AMD Announces BIOS Fix for Ryzen 3000 Boost Clocks, Update Comes September 10

[nicalandia]

AMD Announces BIOS Fix for Ryzen 3000 Boost Clocks, Update Comes September 10
https://www.tomshardware.com/news/amd-ryzen-3000-boost-frequency-bios-fix,40308.html

“AMD is pleased with the strong momentum of 3rd Gen AMD Ryzen™ processors in the PC enthusiast and gaming communities. We closely monitor community feedback on our products and understand that some 3rd gen AMD Ryzen users are reporting boost clock speeds below the expected processor boost frequency. While processor boost frequency is dependent on many variables, including workload, system design, and cooling solution, we have closely reviewed the feedback from our customers and have identified an issue in our firmware that reduces boost frequency in some situations. We are in the process of preparing a BIOS update for our motherboard partners that addresses that issue and includes additional boost performance optimizations. We will provide an update on September 10 to the community regarding the availability of the BIOS.”

 

[nicalandia]

Any news on when the Ryzen 9 3950X will be avilable? I was guessing 09/07/2019 but I guess it will be after the 10th

 

[EXCellR8]

In other words, enough people complained about ~0.1-0.2ghz that they are forced to fix it. The firmware thing sounds like a redirect tactic though...

I understand the variables and conditions but hopefully they learn from this and realize that if end users don't see 1:1 performance numbers that there is going to be certain repercussions.

 

[nicalandia]

In other words, enough people complained about ~0.1-0.2ghz that they are forced to fix it. The firmware thing sounds like a redirect tactic though...

I understand the variables and conditions but hopefully they learn from this and realize that if end users don't see 1:1 performance numbers that there is going to be certain repercussions.

I wonder who has lost more performance, Amd Ryzen 3000 owners with less performance due to less ~0.1- 0.2ghz(per single core max boost) or Intel owners with all of the vulnerability patches?

 

[Kedas]

reduces boost frequency in some situations

So is CINEBENCH such situation? otherwise it won't change it.

I think we can exclude cooling from the equation.

 

[Abwx]

In other words, enough people complained about ~0.1-0.2ghz that they are forced to fix it.

They could had it fixed with a simple disclaimer if they wanted...

 

[EXCellR8]

I wonder who has lost more performance, Amd Ryzen 3000 owners with less performance due to less ~0.1- 0.2ghz(per single core max boost) or Intel owners with all of the vulnerability patches?

Well, to be fair, most Intel users would answer that by stating AMD is #2 regardless of any performance lost.

 

[BTRY B 529th FA BN]

A New Hope

 

[fleshconsumed]

Hooray for AMD fixing a rounding error!

 

[nicalandia]

A New Hope

What hope would that be? I mean this is literally for 0.1-0.2 Ghz in Single Core, because as soon as people start using their CPU for more than playing solitary and watching Netflix all boost CPU boost kicks in.

 

[BTRY B 529th FA BN]

What hope would that be? I mean this is literally for 0.1-0.2 Ghz in Single Core, because as soon as people start using their CPU for more than playing solitary and watching Netflix all boost CPU boost kicks in.

Luke will turn his father from the dark side

 

[Zucker2k]

Hooray for AMD fixing a rounding error!

I expect there'll be hell to pay if the latest bios boosts "correctly" but performance remains the same

 

[nicalandia]

I expect there'll be hell to pay if the latest bios boosts "correctly" but performance remains the same

But performance will remain the same since it's only 100-200 Mgz difference at most and for ST apps, Whoever thinks there will be a difference is naive

 

[Ajay]

But performance will remain the same since it's only 100-200 Mgz difference at most and for ST apps, Whoever thinks there will be a difference is naive

Right, but they pull their butts out of the fire by being able to match their claims wrt clock speeds. Ends a PR nightmare for AMD. They did the right thing.

 

[EXCellR8]

Thus my point... I guess I could have elaborated. It's not about the actual performance as it is so much about marketing. The difference in actual speed in negligible in this case of course, but when most people buy something they want 100% of the potential and not 97%, respectively. I don't see consumers being wrong in stating that, whether they are happy with the product or not.

Personally, I am very happy with my 3700X regardless of what the box says it will boost to; it gets close on an older board and I take that as a massive win.

 

[scannall]

I figured there was a little more in the BIOS. I love my 2600X, and AMD products in general. But, they always have teething issues with new products out of the gate. Intel does as well, but they haven't had anything truly new since 2015 so their BIOS is pretty smooth by now.

 

[Gideon]

I'm sceptical, my 3700x already hits 4400 briefly in ST workloads. There usually aren't free lunches, therefore I'm afraid this might end up hurting lightly threaded workloads. You know, everything, besides closing every other program and running Cinebench 1T

 

[PotatoWithEarsOnSide]

Winter is coming. Ambient temperatures will drop by a few degrees, and our clocks shall go up by the "missing" 0.025GHz.

 

[amd6502]

Winter is coming. Ambient temperatures will drop by a few degrees, and our clocks shall go up by the "missing" 0.025GHz.

Case temps building up in un-airconditioned environments can be a big problem. 7nm seems particularly vulverable to it. der8auer estimates per 20C increase in cpu temp there is a considerable wattage increase for the CPU plus typically a 75-100 MHz drop for boost freqs.

20C differences can be easily encountered, even with the same stock cooler. Person A is running in a 20C or less room, with aggressive fan profile, good case ventilation and has their CPU sitting at 50C under full load. Person B doesn't use AC, runs his computer in a 80F~27C room, has poor case ventilation and possibly a poor fan profile, and has their CPU at ceiling temperatures 70C-75C.

 

[lopri]

I am more concerned about CPU longevity after the "fix." I do not know what kind of solution is going to be applied, but if the current boosting behavior is an indication pushing more voltages more often might be a quick fix. The problem is that many of these chips are already supplied ~1.5V for max boost out of box. With more volts more often a long term longevity of the chips can be an issue.

 

[Andrei.]

I am more concerned about CPU longevity after the "fix." I do not know what kind of solution is going to be applied, but if the current boosting behavior is an indication pushing more voltages more often might be a quick fix. The problem is that many of these chips are already supplied ~1.5V for max boost out of box. With more volts more often a long term longevity of the chips can be an issue.

The issue is that the voltages are REDACTED. I'm running my 3700X via PBO & BCLK at up to 4.45GHz ST at only 1.387V versus the default 1.487V the CPU requests. AMD has a lot of clock headroom if they actually trim off the voltage margins instead of pushing for higher voltages creating a heat feedback effect and thus having to scale down to lower frequencies.

We have a zero tolerance policy for profanity in the tech sub-forums.

Iron Woode
Super Moderator

 

[nicalandia]

I am more concerned about CPU longevity after the "fix." .

The 3600(none X) already cut the Longevity of the first OG Zen just by being released, heck even going from 1700 to 3600 is now an upgrade for many people... So when Zen 3 is release people will be already feeling that their Zen2 is not up to task

 

[DrMrLordX]

The issue is that the voltages are REDACTED. I'm running my 3700X via PBO & BCLK at up to 4.45GHz ST at only 1.387V versus the default 1.487V the CPU requests. AMD has a lot of clock headroom if they actually trim off the voltage margins instead of pushing for higher voltages creating a heat feedback effect and thus having to scale down to lower frequencies.

That seems basically correct. There's really no problem pushing for higher voltages when current levels (and thus, heat) are low. But in all-core situations, I think a lot of these chips - well, 3700x and up anyway - are quite capable of 4.3-4.4 GHz all-core at reasonably low voltages. Especially in workloads that are non-AVX/AVX2. And it surprises me that such behavior isn't already baked into AMD's boost algorithm. It seems really conservative.

It's also too easy to mess up boost behavior with LLC settings and voltage offsets.

 

[JustMe21]

I just hope it was more of refinements they had to work out to get them to the top end speed instead of having to up any voltages.

 

[Markfw]

That seems basically correct. There's really no problem pushing for higher voltages when current levels (and thus, heat) are low. But in all-core situations, I think a lot of these chips - well, 3700x and up anyway - are quite capable of 4.3-4.4 GHz all-core at reasonably low voltages. Especially in workloads that are non-AVX/AVX2. And it surprises me that such behavior isn't already baked into AMD's boost algorithm. It seems really conservative.

It's also too easy to mess up boost behavior with LLC settings and voltage offsets.

You did see that I accidentally set the vcore to 1.1 and the speed to all core 4.1 and its running cooler and faster than stock ! Today I may try for 4,2 or more ! (I forget where I posted this)