Intel Core i9 7960X Linux Benchmarks

[chuckula]

Another very powerful product from the #1 contributor to the Linux kernel.
People who complain about the price over AMD ought to remember that AMD doesn't contribute anything to Linux outside of (sometimes) providing drivers for its own hardware. The trivial markup for a 16 core Skylake X -- which is also clearly a better value than cheap Threadrippers -- shows that you not only care about quality and innovation but that you want Linux to thrive into the future instead of dumping money on the company whose founder and CEO took kickbacks to testify in favor of Microsoft in federal court.

Thanks for the solid review too Mike!

Incidentally, now that these products are all out on the market, a writeup about the Open Porous Media benchmarks that use these new chips would be helpful.

[sdack]

Another very powerful product from the #1 contributor to the Linux kernel.
People who complain about the price over AMD ought to remember that AMD doesn't contribute anything to Linux outside of (sometimes) providing drivers for its own hardware. The trivial markup for a 16 core Skylake X -- which is also clearly a better value than cheap Threadrippers -- shows that you not only care about quality and innovation but that you want Linux to thrive into the future instead of dumping money on the company whose founder and CEO took kickbacks to testify in favor of Microsoft in federal court.

Thanks for the solid review too Mike!

Incidentally, now that these products are all out on the market, a writeup about the Open Porous Media benchmarks that use these new chips would be helpful.

I have the suspicion with kernel 4.14 will Intel leap only further ahead, because of its support of the new PCID feature, which supposedly all AMD CPUs currently lack. If so do I hope to see a retest and we may see AMD falling behind even faster than one might have thought (or liked).

[lem79]

It took Threadripper for Intel to even consider launching these chips. Settle down Chuck

[schmidtbag]

Why were several i7s include in this comparison but the Ryzen 7 omitted? It's not like Ryzen is already old or are out of the competition. It would have been nice to see where it currently holds itself.

To elaborate on Michael's comment, the i7s listed use the same socket and chipset as the i9s. They're sort of an exception.

[chuckula]

Another point... given that this article was using the rather poorly optimized Ubuntu for tests I look forward to a distro shootout including Clear Linux and OpenSuse/Fedora to see what these chips are truly capable of doing.

[sdack]

To elaborate on Michael's comment, the i7s listed use the same socket and chipset as the i9s. They're sort of an exception.

The question is not about why several Intel chips were included, but why the Ryzen 7 was omitted. So if you have to offer an excuse as to why Phoronix doesn't like to give their readers a more rounded picture (by including more different CPUs than just more of the same) then I'd like to read about it.

[Michael]

The question is not about why several Intel chips were included, but why the Ryzen 7 was omitted. So if you have to offer an excuse as to why Phoronix doesn't like to give their readers a more rounded picture (by including more different CPUs than just more of the same) then I'd like to read about it.

I just didn't have time to include it when was already pressed for time getting the tests done in time for the launch day.... There will be Ryzen 7 tested obviously with the upcoming Coffee Lake tests as that's 'normal' desktop processors being tested.

[jhenke]

These chips are very interesting as AMD is forcing Intel here to do things they really do not want to do. But have to for the simple sake to be the fastest. As a German IT news site remarked (golem.de), these chips actually use the die from Intel current Xeon line-up, they are Xeons for the desktop socket. This is interesting as Intel is forced to sell them ~500-1000€ cheaper (Xeon vs i9 brand), but they are still very expensive for Intel to produce, as the die area is pretty big. Whereas AMD can sell almost the same performance for around half the price, because they use smaller dies connected on the same carrier.

[duby229]

These chips are very interesting as AMD is forcing Intel here to do things they really do not want to do. But have to for the simple sake to be the fastest. As a German IT news site remarked (golem.de), these chips actually use the die from Intel current Xeon line-up, they are Xeons for the desktop socket. This is interesting as Intel is forced to sell them ~500-1000€ cheaper (Xeon vs i9 brand), but they are still very expensive for Intel to produce, as the die area is pretty big. Whereas AMD can sell almost the same performance for around half the price, because they use smaller dies connected on the same carrier.

I wonder how much the interposer adds to AMD's cost though? I understand it to be a type of silicon die and it's huge. I'm just wondering if my impression is correct?

[sdack]

I just didn't have time to include it when was already pressed for time getting the tests done in time for the launch day.... There will be Ryzen 7 tested obviously with the upcoming Coffee Lake tests as that's 'normal' desktop processors being tested.

May I ask that you take a closer look at 4.14 before you start doing your Coffee Lake tests? If it's true and 4.14 does indeed give a noticeable advantage to Intel CPUs (due to their support of PCID and AMD's lack thereof) then you might be able to save some time by combining the tests and also to give us a look ahead into how the CPUs will compare in the future.