Intel's new Microcode 0x129 Performance Impact TESTED!

I'm just happy to see these finally coming out. On a side note AI summary is getting wild. "Intel released a microcode update (0x1 129) for the 14900K CPU to address voltage stability issues. The update resulted in a slight performance drop in some tests, such as Time Spy Extreme (700 points), but no significant changes in most other benchmarks. The update also reduced voltage spikes during single-core workloads, potentially improving stability. However, the author notes that it's too early to say if the update fully resolves the voltage issues and recommends waiting for further testing."

Let's not thank them, this is their job, they made a defective product and had to correct it. All we say is "do better next time." We owe no corporation, company, or otherwise any loyalty, special treatment, you want to make money? You have to earn it.

Used market for 13/14th gen is going to be filled with enough landmines to make even the Vietnam War blush...

16:08 Phil was so proud of this shot

>It's time to RMA that, however painful that process is gonna be I'm in the middle of a RMA for my 14900K and Intel loves nothing more than employing dilatory tactics. It’s been weeks of drawn out emails, taking photos of the cpu, sending invoices and receipts, and they’ll do anything other than help me. I would have gone AM5 had I known how badly this was going to turn out.

We do not owe Intel any thank you for fixing what they broke in the first place.

Even if the new microcode does fix the problem, it does nothing to fix Intel's reputation. They knew about this for over a year, did not communicate this to customers, and are now apparently rejecting RMAs of CPUs bought from Amazon and Microcenter because they're "not genuine".

aaaaaaand since this "fix" is going to require months of failure data to see whether it's had any impact on reducing the rate of degradation - ONLY FROM THIS POINT IN TIME ONWARDS - this issue is guaranteed to continue being unresolved and a total PR fustercluck for Intel for months to come

My only question is if they knew this was a thing for basically 2 years why have we only just now got a microcode update

My sympathy is with the poor devs who had to make and push out this microcode so fast. And the terror of whoever had to QA it.

13:30 “there’s 28 threads in a 14900K” small correction, there’s 32 threads in a 14900K.

Honestly intel needs to do more. You can't expect the majority of people to even know what a bios is let alone risk bricking a pc to fix their error. It's like taking your car to get the tires changed and then the garage telling you to torque everything down to spec.

They should also release a testing executable to perform stability tests. By the amount of data they have at this point, they should know what to look for.

I hope someone comes up with a stress test that that will difinitively show if a cpu is bad or not.

Small side note here. In physics Power is always produced by multiplying voltage times current or amps (ampere as the french say). So, we have V * A = W(atts). That means, there higher the voltage for a current is, the more watts you have to cool. Before I RMA'd my 13700K, I was watching Buildziod's voltage analyses video and I thought to myself, what if I fiddle with the resistance and voltage directly in my VR-settings? So I did. I set a upper voltage limit to the VR (voltage regulator, which you can fiddle with since Z690) to 1.380V, which means, my 13700K got an upper voltage limit to 1.380V to run at stock speed. I also lowered the resistance settings in the VR-settings from 40 / 90 AC to DC loadline to 10 / 10 AC to DC loadline. And I set the current limit to Intel's specification to 307 Amps. With benchmate I got my 13700K from 29600-ish point in Cinebench23 multicore point to 30200-ish points at stock and 30962 point with a +300MHz overclock and +200MHz on the P-cores and E-cores, but I had to adjust the voltage limit up to 1.450V. However, within Intel's spcification and my self-emposed voltage limit, Cinebench23 did'nt even get remotely close to the power limit of 253 Watts. Highest measured was 233 Watts and the highest core temp was at 82°C. That is a world away from the previous 104°C.

One comment: 13900's and 14900's have a wide range of VID settings depending on the quality of the silicon. The performance impacts are expected to be much more significant for the chips at the top end of the scale of 'how much voltage they need' than the ones that win the silicon lottery. The better cpus might not get capped at all since they are down 0.1v in a safe zone anyway. Might want to test this with a few more CPUs unless you know this one to be on the high VID side.

MSI of all companies updating their bios first? That's a surprise.

Remember, we just want strong Intel and AMD, and if it were possible, others, competition is key. Yeah so I think it’s reasonable that people expected a drop in performance, without being sure it would.

I remember seeing Falcon Northwest ads in 3DMark 2003, lol.

Very interesting Data. Just goes to show, the high voltages weren't at peak power loads, and not peak workloads (ie. multicore). And with the minimal performance drop with the new patch, you can really see that these high voltage spikes weren't really necessary. They didnt' give you more cinebench performance, they gave you "better performance while loading" ie, when no one really cares or notices. These voltages were to hit clock speeds that were largely marketing bulletpoints, which minimal performance benefits. However what they did do is degrade the silicon. Hopefully Intel learns their lesson.