matter for gaming scenarios, as neither the CPU nor the GPU idle at any point during gameplay.
With all that being said: I think Qualcomm did a great job and ARM on laptops (outside of Apple) might finally be here to stay. But they won't replace x86 laptops anytime soon, and it'll take even longer to make a dent in the PC gaming market because DIY suddenly becomes very relevant. So I don't think ("PC") gaming handhelds should move to ARM anytime soon.
I think that's what we see with apple silicon, right?
Ehhh, kinda. Intel E-cores kinda throw off the balance a bit, but generally yeah.
It's not that uncommon in specialty hardware with CPU instructions extensions for a different architecture made available specifically for translation. Some stuff can be quite efficiently translated on a normal CPU of a different architecture, some stuff needs hardware acceleration. I think Microsoft has done this on some Surface devices.
Translating can still be more efficient.
You would need some ISA that greatly benefits from translating. Like ELBRUS.
load more comments
(4 replies)
As the article says, there is no graphics driver yet, so nobody is experimenting with these chips in the gaming world yet in that sense 😉
Maybe somebody is prototyping a Windows platform in the meantime, and I haven't seen the benchmarks, but I would be surprised if these chips could outperform AMD's similar APU packages.
Isn't it using adreno GPU? Freedreno exists for a long time.
Or maybe they will compile natively for arm.
It would be viable with foss games
most games are not foss, and it makes sense.
(games are more like works of art ranter than software after all, so it just doesn't matter)
My games are
Foss gaming is a thing
yes, but most games aren't.
there's no reason to avoid good indie games just because they're not foss, unless you're a toxic fossbro or something
When did Qualcomm start giving a shit about Linux? They've been on my "hardware and chipsets to avoid if possible" list for pretty much ever.
Since they started targeting the PC segment with these chips to take on Apple's insanely priced m-class chips, and Amazon and Google's custom ARM datacenter chips.
They partnered with Canonical to do the first run of development for kernel support in the past year, and now it sounds like they're moving to get the graphics driver developed and upstreamed.
Graphics driver for sc8280xp are already a thing. There are more issues in convenience daily driving linux, currently. From the top of my head:
- firmware update path
- dtb update/loading path
- no virtualization
- no universal dock compability
- missing HDMI/DP features
I suspect that these issues are common between their ARM chips and will be addressed for both chips almost simultaneously. But I have no real idea on kernel development. And their documentation is only shared with linaro so one can only guess.
always? Android runs a linux kernel, and they support all kinds of embedded systems that run Linux.
Until recently, that "support" had been a barely supported forks of the linux kernel that were barely updated, and was so locked down that custom rom support was a pipedream on snapdragon processors. Which to be fair, is par for the course on most ARM chipsets (It's the reason you see a lot of custom roms for android have extremely old and outdated kernels)
I'm glad to see more ARM companies moving towards working with upstream projects, and not just making working on their stuff a PITA to protect "Trade Secrets" or some bullshit like that.
I'm sorry for leaving out the word "desktop". I'm well aware that Android runs the Linux kernel and that many embedded systems run Linux.
Possibly I conflated them with Broadcom, but I feel sure I recall Qualcomm's lack of openness being problematic in the past also.
Edit - yeah, folks jumping through hoops for their wifi at least as recently as Ubuntu 20.04. https://askubuntu.com/questions/1277359/my-qualcomm-atheros-qca9377-wireless-adopter-is-not-working-in-ubuntu-20-04-lts
ah yeah. maybe less well known, but i had a dev kit from Qualcomm that came with Ubuntu
You are very wrong here. They open-source a lot of things and they even used to have their own open-source modified version of Android for their phone chips.
OK, correction accepted. I probably did conflate them with Broadcom. Someone should let those ubuntu folks know though.. ;)
Oh it's ok. Broadcom is a very bad company in terms of open-source and Linux support. Their most known products are WiFi modules for laptops. Qualcomm on the other hand is probably one of the most open-source friendly commercial companies and it's known for very popular mobile processors such as the Snapdragon series.
I wouldn't call Qualcomm great for foss. It just better than absolutely terrible. Also Broadcom is a terrible company all around. They buy others and then wring them dry.
If the X Elite mainline kernel support pans out, Qualcomm may become top tier in terms of support. It would certainly make them the most important Linux ARM chip. We will see.
You mean like what they're doing to VMware and canning perpetual licenses the second they took over? I guess in some ways they are actually great for FOSS, because I've never seen more interest by Enterprise in Proxmox before they made that decision.
They are shit, but not as shit as Broadcom. The problem with Qcm is their monopolistic behavior and closing details on RF part of chips.
Are there Benchmarks for the CPU yet? Still can't tell if the claims they made are for real or are just marketing bs.
Even if they're marketing BS, some decent mainline support for Qualcomm chips is always welcome. It'll help projects like postmarketOS and mobile Linux distros massively to have more usable Qualcomm code.
It probably wouldn't help phones but I wouldn't complain if it did
There are quite a few YouTubers with press units making benchmark comparisons to M2 and M3 Macs. Overall, it stands up pretty well.
Can you give me some links please?
Are you aware that YouTube has a search function?
Oh no never heard of that is that a new feature? How dose it work?
A couple years ago YouTube decided to F up their search. It used to be mostly things you are searching for, now it's:
20% thing you searched for,
20% Shorts,
10% people also watched,
10% related [extremely tangentially]
10% For You
and 30% ads.
I don't blame anyone for wanting someone to suggest a link after YT's search became hot garbage.
They haven't widely released the chip yet, but they made a lot of public claims during trade shows, which I'm sure you can find.
Yeah, that would be the marketting bs, probably.
That's not true at all friend, they released this week: https://www.tomsguide.com/computing/laptops/copilot-pcs-are-here-11-snapdragon-x-elite-laptops-you-can-buy-right-now
From my small experience with Qualcomm in the past, I'm not too hopeful. In a company I used to work for, we wanted to use one of their SoC with Linux, which they claimed they supported. It was many years ago. But was full of closed binary blobs which even when signing NDAs, we couldn't get the source for. We're talking user-space drivers, sensors offloaded to a separate core with closed source firmware etc. It's Linux, but it's not Linux in spirit, it feels so closed and proprietary and secretive. They're coming from Android, which google architecturally enabled vendors to close their drivers by utilizing HAL. It's the single most significant blow to Linux by any corporation so far. It enabled thousands of vendors to close their shitty driver in user-space and not maintain it for newer kernels (kernel driver is just an IO proxy for user-space drivers). I get that without it, there wouldn't be Android phones we have today, but I expected them to slowly open up. 10+ years later, almost nothing changed, in fact - things seem worse to me.
view more: next ›