AMD CEO Lisa Su Talks About 3rd Gen Ryzen Boost Issue in Q3 Earnings Call

[btarunr]

AMD CEO Dr Lisa Su in response to a question, spoke about 3rd generation Ryzen processor boost issue. Dr. Su was responding to a question by Mitch Steves of RBC Capital on whether she had comments on "the software side" of 3rd gen Ryzen, and articles in the press still popping up about them despite AMD's fix. This was interpreted by the AMD CEO as a question specific to the Precision Boost controversy surrounding 3rd gen Ryzen chips, in which processors would seldom/never hit the advertised maximum boost frequency. AMD tried to address this by issuing updates to its processor microcode under AGESA Combo 1.0.0.3 ABBA, distributed through motherboard BIOS updates. The new microcode is supposed to increase the maximum turbo clock-speeds for "the vast majority" of users.

In her response, Dr. Su began by stating that the company is pleased with the sales of these processors. She then mentioned that AMD is working with its motherboard partners and ODM partners to "improve the optimization of the maximum boost frequency." She notes that the issue has been "largely addressed over the last couple of weeks" (referring to 1.0.0.3 ABBA). She goes on to state that AMD sees its response to the boost issues as more of an "optimization," rather than a "major update," possibly trying to allay investor fears that AMD is firefighting a costly problem with its products. "We're going to continue to improve the platform," she adds, possibly referencing the upcoming AGESA 1.0.0.4 Patch B microcode that's beginning to ship out by motherboard vendors. The earnings call can be accessed here. The specific question can be found at 47:00.



View at TechPowerUp Main Site

 

[DeathtoGnomes]

"optimizations are always being done" should have been her response on this type of call to such a generalized question.

 

[InVasMani]

"AMD shares fell around 1.96% on Tuesday and closed the trading day at $33.03. On a YTD (year-to-date) basis, AMD has risen around 78.9% as of Tuesday. In comparison, the S&P 500 has risen around 21.14% YTD. Intel (INTC) and Nvidia (NVDA) have risen 22.4% and 52.3%, respectively, YTD." Seems only AMD pull of such results YTD has risen more than Nvidia and Intel on S&P500 combined and still somehow gets trashed badly by irrational stock analysts such a Jim Cramer go figure. To be fair idk if he was one of the ones that was leery on it this time, but **** him anyway cause that guy was laughing at Intel being a monopoly and named dog Nvidia and had the audacity to label stock investers keen on AMD as homers right before the stock skyrocket in value and curb stomped Intel's entire CPU lineup convincingly enough.

 

[spnidel]

asking questions like that... it's like they do zero research regarding companies and their products, and invest money into stocks of said company with barely any knowledge
seriously, what the fuck?

 

[Deleted member 185088]

There is a very interesting article in Anandtech about AMD's Ryzen boost. It seems that some people don't understand the difference between AMD and Intel's Turbo modes.

Reaching for Turbo: Aligning Perception with AMD’s Frequency Metrics

www.anandtech.com

 

[TheLostSwede]

asking questions like that... it's like they do zero research regarding companies and their products, and invest money into stocks of said company with barely any knowledge
seriously, what the fuck?

Of course. If you've ever been to any meeting/conference where you have financial people, you'll find that they're the dumbest people in the room...
I've sat through far too many of those kind of events. The only thing more annoying are certain Germans at tech events, as they always try to prove that they know more than the company does about their own products...

 

[voltage]

Intel does wrong, and seemingly the entire internet explodes with hate. amd does wrong, and seemingly everyone buys their products. after I got burned by amd with them lying about core count on bulldozer, and also seeing my best friends brand new platform cant boot because of a bad motherboard firmware (3 weeks ago). ill remain one of the few not supporting amd this time.

That said, I am in need of a new system, badly. ill wait it out for DDR5

 

[DeathtoGnomes]

Intel does wrong, and seemingly the entire internet explodes with hate. amd does wrong, and seemingly everyone buys their products. after I got burned by amd with them lying about core count on bulldozer, and also seeing my best friends brand new platform cant boot because of a bad motherboard firmware (3 weeks ago). ill remain one of the few not supporting amd this time.

That said, I am in need of a new system, badly. ill wait it out for DDR5

i disagree, AMD does enough wrong, as much as Intel, they just dont do the same wrongs as each other.

 

[EarthDog]

She goes on to state that AMD sees its response to the boost issues as more of an "optimization,"

Does anyone else find it odd that "optimizations" are needed to have the CPU at the specs THEY THEMSELVES put on the box??? Feels like the cart before the horse to me! I know this isn't a huge deal, but it baffles me how an engineer can sit there and see what is happening on the CPUs and let them out without most of them being able to reach specification. How does that happen?

That said, I am in need of a new system, badly. ill wait it out for DDR5

Why, if you need a system "badly" would you wait for DDR5? What possible benefits does it have for the common user that you need to wait for it while? If you need something now, get it. DDR5 isn't going to mean squat except for more bandwidth (kitchen sink through a fire hose, anyone) and lower voltage.

 

[kapone32]

Intel does wrong, and seemingly the entire internet explodes with hate. amd does wrong, and seemingly everyone buys their products. after I got burned by amd with them lying about core count on bulldozer, and also seeing my best friends brand new platform cant boot because of a bad motherboard firmware (3 weeks ago). ill remain one of the few not supporting amd this time.

That said, I am in need of a new system, badly. ill wait it out for DDR5

It is not that Intel does wrong. It is the fact that Intel have not have any positive news on the CPU front since the launch of Ryzen. There are things like the security flaws that keep popping up every few months. The failed 10NM node (that was supposed to launch the same time as Ryzen). Do you remember that when TR4 second gen was launching they tried to take the wind out of the sails of AMD by showing a 28 Core CPU running at 4.5 GHZ but using a home air conditioner to keep it viable at those clocks that was found out after the fact? The other point on this thread is even though people could not hit advertised boost clocks on release the 3rd Gen Ryzen CPUs are faster than 2nd Gen and the 2nd is faster than the 1st. Then there have been AGESA updates to improve the performance of those CPUs. Now we have the 9900KS (I feel they had to release this in response to the 3950X) the exact same thing as the 9590 from AMD vs the 8350. As far as Motherboard woes every platform has lemons and some brands are better than others. The simple thing for your friend to do is to RMA the board instead of lamenting on it for 3 weeks.

 

[sutyi]

Does anyone else find it odd that "optimizations" are needed to have the CPU at the specs THEY THEMSELVES put on the box??? Feels like the cart before the horse to me! I know this isn't a huge deal, but it baffles me how an engineer can sit there and see what is happening on the CPUs and let them out without most of them being able to reach specification. How does that happen?

Why, if you need a system "badly" would you wait for DDR5? What possible benefits does it have for the common user that you need to wait for it while? If you need something now, get it. DDR5 isn't going to mean squat except for more bandwidth (kitchen sink through a fire hose, anyone) and lower voltage.

I think internal testing works / worked differently at AMD as far as consumer motherboards go. Considering the boost algorithm was behaving all-over the place (not even VRM dependently in some cases) relation on what brand of you had stuck that Ryzen 3000 series CPU in, I'm highly suspect of boardpartners doing weird things to the boost behavior maybe sacrificing low nT boost clock in favor of high nT loads and clock. I'm hope they now standardized out of the box stock clock behavior with 1003ABBA + SMU update onwards.

On one hand most people dont give much fluff about 25-50MHz 1-2C boost, on the other this should've been resolved before actual mass release... especially when some parts did not meet clocks that were written on the box.

 

[EarthDog]

I think internal testing works / worked differently at AMD as far as consumer motherboards go. Considering the boost algorithm was behaving all-over the place (not even VRM dependently in some cases) relation on what brand of you had stuck that Ryzen 3000 series CPU in, I'm highly suspect of boardpartners doing weird things to the boost behavior maybe sacrificing low nT boost clock in favor of high nT loads and clock. I'm hope they now standardized out of the box stock clock behavior with 1003ABBA + SMU update onwards.

On one hand most people dont give much fluff about 25-50MHz 1-2C boost, on the other this should've been resolved before actual mass release... especially when some parts did not meet clocks that were written on the box.

You can bet your arse that if this was on the AIBs, AMD would have said something to the like. I don't think the AIBs had anything to do with this at all. If so, you would have seen them come out with their own BIOS' on their own time instead of AMD distributing a mass AGESA update for all and specifically addressing it as an issue.

 

[Reaperxvii]

"AMD shares fell around 1.96% on Tuesday and closed the trading day at $33.03. On a YTD (year-to-date) basis, AMD has risen around 78.9% as of Tuesday. In comparison, the S&P 500 has risen around 21.14% YTD. Intel (INTC) and Nvidia (NVDA) have risen 22.4% and 52.3%, respectively, YTD." Seems only AMD pull of such results YTD has risen more than Nvidia and Intel on S&P500 combined and still somehow gets trashed badly by irrational stock analysts such a Jim Cramer go figure. To be fair idk if he was one of the ones that was leery on it this time, but **** him anyway cause that guy was laughing at Intel being a monopoly and named dog Nvidia and had the audacity to label stock investers keen on AMD as homers right before the stock skyrocket in value and curb stomped Intel's entire CPU lineup convincingly enough.

I mean as far as the stock market is concerned it's a s^!t show anyways. While literally everyone is at fault (Nvidia is trading at I think 120 dollars a share? But its valued at 200 something?). Not jabbing at nvidia its literally almost every company currently. Look at their revenue per share then the actual stock price.

Like what a company's revenue per share is is what the stocks "supposed" to be, and when the next recession hits and stocks go way back down to where they're supposed to be alot of people are gonna be sol.

 

[neatfeatguy]

Does anyone else find it odd that "optimizations" are needed to have the CPU at the specs THEY THEMSELVES put on the box??? Feels like the cart before the horse to me! I know this isn't a huge deal, but it baffles me how an engineer can sit there and see what is happening on the CPUs and let them out without most of them being able to reach specification. How does that happen?

I've had cars that don't get the "estimated" MPG and I've had cars that get well over the "estimated" MPG. Then again, without lots of variables being taken into account and having a way to remove them and then testing the cars in a controlled, ideal environment, who's to say it's not the driver that's having the issue....maybe I'm too heavy footed from a dead stop to 55 MPH limit? Maybe I don't utilize cruise control to help maintain a constant 55 MPH speed? Maybe the headwinds are blowing really hard as well? Maybe the 87 octane fuel being used isn't ideal and should be 92 octane?

My point is, not everyone has the exact same build - it's just not possible. Slight variances in different hardware being used, different software installed, different BIOS in use and so on and so forth.

For anyone to expect things to be perfect through all the variations of hardware and software out there, you're crazy. Ideally, it would be nice if all the CPUs worked without fail.

Anyway, maybe at AMD they didn't run into issues with the handfuls of hardware setups they used for testing. But once things are released to the wilds and no longer in a controlled environment it's very possible for some issues to crop up.

 

[EarthDog]

My point is, not everyone has the exact same build - it's just not possible. Slight variances in different hardware being used, different software installed, different BIOS in use and so on and so forth.

For anyone to expect things to be perfect through all the variations of hardware and software out there, you're crazy. Ideally, it would be nice if all the CPUs worked without fail.

Anyway, maybe at AMD they didn't run into issues with the handfuls of hardware setups they used for testing. But once things are released to the wilds and no longer in a controlled environment it's very possible for some issues to crop up.

Indeed... however, this absolutely should have come out in testing. If you think they only tested a 'handful' and called 50K other CPUs and 50+ mobos good, you are sorely mistaking. This happened on high-end boards and potato boards. The behavior did not discriminate. Under their "ideal" conditions specifications were not met.

This is something that should have been caught #1 - a lot earlier, #2 by AMD themselves. Looking around the web, it seems like more people than not had this issue.... how could that NOT be caught earlier???

 

[neatfeatguy]

Indeed... however, this absolutely should have come out in testing. If you think they only tested a 'handful' and called 50K other CPUs and 50+ mobos good, you are sorely mistaking. This happened on high-end boards and potato boards. The behavior did not discriminate. Under their "ideal" conditions specifications were not met.

This is something that should have been caught #1 - a lot earlier, #2 by AMD themselves. Looking around the web, it seems like more people than not had this issue.... how could that NOT be caught earlier???

By that thinking then Nvidia and AMD should have zero driver issues when they release them to the public.

 

[EarthDog]

By that thinking then Nvidia and AMD should have zero driver issues when they release them to the public.

No (apples and oranges). Nothing is perfect. But that said, this was a systemic problem that affected a majority of users and required a new AGESA/Microcode by AMD to correct the issue. This should have been caught a lot earlier and by AMD instead of the public.

 

[DeathtoGnomes]

ndeed... however, this absolutely should have come out in testing. If you think they only tested a 'handful' and called 50K other CPUs and 50+ mobos good, you are sorely mistaking. This happened on high-end boards and potato boards. The behavior did not discriminate. Under their "ideal" conditions specifications were not met.

While applying for a job as a play tester, I assumed that there was a broad range of systems to test on, I was sorely mistaken, I could count them without taking my shoes off. Applying this experience to CPU testing assumptions, I would guess they are testing on that handful of motherboards, leaning heavily on the more common build configs and my guess is that they do use less than 25 motherboards.

 

[neatfeatguy]

No (apples and oranges). Nothing is perfect. But that said, this was a systemic problem that affected a majority of users and required a new ASGESA/Microcode by AMD to correct the issue. This should have been caught a lot earlier and by AMD instead of the public.

I don't see it as apples and oranges. There are just too many variables to account for with all the hardware, software, firmware and OS crap that's out there. If Nvidia and AMD does testing and they encounter no system or game breaking issues with their GPU drivers and release them....but something else completely causes a shit storm (like drivers not having fans spin up and GPUs failing) that they didn't see in all their testing, it happens. Some other unknown configuration/variable out in the world is causing issues they didn't see. Do they end up finding and eventually correcting the issue? Sure.

As for AMD, unless you're part of their engineering team and know for sure, it's just an assumption on your part that AMD dropped the ball. We can draw conclusions all day, but we'll just never know. Perhaps there was something new they didn't account for and missed and results worked correctly from all their other testing. From things I've read it wasn't something that impacted a majority of owners. Some had issues, some didn't and some has CPUs that boosted above the suggested rage - and this is just from a small sampling of folks that have posted. You have that whole other group of people that never looked or reported issues or no issues. It's not fair to claim it was a majority of people.

Wasn't it the 2080Ti cards that some in the first batch were failing and others weren't? You can't claim that a majority of users on the 2080Ti were having issues with faulty cards without taking note from every single person that purchased it upon release. If you only take a small pool of people that are verbal about their failures and you don't get the big picture and you're left with skewed results and incorrect conclusions get drawn. Where there issues? Sure. Did they get fixed, it appears so.

Just like with AMD's Zen 2 - where their issues? Apparently. Some folks had boosting issues and AMD has been addressing them and fixing them.

I'm not defending AMD and saying they're in the clear, but perhaps this whole issue will allow AMD to re-evaluate their testing procedures to make sure it doesn't happen again.

I guess I'm just hoping folks will calm down and relax some. Then again, it is entertaining to read through the posts of some folks on these topics, get so worked up.

 

[EarthDog]

Applying this experience to CPU testing assumptions,

And here is where you go wrong.

From things I've read

Did you miss der8's testing and his graphs?

The problems with the 2080ti showed up months after release, no? Configuration wasnt relevant. You drop an amd chip in any mobo and it's a crapshoot if you received the specs on the box. Different situations to me.

I also want to note that I'm personally not fired up. It is what it is. I said it before in the discussion threads this isnt a huge deal (nvidia cards dieing happened to a much smaller percentage, but still worse). But make no mistake about it, this is AMDs snafu. If it wasnt, the same chips would work in the same boards...but some worked and some didnt. It wasnt the boards.

 

[cygnus_1]

I wonder if the question was really geared more toward the issue with the hardware random number generator.

 

[thesmokingman]

Intel does wrong, and seemingly the entire internet explodes with hate. amd does wrong, and seemingly everyone buys their products. after I got burned by amd with them lying about core count on bulldozer, and also seeing my best friends brand new platform cant boot because of a bad motherboard firmware (3 weeks ago). ill remain one of the few not supporting amd this time.

That said, I am in need of a new system, badly. ill wait it out for DDR5

Yea, like um that's just your opinion.

 

[Cheeseball]

I wonder if the question was really geared more toward the issue with the hardware random number generator.

This is the only reason why I'm not using WireGuard for remote connections on my 3700X build. I could just set nordrand while booting into Fedora but OpenVPN works too.

 

[dicktracy]

Intel does wrong, and seemingly the entire internet explodes with hate. amd does wrong, and seemingly everyone buys their products. after I got burned by amd with them lying about core count on bulldozer, and also seeing my best friends brand new platform cant boot because of a bad motherboard firmware (3 weeks ago). ill remain one of the few not supporting amd this time.

That said, I am in need of a new system, badly. ill wait it out for DDR5

That's more due to the fact that AMD fans are usually more vocal than Nvidia and Intel ones. This isn't even an opinion. It's a fact. On youtube, reddit, tech site comment section like here, most outrage and trolls are from the AMD side.

 

[medi01]

Intel (INTC) and Nvidia (NVDA) have risen 22.4% and 52.3%

lolwhat?

Nvidia stock sent crashing after missed earnings

Nvidia issued disappointing guidance on its latest earnings call after demand for the companies computer chips dried up.

www.cnbc.com