News AMD previews Ryzen 3rd generation at CES

[Hitman928]

More info to come. . .

 

[IRobot23]

That s about 100% sure, in the range of 15% excess power out of the 65W or so it consumed in the demo.

The 75W stated by AT is the estimated delta measured at the main assuming that system idle power was 55W, actually the delta should be a little higher at 84W, wich is 65W at the SoC level.

15% higher score than a 2700X at 65W is perfectly in line with 2x the perf/watt at isofrequency but not isoperf, the 15% delta in perf imply 0.5 x 1.15 = 0.575x the power, wich is exactly the ratio between this SKU and a 2700X, so much that this point to the test being done at 4GHz wich is the 2700X all core turbo in CB.

FTR the numbers stated by AMD imply that power increase by at least 16% from 4 to 4.2GHz, if it were such a frequency clocking the Zen 2 there would be no way that it could be at only 65W when we add an excess voltage, 15% higher throughput and an I/O still in 14nm.

Then i don't see a problem with 16C... I mean we have plenty of headroom for more W. Even if AVX512 uses 40% more power ~ 90W for CPu die x2 = ~ 180W + 55W +10W for I/O

If I/O is low latency (high freq) then probably it will use more power ~20W which means 55W for CPU * 1,4 for AVX 512 = 77W * 2 = 154W (more heat more power + x1,05) + ~ 25-30W for Uncore / iF

So basically we have +15W as ~i9 9900K running AVX2 at 4,3GHz+LLC4GHz?

 

[PotatoWithEarsOnSide]

I tend to enjoy playing the games that I play, not watching the frame counter fluctuate.
Saying that, I'm pretty crap at playing games so maybe I should start watching the frame counter.

 

[DrMrLordX]

I mean according to AMD the first ZEN was already supposed to blow intel out of the water,they made it look sooooooo gooooood against intel hedt in their presentation.

It was "so good" against Intel HEDT, in the real world. $999 6800k got walked on price/performance.

I think TheELF was simply pointing out that we were shown Cinebench only. And it is reasonable to assume that AMD has chosen the best case scenario for their chip.

CBR15 was the first head-turning benchmark for Zen. The next one was Blender. So I really would like them to bench it using the same Blender demo they used for Zen. There's a history there, and it's important for us to know how much things have changed between generations of the Zen uarch. Plus Blender can be compiled to use AVX2 (if only The_Stilt were still here).

But a proper (CES spirited) test for new chip would have been some CPU limited gaming scenario, AMD has just moved I/O and MC further away from cores, that's where THE real questions are.

We didn't get game benches out of them when they first demoed Zen so . . . are you surprised? Also I think it's funny that people actually think a fp-friendly benchmark is now a "best case" scenario for AMD processors. My, how times have changed.

 

[IRobot23]

I tend to enjoy playing the games that I play, not watching the frame counter fluctuate.
Saying that, I'm pretty crap at playing games so maybe I should start watching the frame counter.

Well I do too. I play MMO where CPu is quite big thing in some scenarios... Well ryzen does great job against my old CPU, I enjoy it a lot. Anyway when I build PCs then FPS do matter to me, cause every one wants to know what he is buying.

How every I don't like to build PCs with Intel CPUs unless its like 6C/6T, cause it is quite expensive and power hungry, ... hard to cool. Good MB with some features and good VRM heat sink for longevity (I don't want to kill a board in a year or two). WHile R5 2600/X is quite great with some faster ram it flies, but people want even more.

 

[Arzachel]

Wow, I was just playing cinebench last night. Really had a lot of fun. Cant wait to get home from work tonight and play it again. Guess I can ignore my whole steam library now. I have no problem with cinebench as a preliminary benchmark, but it *is* pretty much a best case scenario for AMD. We will just have to wait for more benchmarks, but gaming is still a question, and a primary use case for many (most) users. Final clocks may be better, but actually the only thing I am impressed with so far is the power consumption. By the time it comes out it will be almost a year after the 9900k, and to only match it in a best case artificial benchmark is underwhelming to me.

I'm not sure how to tell you this but there are people out there using computers for things other than videogames.

 

[jpiniero]

I'm not sure how to tell you this but there are people out there using computers for things other than videogames.

Well, yeah, like Web Browsing, Emails and Office. But you don't need a $500 CPU for that.

 

[PotatoWithEarsOnSide]

You do need one to keep you warm whilst you read your email.

 

[Arzachel]

Well, yeah, like Web Browsing, Emails and Office. But you don't need a $500 CPU for that.

You don't need one for playing games either because you're going to be GPU bottlenecked except for some very specific scenarios. Yet here we are.

 

[jpiniero]

You don't need one for playing games either because you're going to be GPU bottlenecked except for some very specific scenarios. Yet here we are.

It does depend on what GPU you have and what settings you put of course.

 

[SlowSpyder]

Looks like AMD has a winner on their hands. They will sell lots of these and probably continue to eat away market share.

 

[Harry_Wild]

Well, yeah, like Web Browsing, Emails and Office. But you don't need a $500 CPU for that.

Yeah many people do, want the fastest PC for doing web browsing, streaming video and the like. How many Ferrari owners drive 200 mph in the U.S? 1%? Same with top of the line PC owners who do mainstream stuff excluding gaming and video editing.

I happen to be one of them!

 

[Abwx]

Then i don't see a problem with 16C... I mean we have plenty of headroom for more W. Even if AVX512 uses 40% more power ~ 90W for CPu die x2 = ~ 180W + 55W +10W for I/O

If I/O is low latency (high freq) then probably it will use more power ~20W which means 55W for CPU * 1,4 for AVX 512 = 77W * 2 = 154W (more heat more power + x1,05) + ~ 25-30W for Uncore / iF

So basically we have +15W as ~i9 9900K running AVX2 at 4,3GHz+LLC4GHz?

They can set a relatively low base frequency and then let the power management clock the chip at say 125W max power, the all core turbo should be 4GHz for Cinebench and eventualy 10% less for heavier loads.

 

[TheELF]

It's funny that you don't read your own sources thoroughly because if you did you'd know that the 9900K in the TPU review was not maintaining boost clocks, hence making the gap to the 2700X look closer than it really is under proper turbo configurations.

We could also look at techpowerup's O/C result, then we have ZEN 2 at 2057 and 9900k at 2212
So unless AMD pulls another 10% clocks out of their ZENs they will still be slower on 7nm and intel will bring out another 14nm model just because they can still compete on ancient tech.

 

[bsp2020]

Based on the numbers Anandtech gave us (75W) and assuming 15W for the I/O they could probably fit the second chiplet with only a 5% drop in clocks for this workload.

This.

AMD just demoed that they can provide 2x performance of Intel's highest performing chip in mainstream platform but some people think it is not good enough... The worst case scenario for AMD in 2019 will be AMD's mainstream part (8 core Ryzen 3xxxX) will match Intel's best in everything except some high end gaming scenario. AMD's best mainstream part (12 or 16 cores on AM4) has potential to obliterate anything Intel has to offer under $1000 without charging HEDT premium for motherboard and some people here are not impressed?

 

[EXCellR8]

Eh, people find a way to be unimpressed by anything... especially when it comes to tech snobs.

 

[tamz_msc]

We could also look at techpowerup's O/C result, then we have ZEN 2 at 2057 and 9900k at 2212
So unless AMD pulls another 10% clocks out of their ZENs they will still be slower on 7nm and intel will bring out another 14nm model just because they can still compete on ancient tech.

Your point was that AMD purposefully made the 9900K look bad in order to compare favourably with the Zen 2 sample. Where is your proof?

If the 9900K at 5.1 GHz scores 2212, then at 4.7 GHz it would have scored around 2040, which was exactly what it scored in the demo. Thus it drills holes into your claims about the 9900K being gimped in any way.

 

[PotatoWithEarsOnSide]

Would have been funny to compare how much power that 2200 score would have needed.
"hey, our ES, not on final clocks, is within 15% of a 9900K OCed to the hilt, and we do it at one-third of the power. Would you like us to show you our own flagship model? We'll underclock it as low as we can to obtain the same result, and then show you that we can hit their performance at one-quarter their power."
Now that would have been funny.

 

[moinmoin]

The 3rd gen Ryzen CB benchmark was a big tease really. I don't think it's a coincidence that the result is pretty much identical to 9900k, they very likely picked the frequency just to achieve that and showcased the power consumption to make that the point. What does and doesn't that tell us:

• 7nm consumes less power than 14nm (worth a "duh!")
• Power consumption is significantly below the max TDP of today's AM4 systems. The recent Intel and AMD chips all consumed closely the TDP at best and far more in many cases. The overclocking headroom of the past is commonly used at stock already nowadays. The benchmark indicated the 3rd gen Ryzen will either come with a bigger overclocking headroom again (likely sufficiently clocked to keep Intel chips at bay) or the final spec will be significantly above what we saw.
• We don't know how big the impact of all the enlarged caches is.
• We know nothing about frequencies (and as such IPC).
• We know nothing about latencies.
• We know nothing about use cases outside CB.
• We know nothing about the impact of the FP improvements (as CB doesn't use them).

The showcase of the lidless package at least confirmed that 3rd gen Ryzen uses the same system design as Epyc 2, with what appears to be the same CPU chiplet (meaning it includes the wider design and bigger caches) and an IOC that is bigger than a quarter of Epyc's IOC (meaning it may have additional logic specific for desktop or even preparation for a potential MCM APU as some have already guessed).

 

[coercitiv]

Your point was that AMD purposefully made the 9900K look bad in order to compare favourably with the Zen 2 sample. Where is your proof?

He's probably unaware that stock 9900K does 1760 points in CB15. AMD let 9900K run for it's life, then sent Zen 2 to play tag while running at 60% power.

Reminds me of the people who did not believe 9900K would be obsolete in under 12 months from launch.

 

[LTC8K6]

Well, we are comparing a new architecture 7nm CPU to a several times recycled architecture 14nm CPU.

The recycled 14nm CPU is hanging in there pretty darn good, but it's certainly showing it's age.

 

[dacostafilipe]

Well, we are comparing a new architecture 7nm CPU to a several times recycled architecture 14nm CPU.

We are comparing a new node to a node that made multiple iterations with improvement in power and efficiency. And calling Zen2 a new architecture ... improved architecture yeah, but "new" :/

 

[PotatoWithEarsOnSide]

Anyone that already owns a 9900K won't be disappointed because they own a CPU that should serve them well for a long while yet. However, it's a hard sell to get anyone else to buy a 9900K now that we know Ryzen 3000 has in store.

 

[trollspotter]

Very much so.
What interests me is that with Zen+ a 3% CB improvement resulted in a 5% gaming improvement, primarily because the gains were down to improved memory latencies.
If we're going to see the same with Zen 2, then I suspect that the uplift in gaming will be pretty significant. As you said, the clocks alone will provide a chunky uplift, moreso if the ST boost is up near that 5.0GHz mark.
My only concern really us that IF changes haven't really been talked about much anywhere, as there were rumours of decoupling IF clocks from Memclock. If that is the case, then we're definitely likely to see memory latencies decrease, even if somewhat offset by having gone the MCM route.
I don't envision the Ryzen 3xxx CPUs to be behind in gaming at all, except in some very niche cases.

I remember the Q9550 doing pretty well in gaming with its memory controller glued on as well.

 

[LTC8K6]

We are comparing a new node to a node that made multiple iterations with improvement in power and efficiency. And calling Zen2 a new architecture ... improved architecture yeah, but "new" :/

Zen is a new architecture.
Coffee Lake is an old architecture.

 

[ozzy702]

Looking good thus far. Excited to see room for 16 cores down the road and excited at that power usage. As has already been mentioned, the real wildcard is latency. If AMD can narrow the gap on latency then the added frequency and IPC should help a great deal in gaming. If not, it still won't match the 9900k, although it will be close enough that nobody should be purchasing a 9900k once Zen2 is released.