Review AMD's New Binning Strategy on Ryzen 3000: Core-by-Core Turbo Analysis

[TJ Hooker]

My thoughts exactly, expect the class action law suits to be filed shortly. This sounds an awful lot like the Nvidia 970 graphic card debacle where Nvidia knew they were making false claims. Maybe TH should rethink their AMD 3000 series recommendations.

Those recommendations were presumably based on the price and performance (as measured by TH in their benchmarks) of the CPU. Given that neither of those have changed, why do you think they should change their recommendations?

 

[redgarl]

I will not pretend to anything but here is my point of view. If AMD starts with biased or double standard specs whether it had been done in the past or not then the company will not see my as a customer.
Transparency is prime gold so uttershittery components does not happen down the road (fast).

I thank you very much Tom's hardware people for test, research, and honesty.
I was considering 3950x but in lights of this article, it's a cold shower enough to wait for i9 10k series and see what comes up. After all i9 9900k is not that far away in spite of "Old tech".

Honesty? You just buying it?....!!!

 

[jimmysmitty]

Ah cmon... so far, only Toms did mention anything about this and the BIOS are in shamble. I would wait a month for things to settle a little bit. We are talking about the launch of a new uarch and a new chipset. Did toms tested this with x470/x370 or b450/b350 motherboards? Answer is no.... BUT OMG, TOMS IS RIGHT huh... Remember the 3200MHz RAM speed on x370 boards at launch? And now look at it today, I am at 3333MHZ on a 1700x.

If AMD came with those numbers it is because they came to those numbers in their labs. Like you said, it would be suicide to advertise false specs.

By the way, that old article is even more relevant to show how much Toms article are similar to INFLUENCERS videos nowadays. I don't trust anything here anymore, but it does give me a second opinion.

If AMD verified the findings then it is legitimate findings. BIOS updates almost never change potential clock speed unless there was a major issue and if so AMD would have been better off postponing the launch to fix said issue. However since they did verify the findings and launched the product my guess is they got it to where they could and the only improvements will be seen from steppings and process improvements.

I am not saying a suit should be filed, I disagree with them. I disagreed with the class action and fines against Microsoft, Windows is their product and they should be able to include basic applications so long as users can still install third party applications. I even disagree with the one about AMDs Bulldozer although I understand the reasoning behind it and how it could be construed as such anyone who purchased the product should have done real research and would have found that the design was different than most other core designs at the time.

TH tested with the best available products, a very expensive board the MSI X570 Godlike, for the CPU. I would think the one with the best potential for clocking the highest would be an over priced board with the chipset designed for said CPU rather than an older one designed for the CPU before it.

And the article has no bearing on this at all. Was it a bad article? Damn right it was. But like most major publications eventually people get in that are not quite fit for the market they are in. It, however, does not invalidate this claim. This is not coming from just TH but pretty much every site is having issues hitting specified boost clocks and even well known OCers are hitting walls with clock speeds and OCing.

 

[germanium]

Makes me glad I stuck with Intel. At least when I set my 9900k to an all core over clock it is capable of running all cores at that speed no problem.

I found it to actually more stable than at stock speed as it tended to undervolt while still running max clock on single core with stock settings on my motherboard. Sometimes as low as .667 volts while single core running at 5GHz. I'm running 5GHz all core boost at 1.35 volts with hyperthreading or 1.30 volts without hyperthreading.

 

[Soaptrail]

My thoughts exactly, expect the class action law suits to be filed shortly. This sounds an awful lot like the Nvidia 970 graphic card debacle where Nvidia knew they were making false claims. Maybe TH should rethink their AMD 3000 series recommendations.

Why would Tom's rethink the recommended CPU for the 3000 Ryzen's? They based their recommendations on the testing they did and not the advertised speeds.

 

[Gurg]

If AMD verified the findings then it is legitimate findings. BIOS updates almost never change potential clock speed unless there was a major issue and if so AMD would have been better off postponing the launch to fix said issue. However since they did verify the findings and launched the product my guess is they got it to where they could and the only improvements will be seen from steppings and process improvements.

I am not saying a suit should be filed, I disagree with them. I disagreed with the class action and fines....

If your product will only perform at 4.1-4.2, then it is false advertising to claim 4.4-4.6. This would not be an issue unless AMD had not claimed better performance than the vast majority of these CPUs could deliver. If it was just limited to a handful of CPUs then the manufacturer should offer an exchange.

I was given a full refund on the Nvidia 970s I purchased which I rolled over toward the purchase of two 980s after the issues were identified. If you are unhappy with the AMD 3000 series CPUs then most retailers should give a refund within at least 30 days of purchase as the product does not perform as represented. If you keep it now that this is known then your refund options should be limited.

TJHooker said: "Those recommendations were presumably based on the price of the CPU and the performance (as measured by TH in their benchmarks). Given that neither of those have changed, why do you think they should change their recommendations? "
Soaptrail said: "Why would Tom's rethink the recommended CPU for the 3000 Ryzen's? They based their recommendations on the testing they did and not the advertised speeds."

TH and any reviewer that posted the AMD higher 4.4-4.6 claim should edit in a revision/disclaimer stating that there are problems with that claim and a link to this article. That is a fair and responsible disclosure.

 

[jeremyj_83]

If your product will only perform at 4.1-4.2, then it is false advertising to claim 4.4-4.6. This would not be an issue unless AMD had not claimed better performance than the vast majority of these CPUs could deliver. If it was just limited to a handful of CPUs then the manufacturer should offer an exchange.

How is it false advertising. They state a base clock and a boost clock. The only clock speed that you are guaranteed to get is the base clock.

 

[Giroro]

AMD shouldn't be allowed to sell 4.35 GHz chips as 4.4 GHz.
If we let them get away with it, then what's stopping them from marketing the same silicon as having a 5GHz boost, or why stop there, 10 GHz?

Look at how misleading the storage industry is. The last thing I want is for CPU makers to to feel comfortable inventing numbers vaguely in the ballpark, then counting on customers not knowing any better.

 

[TJ Hooker]

How is it false advertising. They state a base clock and a boost clock. The only clock speed that you are guaranteed to get is the base clock.

That's not how boost clocks have worked in the past. Yes, there may be some cases where you may not reach the maximum boost (or may not reach it for very long) depending on the cooling/power of your platform (moreso for mobile chips), but I think this is the first time we've seen the CPUs themselves incapable of reaching their rated boost clocks.

 

[jeremyj_83]

That's not how boost clocks have worked in the past. Yes, there may be some cases where you may not reach the maximum boost (or may not reach it for very long) depending on the cooling/power of your platform (moreso for mobile chips), but I think this is the first time we've seen the CPUs themselves incapable of reaching their rated boost clocks.

It is possible that it hits 4.4GHz but the pulling time is to long to register it, ie hits 4.4GHz for all of 30ms. When I think of boost clocks I look at it like what your ISP says for download speed, up to 1Gbps. If you are only getting 900Mbps they are covered since it says the magical "up to." In the Ryzen 3000 series perhaps you need more than a H115i to get the max boost speed. The fact that AMD is even able to get to 4.35GHz is surprising since Intel can only get to 4.1GHz on their 10nm but 14nm can hit 5.0GHz. Intel's 10nm and TSMC's 7nm are basically equal to each other in terms of density. I know it sounds like I am trying to be an AMD apologist, that isn't the case at all. I do wish that AMD would state something like Intel's Precision Boost 3.0 for the single core.

 

[TJ Hooker]

It is possible that it hits 4.4GHz but the pulling time is to long to register it, ie hits 4.4GHz for all of 30ms. When I think of boost clocks I look at it like what your ISP says for download speed, up to 1Gbps. If you are only getting 900Mbps they are covered since it says the magical "up to." In the Ryzen 3000 series perhaps you need more than a H115i to get the max boost speed. The fact that AMD is even able to get to 4.35GHz is surprising since Intel can only get to 4.1GHz on their 10nm but 14nm can hit 5.0GHz. Intel's 10nm and TSMC's 7nm are basically equal to each other in terms of density. I know it sounds like I am trying to be an AMD apologist, that isn't the case at all. I do wish that AMD would state something like Intel's Precision Boost 3.0 for the single core.

Sure, I suppose it's possible it's hitting max boost clocks for periods of time too brief to observe, but again, that's never how it's worked in the past. With regards to needing an H115i, that's the exact cooler they used in this article and still failed to hit max boost clocks.

The technical considerations of hitting X GHz on a Y nm node are irrelevant to the consumer buying a CPU.

 

[mlee 2500]

Okay, the CPU have been released for a mere 3 weeks. How do we know it is not related to BIOSes? You lost all credibility with Just Buy It!

Because AMD confirmed it? They make the Ryzen CPU.

 

[mlee 2500]

I ran a 6 minute mile once, about a decade ago, so I still tell everyone that I can.

 

[jimmysmitty]

It is possible that it hits 4.4GHz but the pulling time is to long to register it, ie hits 4.4GHz for all of 30ms. When I think of boost clocks I look at it like what your ISP says for download speed, up to 1Gbps. If you are only getting 900Mbps they are covered since it says the magical "up to." In the Ryzen 3000 series perhaps you need more than a H115i to get the max boost speed. The fact that AMD is even able to get to 4.35GHz is surprising since Intel can only get to 4.1GHz on their 10nm but 14nm can hit 5.0GHz. Intel's 10nm and TSMC's 7nm are basically equal to each other in terms of density. I know it sounds like I am trying to be an AMD apologist, that isn't the case at all. I do wish that AMD would state something like Intel's Precision Boost 3.0 for the single core.

Well known overclockers using better cooling still can't hit the speeds.

I think its an issue in the design or process.

Until we see it all I know about Intels 10nm is it is more dense than TSMCs 7nm. Who knows until it launches (if we even get it on desktop). They are very different process tech even in basic design. Intel was going hard on it where most others have taken the easiest way to get to the next nodes.

I will wait to judge Intels 10nm when we have full product lineups just as I did for Ryzen. All we have for anything there is rumors and speculation.

That said, the best thing AMD could have done is limit the boost clock rated speeds on the products. That way if it was getting higher than that it would have been more praise instead of what it is now which is controversy. I wouldn't be surprised if their marketing team got them into this as marketing teams are really good at that.

 

[Ncogneto]

Yawn.............................

Much a do about nothing. If your buying a CPU for advertised clock speed, your a moron anyways. You buy it for how well it performs, end of story. If you wanted clock speed, buy the Bulldozer.

IF Tom's was smart the would test about a 1/2 dozen of the same AMD processor and test them all in an identical test bed. If the variances between them are significant, then you have an issue.

 

[jimmysmitty]

Yawn.............................

Much a do about nothing. If your buying a CPU for advertised clock speed, your a moron anyways. You buy it for how well it performs, end of story. If you wanted clock speed, buy the Bulldozer.

IF Tom's was smart the would test about a 1/2 dozen of the same AMD processor and test them all in an identical test bed. If the variances between them are significant, then you have an issue.

Man have I got news for you. There are countries where something as little as "advertised clock speed" not meeting would be a big problem. Where consumers have vastly more rights than corporations.

It is an issue. Is it a major one? No. But everyone who buys a CPU buys on reviews and stated specifications. What it they stated 8 cores but it was really 6 cores yet performed just as well? Would you still be OK that you got a 6 core CPU that was claimed to be 8 cores?

 

[gerardfraser]

Not all cores in AMD's Ryzen 3000-series processors can reach the boost frequency.

AMD's New Binning Strategy on Ryzen 3000: Core-by-Core Turbo Analysis : Read more

The article is misleading and really not 100% accurate for 3600X.

I have a 3600X which can hit 4525Mhz on all cores and up to 3CCX or 6 cores same time.
Hell I did a short single thread run and 3600X Hitting 4500Mhz Cinebench 20 and over 4400Mhz Max boost on 3CCX in a minute video.

I could be wrong or misunderstood the article or did not read what it was trying to say but I disagree with some areas where it is not true for me.

 

[MadmanRB]

I will not pretend to anything but here is my point of view. If AMD starts with biased or double standard specs whether it had been done in the past or not then the company will not see my as a customer.
Transparency is prime gold so uttershittery components does not happen down the road (fast).

I thank you very much Tom's hardware people for test, research, and honesty.
I was considering 3950x but in lights of this article, it's a cold shower enough to wait for i9 10k series and see what comes up. After all i9 9900k is not that far away in spite of "Old tech".

Sure let a website well known for its intel bias in the past influence you.
Its well known Tomshardware has been a intel shill for like forever, especially when they made that article not so long ago where they made no mention of the new AMD chips and only featured intel processors.
Plus I must remind everyone this is early days for the zen2 platform so these early test results only tell a small story.

 

[TJ Hooker]

The article is misleading and really not 100% accurate for 3600X.

I have a 3600X which can hit 4525Mhz on all cores and up to 3CCX or 6 cores same time.
Hell I did a short single thread run and 3600X Hitting 4500Mhz Cinebench 20 and over 4400Mhz Max boost on 3CCX in a minute video.

I could be wrong or misunderstood the article or did not read what it was trying to say but I disagree with some areas where it is not true for me.

Yes, I'm afraid you seem to have misunderstood the article. The fact that your CPU can hit advertised boost clocks on any core does not change the fact that some can't (or hit them on even any single core).

Also, in your video I'm only seeing 1 core hit 4525 MHz. In the description you say "Set Override in Bios", do you mean you turned on PBO?

 

[TJ Hooker]

Sure let a website well known for its intel bias in the past influence you.
Its well known Tomshardware has been a intel shill for like forever, especially when they made that article not so long ago where they made no mention of the new AMD chips and only featured intel processors.
Plus I must remind everyone this is early days for the zen2 platform so these early test results only tell a small story.

Why do you even visit the site if you think TH is just a bunch of Intel shills?

 

[redgarl]

As of now, it is all a question of BIOSes... this piece is a joke. You have nothing except... the result on the same AND SINGLE x570 board.

How do you know it is not the board that is the limiting factor? I will tell you, YOU DON'T!

View: https://youtu.be/27sm05s6QSk


This guy was able to get to 4.55GHZ on Single core with a 3900x with a x370 Gaming Pro Carbon. So here goes your whole article...

 

[MadmanRB]

Why do you even visit the site if you think TH is just a bunch of Intel shills?

Because tomshardware used to be good, less bias in the past.
unfortunately its still searched a lot, and tomshardware seems to be going for clickbait rather than anything else.

 

[TJ Hooker]

As of now, it is all a question of BIOSes... this piece is a joke. You have nothing except... the result on the same AND SINGLE x570 board.

How do you know it is not the board that is the limiting factor? I will tell you, YOU DON'T!

View: https://youtu.be/27sm05s6QSk


This guy was able to get to 4.55GHZ on Single core with a 3900x with a x370 Gaming Pro Carbon. So here goes your whole article...

So you criticize TH for using a sample size of one, but then use a sample size of one as your own evidence?

Also, the rated boost clock for a 3900X is 4.6GHz, so your example is actually just another instance of the CPU failing to hit max boost clocks...

Edit: This article isn't 'no Ryzen 3000 will hit max boost on any core'.
It's 'Ryzen 3000 may be capable of hitting max boost clocks on only some of its cores (if any)'.
So even if there cases out there where a Ryzen 3K will hit max boost on any/all of its cores, that doesn't necessarily negate or disprove anything in this article.

 

[MadmanRB]

So you criticize TH for using a sample size of one, but then use a sample size of one as your own evidence?

Also, the rated boost clock for a 3900X is 4.6GHz, so your example is actually just another instance of the CPU failing to hit max boost clocks...

Yes but you could go for articles that use multiple samples too, you can bet tech jesus might try this out or Jay.

 

[TJ Hooker]

Yes but you could go for articles that use multiple samples too, you can bet tech jesus might try this out or Jay.

What is being shown in this article is pretty consistent with what I've seen in other reviews. Der8auer had like a dozen Ryzen 3K chips and none of them were hitting rated boost clocks.

Edit: And as someone already mentioned in this thread, the fact that virtually every review is finding that max all core OC is at least 100-200 MHz less than max boost also supports the finding that not all cores in every chip are capable of running at max boost.