AMD Updates Ryzen Product Pages to Elaborate on "Max Boost Clocks"

[Darmok N Jalad]

My impression of “boost clocks” is that it is the theoretical maximum frequency the processor is allowed to reach, and it may only do so for split seconds. The observable, sustainable boost clock is somewhere less than that, but still significantly more than the all-core clock that is rated to stay within TDP. The boost clock is intended to handle brief workloads quickly, but you are still bumping into the core’s base design limitation. The boost clock is not a guaranteed stable and sustainable speed. This is what I have observed on any modern CPU I’ve owned, going all the way back to Westmere. I rarely have seen an Intel CPU get to the boost on a monitor program, but I have seen it more so with AMD. It’s just with Intel, you seem to be able to force the issue manually a whole lot easier.

 

[bug]

My impression of “boost clocks” is that it is the theoretical maximum frequency the processor is allowed to reach, and it may only do so for split seconds. The observable, sustainable boost clock is somewhere less than that, but still significantly more than the all-core clock that is rated to stay within TDP. The boost clock is intended to handle brief workloads quickly, but you are still bumping into the core’s base design limitation. The boost clock is not a guaranteed stable and sustainable speed. This is what I have observed on any modern CPU I’ve owned, going all the way back to Westmere. I rarely have seen an Intel CPU get to the boost on a monitor program, but I have seen it more so with AMD. It’s just with Intel, you seem to be able to force the issue manually a whole lot easier.

That's just the thing, everyone's going by their own impressions. Till now, CPUs used to reach advertised speeds, hold them, and could easily exceed them (even if a little OC was required).
Because of clever clocking (or whatever reason), the new Ryzens seem to have like no headroom at all and in the absence of a clear definition of what boost means, we're left scratching our heads because it seems neither impressions nor experience apply anymore.

 

[bogami]

I must point out that the presentation of the Turbo speed is very unrealistic because it looks like AMD is basically better which is misleading, if it has some intel good, it's OC capability and high quality silicone that makes it possible. What helps you get 60x12A or 14A of power (enough to fire an AMD processor four times) if the silicon can't handle the pressure and only added 100hz to its 100% smaller architecture, and not always. So the fact is that they founded the 3000 series for sale of waste cores !! That the waste kernels are being sold here and in order to achieve the better quality required they have delayed production, so the manufacturers are incapable of producing what they boasted to the customer and extort the product we have! Whether the R9 3950X is based on higher quality production is a question. And the TR4 shows good after the first results, but these are the north products that lead more because of the number of cores. I have been using the i7 3770K 22Nm at 4.8 Gh (all cores) since its release and it works smoothly and I still have some reserves. AMD has only one core at 4.8 Gh and hardly ever. I'm still going over 7Nm of the new AMD quad-core processors. So the 3000 series is a dumpster for AMD and how much of a fuss about these lies about raising the performance makes me seriously wonder if I will buy the R9 3950X, which has a kernel count and a little more reasonable price, everything else is conditional.

 

[Nordic]

Ok, I get it, you're bitter.
Try this: fire up SuperPi (single threaded) and see whether you reach the advertised speeds. If you don't, then yes, AMD has shafted you.

Going by that definition then I think every ryzen owner got shafted. I still get 4.2ghz, and only sometimes 4.3ghz when running single threaded benchmarks. I have not tried super pi yet though. I will report back.

Edit: I am having issues. I accidentally did something to hwinfo to make it only show the clock multiplier, not the clockspeed. When I open superpi it just sits there saying analyzing devices.

 

[Zareek]

See, I feel the opposite of most of the people complaining. When I read reviews and read the marketing materials, I thought the 3800X would only see all core speeds of 3.9GHz consistently and occasionally see boosts higher than that. So when I run a benchmark like Cinebench and see all cores at 4.2 to 4.3Ghz, I say wow I'm getting a 300-400Mhz overclock without doing a thing. When I run F1 2018 and all 8 cores are clocked at 4.3Ghz for a 6 hour gaming session it's the same thing, free performance boost. I will admit it is not always easy to catch the cores hitting 4.5Ghz but I've seen 2 at the same time go that high, single boosts to 4.55Ghz and frequently I see 2 to 4 cores hitting 4.4Ghz together. I wish I knew what factors allowed the higher clocks so I could align the stars myself but I don't feel like I got ripped off. I feel like the people who bought 3700X got a better deal because they are basically on par in 90% of usage cases. I paid almost 20% more to essentially get 2 to 3% better performance, I can live with that.

 

[TheLostSwede]

Ok, I get it, you're bitter.
Try this: fire up SuperPi (single threaded) and see whether you reach the advertised speeds. If you don't, then yes, AMD has shafted you.

I don't, at least not without PBO and XFR enabled with every auto OC setting maxed.
4.4GHz is the max I see otherwise.
Not bitter, I just feel like I wasted money compared to if I'd gotten a 3700X.
The lack of acknowledgement of this issue from AMD is what bugs me.

 

[biffzinker]

Ok, I get it, you're bitter.
Try this: fire up SuperPi (single threaded) and see whether you reach the advertised speeds. If you don't, then yes, AMD has shafted you.

7Zip benchmark works for the single core boost test.

 

[TheLostSwede]

See, I feel the opposite of most of the people complaining. When I read reviews and read the marketing materials, I thought the 3800X would only see all core speeds of 3.9GHz consistently and occasionally see boosts higher than that. So when I run a benchmark like Cinebench and see all cores at 4.2 to 4.3Ghz, I say wow I'm getting a 300-400Mhz overclock without doing a thing. When I run F1 2018 and all 8 cores are clocked at 4.3Ghz for a 6 hour gaming session it's the same thing, free performance boost. I will admit it is not always easy to catch the cores hitting 4.5Ghz but I've seen 2 at the same time go that high, single boosts to 4.55Ghz and frequently I see 2 to 4 cores hitting 4.4Ghz together. I wish I knew what factors allowed the higher clocks so I could align the stars myself but I don't feel like I got ripped off. I feel like the people who bought 3700X got a better deal because they are basically on par in 90% of usage cases. I paid almost 20% more to essentially get 2 to 3% better performance, I can live with that.

The bizarre thing is that you have a smaller cooler and a board that supposedly can't deliver as much power through its VRM, yet you're getting the correct boost speeds. Is this with any of the overclocking features enabled, or just default settings?

 

[tabascosauz]

Meh. I don't get the whole hubbub from the laymen. If the 3700X can stay at 4200MHz on not only one core but all the way up to 16T where it still stays at roughly 4200MHz, that's a win for me. I'm tired of shelling out premium money and still having a fair amount of performance locked outside my reach simply because Intel's products can't qualify under the company's own longstanding definition of TDP without trying to wriggle their way out into the pure comedy that is P-states. Ryzen 3000's sustained performance is impressive and on the money, that's what matters. All of the performance is there on the table for us who won't/can't bring in water to tame the comical power draw and heat output that said 9900K shits out at its advertised clock speeds. Not to mention that the 9900K still sits at that price, lmao.

That said, the lawyer in me also knows exactly what's coming - I mean, AMD made a bit of a blunder here. Lots of people still are fooled though; class action lawsuits accomplish exactly zero aside from generously lining the pockets of participating law firms. It'll be amusing to see the reactions of the vengeful "AMD has betrayed my trust" type as they realize that they don't actually stand to gain anything from the legal proceedings.

I suppose that AMD wanted to soften enthusiasts' ego damage upon seeing the clock speed hit from crossing 10nm, but even the advertised clocks are not that high. If they had just advertised a conservative 4200MHz for the 3700X for example, and touted efficiency a bit more and "just wait for the reviews, you'll see how good it is", they would have been just fine. Ryzen 3000 speaks for itself really; actually marginally better IPC than Intel at sane power and temperatures, so the 9900K's 5GHz boasting would still be smoke and mirrors.

Intel will have to go through the same "clock shock" when it gets down below 10nm, so I don't see what the big deal is.

 

[Zareek]

The bizarre thing is that you have a smaller cooler and a board that supposedly can't deliver as much power through its VRM, yet you're getting the correct boost speeds. Is this with any of the overclocking features enabled, or just default settings?

No overclocking still... This is likely to be the first CPU I've owned in over twenty years that I won't be overclocking. I don't really see much point in it, that might change but I doubt it.

 

[TheLostSwede]

No overclocking still... This is likely to be the first CPU I've owned in over twenty years that I won't be overclocking. I don't really see much point in it, that might change but I doubt it.

Not saying you should. What I'm saying is, your CPU behaves as expected with "lesser" hardware compared to mine. I should be seeing at least the same boost speeds as you, but I'm not...

Meh. I don't get the whole hubbub from the laymen. If the 3700X can stay at 4200MHz on not only one core but all the way up to 16T where it still stays at roughly 4200MHz, that's a win for me. I'm tired of shelling out premium money and still having a fair amount of performance locked outside my reach simply because Intel's products can't qualify under the company's own longstanding definition of TDP without trying to wriggle their way out into the pure comedy that is P-states. Ryzen 3000's sustained performance is impressive and on the money, that's what matters. All of the performance is there on the table for us who won't/can't bring in water to tame the comical power draw and heat output that said 9900K shits out at its advertised clock speeds. Not to mention that the 9900K still sits at that price, lmao.

That said, the lawyer in me also knows exactly what's coming - I mean, AMD made a bit of a blunder here. Lots of people still are fooled though; class action lawsuits accomplish exactly zero aside from generously lining the pockets of participating law firms. It'll be amusing to see the reactions of the vengeful "AMD has betrayed my trust" type as they realize that they don't actually stand to gain anything from the legal proceedings.

I suppose that AMD wanted to soften enthusiasts' ego damage upon seeing the clock speed hit from crossing 10nm, but even the advertised clocks are not that high. If they had just advertised a conservative 4200MHz for the 3700X for example, and touted efficiency a bit more and "just wait for the reviews, you'll see how good it is", they would have been just fine. Ryzen 3000 speaks for itself really; actually marginally better IPC than Intel at sane power and temperatures, so the 9900K's 5GHz boasting would still be smoke and mirrors.

Intel will have to go through the same "clock shock" when it gets down below 10nm, so I don't see what the big deal is.

The "hubbub" as you put it, is about the fact that AMD is marketing their products as something that's unachievable for a lot of their customers and doesn't provide any reasoning as to why it's happening.

Look, as I keep saying, I'm not upset with the performance, what bugs me is the fact that I paid extra for something that doesn't deliver what it says on the tin. In fact, I seemingly paid more for something that performs less, as there are 3700X chips that outperform my 3800X in benchmarks.

As far as my system goes, it's quiet, it runs cool (if you don't consider that the chipset hits 80C+ at times) and it's powerful, it's just not what I apparently paid for when it comes to the CPU.
Some people aren't having any issues, so there's nothing to complain about, but quite a lot of people are, so AMD needs to step up and explain what's going on and then do something about it.

 

[Vario]

A few in this thread claim that the AMD XFR is better than Intel's turbo. But they also claim Ryzen isn't boosting to the speeds it is marketed to, or it is not consistently. So what makes XFR superior to the Intel turbo? With the Intel, when I have load, the CPU boosts to the turbo multiplier I set in bios or to the preset Intel value if I didn't change the turbo multiplier or own a non K. When I don't have load, it downclocks. What is there to improve on this?

 

[londiste]

A few in this thread claim that the AMD XFR is better than Intel's turbo. But they also claim Ryzen isn't boosting to the speeds it is marketed to, or it is not consistently. So what makes XFR superior to the Intel turbo?

Technical details of the boost mechanics. Works faster and with smaller frequency increments.

 

[EarthDog]

A few in this thread claim that the AMD XFR is better than Intel's turbo.

Who claimed that?

Technical details of the boost mechanics. Works faster and with smaller frequency increments.

w31ght wut? I mean if true, yipee... but... it has to work for everyone.... like Intel's does. AMD's boost plays too many reindeer games and for some (dare I say many?) it doesn't reach its boost clock at all. I've run Super Pi (like someone suggested earlier) and set affinity to one core and it still didn't boost to what the package said. That core maxed out at 4.35 GHz, not 4.4 GHz. It's .5 multi short. While it isn't a big deal, it isn't doing what it is supposed to.

 

[einard]

What is your MB? Yeah PBO an autoOC does nothing, I am patiently waiting for the next BIOS revision, hopping for higher boosts.

B450 Gaming Plus. Which to my knowledge is identical to Tomahawk, slightly lesser VRM heatsinks.

Tried disabling PBO (instead of the default auto), had not effect. Tried using offset. First -0.1, didn't boot. Tried-0,5 and then I saw something very wierd.
The idling voltages are still 1.46ish. Frequency is down to 4.25, which also is my max all singlecore boost. Down from 4.275.
The wierd thing is that when I load up CB20, I get 1.26v and 3.95-3.975 on all cores (old was 3.9). So by setting a negative voltage I've -increased- my all core voltage and clock.
Whats even more wierd is that temperatures went down on the CB run. 60C. Which doesn't make sense at all since both frequency and voltages are higher..

Something isn't right at all. Early adopter issues hopefully. Luckily it's stable, so I don't mind waiting for some amd finewine bios/driver update. Atleast the -0.5 voltage lowered temps alot, so the fan isn't spooling up and down so much.

 

[Turmania]

I'm an old fashioned simple lad.when I buy something according to the box, I expect it to perform like it. You write 4.5 ghz boost I expect it to do so with all cores.I don't buy an 8 core processor just so 1 core reaches advertised speeds.it's shame and scam.

 

[Dragonsmonk]

I'm an old fashioned simple lad.when I buy something according to the box, I expect it to perform like it. You write 4.5 ghz boost I expect it to do so with all cores.I don't buy an 8 core processor just so 1 core reaches advertised speeds.it's shame and scam.

So you have not checked into any of the CPU's you had for the last.... 5 years +?

 

[EarthDog]

I'm an old fashioned simple lad.when I buy something according to the box, I expect it to perform like it. You write 4.5 ghz boost I expect it to do so with all cores.I don't buy an 8 core processor just so 1 core reaches advertised speeds.it's shame and scam.

Respectufully, it isn't so much old fashioned as you not knowing how the processors work.

It isn't a scam, that is for sure...






What is it with users at this site calling things scams that aren't even close to being one??? Do you people drink from the same lead infused water???

 

[Turmania]

Respectufully, it isn't so much old fashioned as you not knowing how the processors work.

It isn't a scam, that is for sure...






What is it with users at this site calling things scams that aren't even close to being one??? Do you people drink from the same lead infused water???

TDP ratings and all the turbo speeds at this point of time is shambolic, misleading and yes scam. Customers need better clarification from boxes and online retailers when buying a modern cpu these days. Something needs to be done about it.

 

[EarthDog]

TDP ratings and all the turbo speeds at this point of time is shambolic, misleading and yes scam. Customers need better clarification from boxes and online retailers when buying a modern cpu these days. Something needs to be done about it.

Nope and Nope. For Intel, it works as described... always. Ryzen 3.........not so much.

We'll agree to disagree. But education on the product you are going to buy is key.

 

[yeeeeman]

It seems that you are a part of those 99% aswell !

Most guys don't ever get close to max advertised boost clocks even when running single core loads with everything else closed .
The only way to hit advertised boost clocks is to move around the mouse when PC is idle , this is what AMD means for '' optimal conditions '' and this is where the marketing BS comes to play !

The debate here is to know if it is ethical from AMD to advertise clocks you can hit only when you play with your mouse at idle instead of advertising clocks you can hit under '' normal '' single core loads ?
The answer is hopefully obvious , hence why many peoples do rightfully so complain about not getting what they paid for !

Dude, you just proved my point. The heck are you talking about with mouse at idle shit? There are people who are hitting 4.6ghz with 3900x on single core...

What? Where does that definition get clarified? Only one core on the package being even capable and its luck of the draw whether you might use it? That sounds like... well I don't know, a whole lot of WTF.

It doesn't get clarified anywhere. It is just an opinion bases on real reasons why and cannot maintain their promise of boost clocks. The fact that they can't reach max boost single core clock on all cores is the logical explanation. Otherwise it doesn't make sense that when you are on idle and run a single cinebench run you don't get full boost clocks.

 

[TheMadDutchDude]

I haven’t seen mine hit it’s max frequency, and it’s under custom water cooling.

I’m not bothered. I run it at 4.2 on all cores because it requires 0.1v less than 4.3 and it makes my system so much quieter. You certainly don’t notice the 100 MHz lower clocks, but the noise drop is fantastic.

 

[hzy4]

I haven’t seen mine hit it’s max frequency, and it’s under custom water cooling.

I’m not bothered. I run it at 4.2 on all cores because it requires 0.1v less than 4.3 and it makes my system so much quieter. You certainly don’t notice the 100 MHz lower clocks, but the noise drop is fantastic.

what vcore do you run it at?

 

[TheMadDutchDude]

1.25v for 4.2 GHz. Not the best chip, as always. I’ve had no luck on the three Ryzen chips I’ve had thus far.

 

[EarthDog]

1.25v for 4.2 GHz. Not the best chip, as always. I’ve had no luck on the three Ryzen chips I’ve had thus far.

If that isn't the best... I've got a dud. It takes over 1.3V for me to run 4.25 GHz on a 3700X...temps break 90C in AIDA64 stress test (default) with Corsair H150i, lol.