Evildead666:

As long as it says "Up to" that's it. Internet bandwidth from ISP's. "Up to 100Mbit Internet". They guarantee the base clock, and then the boost clock is "Up to". No one would have a leg to stand on. I'm happy with my 3800X. I also believe that we are still a way away from the Final Bios versions, for any AM4 Motherboard. What we have today, is still not the final clocks etc. As for the "Review Bios", you'd have to prove that they deliberately held back the "proper" Bios, and I don't think that is going to be done either. I suppose when you don't have the deep pockets of Intel, you can't have the same sort of CPU launch.

https://www.amd.com/en/products/cpu/amd-ryzen-7-3700x#product-specs There is no mention of up to on this page at all and even the "max boost clock" the "max" part was added 15 days ago: https://www.reddit.com/r/hardware/comments/cp9zug/amd_has_updated_the_ryzen_product_pages_to_be/ AMD posted a video where they say you should be able to do 200mhz+ beyond the highest clock printed on the box. As it currently stands I can't even hit the box speed, I'm capped nearly 100mhz below it on a $380 motherboard with liquid cooling. Couple all that with every single board getting lower clocks on newer bios's compared to review/release ones and ASUS saying this and I don't really see how you can take it any other way. Again, it doesn't make the processor bad - but it needs to be called out so they stop doing it in the future.

Man i feel bad for everyone affected by this issue. And i agree with Denial, AMD should be called by this type of "issues" the same way we call other companies when they pull silly stunts like this.

Denial:

3900x and I've yet to see boost clocks or even similar clocks to what the reviewers were getting on initial bios. Do I care that much? Not really - chip still performs well and better than my 7820x - should AMD be called out on it though? Absolutely. The larger point here is exactly what loophole said - if any other company had done this people would be losing their crap but AMD does it and half the community just handwaves it away, some people trying to make it out as a good thing. As I said in the other thread it sets a precedent for AMD to get away with things like this as they have for years now.

Care to share some of your results?

Evildead666:

I'm happy with my 3800X. I also believe that we are still a way away from the Final Bios versions, for any AM4 Motherboard. What we have today, is still not the final clocks etc. As for the "Review Bios", you'd have to prove that they deliberately held back the "proper" Bios, and I don't think that is going to be done either. I suppose when you don't have the deep pockets of Intel, you can't have the same sort of CPU launch.

So am I, it mostly boosts to 4325MHz in games and the maximum single core boost I've seen is 4550MHz (if iCUE is disabled). I could also run an all-core OC @ 4400MHz to get an even boost in games.

Netherwind:

Care to share some of your results?

We don't need his results. Something like 96% of 3900x's don't hit their advertised boost.

Evildead666:

As long as it says "Up to" that's it. Internet bandwidth from ISP's. "Up to 100Mbit Internet". They guarantee the base clock, and then the boost clock is "Up to". No one would have a leg to stand on. I'm happy with my 3800X. I also believe that we are still a way away from the Final Bios versions, for any AM4 Motherboard. What we have today, is still not the final clocks etc. As for the "Review Bios", you'd have to prove that they deliberately held back the "proper" Bios, and I don't think that is going to be done either. I suppose when you don't have the deep pockets of Intel, you can't have the same sort of CPU launch.

If your defense is, "Hey, the ISPs do it," you've already lost.

Reardan:

We don't need his results. Something like 96% of 3900x's don't hit their advertised boost.

Oh, it's that bad...

Netherwind:

Care to share some of your results?

I'll do some extensive testing at some point. I just know that when I first got my x570-E with original bios, I saw within 25mhz of advertised boost - with the latest bios (at the time before the board died) that dropped to 125mhz away from advertised. I then switched to gigabyte aorus pro and now its roughly 100mhz. I don't stare at it or record it throughout the day but in my quick testing with a few different apps that's the behavior I've seen.

Reardan:

We don't need his results. Something like 96% of 3900x's don't hit their advertised boost.

3800x, the processor that both Netherwind and Evildead666 have on average are the best at hitting their advertised speeds - mine is the worst - feels bad =( Like I said it's not that big of a deal but when you give a little companies tend to take a lot. If these processors weren't capable of hitting those speeds reliably they should have never been advertised as that speed.

It seems like AMD has pushed their processors at launch to beat Intel in various benches get the sales an dial it back to avoid an avalanche of cpus that died prematurely... Bad form

Denial:

I'll do some extensive testing at some point. I just know that when I first got my x570-E with original bios, I saw within 25mhz of advertised boost - with the latest bios (at the time before the board died) that dropped to 125mhz away from advertised. I then switched to gigabyte aorus pro and now its roughly 100mhz. I don't stare at it or record it throughout the day but in my quick testing with a few different apps that's the behavior I've seen.

Sorry to hear that you've had problems man. An ASUS fan made something cool and modded a AGESA 1.0.0.3 BIOS with the 1.0.0.2 boost behavior. Hope everyone could get something like that in the end 🙂

Evildead666:

Go for it, see how far you get. Which Ryzen 3000 cpu did you get ?

I can buy any model [but best to buy the most expensive] and sue them, i bet there is squad of greedy lawyers that only look for a reason to sue anyone and ill be happy to get a small cut from it. Free money is free money. 🙂

Evildead666:

The X570 chipset wasn't designed. If you read a bit more, you would know that it is just the IO chip from the Ryzen chips themselves, because ASMedia didn't have the actual chipsets ready in time... Nothing was "squeezed in" to the chipset, it just doesn't have any power saving states, as it wasn't meant to be a motherboard chipset....so it runs pretty fast and hot.

They used the IO chip as basis but its semi-custom chip, and it can be easily proved by comparing the amount of PCIe lanes coming from CPU x24 vs Chipset x16

Denial:

3900x and I've yet to see boost clocks or even similar clocks to what the reviewers were getting on initial bios. Do I care that much? Not really - chip still performs well and better than my 7820x - should AMD be called out on it though? Absolutely. The larger point here is exactly what loophole said - if any other company had done this people would be losing their crap but AMD does it and half the community just handwaves it away, some people trying to make it out as a good thing. As I said in the other thread it sets a precedent for AMD to get away with things like this as they have for years now.

Agree 100%. I have the 3700X and it's never hit its boost clock either. My X570 board shipped with a newer bios from the get go. It's still a good chip and I'm not going to sell it and replace with an Intel chip, but they should have rated the chips lower so people would know what to expect. Bottom line, it's simply not a 4.4GHz chip. Every time I've brought up the boost clock issue, people jump all over me in defense of AMD. They shouldn't get a free pass because they are the good guy underdog. If anything, AMD has put a sour taste in my mouth after giving them a shot for the first time. Everything leading up to the launch painted a different picture. Thought we would see single cores boosting 200MHz over max boost, but it turns out they can't even hit max boost for most users. Then it became "ideal conditions" are necessary to hit max boost. They did what they did to have a good launch and they succeeded. We'll see what AMD responds with, but they've been pretty good at talking in circles so far and the faithful have no problem just taking their word for it. Maybe I'm just experiencing a bad case of the "observer effect".

I find this a bit suspicious - why would AMD suddenly care about this now, or at least not make a statement about it? Do they seriously think nobody is going to notice that their CPU doesn't reach it's advertised boost clocks, even if given ample cooling and power delivery? Something isn't adding up here. Also from what I recall, people expressed concerns about the high boost voltages, where AMD was like "eh it's not problem". So... did they change their mind?

fantaskarsef:

I wonder what they mean in terms of longevity. Will the CPU die withing 2 years with the more aggressive boost? Within 5? Hard to know if one wants the higher clocks or not with not knowing the facts or at least, the expectations AMD puts into this.

Primarily stability. A CPU doesn't typically just suddenly die when you push it too hard, but errors start to crop up more often. I have an old 6-core CPU that was once overclocked to 4.7GHz. It's in a BOINC rig, running at 100% 24/7 (during the winter), and over time its reliability degraded. It's now at I think 4.3GHz to remain stable. I predict by next spring, it'll be around 4.0.

Not an issue and not a surprise since we all knew Ryzen 2 is pre-overclocked thus reduced OC headroom thus silicon lottery uncertainty area closer. 10? 50? 100 MHz away from stated max boost? - will that level of variance be felt by anyone but the most anal user? Should Marketing be tested?

sverek:

Boost is automatic overclocking, it does work out of the box. Basically AMD is already overclocking CPU for us, no much gain by overclocking it manually. Still, would be awesome to have an option to "disable" this safe feature and go for higher clocks.

Well, here is another OC option. Manually setting all cores to some clock at something like 1.3~1.35V. That surely is nice OC for productivity. And one can set custom power states to keep CPU downclocking when unused. (That's if it works on new Agesa.)

Mesab67:

Not an issue and not a surprise since we all knew Ryzen 2 is pre-overclocked thus reduced OC headroom thus silicon lottery uncertainty area closer. 10? 50? 100 MHz away from stated max boost? - will that level of variance be felt by anyone but the most anal user? Should Marketing be tested?

Considering how many people here think paying hundreds of dollars extra for a 6% improvement in games is worth it, losing up to 100MHz after a BIOS update is probably enough to make them swear upon their mother's grave to never shop AMD again.

schmidtbag:

Considering how many people here think paying hundreds of dollars extra for a 6% improvement in games is worth it, losing up to 100MHz after a BIOS update is probably enough to make them swear upon their mother's grave to never shop AMD again.

Is this 6% consumed completely/significantly by a 50MHz reduction? Is the "hundreds of dollars extra" consumed completely/significantly by a 50MHz reduction? Are we over-obsessing to a degree?

Mesab67:

Is this 6% consumed completely/significantly by a 50MHz reduction? Is the "hundreds of dollars extra" consumed completely/significantly by a 50MHz reduction? Are we over-obsessing to a degree?

Is it over obsession or just a desire to keep these companies in check? Look at the mobile industry - where SoC vendors were creating specific profiles so that when benchmarking apps were detected the processor would go to a boost state that isn't normally available. It wasn't until Anandtech called them out repeatedly that it stopped. How about Nvidia, with the 970 fiasco - 3.5GB+.5GB partition - only effected some games and even in those games it created slightly more stutter than you'd get hitting the 4GB barrier anyway.. class action suit and tons of negative press later and I'm willing to bet Nvidia doesn't make that mistake again. Same thing needs to happen to AMD or they'll just keep doing it.

Mesab67:

Is this 6% consumed completely/significantly by a 50MHz reduction? Is the "hundreds of dollars extra" consumed completely/significantly by a 50MHz reduction? Are we over-obsessing to a degree?

Yes, we are over-obsessing. That's pretty much my point - people are willing to spend 60% more money for a 6% performance increase (even if only in specific workloads), and people are willing to fight tooth an nail over a 3% performance loss in specific environments. Neither makes sense as far as I'm concerned. But...:

Denial:

Is it over obsession or just a desire to keep these companies in check? ... Same thing needs to happen to AMD or they'll just keep doing it.

This is a good point. If AMD really has tried to be sneaky by silently limiting the boost clocks (I'm still not convinced this is their decision), shame on them, and they need to be called out for it. So far, I'm having a hard time believing if that is the case. Like I said before, they can't possibly be dumb enough to sell a product that is currently only accessible and appealing to enthusiasts, and think that those people aren't going to notice that they can't reach boost clocks, regardless of what they have. I'll give them a pass for their "up to" boost clock claims, where it's up to the user to provide the appropriate conditions to achieve them. But the thing is, if AMD really did make this tweak, does that not mean they changed the definition of what the CPU boosts up to? At that point, it would still be false advertising, especially if they didn't announce it (which, according to this article, they haven't). Unless there are reports of many of these CPUs prematurely failing, I just don't see how they could possibly think limiting the boost clock is a good idea, and, one they can get away with. Seems like doing this is just asking for a lawsuit.

Mesab67:

Is this 6% consumed completely/significantly by a 50MHz reduction? Is the "hundreds of dollars extra" consumed completely/significantly by a 50MHz reduction? Are we over-obsessing to a degree?

I don't think so. That's like a gas station not making sure their pumps are regulated to dispense the right amount of gas. Maybe you're only losing 10 miles/KM's on a full tank which is not super noticeable, but it's still not what you paid for. Worst part is, this isn't AMD making a regulation mistake. They knew about this at launch and talked around it. Intel is about to release their 9900KS with a 5.0GHz all core bin. Do you think people will let them slide if it doesn't hit the mark. Nope. AMD messed up. There's enough data to support that and this will be expanded upon. Ryzen 3000 is a very good series, but it was overrated for sales purposes. They should be called out and hopefully they owe up to it instead of talking in circles around the issue.

K.S.:

It's different with PBO, Turbo etc. There are pre-encoded frequencies embedded in the microcode instructing the CPU what to clock at. If I say Boost is (4.5) My marketing material referring to the max boost, one can easily correlate that with what is encoded into the micro-code for the cpu. Specific frequencies. Internet Service is literally a situation that relies on "Loop-Qualifications" and or varying factors beit the type of transport method used to deliver service, attenuation can vary in depending on that. Once settled in that area - you have a "Set Range" provisioned on the line that is the amount of Bandwidth the customer can "Saturate" - Up to the max because of generally a combo of environmental, networking & just acts of god factors.

PBO has a big disclaimer (That I havent read) so i expect that to be out of bounds of any official specifications or boost clocks. As for Internet ISP stuff, i would just like to say that after years of telling the ISP's that the line was bad, and having only 1,5Mbit where my neighbour has 3.3Mbit, was fruitless. At 1.5Mbit ( at best, usually 1Mbit ) Netflix and Prime allow 140p. I dare you to look at a Youtube video in 140p and tell me its watchable on any normal screen, even a TV. Then the ISP recently (6 Months or so) put Fibre to the end of the road, and I have 20Mbit Max now. Its not possible to adequetly describe with words. 🙂 or Emojis (First World Problem, I know, but Mind Blowing none the less.)

MegaFalloutFan:

I can buy any model [but best to buy the most expensive] and sue them, i bet there is squad of greedy lawyers that only look for a reason to sue anyone and ill be happy to get a small cut from it. Free money is free money. 🙂 They used the IO chip as basis but its semi-custom chip, and it can be easily proved by comparing the amount of PCIe lanes coming from CPU x24 vs Chipset x16

Binning ?