Hacker News Reader

> Companies almost always develop to the most popular console, and put in the barest minimum of effort to port it to PC and elsewhere.

I am still not over those statements from Microsoft from the first Xbox era and Fable and how it would speed up dev. and how we would get games on both platforms at the same time and then how they didn't release Fable 2 on PC even though they said DirectX blah blah. /rant

>This kind of shifts the price / perf curve way out on PC, and a lot of people are unhappy about it, but publishers don't really care anymore.

Disagree. Publishers care about PC today more than 15-20 years ago.

Back then, there was real risk of the entire PC gaming market disappearing with the exception of FPS, real-time strategy, and 4X. Once FPS (especially online multiplayer) became viable on consoles, and RTS and 4X's shares of the overall market shrank bigtime, the PC market was saved by MMORPGs, then Steam, then indie games, and so on.

Today you have Microsoft finally taking seriously its vow from way back when (as johnchristopher noted) to meld PC and console development and publishing, Sony unexpectedly making a PC game push, and the rising tide from Steam Deck becoming a viable platform benefiting both the mainstream PC and Linux gaming markets.

Consoles remain the largest category as a market, but the PC market compares well with any single console. There's a reason why Madden NFL returned to PC after a decade away.

Considering how much money people drop on gaming PCs these days, why is that sorta hardware advantage not available on PC as well? You'd think there would be a market for it, if offers such an advantage.

People may drop more money on PC than the average console person drops on their co sole, but the number of people doing this to their PC is vanishingly small. So per capita it’s better, but in aggregate spending a bunch of dev time on such a small market isn’t smart.

Put succinctly: even if this hardware advantage were available on PC, game devs wouldn’t optimize for it.

As I understand it, you can only get that advantage by having an SoC, where the CPU, RAM and GPU are all one unit (like Apple's M1 chip). If you want a system where to upgrade the RAM, you have to upgrade the whole system - well that already exists, it's called consoles.

Do you think Apple is working on finding a way to be easier to develop for because of their SOC strategy? Like, hey we are just like a game console, easier than PC!

The RAM being "on unit" has no impact here. It just requires the CPU & GPU to share an IOMMU & share the memory bus. Whether or not that RAM is soldered or not is totally irrelevant.

Laptops have had unified memory for a ~decade as a result, but once you go to a discreet GPU instead of integrated then there's no feasible way to share the IOMMU & memory bus, at which point it's not practical. And then up-gradable graphics or just 400W GPUs like the 3090/4090 behemoths are incompatible without extreme cost.

> The RAM being "on unit" has no impact here. It just requires the CPU & GPU to share an IOMMU & share the memory bus. Whether or not that RAM is soldered or not is totally irrelevant.

Sure, but to get the same memory bandwidth as a M1 Max you'd need 8-channel DDR5 memory, so your socket would look like Epyc and consume 50W+ just for data movement.

GDDR allows moving much more data than regular DDR (or even LPDDR) but both GDDR and LPDDR5X require much tighter signaling requirements that preclude socketed memory. And that means much higher watts-per-byte-transferred for socketed memory, in practice. Like dozens and dozens of watts higher for a typical configuration.

> Laptops have had unified memory for a ~decade as a result

There's unified memory, and there's unified memory.

Fusion HSA and Intel iGPUs have always required you to designate memory as either graphics or CPU side so that it knows which cache controller to run it through, and this has implications for memory and cache visibility.

ctrl-f 'garlic' and 'onion': https://www.eurogamer.net/digitalfoundry-how-the-crew-was-po...

https://forum.beyond3d.com/threads/amd-kaveri-apu-features-t...

Money shot: "One issue we had was that we had some of our shaders allocated in Garlic but the constant writing code actually had to read something from the shaders to understand what it was meant to be writing - and because that was in Garlic memory, that was a very slow read because it's not going through the CPU caches. That was one issue we had to sort out early on, making sure that everything is split into the correct memory regions otherwise that can really slow you down."

In contrast, the newer stuff like PS5/Xbox Series family and Apple Silicon family has true zero-copy where it's all just flat memory and all writes are immediately visible to all clients of the memory controller. You can interleave GPU and CPU and Neural Engine tasks without forcing explicit sync into memory to flush the cache/bus. I think most people would say that's meaningfully different.

Both are "unified" but one is more unified than the other.

Mostly because PCs are modular on axes that aren’t that useful today but also can’t easily be changed.

You can’t have the same memory with the same latency used for both CPUs and GPUs because GPUs don’t have DIMM and CPUs do, for example.

You also can’t guarantee a fast enough SSD to be able to delay loading assets until they are needed.

It only works because on modern consoles the RAM is soldered directly on the board and so the data doesn’t have to travel across PCIe lanes. Doing this on a normal PC would have horrific performance. ARM Macs do have this architecture but you pay for it with not being able to replace any individual components for upgrades or even repairs.

you can have sockets/slots for your RAM without it needing to be on the PCIe bus

any PC would be an example

AMD HSA (which afaik current-gen consoles are based on) is available on PC APUs also. Although it is pretty much completely undocumented mess on software side.

Gaming consoles have worse performance than PCs.

> On console the ram is shared, and moving data to / from ram can be accomplished via onboard streaming hardware. PC lacks this, so they try to load everything in one shot to avoid studders. That's why you see so many games these days that need 12-16gb of vram.

Uh, no? The majority of VRAM data is texture data, and there's absolutely no point or benefit to sharing that with the CPU. So unified vs. not is irrelevant and plays no role.

Also consoles have been (and still absolutely do) stream texture data in regardless. Hell texture pop-in was easily the biggest graphics issue of the outgoing generation of consoles. But even on the current generation 16GB of shared memory is absolutely not enough to contain all the texture & geometry for a modern open world game. It still absolutely streams data, which is why Sony invested so much into texture streaming tech with the PS5 still.

The point of shared memory is that you can skip a whole memcpy. You load from drive into shared RAM and use it immediately. Preferably it's stored on disk in the same format that the GPU can use so you don't have to parse.

Compare that to a PC with a discrete GPU, you have to load the texture/mesh into RAM and (possibly memcpy to a mapped buffer that the GPU can see) and then issue a copy to move it over PCIe to dedicate GPU vram.

That only marginally improves asset streaming performance. Both CPU RAM and a pcie x16 slot have much more bandwidth than any nvme drive will (which is only x4 pcie, after all, even ignoring the limitations of the NAND flash & controller itself) so that extra memcopy isn't significant. And has absolutely no impact whatsoever on general VRAM usage as was claimed anyway.

Also you don't have to load it to system RAM first. DirectStorage exists (kinda)

Well, that and you’re optimizing for a small number of console models vs. a large number of PC configurations.

Further, the piracy situation on PC is much worse (from the developer’s POV) than on consoles.

These factors might explain why developers wouldn’t be chomping at the bit to make well-optimized PC games.

I've been wondering something. Has the PC ever been the majority platform for video games?

It depends on how you define it. I'd define it as having more than half of all game software sales (including DLC, subscriptions, microtransactions, etc.). You could also include hardware spending, but most people bought computers to do other things. You could also use time spent playing games, and in that case, the PC might win because of Solitaire and Minesweeper.

I'd define “PC” as a machine running MS-DOS or Windows. Also, I'd include any machine with a built-in compatibility layer, like the Steam Deck. I wouldn't include other personal computer platforms.

I've been searching the web, but I haven't been able to find the stats on this.

On console the ram is shared, and moving data to / from ram can be accomplished via onboard streaming hardware. PC lacks this

This hasn't been true for years now.

If you're talking about 'direct storage', it's in no way the same. That runs on the gpu, using perf that could otherwise go to rendering (I think DF saw a 10% drop in fps on some games?). Frankly, this isn't going to be fixed until GPUs have dedicated hardware in place, much the same as on console. Maybe that happens with the 5xxx series, we'll have to wait and see.

> something like 80%+ of rockstar's revenue was from console

GTA V had a massive head start on PS3 before it was released on PC, and then re-released on PS4, and then re-re-released on PS5.