Prime95 and Skylake

[whoololon]

According to this article, Intel's 6th generation processors suffer from a flaw that causes the PC to freeze when subjected to certain complex workloads.
The report points to programs like Prime95 as triggers for the error, and a solution that involves updating the BIOS.
Normally, for a "regular" user this does not represent much of a problem, since what is being talked about are advanced calculation applications.

[Yorus]

Well, that's good to know, because at work I use Prime95 a lot to test micros and cooling.

R Tape loading error, 0:1

[Espinetenbolas]

Between that and the Skylake PCBs bending with big heatsinks, Intel is really showing off with this generation :facepalm:

[cobito]

Actually, I am surprised that this kind of thing doesn't happen more often. That a hardware design flaw in such incredibly complex integrated circuits is news is something to take your hat off to.

Skylake was becoming a plausible option to renew my ancient platform this summer; let's see how the matter is resolved and we don't have to wait for the next tick to bring the bugs fixed.

[whoololon]

Hehe, the "bendgate" of skylake is quite interesting, there is a pretty good article on the subject, incorporating notes from various cooler manufacturers, that is, it is in modern Teutonic and the Google translation is really quite sad.

Regarding the failure, given the problem and the possible solution (much less traumatic than the TSX cap in Haswells), I would classify it more as an anecdote than news, even more so when it only affects a small number of users (and maybe Yorus. ;D )

That said, the institutions and companies that bought them specifically for this purpose, it probably didn't make them laugh much.
…and this happens with an AMD that is playing it all or nothing in the next 2 years, I'm scared if Intel is finally left alone.

Anyway, let's hope that by the time I have to change my equipment, in about 6 years, they have already released the photonic processors or at least the 5 nm processors in stacked configurations.

[cobito]

@whoololon:

I'm afraid that Intel will eventually be left alone.

I think that today it is increasingly less relevant that Intel is left alone. What I see in a few years is that x86 will not have a monopoly on the "PC" but that other players will appear and may even be strong.

On the one hand we have ARM which is currently making do in mobile devices. ARVv7 is timidly entering the desktop since, apart from Raspberry Pi which is a very basic board with very limited power, there are a lot of alternatives for less than 200€ that if they don't match, they exceed in performance the x86 platforms of the same price. And when aarch64 becomes popular, I am sure that a lot of options will appear to build a desktop with a very decent power, at a low price and with moderate consumption.

On the other hand we have RISC-V, which although it is nothing more than an ISA, some heavyweights of the industry have shown their interest which indicates that it is possible that the architecture will eventually materialize.

Regarding software compiled for x86, with Microsoft's ARM tablets we saw how "fast" and "easy" it was to port a complete operating system and key programs to ARM. On the other hand, Valve has already ported the Source engine and some games to ARM to run on Nvidia's platforms (I can't remember the name right now).

So if Intel is left alone in x86, I don't think it will be a great tragedy. Moreover, it may be that this makes new and interesting options flourish apart from the veteran architecture.