While ofc I don't have their code I'm a systems engineer (with a masters degree and phd student if that actually matters 😛) working on RA/AI and a hobbyist game dev so I'd bet 1 month of wages it's a 10 min fix. (Unless it's on some sort of drm crap :S)
i'd actually believe its part of their anti-hack code.
But... I still don't think Bungie is really obligated to make a patch because of AMD's mistake, especially since AMD already said they were working on it. It'd have been nice for Bungie to just release a patch anyway, but I don't fault them for not bothering.
Not saying that AMD shouldn't patch it via software or bios updates or both, but honestly, updating the AMD software is something a lot of people don't do, and even less update firmware.
The correct way to go about it, in my opinion, is for bungie to update their game, and AMD provide the updates they are providing. This eliminates the issue on both fronts, and the consumers benefit. Typically i don't go around looking for chipset driver updates because a game is not working....
Not saying that AMD shouldn't patch it via software or bios updates or both, but honestly, updating the AMD software is something a lot of people don't do, and even less update firmware.
The correct way to go about it, in my opinion, is for bungie to update their game, and AMD provide the updates they are providing. This eliminates the issue on both fronts, and the consumers benefit. Typically i don't go around looking for chipset driver updates because a game is not working....
Well, here's the crux of the matter:
Destiny 2 was just one example. systemd in Linux was a much bigger problem, but also still arguably niche. The problem is hardware-specific, so, it is unfortunately entirely up to AMD to fix this, because the problem will come back for other programs that may use rdrand. It's just coincidence that so far, only 2 programs have been identified. There could be more.
Well, here's the crux of the matter:
Destiny 2 was just one example. systemd in Linux was a much bigger problem, but also still arguably niche. The problem is hardware-specific, so, it is unfortunately entirely up to AMD to fix this, because the problem will come back for other programs that may use rdrand. It's just coincidence that so far, only 2 programs have been identified. There could be more.
And that's kind of my point though. What will fix it for the majority without people even knowing? Bungie applying a fix to to the game, and whatever needs to be changed in linux.
Again i'm not saying that AMD shouldn't provide a fix, i'm just saying that they aren't the only ones who should as it's unreasonable to expect everyone with a Zen 2 processor to update their bios or chipset drivers. You HAVE to update destiny 2 to play it, so if bungie provides a fix, in that one scenario, not a single person will be affected from then on. Maybe i feel that if i was a developer/publisher that i'd want to make certain people can buy and play my games more then bungie if they don't provide a fix, but if they don't, that really doesn't look good for them.
One other possible solution i guess would be if AMD provides microsoft with the latest chipset driver as a recommended automatic update through windows update.....but that still means people have to make sure their system is up to date and are not blocking windows updates, whereas, again, Destiny 2 HAS to be updated, to play.
It's still by no means a fix for all situations, a bios update realistically is the only way to do that.
But put simply, it doesn't hurt to provide the fix on multiple fronts, and it only helps Bungies customers by them providing a fix for their game, regardless of anything else. Otherwise in 5 years people could still be having this issue if they haven't updated to the latest chipset drivers or bios.
If this was a GPU driver issue i probably wouldn't say that bungie really needs to do anything, because people are use to updating their GPU drivers. Chipset drivers and bios updates however, are a different story.
And that's kind of my point though. What will fix it for the majority without people even knowing? Bungie applying a fix to to the game, and whatever needs to be changed in linux.
Well, you're kinda missing the point here:
rdrand is an already well-established instruction since 2012. AMD got it right the first time around, even for Zen. It's not the same thing as other forms of random numbers, so, even though so far we only know of 2 cases, I bet you there are much more than that. The thing is, Zen2 hasn't really been widely adopted yet, especially for high-end workstations or servers (that's what TR, Epyc, and Xeon are for) and that's where rdrand is more likely to be used. Random numbers are ideal for security purposes, which is why systemd in Linux uses it and, as stated by Astyanax, likely contributes to Destiny 2's anti-hack system.
Again i'm not saying that AMD shouldn't provide a fix, i'm just saying that they aren't the only ones who should as it's unreasonable to expect everyone with a Zen 2 processor to update their bios or chipset drivers. You HAVE to update destiny 2 to play it, so if bungie provides a fix, in that one scenario, not a single person will be affected from then on. Maybe i feel that if i was a developer/publisher that i'd want to make certain people can buy and play my games more then bungie if they don't provide a fix, but if they don't, that really doesn't look good for them.
I don't disagree, but, seeing as there is a protocol for rdrand that Zen2 is not following, I don't exactly fault Bungie for basically saying "not my problem". It's shitty customer support on their end, but in principle, they're not wrong.
One other possible solution i guess would be if AMD provides microsoft with the latest chipset driver as a recommended automatic update through windows update.....but that still means people have to make sure their system is up to date and are not blocking windows updates, whereas, again, Destiny 2 HAS to be updated, to play.
I think most people update the chipset driver through Windows update anyway. I don't think there are too many OEM PCs with Zen2 CPUs yet.
It's still by no means a fix for all situations, a bios update realistically is the only way to do that.
Agreed, though, that's even harder for the average person to do.
If this was a GPU driver issue i probably wouldn't say that bungie really needs to do anything, because people are use to updating their GPU drivers. Chipset drivers and bios updates however, are a different story.
I don't disagree. It's not an easy situation no matter how you look at it, but, at least AMD is doing something.
I don't disagree, but, seeing as there is a protocol for rdrand that Zen2 is not following, I don't exactly fault Bungie for basically saying "not my problem". It's shitty customer support on their end, but in principle, they're not wrong.
its the state of the register the instruction uses being out of sorts once the ccx has powered down, rather than it actually outright being wrong. if you never power off the chip by fully disabling CC power states, i suspect destiny 2 wouldn't have any issues - trouble is CC states are how ryzens do their boosts.
That's rich coming from you. You ignore the first part of the sentence that basically agrees AMD is at fault, yet the part that doesn't really have anything to do with AMD makes me have too much fanboyism?
Destiny 2 is not a very successful game. This is why they eventually gave it away for free.
I can't get windows to boot with the 12 core ryzen. It just want's to restart and notify that there was an unexpected error. Done this with 2 new rig builds. I wish I can get Win 10 to boot properly.
Thats a curious one Biffo, you should try exchanging your storage devices if htey happen to be brand new
Are your boards on the latest bios releases?
It was an absurd statement and you continue to make it absurd. That's like saying, Warframe isn't popular because its free. To defend AMD and say, oh well it isn't popular, so who cares. This is not some indie game. It IS a popular game from a triple A developer. Anyone should be annoyed when a company doesn't test their stuff before it comes out and then takes THREE weeks to fix it.
Just because you like a game, doesn't make it popular. For a AAA title, the most attention Destiny 2 has got in months was it not playing properly on Ryzen 3000. That's not signs of a popular game. Warframe, to my knowledge, was always free-to-play from its release date. There's a REALLY big difference between a game that was always free-to-play vs a game that had a price tag and then became free, because nobody was playing it.
Anyway, I didn't say this to defend AMD, but you're so caught up in your own biases that it's the only conclusion you could come up with. I don't know who at AMD pissed in your morning cereal, or, who at Intel bought you a new Ferrari, but you need to tone down the hate a notch.
I already agreed that people should be annoyed at AMD didn't do proper testing, but you still seem to ignore this.
And AMD's older stuff worked instead of their new stuff. Its totally ridiculous they missed this. This is why AMD continues to be made fun of when it comes to gaming. And people like you are not helping them.
Ok, now I'm going to come to AMD's defense:
Have you ever written a program yourself? Ever noticed how for a while everything seems to work, and then for some unknown reason, a behavior you knew for a fact worked fine suddenly doesn't under very specific circumstances? It could be so rare that you aren't the one who noticed; someone else did (hence bug reports). Designing a logic circuit is analogous to this. x86 CPUs aren't simple, hence the first "C" in their CISC denomination. Both Intel and AMD have hundreds of instructions. Zen2 was a drastic enough change, with rdrand being obscure enough, that it's not really surprising that this problem happened. Is AMD at fault here for not doing rigorous testing? 100% yes. But at the same time, people make mistakes, and they're more prone to it if the mistake is obscure and rare. It's not like AMD broke this, knew about it, and hoped nobody would notice. This is why I'm not upset at Intel for their slew of security vulnerabilities - they set things up in a way that was supposed to offer better performance, and understandably, didn't realize the very specific way where that could be exploited. You don't hear me talking crap about Intel for making a much more serious mistake.
Hopefully, AMD will learn their lesson from this and create a program that actually tests each of their instructions, rather than use something like CPU-Z to prove if they just simply exist.
TL;DR:
It was a mistake. A minor one, but still an inconvenience and an annoyance. Their solution wasn't the most ideal and it took them a while to reach it, but at least they did. It's not that big of a deal.
AMD has known for 5 years almost that their RDRAND instruction doesn't work properly following a sleep/resume cycle, I'm sure of it.
the first instances appeared with Excavator apu's on linux htpc's.
fixworkaround was to disable RDRAND https://cdn.discordapp.com/attachments/594661727057215510/605687799714414605/attachment.png