Discussion Ryzen 3000 series benchmark thread ** Open **

Page 24 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

beginner99

Diamond Member
Jun 2, 2009
4,195
248
126
(for now we can call them out for bad launch BIOS support).
bascially typical AMD launch. It's even more puzzling as some of the insiders have said that the mobos have literally been availabel for 2-3 months before launch. So I guess AMD has a problem on the software / bios side of things. I suspect they did a respin pretty late for better clocks and hence it isn't all sorted out yet.

Still given that they could have just put a safe number on the package. No one would complain if it boosts higher than advertised. The 4.6 ghz on the 3900x seems like they just had to put a higher number on there.
 

Zucker2k

Senior member
Feb 15, 2006
828
121
136
There's a low clock ceiling of about 4.4GHz-ish which some individual cores seem to be able to breach. In my opinion, Zen2 is mostly a 4.3GHz product, if AMD had gone conservative with their boost algorithms. Zen+ would probably had exhibited the same problems we're witnessing here but AMD wisely held back at 4.35GHz. Look at the voltages they're pumping into these 7nm chips; 1.47v for the 3900x, for example. I don't know any process that'll scale up high (4.6GHz+) on those voltages on air. In light of the above, is it any wonder that the lower core products have no issues clocking up to their advertised boost clocks? I think not. The CCX design, and die layout all seem to be accessories to the heat issues as well. Hot-spotting, anyone?
 

Kenmitch

Diamond Member
Oct 10, 1999
7,362
282
126
Ryzen Master has a new ECO option available for the 3xxx series of chips.

Has anybody seen any reviews that have tested it?

I decide to play around with it myself to see what it does.

Here's the normal stock setting behavior of my 3600

1563895578847.png

The ECO mode results

1563895616410.png

Guess I'll have to do some more playing around to see how it effects other use scenarios.
 
Last edited:

Thunder 57

Senior member
Aug 19, 2007
760
264
136

moinmoin

Senior member
Jun 1, 2017
915
402
106
Feb 23, 2017
636
578
106
There's a low clock ceiling of about 4.4GHz-ish which some individual cores seem to be able to breach. In my opinion, Zen2 is mostly a 4.3GHz product, if AMD had gone conservative with their boost algorithms. Zen+ would probably had exhibited the same problems we're witnessing here but AMD wisely held back at 4.35GHz. Look at the voltages they're pumping into these 7nm chips; 1.47v for the 3900x, for example. I don't know any process that'll scale up high (4.6GHz+) on those voltages on air. In light of the above, is it any wonder that the lower core products have no issues clocking up to their advertised boost clocks? I think not. The CCX design, and die layout all seem to be accessories to the heat issues as well. Hot-spotting, anyone?
What relevance is a clockspeed when it hits its performance levels irrespective of what number you get to jerk off to?
Too many idiots thinking that clockspeed is the ultimate arbiter of performance. Newsflash: it isn't. Bulldozer was hitting high clocks with crap performance, and Sunny Cove supposedly hits great performance with "crap" clocks.
It doesn't matter where your performance comes from; if the performance is there, the performance is there. Ryzen 3000 is there (or thereabouts when it comes to gaming).
 

Thunder 57

Senior member
Aug 19, 2007
760
264
136
The effective CPU speed index measures performance with the following weights: 40% single-core, 58% quad-core and 2% multi-core.
Source

Before Ryzen was released the ranking was based on: 30% Single core performance 60% Quad core performance 10% multi core performance
Internet Archive

ROFLcopter
If anything multicore should be going up at the expense of single core. Maybe 20/60/20 or even 20/50/30 since there are so many 6+ core CPU's out these days.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
18,178
2,133
136
The effective CPU speed index measures performance with the following weights: 40% single-core, 58% quad-core and 2% multi-core.
Source

Before Ryzen was released the ranking was based on: 30% Single core performance 60% Quad core performance 10% multi core performance
Internet Archive

ROFLcopter
At least Ryzen is now much stronger on single core, but yes, sounds like Intel again and money can buy anybody (almost)
 

ondma

Senior member
Mar 18, 2018
468
144
61
Yea, intel has bought them off for sure. That is why the first thing you see on their homepage is huge pictures of 2 ryzen processors.
 

moinmoin

Senior member
Jun 1, 2017
915
402
106
At least Ryzen is now much stronger on single core, but yes, sounds like Intel again and money can buy anybody (almost)
It looks like only delaying the inevitable, Ryzen 4k may well be surpassing 14nm++ Intel in single, quad (thanks to further CCX optimizations) and multi core all at once, then the balance between all of them is moot.

The result of this balance tweak is pretty funny though even for Intel chips looking through the list sorted by "Avg. bench %" with fun stuff like i3 8350K and i5 8400 being on the same level.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
18,178
2,133
136

Thunder 57

Senior member
Aug 19, 2007
760
264
136
It looks like only delaying the inevitable, Ryzen 4k may well be surpassing 14nm++ Intel in single, quad (thanks to further CCX optimizations) and multi core all at once, then the balance between all of them is moot.

The result of this balance tweak is pretty funny though even for Intel chips looking through the list sorted by "Avg. bench %" with fun stuff like i3 8350K and i5 8400 being on the same level.
At which point they'll change it so multicore is 90% to make sure Zen pays an inter-CCX penalty :p.
 

Zucker2k

Senior member
Feb 15, 2006
828
121
136
But I thought the consensus was that AMD has the ipc advantage now; apparently not. :confused_old:
 

Zucker2k

Senior member
Feb 15, 2006
828
121
136
What relevance is a clockspeed when it hits its performance levels irrespective of what number you get to jerk off to?
Too many idiots thinking that clockspeed is the ultimate arbiter of performance. Newsflash: it isn't. Bulldozer was hitting high clocks with crap performance, and Sunny Cove supposedly hits great performance with "crap" clocks.
It doesn't matter where your performance comes from; if the performance is there, the performance is there. Ryzen 3000 is there (or thereabouts when it comes to gaming).
Forget clocks, the performance came about by sorcery.

Don't change the topic. The question at stake is about unreachable boost clocks and potential risks to AMD's reputation and possible legal jeopardy over false advertising.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
18,178
2,133
136
Forget clocks, the performance came about by sorcery.

Don't change the topic. The question at stake is about unreachable boost clocks and potential risks to AMD's reputation and possible legal jeopardy over false advertising.
This is a benchmark thread. And the statement "the performance came about by sorcery. " is just trolling. And "legal jeopardy over false advertising" ? again, trolling. It has been shown they do it in some cases, just not as many as people like;
 

Zucker2k

Senior member
Feb 15, 2006
828
121
136
This is a benchmark thread. And the statement "the performance came about by sorcery. " is just trolling. And "legal jeopardy over false advertising" ? again, trolling. It has been shown they do it in some cases, just not as many as people like;
How come you missed the part where the person I responded to said everybody who doesn't agree with him that clocks are not the ultimate arbiter of performance is an idiot? Here, I'll quote it:

Too many idiots thinking that clockspeed is the ultimate arbiter of performance
That whole post is neither here nor there because performance is always hinged on clockspeed. One user even calculated exactly what the contentious 100MHz amounts to - 2.2%. So yeah, that means something to some people, but they should all discard their opinions for one that says Zen 2 hits its "performance levels" (whatever that is) and anybody who thinks otherwise is an idiot.

Clock speed IS indeed the ultimate arbiter of performance. IPC is more or less fixed in silicon. The main thing that can change how that ipc interacts with code is clockspeed, all else being equal. That is why this is important; along with whatever the implications are for the manufacturer vis a vis the enthusiast community.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
18,178
2,133
136
How come you missed the part where the person I responded to said everybody who doesn't agree with him that clocks are not the ultimate arbiter of performance is an idiot? Here, I'll quote it:



That whole post is neither here nor there because performance is always hinged on clockspeed. One user even calculated exactly what the contentious 100MHz amounts to - 2.2%. So yeah, that means something to some people, but they should all discard their opinions for one that says Zen 2 hits its "performance levels" (whatever that is) and anybody who thinks otherwise is an idiot.

Clock speed IS indeed the ultimate arbiter of performance. IPC is more or less fixed in silicon. The main thing that can change how that ipc interacts with code is clockspeed, all else being equal. That is why this is important; along with whatever the implications are for the manufacturer vis a vis the enthusiast community.
You are not making any sense. In many cases, the Ryzen 3000 series beats even the 9900k at a lower clockspeed. So how can you possibly say that "Clock speed IS indeed the ultimate arbiter of performance " ?
 

Vattila

Senior member
Oct 22, 2004
419
185
136

Zucker2k

Senior member
Feb 15, 2006
828
121
136
You are not making any sense. In many cases, the Ryzen 3000 series beats even the 9900k at a lower clockspeed. So how can you possibly say that "Clock speed IS indeed the ultimate arbiter of performance " ?
You're taking things out of context by comparing two separate architectures. Which is faster: A 3900x @4.5GHz or a 3900x @4.6GHz? Never mind the answer. I give up.
 
Feb 23, 2017
636
578
106
Which is faster?
A 3900X that hits 196fps at 4.5GHz or a 3900X that hits 196fps at 4.4GHz?
Makes no difference where the performance comes from. If X is the product of Y and Z, with X being constant, it doesn't matter what Y and Z are individually. They hit max performance out of the box, which ultimately means that Y and Z are irrelevant in the overall context of things.
We could be talking about Intel CPUs here for what it matters.
I'm tired of people hanging onto any little shred of information as a means to burn AMD when it is the bigger picture that counts. The bigger picture is that these CPUs are competitive, and in very many cases the performance leaders. How they got there is neither here nor there in the grand scheme of things. Sure, from a technical point of view it is interesting, but certainly not something with which to scold them.
 

Similar threads



ASK THE COMMUNITY

TRENDING THREADS