I didn't get anywhere updating my Gigabyte AORUS x470 mobo to 12-core Gen3, kept getting blue screens and bugcodes.

Absolutely no one needs PCIe4.0, yet. It's normal that new tech is introduced over time, and it's perfectly understandable companies want to profit from the R&D so AMD want to sell X570. I think the validation might be plausible, the traces and that all BS they talk about. But it's not like you can't stick any new processor in any more!

asrock seems not following this trend, at least with x470 taichi. But a new bios faced up,3.60, that it seems to be specifically for ryzen 3000, updating agesa to 1.0.0.3 abb and addressing destiny 2 performance issue with ryzen 3000, never saw a bios addressing an issue of a specific game. Asrock even is against the upgrade on platforms different from ryzen 3000.

its not simply for destiny 2, it corrected a chip flaw via microcode that would render the RDRAND results non-random.

Astyanax:

its not simply for destiny 2, it corrected a chip flaw via microcode that would render the RDRAND results non-random.

is this explicable in a simple way for me to understand? :P

love how there were pages talking about pcie 4.0, the new motherboard and PCB design necessary to insure it's speed and quality and then people thinking any MB even old ones will suddenly support it with a bios update if you are annoyed by the drop (not really one it was never physically supported) you can read Hilbert great review on the 1st available pcie 4.0 nvme ssd (not very good I'm keeping my 970pro) https://www.guru3d.com/articles_pages/corsair_mp600_pcie_4_nvme_ssd_review,10.html

Who really cares? Nobody needs PCI-E 4 support in a long while. Gigabyte boards are fine. They tried their best to get this feature out, but it was not possible in the end. Nerds are so whiny I can't stand them.

rl66:

Intel have done it too... remember the entry level X79 that were in PCIe 2.0 despite it was capable of PCIe 3.0.

What happened with that was the X79 is a native PCIe 2.0 platform and it was motherboard manufacturers who enabled 3.0 via a bios, not Intel. The problem arose when nVidia released the GTX 680 and while testing on as many motherboards as they can and they found that not all of them adhered to a strict enough signal timing level and thus they found it impractical or impossible to tune their cards/drivers to perform at a level of stability high enough to warrant 100% certification.

Andrew LB:

What happened with that was the X79 is a native PCIe 2.0 platform and it was motherboard manufacturers who enabled 3.0 via a bios, not Intel. The problem arose when nVidia released the GTX 680 and while testing on as many motherboards as they can and they found that not all of them adhered to a strict enough signal timing level and thus they found it impractical or impossible to tune their cards/drivers to perform at a level of stability high enough to warrant 100% certification.

On Intel X79 motherboard (the complete board not only the chipset) you can select "Gen 3" or "Gen 2" on the PCIe setup and "Gen 3" is default setting. On the Asus budget X79 you can't set it at all and it is Gen 2 while 1st version of the bios were allowing it in gen 3 (but ok the whole board was crap whatever the bios 🙂 ).

It's better to disable it altogether than messing up your most important product launch because your board partners are trigger - happy with signaling requirements. It makes absolute sense.

kilyan:

is this explicable in a simple way for me to understand? 😛

RDRAND is a command that the CPU supports which helps generating random numbers. Random number generation is very important, as it is the basis of all encryption (most modern encryption really is multiplying two random numbers using the result as an encryption cipher, and then making one of the original numbers public, and the second one private, so you need both to decrypt). Real randomness is impossible but fake enough and it becomes random enough that it doesn't matter. RDRAND is hardware accelerated, so it helps a lot with generating randomness, even though it shouldn't be used only by itself (and it isn't). The issue that Ryzen 3k has (if it's a hardware issue, or it can be eventually fixed, we'll see), is that RDRAND sometimes returns -1, instead of a random number. That's especially terrible since it's normal usage is in combination with other random generation techniques (like M. Twister), so if your software trusts the CPU, it's kind of screwed, because not only you lose a source of randomness, but the - 1 is a known value.

its fixed in ABB

Astyanax:

its fixed in ABB

You mean they disabled it, right?

PrMinisterGR:

You mean they disabled it, right?

no, the chipset driver was the disable, the abb update is the microcode update. without driver or bios you get buggy RDRAND with driver but no bios you get RDRAND and RDSEED disabled with driver and bios you get RDRAND and RDSEED enabled and fixed. Chipset driver fix is disabled when agesa 1.0.0.3ABB is present.

Nice, that's great news, I thought they had a legit problem there for a while.