AMD Ryzen 9 3950X Overclocked To 5 GHz Across All 16 Cores On LN2

Status
Not open for further replies.
The context is that two records are on Cinebench, which is a very favorable benchmark for AMD RyZen, and the third record is on a toy bench, where the 3950X beats by only 7% a CPU released in 2017.

It’s almost as if the way records work is they beat past records...

How many more records do they have to beat until it counts? 5? 7?
 
What I am saying is that the records are far from impressive when one takes them in context. Next quarters will be very funny because I will be watching how people begin looking for excuses when the product is reviewed and far from the current hype.

Does that mean we'll get a break from your incessant yapping if it's not far from the current hype?

Let's nail down some numbers here, what's "far"? Give a percentage, so we'll know exactly what you mean, no weasel word, mealy mouthed bs.
 
The screeenshot shows this isn't an ES.

I don't see anything in the screenshots wccftech has in the article that indicates it isn't a ES. The screen shot of CPU-Z actually leans towards it being a ES. The Screen shot of CPU-Z Shows it as "AMD 100-000000033-01" for the specification (Cinabench shows the same in the video). If it wasn't a ES, it would show it Similar to Ryzen 2700x . CPU-Z shows the 2700x as "AMD Ryzen 7 2700X - Eight-Core Processor" for the specification. (possible this is a limitation of CPU-Z, but who knows) Even the Model number is a 1 in the screen shot, where a 2700x is 8. Considering it won't be released for another 3 months, it is more than likely this was all done on a ES, as it is VERY, VERY unlikely they are handing out full retail versions 3+ months before they are even available.
 
You also have to acknowledge that this thread is about a "mainstream platform" CPU. Whether the price is mainstream or not is irrevelant.
In the end price is the ultimate factor as far as buying something but again we are talking about "mainstream platform" here. Keyword, Platform.

This thread is about the R9 3950X. And to me this $749 CPUs is not mainstream, independently of it shares the same socket than some $90 CPU.

About the HEDT low core model from Intel, are you talking about the 7740x by chance??? If so, then that CPU is a HEDT CPU without the benefits that the HEDT platform features lol!

No. I was talking about something like the 7800X.
 
This thread is about the R9 3950X. And to me this $749 CPUs is not mainstream, independently of it shares the same socket than some $90 CPU.

What it is "to you" is irrelevant. It is a CPU, on a mainsteam platform, with dual-channel memory (admittedly very fast), beating out a significantly more expensive CPU on an HEDT platform using quad-channel memory. You're just pissy that AMD beat some records. If this was an Intel CPU on an Intel mainstream socket you'd be all happy and claiming victory. You are trying everything you can to spin this into something else and its pretty bloody obvious.
 
I don't see anything in the screenshots wccftech has in the article that indicates it isn't a ES. The screen shot of CPU-Z actually leans towards it being a ES. The Screen shot of CPU-Z Shows it as "AMD 100-000000033-01" for the specification (Cinabench shows the same in the video). If it wasn't a ES, it would show it Similar to Ryzen 2700x . CPU-Z shows the 2700x as "AMD Ryzen 7 2700X - Eight-Core Processor" for the specification. (possible this is a limitation of CPU-Z, but who knows) Even the Model number is a 1 in the screen shot, where a 2700x is 8. Considering it won't be released for another 3 months, it is more than likely this was all done on a ES, as it is VERY, VERY unlikely they are handing out full retail versions 3+ months before they are even available.

I gave some background on what is an engineering sample and what is a qualification sample in this old post. Engineering samples begin with a number indicating the generation. E.g. 2S1404E2VJUG5_20/14_N is a second gen engineering sample for a 64C LP server Zen2 chip (Rome), whereas 5D0108BBM8SH2_37/34_N is the first gen of an engineering sample for a 12C desktop Zen2 chip (Matisse). Qualification samples begin with a "Z", as ZS1711E3VIVG5_24/17_N, which is the QS associated to the above second gen ES server chip.

As CPU-Z shows the overclocked chip is a "AMD 100-000000033-01". This is a preproduction sample.

An engineering sample of the R9 3950X would have the next string #D2113BBMGPH2_43/33_N with # being a number.
 
Last edited:
I gave some background on what is an engineering sample and what is a qualification sample in this old post. Engineering samples begin with a number indicating the generation. E.g. 2S1404E2VJUG5_20/14_N is a second gen engineering sample for a 64C LP server Zen2 chip (Rome), whereas 5D0108BBM8SH2_37/34_N is the first gen of an engineering sample for a 12C desktop Zen2 chip (Matisse). Qualification samples begin with a "Z", as ZS1711E3VIVG5_24/17_N, which is the QS associated to the above second gen ES server chip.

As CPU-Z shows the overclocked chip is a "AMD 100-000000033-01". This is a preproduction sample.

An engineering sample of the R9 3950X would have the next string #D2113BBMGPH2_43/33_N with # being a number.

It is more than likely this is just another ES chip that is not following their traditional id method. This chip, (AMD 100-000000033-01). Has a base clock of 3.29 Ghz and a max turbo of 4.29 Ghz.

https://browser.geekbench.com/v4/cpu/13495867

Then you have this:

"However, equally we must note that the benchmarked AMD chip could be an early (engineering) sample of the 3950X, as the clock speeds are running slightly slower than we’ve been promised by AMD, with the base clock at 3.3GHz rather than 3.5GHz, and boost to 4.3GHz (as opposed to a 4.7GHz maximum, albeit that won’t be across all cores).

And that means the finished AMD product which launches in September could potentially be even faster, and might just about equal the 9900K even in single-core performance. Note that the 9900K is an 8-core chip, and can be had at the time of writing for around $500 (around £395, AU$725)."


https://www.techradar.com/amp/news/amd-ryzen-9-3950x-leak-shows-it-dominating-an-18-core-intel-cpu

So, which is more likely: you are correct, it isn't a ES chip, as you claim, and AMD has given the world false specifications for their new flag ship desktop processor, Or you are wrong and it is an ES chip?

Which brings us back to my original statement: We don't know if this is an ES chip or not.
 
Last edited:
damn, thats freakin awesome. How much it will be able to do under water, I wonder!

My hope would be 4.7GHz across all cores since this is the turbo frequency.
 
My hope would be 4.7GHz across all cores.

I believe my Ryzen 2700X only does 4.1 across all cores manual OC where in auto mode it single core boosts to 4.3. Maybe I could get to 4.3 with lower ram (2933 vs 3200) but haven’t tried it.

Based on my limited experience I’d be happy/expect 4.5/4.6 under water. That’s a lot of cores and more dense die.
 
I believe my Ryzen 2700X only does 4.1 across all cores manual OC where in auto mode it single core boosts to 4.3. Maybe I could get to 4.3 with lower ram (2933 vs 3200) but haven’t tried it.

Based on my limited experience I’d be happy/expect 4.5/4.6 under water. That’s a lot of cores and more dense die.

How to you take your desktop underwater without it shorting out?
 
When you yourself define what is, and what is not, "mainstream", you know you can win ALL arguments. Well played sir.

Some of us have the bad habit of thinking by ourselves. The marketing department of X company can began tomorrow to say that a $10000 chip is mainstream. I am not going to bite...

It is more than likely this is just another ES chip that is not following their traditional id method.

Because you say so.
 
Some of us have the bad habit of thinking by ourselves.
What must a processor cost in order to be declared as "HEDT" by you?
Also, is there a price of a motherboard that determines it, too, is HEDT rather than mainstream?
And finally, how often do you re-calculate these numbers?
 
What must a processor cost in order to be declared as "HEDT" by you?
Also, is there a price of a motherboard that determines it, too, is HEDT rather than mainstream?
And finally, how often do you re-calculate these numbers?

He doesn't realize that budgets are different for different people.....
 
Because you say so.

Nope! The information out there supports it. The speed of the chip you are claiming is not an ES sample is 6% slower base clock, and 9% slower boost clock then what AMD has released as the specs of the 3950x (I guess AMD could be lying and is prepping themselves for massive lawsuits). Then you add in that they are getting near 6Ghz on ES chips of the 3900x as stated here from Gamer's Nexus at 2:13:



Again, nearly 6Ghz on an ES sample. I guess adding 4 cores could cause the LN2 overclocking to drop 15% when comparing a non ES sample of the 3950x to a ES of the 3900x. But that really doesn't sound very realistic and doesn't make a lick of sense, except maybe to you.

Basically EVERYTHING but your opinion points to this being an ES of the 3950x, which your opinion is being based off the chip ID number, that AMD has full control over and can change/use what every they chose, even if it isn't in line with previous id formats. And they don't have to get your approval to do so.

But if you want to claim that It's just because "I say so" and ignore all the other facts out there; GO FOR IT! But grab a towel, because there is a high possibility you will need it to wipe the egg off your face come September, when we start seeing the real results of a non ES chip of the 3950x rather than these results here.

Nothing you have said change the fact that we really don't know if the results here are based off a ES or production/retail chip. But the facts support that it is more than likely an ES chip of the 3950x, other than the id label per your explanation. So, go ahead, keep your eggs in that basket. One that is only held together by that ID number because all the other information available supports that you are most likely wrong about it not being an ES chip. But, there is the slim chance, you are correct. Hence, why I said, WE REALLY DON'T KNOW, which has been my position all along, but you want to argue the point, based only off the ID number.
 
Last edited:
That's a good point. I mean if Intel released a $750 mainstream processor in 2019 I can only imagine how many people would complain about the high price.

Technically, Intel has had $1,000 mainstream processors. This is back when Intel's segmentation was different than it is today. Its Pentium 4 Extreme Edition CPU's were really for anyone who could afford them. They didn't have an HEDT segment as we know it now. I'm sure some people will try and point out that few people bought them, and that's true. However, few people buy Core i9 9900K's either. I'll bet that far more people tend to spend $250-$300 on a CPU than $500+. K-series parts have always been for the enthusiast which isn't really all that mainstream.
 
If Intel released a $750 mainstream processor in 2019 it would be built in 14nm++, have 10 or less cores

I agree with your first points.

and wouldn't break the records of a $1700 processor.

This one I would have expected Intel's marketing to find some applications where the 10C CPU would break some record..
 
Last edited:
Nope! The information out there supports it. The speed of the chip you are claiming is not an ES sample is 6% slower base clock, and 9% slower boost clock then what AMD has released as the specs of the 3950x (I guess AMD could be lying and is prepping themselves for massive lawsuits). Then you add in that they are getting near 6Ghz on ES chips of the 3900x as stated here from Gamer's Nexus at 2:13:



Again, nearly 6Ghz on an ES sample. I guess adding 4 cores could cause the LN2 overclocking to drop 15% when comparing a non ES sample of the 3950x to a ES of the 3900x. But that really doesn't sound very realistic and doesn't make a lick of sense, except maybe to you.

Basically EVERYTHING but your opinion points to this being an ES of the 3950x, which your opinion is being based off the chip ID number, that AMD has full control over and can change/use what every they chose, even if it isn't in line with previous id formats. And they don't have to get your approval to do so.

But if you want to claim that It's just because "I say so" and ignore all the other facts out there; GO FOR IT! But grab a towel, because there is a high possibility you will need it to wipe the egg off your face come September, when we start seeing the real results of a non ES chip of the 3950x rather than these results here.

Nothing you have said change the fact that we really don't know if the results here are based off a ES or production/retail chip. But the facts support that it is more than likely an ES chip of the 3950x, other than the id label per your explanation. So, go ahead, keep your eggs in that basket. One that is only held together by that ID number because all the other information available supports that you are most likely wrong about it not being an ES chip. But, there is the slim chance, you are correct. Hence, why I said, WE REALLY DON'T KNOW, which has been my position all along, but you want to argue the point, based only off the ID number.


No. The information doesn't support that it is an ES. Rather at the contrary the information shows it is not an ES. I already explained which is the codename for ES, and why the codename for this chip proves it isn't an ES.

Clocks can be increased or decreased. I can downclock a 1800X at 3.5GHz and benchmark it, and it continues being a 1800X, not an ES. This R9 chip could be downlocked for lots of reasons, including cooling or motherboard stability issues. In fact AMD recommends downclocking IMC of Naples chips before benchmarking with Linpack (another day I will explain you why AMD recommend that).

We know more now. We know that "100-000000033-01" is the Ordering Part Number of the R9-3950X. Below a list of the Ordering Part Number for Rome SKUs.

D9W-s5TU8AAm5Ms?format=png.png
 
If I wait a bit I will get a 6ghz 10nm Desktop part..........let's begin the wait grand master juanchi
 
It is not an engineering sample.



It is a qualification sample. Clocks must be surely final. I am asking about the steeping.

You are talking about engineering samples. I am talking about qualification samples.

.....

A qualification sample is often identical to the commercial version of the chip, usually sent out to OEMs to let them validate their hardware, BIOS, ACPI etc etc, with the final chip. AMD denotes qualification samples with a Z in the first string of the codename.

For R7 Ryzen:

1D2801A2M88E4_32/28_N --> (first batch engineering sample with 2.8GHz base clock and 3.2GHz turbo)

2D3151A2M88E4_35/31_N
--> (second batch engineering sample with 3.1GHz base clock and 3.5GHz turbo)

ZD3601BAM88F4_40/36_Y --> (qualification sample with 3.6GHz base clock and 4.0GHz turbo)

R7 1800X
--> (commercial chip with 3.6GHz base clock and 4.0GHz turbo)

No. The information doesn't support that it is an ES. Rather at the contrary the information shows it is not an ES. I already explained which is the codename for ES, and why the codename for this chip proves it isn't an ES.

Clocks can be increased or decreased. I can downclock a 1800X at 3.5GHz and benchmark it, and it continues being a 1800X, not an ES. This R9 chip could be downlocked for lots of reasons, including cooling or motherboard stability issues. .....

To be clear.

If it's a QS the clocks are almost exactly what they will be at launch. Unless they're not.
 
No. The information doesn't support that it is an ES. Rather at the contrary the information shows it is not an ES. I already explained which is the codename for ES, and why the codename for this chip proves it isn't an ES.

Clocks can be increased or decreased. I can downclock a 1800X at 3.5GHz and benchmark it, and it continues being a 1800X, not an ES. This R9 chip could be downlocked for lots of reasons, including cooling or motherboard stability issues. In fact AMD recommends downclocking IMC of Naples chips before benchmarking with Linpack (another day I will explain you why AMD recommend that).

We know more now. We know that "100-000000033-01" is the Ordering Part Number of the R9-3950X. Below a list of the Ordering Part Number for Rome SKUs.

View attachment 168653

Did you get that towel like I told you to? I hope you are not letting that egg on your face dry... And we didn't even have to wait till September. :D :D :D

Would be nice if you gave us a link where you copied that from, so we could, you know, see what it is actually referring to, and what they are saying in such an article. a screen shot of a chart means nothing without context. As a google search for such sku's gives us nothing. I do find it interesting that we are talking about Ryzen 2, and you keep throwing Rome data at us. Where is the Ryzen 2 info.. not Rome's?
 
Last edited:


The Notebookcheck article claims that it is an ES, but both the hwbot submission and the Geekbench database prove it isn't an ES. :rolleyes:

If I wait a bit I will get a 6ghz 10nm Desktop part..........let's begin the wait grand master juanchi

I don't known anyone expecting 6GHz 10nm Intel parts, but I know people believed AdoredTV 'leak' about a 5.1GHz 16C Zen2 $499 chip.

To be clear.

If it's a QS the clocks are almost exactly what they will be at launch. Unless they're not.

You are quoting that I wrote in on other threads about other chips. I have explained in this thread that this chip is neither an ES nor an QS. It is a retail chip with the Ordering Part Number "100-000000033-01".

Lyin' juan caught again. He'll ignore this though as usual and somehow it'll be AMDs fault and intel is moar betterer at some moving goalpost metric.

You are making up stuff again. I explained in #169 why this "AMD 100-000000033-01" chip isn't an engineering sample. And I added in #192 that the codenames 100-000000... are the Ordering Part Numbers for the retails chips.

Did you get that towel like I told you to? I hope you are not letting that egg on your face dry... And we didn't even have to wait till September. :D :D :D

Would be nice if you gave us a link where you copied that from, so we could, you know, see what it is actually referring to, and what they are saying in such an article. a screen shot of a chart means nothing without context. As a google search for such sku's gives us nothing. I do find it interesting that we are talking about Ryzen 2, and you keep throwing Rome data at us. Where is the Ryzen 2 info.. not Rome's?

I posted a leak with the full Rome lineup that will be announced soon. It isn't in any article. The table proves that 100-0000... codenames are the OPN of the actual chips.
 
Last edited:
Are there also two AMD fanboys/shills that annoyingly post negative intel stuff in every positive intel thread?

I mean if there ever was a positive intel thread.
 
Status
Not open for further replies.
Back
Top