AMD Ryzen (Summit Ridge) Benchmarks Thread (use new thread)

[Glo.]

What file do we use to replicate their test?

I thought there was a file that AMD provided. Must've made a mistake there.

 

[TheELF]

About the power thing.
Just take a look at some x99 mobo reviews/benchmarks and you can see that there are boards that draw more power then others,boards that boost the CPU more then others and so on.
Lisa clearly stated out of the box for intel but did not give a mobo model.
Looking at total system power you can't tell one from another you can't tell if the board is drawing too much power or not.
Just looking around in this forum, how often have you seen people saying that you can and should undervolt CPUs and GPUs because the manufacturers use way too much power?

 

[TheELF]

What file do we use to replicate their test?

I think you're supposed to render the animation with blender and then convert it with handbrake.

 

[witeken]

Look, man: it seems that you lost this argument long ago--why make it worse?

The only thing here who can judge about "winning" or "losing" an argument is reality itself, not people's silly, biased opinions.

I'ma leave you with this quote from someone with more intelligence than this whole forum combined.

“I can live with doubt and uncertainty and not knowing. I think it's much more interesting to live not knowing than to have answers that might be wrong.”
― Richard Feynman, The Pleasure of Finding Things Out: The Best Short Works of Richard P. Feynman

The first principle is that you must not fool yourself — and you are the easiest person to fool.

 

[sirmo]

I only said a minor speed bump for Zen would put the two systems in that demo at equal power usage levels. I did not say anything about the 6900K clocks.

Isn't' the same true for any CPU? (higher clock higher power consumption)

 

[coercitiv]

Isn't' the same true for any CPU? (higher clock higher power consumption)

Yes it is. What's your point?

 

[IEC]

Dare I board the hype train yet?

There are certainly plenty of unknowns, but if we're talking performance at the level of a 6900K or better and cost is closer to $400 or $500... I'll buy it.

Curious to see what the top end of PCI-e lanes on AM4 will be, as well.

 

[sirmo]

Yes it is. What's your point?

That's what I was trying to figure out about your post. lol

 

[Dresdenboy]

Yea... no single threaded benches yet... a little weird.

Nowadays ST performance relies the most on boost technologies. If they switched off the chips boosting capabilities, ST results would just be off.

 

[superstition]

“I can live with doubt and uncertainty and not knowing. I think it's much more interesting to live not knowing than to have answers that might be wrong.”
― Richard Feynman, The Pleasure of Finding Things Out: The Best Short Works of Richard P. Feynman

That quotation seems fallacious in more ways than one.

The first principle is that you must not fool yourself — and you are the easiest person to fool

Also fallacious.

I'ma leave you with this quote from someone with more intelligence than this whole forum combined.

I can come up with better adages than the ones you posted and I don't know much about physics. Apples and oranges. People can be brilliant in some ways and marginal in others.

 

[krumme]

The enthusiast market is where i5 and i7 is selling now. Thats the interesting segment.The HEDT market is miniscule and small niche with highest margins but lowest total profit. What this cpu needs to be is fast enough to compete with highest performance i7 for the desktop market. It seems its more than capable of doing so with the high base freq possible.
It might be so fast amd can price it at 500 usd but imo that will only be for the topend bin only as the interesting market is 150 to 400 usd.

Now one can wonder how many ccx will be allocated for desktop when the servermarket is attacked q2?

Btw It looks to me some people trading stocks knew the ipc and freq beforehand looking at the crazy buy up of amd stock recent months and days.

 

[moonbogg]

It's actually not a valid point. To demonstrate the inanity of including the delta:

Zen idle: 0W
Intel idle: 106W

Zen load: 94W
Intel load: 191W

Zen delta: 94W
Intel delta: 85W

Utterly pointless.

Everyone seems to be missing his point, unless its me who is missing something. I understand that lower power at both idle and load is, of course, better. But we are looking at the CPU, and the CPU's both consume essentially zero watts at idle. The power consumption figures at idle reflect total system power consumption for both systems, of which the CPU's contribute nothing to. There-to-the-fore, the Intel system's power consumption increased less upon loading than the AMD system, so the Intel CPU itself is more power efficient than the AMD CPU.
Nobody may indeed care about this though because the overall system power consumption is, for some reason, lower on the AMD rig for both idle and load, but this must be attributed to other factors besides the CPU's. This is why using power delta is the only metric that can be relied upon when complete knowledge of every other single aspect of the system is not possessed.

Make sense?

We are also ignoring the fact that the Intel CPU was operating at higher frequencies than the AMD chip, which was limited to 3.4 across all cores. Now me personally, I don't give a flea's crap about any of this. I just want to know how Zen OC's and what it can do for the next BF game.

 

[superstition]

This is why using power delta is the only metric that can be relied upon when complete knowledge of every other single aspect of the system is not possessed.

As long as people don't assume the delta data tells them how efficient the part is at idle.

If one knows, with certainty, that the CPU doesn't use power when idling that's different, of course.

I guess part of it is philosophical, too. Some might consider the minimum chipset power consumption to be part of the CPU power consumption. A good example is the original Atom platform, where the CPU consumed half the power of the chipset.

Not taking the chipset/graphics into account with that Atom would make the platform seem a lot better than it was. Instead, it was quite silly.

 

[kraatus77]

nope because nobody is going use cpu only so even if it consumes 0w outside mobo, it's pretty much useless.

afterall, you need full system in order to use it, and that's where zen wins. so on perf/w metric zen wins. delta is useless because chip alone can't do crap itself.

also zen is on non retail mobo and still not finished. so there's always chance of that delta going in zen's favor.

 

[moonbogg]

As long as people don't assume the delta data tells them how efficient the part is at idle.

If one knows, with certainty, that the CPU doesn't use power when idling that's different, of course.

I guess part of it is philosophical, too. Some might consider the minimum chipset power consumption to be part of the CPU power consumption. A good example is the original Atom platform, where the CPU consumed half the power of the chipset.

Not taking the chipset/graphics into account with that Atom would make the platform seem a lot better than it was. Instead, it was quite silly.

Right. Some of us are focusing specifically and only on the CPU in isolation to judge it based on its isolated merits, but CPU's don't function in a vacuum, so for practical purposes and based on the preview, the Zen "system" is more power efficient while the Intel "CPU" is more power efficient. Most people care (if at all) how much power is being sucked from their wall (or laptop battery), not what the CPU itself is doing.

 

[superstition]

delta is useless because chip alone can't do crap itself.

Unless there are differences in chipset consumption. If a processor can be installed on different chipset versions then the specific chipset consumption is certainly a factor. With the FX 8 core processors, for instance, there was the 990 and 970. And, there is the issue of the power delivery system and other things. As far as I know one minor drawback of a highly-robust VRM/MOSFET system is higher power draw.

Back in the day the same AMD chip could be installed into an ATI board with integrated ATI graphics, an Nvidia board, etc. Each had different levels of power consumption.

Have there been cases where chipset die shrinks happened, where a CPU could be installed in more than one compatible chipset where one uses a larger node than another?

 

[R0H1T]

Everyone seems to be missing his point, unless its me who is missing something. I understand that lower power at both idle and load is, of course, better. But we are looking at the CPU, and the CPU's both consume essentially zero watts at idle. The power consumption figures at idle reflect total system power consumption for both systems, of which the CPU's contribute nothing to. There-to-the-fore, the Intel system's power consumption increased less upon loading than the AMD system, so the Intel CPU itself is more power efficient than the AMD CPU.
Nobody may indeed care about this though because the overall system power consumption is, for some reason, lower on the AMD rig for both idle and load, but this must be attributed to other factors besides the CPU's. This is why using power delta is the only metric that can be relied upon when complete knowledge of every other single aspect of the system is not possessed.

Make sense?

We are also ignoring the fact that the Intel CPU was operating at higher frequencies than the AMD chip, which was limited to 3.4 across all cores. Now me personally, I don't give a flea's crap about any of this. I just want to know how Zen OC's and what it can do for the next BF game.

Except at idle it is consuming some amount of power & it isn't anywhere near zero watts, what you're hinting towards is likely when the system is asleep, bear in mind there are other processes running in the background & so even if there was no foreground application the CPU would still be hovering in the single digits, so far as power consumption is concerned.

Though I do agree that with the turbo off for AMD we'll likely see more load, & better performance, from zen & increased power consumption. The delta load doesn't show much, there are just too many variables there to consider.

 

[moonbogg]

Except at idle it is consuming some amount of power & it isn't anywhere near zero watts, what you're hinting towards is likely when the system is asleep, bear in mind there are other processes running in the background & so even if there was no foreground application the CPU would still be hovering in the single digits, so far as power consumption is concerned.

Though I do agree that with the turbo off for AMD we'll likely see more load, & better performance, from zen & increased power consumption. The delta load doesn't show much, there are just too many variables there to consider.

Yeah I don't know. I was just trying to understand another member's point regarding their comments on power delta. They sort of got ripped apart by people who seemed to have missed their intended point.
As far as I'm concerned, I won't be going SLI anymore, so I will have plenty of PSU headroom for the CPU to guzzle as much power as it damn well pleases. Just hand me the many FPS please is all I ask of the CPU.
Also, these are high end desktop CPU's being compared. For what they are, the power consumption is close enough to be called a tie IMO. They are practically equal in terms of power consumption.

 

[superstition]

Yeah I don't know. I was just trying to understand another member's point regarding their comments on power delta. They sort of got ripped apart by people who seemed to have missed their intended point.

Insulting a person's IQ is a recipe for negative feedback. In fact, we were told that the guy he quoted is more intelligent than all of us in combination, so the ante was upped.

 

[zinfamous]

The only thing here who can judge about "winning" or "losing" an argument is reality itself, not people's silly, biased opinions.

Well, that's interesting...

Here is some wisdom from another really smart dude:

One who throws stones into pool should not complain about pool filled with stones.

--zinfamous

look, going OT, so I'll just leave it at that.

 

[moonbogg]

You guys need to be more like Bogg. You can't hurt my feels.

 

[krumme]

You guys need to be more like Bogg. You can't hurt my feels.

Happens for most autistic people.

 

[realibrad]

Yeah I don't know. I was just trying to understand another member's point regarding their comments on power delta. They sort of got ripped apart by people who seemed to have missed their intended point.
As far as I'm concerned, I won't be going SLI anymore, so I will have plenty of PSU headroom for the CPU to guzzle as much power as it damn well pleases. Just hand me the many FPS please is all I ask of the CPU.
Also, these are high end desktop CPU's being compared. For what they are, the power consumption is close enough to be called a tie IMO. They are practically equal in terms of power consumption.

His original point was that the delta told a story of a more efficient chip for Intel, and that is not inherently true. AMD could have a very efficient chip, and a horribly inefficient chipset. It could also be true that AMD could be using more power on their CPU and the chipset is super efficient. Until we get more information, then we will just be guessing.

I jumped in when he highlighted the delta as if that alone was significant and its not.

As you pointed out, CPUs are not the only part of the system. There are many things going on, but he decided to try and draw conclusions instead of raise a question. I think its reasonable to want to know more and not assume everything we see is the full story, but I also think its incorrect to say that the delta shows AMD has lower perf/watt.

He then doubled down and said that because Anand used delta, that it proves his point which it does not because they use it in different ways. He got ripped because of his false claim and how he reacted.

 

[AtenRa]

At the end of the day it is the energy you consume at the wall that you are going to pay and not only the CPU

So one system used less power to finish the same work at less time and this system has the higher perf/watt.

 

[realibrad]

At the end of the day it is the energy you consume at the wall that you are going to pay and not only the CPU

So one system used less power from the wall to finish the same work at less time and this system has the higher perf/watt.

Well sorta. If the CPU is sucking up more power, than its likely harder to cool. A system where the CPU uses less power is more likely to OC better.