Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

[Veradun]

Look at reviews of the 6900K online, the memory configuration is irrelevant in Cinebench

 

[Andrei.]

What are you saying ? It's called apple to apple , Not apple to Orange.

That's a stupid argument. Nobody running those Intel CPUs is going to run them in dual-channel. We want apples to apples real usage not apples to apples "technically correct" usages.

Look at reviews of the 6900K online, the memory configuration is irrelevant in Cinebench

Well alright. Looking to see real benchmarks.

 

[sirmo]

For IPC comparisons you actually would want to show dual channel performance.. apples to apples.

 

[Andrei.]

For IPC comparisons you actually would want to show dual channel performance.. apples to apples.

The memory controller is part of the microarchitecture hence part of what will contribute to IPC, no?

 

[CentroX]

Well 6900K @ 3.7Ghz scores 153pts, you do the math at what clock it ran in order to score 162.

so yeah there we have it. Broadwell level.

ladies and gentlemen, AMD goes 52% and reaches broadwell. Who would of thought that in spring last year when we started discussing zen?

 

[PhonakV30]

http://images.anandtech.com/galleries/5485/AMD Ryzen Tech Day - Lisa Su Keynote-32.jpg

I found important Note :

Power efficiency of consumer client desktop 8-core processors based on cinebench R15 nT score divided by wall power watts during testing.
Measured system wall power : 115w (1700) vs 142w(6900k)
Intel 6900K ==> 1473/142 = 10.37
AMD Ryzen 1700 == >1410/115 = 12.26
AMD Ryzen 1700 offers 18.22% more performance per watt

So I think Ryzen 1800x consumes more than 6900K ? or close ?

 

[Magic Hate Ball]

Stock configurations, with quad channel:

 

[CentroX]

maybe Zen+ will be quad channel?

 

[looncraz]

Anyone else still on Sandy Bridge that is thinking of upgrading to a Ryzen 1800X or 1700X? Mine is at 4.5Ghz and I haven't seen great reasons to upgrade before Ryzen. 4 more cores, upgrade from 1600mhz DDR3 to whatever DDR4. Also more bandwidth via PCI 3.0 16x for my 980 Ti SLI setup.

Just concerned that with my Sandy Bridge @ 4.5ghz, Ryzen @ 4.0ghz is still slower per core. Use is primarily gaming, so work per core is probably more important than the total number of cores.

I have an i7 2600k @ 4.5Ghz. Ryzen will beat it in every metric, no doubt.

~3.8GHz Ryzen should approximate 4.5Ghz Sandy Bridge performance.

 

[Magic Hate Ball]

maybe Zen+ will be quad channel?

If it's AM4 still, which is what they've been saying, probably not.

The space required for quad channel is not worth it in this market segment. Server chipsets for workstations will likely be where you find quads as they expand their product line.

 

[CentroX]

If it's AM4 still, which is what they've been saying, probably not.

The space required for quad channel is not worth it in this market segment. Server chipsets for workstations will likely be where you find quads as they expand their product line.

True. But if Zen+ continues to further improve the IPC, then am onboard as well.

 

[sirmo]

"The memory controller is part of the microarchitecture hence part of what will contribute to IPC, no?"

If you want to know about the performance of the core then no. You'd want to compare apples to apples. Any additional performance coming as a result of additional channels is an advantage in the platform not the actual cores.

I do think AMD should have done quad channel comparison, but to us enthusiast dual channel to dual channel is a more useful comparison.

 

[Absolute0]

http://images.anandtech.com/galleries/5485/AMD Ryzen Tech Day - Lisa Su Keynote-32.jpg

I found important Note :

Power efficiency of consumer client desktop 8-core processors based on cinebench R15 nT score divided by wall power watts during testing.
Measured system wall power : 115w (1700) vs 142w(6900k)
Intel 6900K ==> 1473/142 = 10.37
AMD Ryzen 1700 == >1410/115 = 12.26
AMD Ryzen 1700 offers 18.22% more performance per watt

So I think Ryzen 1800x consumes more than 6900K ? or close ?

Interesting find. According to that, Ryzen 1700 TDP = 65W but power draw from the wall, measured by AMD, = 115 W. And yes the 1800X will certainly be drawing more power if it's turboing to 4.0 instead of 3.7. We know that TDP != power draw but i commented earlier on why AMD is releasing their SKUs with 95W and 140W heatsinks and this makes more sense now.

I have seen opinions earlier that 1800X = 95W because that's AMD's TDP rating and sounded too good to be true; might be that it is too good to be true.

 

[Mezzanine]

Well it only took 10 years for amd to become competitive, it won't be long and Intel will crush ryzen like a coke can.





This is your fifth troll post out of your last 5 posts this past month.



esquared
Anandtech Forum Director

 

[sirmo]

"Interesting find. According to that, Ryzen 1700 TDP = 65W but power draw from the wall, measured by AMD, = 115 W"

Sounds about right.. add fans, motherboard, PSU switching losses + mobo VRM switching losses (think absolute best case is ~90% efficiency).. it could very well be around 65watts

An 80% efficiency PSU has an efficiency curve and a 600 watt power supply may not be optimized for max efficiency at 100 watt loads. So you're talking 75% or so.

 

[CentroX]

Well it only took 10 years for amd to become competitive, it won't be long and Intel will crush ryzen like a coke can.

The management is completely different today. Lisa Su seems extremely competitive compared to past CEOs. They will chase Intel for many years.

 

[Magic Hate Ball]

Interesting find. According to that, Ryzen 1700 TDP = 65W but power draw from the wall, measured by AMD, = 115 W. And yes the 1800X will certainly be drawing more power if it's turboing to 4.0 instead of 3.7. We know that TDP != power draw but i commented earlier on why AMD is releasing their SKUs with 95W and 140W heatsinks and this makes more sense now.

I have seen opinions earlier that 1800X = 95W because that's AMD's TDP rating and sounded too good to be true; might be that it is too good to be true.

Yeah, PSU's aren't 100% efficient, plus there are other components to consider. A video card (even idling), RAM, HDD, fans can even draw a few watts total. Add all that together and say 80-90% efficiency on the PSU?

I'd say 115W is pretty good during a pure CPU benchmark.

 

[krumme]

The memory controller is part of the microarchitecture hence part of what will contribute to IPC, no?

Yep - its not that simple.
Its about getting a realistic scenario and comparison. But what is that?
If you measure performance using 4 channels it goes to say you dont unplug the 2 of them before you do the efficiency bm. And measure at the wall. You cant run the cpu or memory controller without a motherboard....
But hey if people buy a zen they dont buy a 6900 so in real world realistically they buy a 7700 or at most a 6800 instead

 

[MaxDepth]

I'm just thinking about the recent AMD GPU release in that a bunch of the 4GB cards were actually 8GBs, just with a different sticker because of opening day stock shortage. I'm wondering if the 1700 were not the same as the 1700K and just let people figure it out for themselves.

And yes, I am old enough to remember the graphite (pencil) trick with the AMD Durons (Athlons). <sigh> the good old days...

Also unlocked, no XFR = no auto boosting beyond listed turbo. Probably lower binned.

But who knows, you might get lucky and OC as much as an 1700X or 1800X. It's not impossible, though unlikely if they are binning heavily.

We won't know for sure until we see some reviews with OC results.

 

[looncraz]

Interesting find. According to that, Ryzen 1700 TDP = 65W but power draw from the wall, measured by AMD, = 115 W. And yes the 1800X will certainly be drawing more power if it's turboing to 4.0 instead of 3.7. We know that TDP != power draw but i commented earlier on why AMD is releasing their SKUs with 95W and 140W heatsinks and this makes more sense now.

I have seen opinions earlier that 1800X = 95W because that's AMD's TDP rating and sounded too good to be true; might be that it is too good to be true.

My system, at idle, from the wall... ~120W.

i7-2600k... running 1.6Ghz at idle, low load.

At full CPU load it hits 282W @ 4.5Ghz.

Ryzen is crazy efficient...

And, of course, 115W... minus VRM efficiently (~80%), so 92W max to the CPU. Minus the fans, SSD, video card, PSU efficiency... 55~65W left over for the CPU.

It's astonishingly efficient.

 

[BeepBeep2]

Disappointing to see they gimped the Intel systems by only going dual-channel.

Define "gimped" - in most cases you see no increase by going quad-channel. Single channel would have been "gimped".
http://www.pcworld.com/article/2982...ing-truth-about-their-performance.html?page=2

 

[Riek]

Interesting find. According to that, Ryzen 1700 TDP = 65W but power draw from the wall, measured by AMD, = 115 W. And yes the 1800X will certainly be drawing more power if it's turboing to 4.0 instead of 3.7. We know that TDP != power draw but i commented earlier on why AMD is releasing their SKUs with 95W and 140W heatsinks and this makes more sense now.

I have seen opinions earlier that 1800X = 95W because that's AMD's TDP rating and sounded too good to be true; might be that it is too good to be true.

well the idle power consumption was 40W.

So assuming that the other components during the test consumed as much as the cpu in idle mode, you would have 114-40W = 74W delta on the power cable.. PSU efficiency of 90% would give close to 67W.
So close enough i guess.

 

[Shivansps]

so yeah there we have it. Broadwell level.

ladies and gentlemen, AMD goes 52% and reaches broadwell. Who would of thought that in spring last year when we started discussing zen?

I whould hold that until we know if XFR was enabled (no reason not to) and if it was enabled the st clock speed.

 

[Magic Hate Ball]

well the idle power consumption was 40W.

So assuming that the other components during the test consumed as much as the cpu in idle mode, you would have 114-40W = 74W delta on the power cable.. PSU efficiency of 90% would give close to 67W.
So close enough i guess.

I think RAM can consume more power as well: http://www.tomshardware.com/reviews/intel-core-i7-5960x-haswell-e-cpu,3918-13.html

 

[looncraz]

Well it only took 10 years for amd to become competitive, it won't be long and Intel will crush ryzen like a coke can.

It takes years to make adaptations. Intel and AMD will be close for a few years to come, at least. Zen+(2) will supposedly be ~15% higher IPC, and presumably higher efficiency and frequency... on the same AM4 motherboards.

Zen++(3) is a process shrink, higher IPC, maybe higher frequency, maybe even more cores (kind of doubt it).

The real damage to Intel will come from Raven Ridge. A native quad-core APU that will probably clock higher than these first 8-core CPUs.

Zen+ will be a 2018 product, probably mid-year, and will push AMD's IPC above Intel's current best. Intel only plans a 15% TOTAL improvement with IPC and clock-speeds improvements combined. But this comes later this year, so Zen+ will be late to the show.

Intel counted Kaby Lake as having 15% more performance than Skylake, to give you a clue...