AMD AGESA 1.0.0.3ABBA Detailed, Fixes Zen2 Boost Issues

[TheLostSwede]

The numbers don't seem like all this was worth it. So you get +50 Mhz and the clock will hold at the spec for a few seconds to net you less than 1% performance. For what?

Uhm, you've clearly followed this topic then...
I've gained 200MHz boost on the highest boosting core, but I guess that's not worth it to you...
Not long ago, none of my cores would clock over 4,400MHz, now the slowest ones boost to 4,550MHz, but again, I guess that's not worth it either?

 

[Aerpoweron]

So this Bios update make the intel "5GHz means 5GHz" statement baseless now?

 

[TheLaughingMan]

Uhm, you've clearly followed this topic then...
I've gained 200MHz boost on the highest boosting core, but I guess that's not worth it to you...
Not long ago, none of my cores would clock over 4,400MHz, now the slowest ones boost to 4,550MHz, but again, I guess that's not worth it either?

I am glad it corrected whatever issue was going on with your BIOS. I have been watching this very closely since I will be buying one of these chips in the next few weeks. Your numbers seem to be part of the fringe since you gained 200 Mhz. Most of the chips from the online survey that D8bauer put out were only 50 to 75 MHz off. And yes I don't think 50 MHz would be worth it. Yours clearly needed a fix.

I am happy they did something, I am just worried that like with the testing of this issue, once it gets push to the board makes we are going to get mixed results on the level of improvement.

 

[Kapone33]

If nothing else it proves that AMD is listening to their customers. If anyone thinks it is because of Debaur's youtube challenge they would be wrong as AMD has probably been working on this since launch.

 

[Nordic]

The numbers don't seem like all this was worth it. So you get +50 Mhz and the clock will hold at the spec for a few seconds to net you less than 1% performance. For what?

Think about it this way. You are paying for 50/10 mbps internet but are only getting 45/1 mbps? Sure it won't really affect what you are doing, and 99% of people won't notice the difference, but that is not what you are paying for. AMD admitted there was a bug, and has seemed to have fixed it. We will be getting the specifications we paid for and will be happy.

 

[TheLostSwede]

I am glad it corrected whatever issue was going on with your BIOS. I have been watching this very closely since I will be buying one of these chips in the next few weeks. Your numbers seem to be part of the fringe since you gained 200 Mhz. Most of the chips from the online survey that D8bauer put out were only 50 to 75 MHz off. And yes I don't think 50 MHz would be worth it. Yours clearly needed a fix.

I am happy they did something, I am just worried that like with the testing of this issue, once it gets push to the board makes we are going to get mixed results on the level of improvement.

Why are we still discussing his flawed survey? And why am I a fringe case? Holy crap...
No, my CPU couldn't boost 1MHz beyond 4,400MHz until recently and it's an issue several other people have had here, that there has been a hard ceiling which can't be bypassed in any way whatsoever.
This was fixed on Gigabyte boards a few weeks ago and a couple of my cores would boost to 4,525MHz, but the rest would be 4,475MHz at the most. Now, as you can see, all cores boosts to a minimum of 4,550MHz, so this clearly changes things significantly. I.e. I'm now boosting up to 100MHz over AMD's claimed boost speed, on at least one core.

I wish people would stop making crap up on this topic, as people have had a wide range of issues and it's clear AMD is working on solving things. If you don't have one of the CPUs, please don't make assumptions.

So this Bios update make the intel "5GHz means 5GHz" statement baseless now?

Well, depends on how you look at it, but mostly, yes.
Obviously all core boost is still not going to get any higher, but that wasn't promised either.

 

[Eskimonster]

Im seriously surprised, cant wait til Der8auer calls out for another group test scenario again.

 

[thesmokingman]

Why are we still discussing his flawed survey? And why am I a fringe case? Holy crap...

Hey Swede, can you elaborate on the flawed part? I'm a bit out of the loop on that, thanks.

 

[GeorgeMan]

The numbers don't seem like all this was worth it. So you get +50 Mhz and the clock will hold at the spec for a few seconds to net you less than 1% performance. For what?

Avoiding lawsuits.

 

[TheLostSwede]

Hey Swede, can you elaborate on the flawed part? I'm a bit out of the loop on that, thanks.

1. Too small sample size. The only potentially valid results are for the 3700X.
2. Too many variables. It looks like there's a fair amount of user error in the "fringe" results, but he simply filtered those out so...
3. He doesn't quite understand statistics, as he's applying a very weird filter to the results.
4. The way he presents the data is flawed, although I guess that comes down to how it was collected as well. See 2.
5. In many cases "fringe" results were filtered out for no apparent reasons. See 3.
And so on...

 

[thesmokingman]

1. Too small sample size. The only potentially valid results are for the 3700X.
2. Too many variables. It looks like there's a fair amount of user error in the "fringe" results, but he simply filtered those out so...
3. He doesn't quite understand statistics, as he's applying a very weird filter to the results.
4. The way he presents the data is flawed, although I guess that comes down to how it was collected as well. See 2.
5. In many cases "fringe" results were filtered out for no apparent reasons. See 3.
And so on...

Thanks for breaking it down. But it sure got him a jack ton of hits, lol. That was probably more important than adhering to the scientific method.

 

[TheLostSwede]

Thanks for breaking it down. But it sure got him a jack ton of hits, lol. That was probably more important than adhering to the scientific method.

It's not all wrong, but he drew some flawed conclusions imho.
In all fairness, it showed how widespread the issue was, with not getting to the correct boost speeds. However, it also showed that a lot of DIY system builders that follows him, are not that great at building PCs...

 

[thesmokingman]

It's not all wrong, but he drew some flawed conclusions imho.
In all fairness, it showed how widespread the issue was, with not getting to the correct boost speeds. However, it also showed that a lot of DIY system builders that follows him, are not that great at building PCs...

He should have read more Shamino...

In other news, moar Intel hax...

Weakness in Intel chips lets researchers steal encrypted SSH keystrokes

DDIO makes servers faster. It can also allow rogue servers to covertly steal data.

arstechnica.com

 

[Nordic]

Thanks for breaking it down. But it sure got him a jack ton of hits, lol. That was probably more important than adhering to the scientific method.

Honestly, I really don't think that was his motivation. It would be hard to do an actual scientific survey in as short of time. I think what he did was okay because he was very upfront with the flaws with his methodology. It was not meant to be scientific.

What he did do was bring attention and validity to an existing problem. Regardless of if his efforts affected this, amd has a fix. I am EAGERLY waiting for asrock to get it out to us.

 

[Xzibit]

Honestly, I really don't think that was his motivation. It would be hard to do an actual scientific survey in as short of time. I think what he did was okay because he was very upfront with the flaws with his methodology. It was not meant to be scientific.

What he did do was bring attention and validity to an existing problem. Regardless of if his efforts affected this, amd has a fix. I am EAGERLY waiting for asrock to get it out to us.

He did try to pass it as such. Roman in his video said he filtered out the "noise". How is that even possible when their wasn't a validation process for the survey. Anyone could fill in the survey with no proof at all.

 

[hzy4]

Uhm, you've clearly followed this topic then...
I've gained 200MHz boost on the highest boosting core, but I guess that's not worth it to you...
Not long ago, none of my cores would clock over 4,400MHz, now the slowest ones boost to 4,550MHz, but again, I guess that's not worth it either?

I though the Agesa was released just for the board vendors, and the implementation into BIOS updates could take up to the 30th September. Is the Agesa available through AMD? Can you clarify where you got it from?

 

[thesmokingman]

Honestly, I really don't think that was his motivation. It would be hard to do an actual scientific survey in as short of time. I think what he did was okay because he was very upfront with the flaws with his methodology. It was not meant to be scientific.

What he did do was bring attention and validity to an existing problem. Regardless of if his efforts affected this, amd has a fix. I am EAGERLY waiting for asrock to get it out to us.

If you followed Shamino of Asus, you'd know that AMD had been steadily lowering boost from initial release bios to presumably pad the cpus for longevity. Thus this whole controversy is actually blown way out of proportion by Roman. He gave the other team fuel to troll and make this issue much bigger than it actually is.

 

[moproblems99]

Think about it this way. You are paying for 50/10 mbps internet but are only getting 45/1 mbps? Sure it won't really affect what you are doing, and 99% of people won't notice the difference, but that is not what you are paying for. AMD admitted there was a bug, and has seemed to have fixed it. We will be getting the specifications we paid for and will be happy.

Paying for 50/10 and getting 45/1 is bit different than paying for 4600mhz and getting 4550mhz.

Your scenario should have been like: Paying for 50/10 mbps and getting 49.5/9.9 mbps. Don't get me wrong, AMD should have had this figured out before launch but I think you are over dramatizing it just a tad.

 

[Assimilator]

Lemme get this straight...

AMD releases a phenomenal product that gives far more value than the competition at a far lower price...

And people moan and complain and cry oceans of tears because in some cases their CPU only performs at 95% of what it "should" (4.4GHz vs 4.6GHz) or, in the most ridiculous cases, boosts all of 50MHz lower than they believe it should...

FIFTY
WHOLE
MEGAHERTZ.

1/20th of a gigahertz. On a 4GHz CPU, less than 1% of the clock speed. Yet wars have been lost, worlds destroyed, for fifty megahertz... or so it would seem given the hue and cry, because we all know that more megahertz = more performance, right? Right?

Because there's never been a scenario where lower clocks over a period produce better performance than higher clocks causing intermittent throttling, right? Right?

This sort of uninformed, immature, infantile BS is exactly why companies stop listening to their customers. Because it teaches them that when they do, their customers take that opportunity to s**t on them. The net result is that the people who have useful feedback also get ignored, which benefits nobody.

Seriously kids, grow the f**k up. Good on AMD for addressing this for those who are insecure about the size of their manhood, but god damn.

I though the Agesa was released just for the board vendors, and the implementation into BIOS updates could take up to the 30th September. Is the Agesa available through AMD? Can you clarify where you got it from?

Tweaktown Gigabyte Beta BIOS forum.

 

[lexluthermiester]

Uhm, you've clearly followed this topic then...
I've gained 200MHz boost on the highest boosting core, but I guess that's not worth it to you...
Not long ago, none of my cores would clock over 4,400MHz, now the slowest ones boost to 4,550MHz, but again, I guess that's not worth it either?

To be fair, 200mhz out of 4400+mhz isn't enough to make a big deal about. That's less than 5%.

Im seriously surprised, cant wait til Der8auer calls out for another group test scenario again.

He's just trying to drum up business.

 

[Vya Domus]

Will be interesting to see if this is a geniune fix or AMD just relaxing the restrictions

Not sure there's a difference, that's how this works, all you do is modify the restrictions. There is no fix of sorts, the silicon is already shipped.

1/20th of a gigahertz. On a 4GHz CPU, less than 1% of the clock speed. Yet wars have been lost, worlds destroyed, for fifty megahertz... or so it would seem given the hue and cry, because we all know that more megahertz = more performance, right? Right?

People felt cheated apparently because it didn't coincide with the number AMD wrote on the box to the nth decimal point. I am not sure, I wrote pages of comments trying to figure out what this is about since this wont affect the user experience in any tangible way but got no definitive answer, people got really defensive.

They took our megahertz ! That's all I understood, unfortunately, they got what they wanted I guess.

 

[xorbe]

Am I the only one that installs the cpu, slaps in 3200 CL14, and just uses the PC as is?

 

[moproblems99]

1/20th of a gigahertz. On a 4GHz CPU, less than 1% of the clock speed. Yet wars have been lost, worlds destroyed, for fifty megahertz... or so it would seem given the hue and cry, because we all know that more megahertz = more performance, right? Right?

Because there's never been a scenario where lower clocks over a period produce better performance than higher clocks causing intermittent throttling, right? Right?

It will be interesting to see how all the before and after benchmark test thingies play out. Perhaps specific chips will see more of an increase, perhaps not?

That said, I don't understand why they (AMD) created this whole fiasco (I use that word very lightly) in the first place. If it made more sense to have the boost a little lower, then have them boost lower. But print that on the box and materials - don't put the higher numbers because you know damn well stuff like this would happen. I think what happened is they figured out dropping 50mhz or so from boost was able to get other metrics in a better position pretty late in the game. Then they didn't want to lower the numbers on the box by 100 because that would have been too close to the Zen+ numbers or they already had that shit printed.

Either way, this shouldn't have happened and it rubs me the wrong way because I am not sure if they are trying to be sneaky or are just incompetent. Neither are a good look. Ultimately, I bet it won't make much of a difference when it is all average out.

I think it would be awesome if @W1zzard would do one when official bioses come out. Hell, the beta ones will do as I imagine they will never leave beta status.

 

[W1zzard]

think it would be awesome if @W1zzard would do one when official bioses come out. Hell, the beta ones will do as I imagine they will never leave beta status.

Just waiting for asrock taichi version to come out

 

[Arjai]

Around here? You are going to be in the minority. Even I, a casual, have done some tweaking, based on the things I find here.