This thread is permanently archived
Intel Meltdown
| Use amd. It only suffers from spectre.
Or just dont give a shit, wait a bit for the price falling that is expected to happen with intel processors and build yourself a new computer.
| my gaming pc uses a intel i7 sheeeeeit
| Am using intel but bought the processor for 1/3 of its price.. So even if there is this issue. Intel shall fix the for their future processors and give good discounts on the older ones
| Do note that only certain workloads are affected by the patch. This is very naive and optimistic, I know, but that is the case here.
On the Linux side of things, only workloads that rely heavily on accessing the kernel will see a very huge impact on performance, whereas those that does little, say, games, will only see a slight drop.
Of course, this is still a bad news for everyone involved, and the only proper fix is to switch out your CPU.
| Games access the kernel not so much? The kernel commands all hardware. A game uses the gpu. Via a driver/kernel module. Slowdowns, the gpu will render based on what the userspace process is doing. Mind you, the fact that we are dealing with gpu, does not mean cpu is irrelevant. All syscalls and coordination in a computer is done by cpu. Basically the cpu is the computer, all else is periferics.
| >>103884
Let's correct this to the core. The kernel commands all the hardware, this is true. But you are confusing the topic here.
Additionally, all references I use as "kernel" refers to *Linux* not Windows.
By "access" or "call" to kernel I don't mean "using" the kernel. I'm talking loads similar to I/O operations.
Looking at the kernel patches, it seems that as long as apps/programs stay in userspace, their performance is largely unaffected.
| A look at the benchmark done by Phoronix, https://www.phoronix.com/scan.php?page=news_item&px=x86-PTI-Initial-Gaming-Tests shows that gaming performance is largely unscathed.
Benchmark that pings and calls the kernel every now and then (and thus, doesn't stay in the same space), has their performance degraded, some even halved.
But games mostly doesn't do this. They stay in userspace and they /should/ stay that way.
| Is disk access considered userspace? What about network access? Games make heavy use of these two. And I dont think of tetris.
| >>103923
Network access would need to worry more about the latency, and as for disk access, well, I have mentioned before, haven't I? I/O operations are indeed the ones affected by the patch.
However, if you're referring to loading up textures and whatnot, it seems to be largely unaffected.
I have not been researching deep into the issue, and have only glanced at the patches, there could be a specific set of I/O operations affected by this.
| I read the article and also the comments to it. It seems that the test dealt mostly with framerate. To be honest a game is not just framerate and loadtimes and delays from network performance are very important. Framerate comparison is a bit shallow in my opinion when measuring general slowdowns. A slight delay in network packet pdocessing could make a game unplayable. I am a veteran player of starcraft 1/2 and while the computer doesnt suffer framerate drops, the game is unpla...
| ... yable from bad network or when there are vast numbers of units involved in a battle, precisely because it is the processor that handles each unit and their interactions and relaying all that information to and fro the network pipe in realtime.
| >>103934 >>103935
Fair point! I'm not much of a gamer so I don't know the perspective of one. Your comments gave some light to the issue you mentioned.
This thread is permanently archived
| Thoughts