Intel Announces 8th Gen Core CPUs: Claims 40% Boost Over Gen 7, More Cores

[starshipeleven]

I'm fully aware that Intel and AMD are the most popular CPU vendors, but may I remind you that there are other, smaller vendors who might not include anything like PSP?

There is only IBM's Power, everyone else has some kind of security coprocessor bullshit inside.

[chithanh]

Also, Intel ME can be permanently disabled but bitching about Intel is so cool nowadays, right?

Nobody knows what exactly happens when you do that.
Some parts of Intel ME are still active as evidenced by the fact that your system doesn't shut down like when you completely overwrite the ME firmware.

After applying me_cleaner, Intel ME does less than before that's for sure, but who tells you that it doesn't enter a free-for-all debug mode when you overwrite parts of its firmware? Only Intel could tell, but they won't unless you are special friends with them.

[starshipeleven]

Nobody knows what exactly happens when you do that.

All evidence point towards being de-powered or neutralized though.

Some parts of Intel ME are still active as evidenced by the fact that your system doesn't shut down like when you completely overwrite the ME firmware.

ME isn't a monolithic system, it is a modular system. Core functionality initializes the board and keeps the watchdog from killing the system.

After applying me_cleaner, Intel ME does less than before that's for sure, but who tells you that it doesn't enter a free-for-all debug mode when you overwrite parts of its firmware?

A free-for-all debug system with no access to the outside world as the modules to accessing the outside world (ethernet and OS interfaces, and they are known) have been deleted. It can probably still talk with the (unpopulated) debug port (some kind of serial/UART/JTAG) on your mobo, but that's not worth crying about.

[sdack]

Not sure I follow your logic here.
...
So no, benchmarks are most certainly not all synthetic.

No, you don't follow... It's good to go into details, but you'll also have to come out again and not get lost in it, but you need to get back to the bigger picture of what it's all about. When a benchmark uses a simple loop to copy data around and thereby to measure memory performance then this is a synthetic measurement, but at the same time is such a simple task as "real-worldly" as any application, which has to copy data around. Vice-versa is a benchmark, which uses a real-world application for a measurement, just as synthetic when it repeats one particular work-load over and over again. That's not actually a real-world use for most real-world applications. Rather do work-loads in the real world keep constantly changing. In other words are you only "abusing" a real-world application for a benchmark, then leave the sticker on just so you can point to it and say it was a real-world scenario when it's not and lie about how synthetic it actually it is.

So yeah, all benchmarks are synthetic. Once you understand this can you also move on and begin to understand that it's not about a benchmark being synthetic, but rather about understanding what a measurement represents. You yourself don't actually give a reason for why you dislike it. It's an opinion you have and not based on any reason. Perhaps you don't like how SYSmark reduces many different factors down to a single number, which I can agree with when you want more and not less details. However, you shouldn't fall for the illusion of a detailed benchmark, made from real-world applications, to be able to tell you everything about the hardware (or the software) either.

[chithanh]

All evidence point towards being de-powered or neutralized though.
ME isn't a monolithic system, it is a modular system. Core functionality initializes the board and keeps the watchdog from killing the system.

I agree that it looks likely that ME doesn't do anything useful or dangerous any more. But I learned from Peter Stuge's 30C3 talk that even if the firmware is completely overwritten, it will send a single IPv6 packet out of the ethernet port (on video between 17:00 and 18:00, the whole talk is worth watching though).

A free-for-all debug system with no access to the outside world as the modules to accessing the outside world (ethernet and OS interfaces, and they are known) have been deleted. It can probably still talk with the (unpopulated) debug port (some kind of serial/UART/JTAG) on your mobo, but that's not worth crying about.

As I wrote above, we believe but we don't know that access to the outside world is completely stopped, it might still be listening for some special knocking packet on the ethernet port or whatever. And only Intel knows whether it does.