It could be worse. Imagine what would happen if Intel was more

>>61352343
>game AI
>+18%

But intel said that AMD was bad for games!

>>61352391
Ryzen performs splendidly, considering the clocks.
Zen2 will kill even Intel's gaymen chips.
>>61352343
Oyyy veyyy.

>>61352343
After looking at the number of people who really do not fully understand the entire architecture and workloads and thinking that AMD Naples is superior because it has more cores, pci lanes etc is surprising.
AMD made a 32 core server by gluing four 8core desktop dies whereas Intel has a single die balanced datacenter specific architecture which offers more perf if you make the entire Rack comparison. It's not the no of cores its the entire Rack which matters.
Intel cores are superior than AMD so a 28 core xeon is equal to ~40 cores if you compare again Ryzen core so this whole 28core vs 32core is a marketing trick. Everyone thinks Intel is expensive but if you go by performance per dollar Intel has a cheaper option at every price point to match Naples without compromising perf/dollar.

>>61352433
DAMAGE CONTROL

File: 1370053949513.jpg (32KB, 461x346px) Image search: [Google]

32KB, 461x346px

>>61352433
This is straight from those marketing slides, isn't it?

File: angry.jpg (55KB, 297x365px) Image search: [Google]

55KB, 297x365px

>>61352343
REMOVE DIS ANTISEMITIC CHARTS YOU SCHMUCK

>>61352343
Reminder that this was tested with 2400MHz memory and not 2600MHz EPYC supports.

Sykalake-EP was running 2600MHz of course.

>>61352343
We take all competitors seriously, and while AMD is trying to re-enter the server market segment, Intel continues to deliver 20-plus years of uninterrupted data center innovations while maintaining broad ecosystem investments. Our Xeon CPU architecture is proven and battle tested, delivering outstanding performance on a wide range of workloads and specifically designed to maximize data center performance, capabilities, reliability, and manageability. With our next-generation Xeon Scalable processors, we expect to continue offering the highest core and system performance versus AMD. AMD’s approach of stitching together 4 desktop die in a processor is expected to lead to inconsistent performance and other deployment complexities in the data center.

>>61352819
B-b-but muh Platinum membership processors!

File: s1va16ehus7z.png (222KB, 638x465px) Image search: [Google]

222KB, 638x465px

>>61352433
Ryzen uses a sophisticated interconnect that manages to have lower latencies between dies on the interposer than Skylake-EP does between cores. It will actually scale better than the Skylake-EP, especially when you start going 2S. Skylake-EP does support 8S but that's such a niche that no one fucking uses it, 0.05% of buyers.

EPYC also isn't needlessly cut down, if all you need is tons of PCI-E lanes you can just buy the 8 core 1S model for the 64 lanes and full complement of IO.

Ryzen cores actually are not weaker than Intel cores, the difference in performance in muhgames usually just comes down to worse latency than the monolithic quad cores. Read your Agner sometime, that see when Raven Ridge APUs come out.

File: 1499953102756.jpg (75KB, 682x776px) Image search: [Google]

75KB, 682x776px

>>61352343
It gets worse.

AMD has sunk their fangs into Intel and is sucking them dry

>>61353587
>8 core 1S model for the 64 lanes and full complement of IO.
It's 128 lanes for 1S.

>>61352343
>muh performance difference
>actually just within 15%

what really matters is the ecosystem and its about "it just werks". If you pay 200% on nodes one time, it doesn't matter. If you pay less but then your servers die becaue of a immature platform its much worse and could kill your business.

big enterprise is going to go intel. one time expenses are irrelevant.

>>61353663
Big enterprise wants lower TCO, something than Intel can't give them.

>>61353663
This is the same argument Intel uses in its Purely slides: "you're stuck with us so don't change!, imagine the cost!"

AMD shouldn't have made CPUs ever again in that case.

>>61352343

Please take these benchmarks with a bit of salt. They were using "Ubuntu Server "Xenial" 16.04.2 LTS (Linux kernel 4.4.0 64 bit). The compiler that ships with this distribution is GCC 5.4.0.".
I believe (correct me if I'm wrong), that kernel and compiler do not have Zen AND Skylake optimizations. I will say that EPYC will most likely get more performance out of the optimizations than Intel if the desktop platform is anything to go by.

>>61353721
That's irrelevant, nobody uses newer kernels in anything needing stability.

RHEL6 and Ubuntu LTS are your baselines.

File: 1477694251038.png (358KB, 552x543px) Image search: [Google]

358KB, 552x543px

>>61352433
>monolithic diarrhea is better than glued together gold

>>61353663

You have a point, to big companies were 10 minutes of downtime is literally tens to hundreds of thousands lost profit, reliability and support is king.
AMD has to nail their first few early roll outs of EPYC so that the good word can get out on the streets. I have confidence in this, if you look at the last ER they hired a whole bunch of engineers to ramp up into EPYC, so hopefully they got their shit together on this front.
If they get good reviews from big players (like Microsoft and Dropbox) then more people will be willing to buy into AMD, and reap the significantly lower TCOO.

>>61353761
>$999 for the 16c/32t
>$799 for the 12c/24t
>while boasting good clockspeeds

I'm ok with this

File: Screenshot from 13-07-2017 15:55:34.png (23KB, 686x194px) Image search: [Google]

23KB, 686x194px

>>61353761
Intel "TDP"

File: fNRYdhs.png (497KB, 700x792px) Image search: [Google]

497KB, 700x792px

>>61353761
Well duh it's two 8 cores, and the 1800X has a 95w TDP.

>>61353794
note in order that's a 22 core, 28 core, and 32 core.

>>61353776

At what point does 10-30% improvement in your workload become "irrelevant" or a few % efficiency gain?
At what point is a possible updated kernel, supported by AMD's own engineers + the broader community, not stable? At what does it become stable?

>>61353838
Outside of customers running their own custom kernels, everyone runs stock RHEL kernels, which are old but also 120% stable.
Stability is always valued over performance, that 1% is critical.

And personally, for companies dealing with millions each day, I can understand fucking why.

>>61353663

PHB care about the bottom line. Intel is going to hell of a time convincing them to invest in 20-30K systems when thier AMD counterpart offers the same performance/reliability at 1/2 the cost and has lower TCO.

The Intel brand name isn't powerful enough to justify that premium.

Intel marketing is still stuck in 1990s thinking AMD is K6-K7 when Opteron/K8 have proven themselves to a robust enterprise/HPC platform.

>>61353761
>>61353784

>>61353761

Still better than i9 chips that it is meant to go against.

At least you don't need a custom water-cooling system and delid the chip to keep the chip from becoming a blast furnace.

>>61352343
>(2680-2670)/2670 = -0%
I understand rounding 1/267 (0.37%) to 0%, but it's still positive. How did they fuck that up and look at it and go "yup, negative"

I'm actually happy that it only has 2 skus compared to that shitty leak saying that it had 10

>>61354211
Yeah this is a lot easier to understand.

>>61354211
The 12 core doesn't make any sense to me. At that pricepoint, I'd rather splurge some more and get the 16 cores.

File: 1497027833797.jpg (29KB, 628x380px) Image search: [Google]

29KB, 628x380px

>>61353761
>>61354771

Jesus fucking Christ these niggers are persistent.

>>61354864
Intel damage control squad at their finest.

kek that jewtel shill is really sad

File: threadripper cooler.jpg (70KB, 1024x576px) Image search: [Google]

70KB, 1024x576px

Threadripper won't have any problem staying cool.

>>61355030
Hnnnnnnnnnnggggggggg

>>61355030
>the size of that contact place
>separate soldered dies
Oh yes.

>>61355030
>>61355058
now someone needs to delid one of these so we can know how the dies are positioned on the package and place thermal paste in the most efficient way

>>61355229
That german dude will probably do it.

>>61354859
>4Ghz

File: 1499959791737[1].png (697KB, 650x462px) Image search: [Google]

697KB, 650x462px

>>61355244
judging by this picture of epyc's interconnects, TR dies will be side by side horizontally

>>61355293
That's some nice looking 'glue'.

>>61353645
The 1950x's 3000 scores is similar to Intels $2000 xeon.

Jesus Christ.

>>61355293
I N F I N I T Y
F A B R I C

File: TB2rgQ7xblmpuFjSZFlXXbdQXXa_!!2341477237.jpg (349KB, 1060x1047px) Image search: [Google]

349KB, 1060x1047px

>>61355394
They already lost.

Man, a first gen core is making a mockery out of Intel's "biggest datacenter architecture advancement since Nehalem"

I don't even wanna see what Intel's face will be when 7nm EPYC rolls in, more clocks, 50% more cores per CCX, lower power.

I can't imagine

>>61355515
that xeon gold 6150 is a 3 grand cpu

>>61355538
I hope AMD's stock drops when people panic over Q2 earnings.

>>61355538
What if
WHAT IF
AMD adds on-package HBM2 for EPYC2?

>>61355547
You pay for XEON technology, battle tested, 20 years of innovation, market leader, friend

>>61355270
>4Ghz 16-core already
Don't even need to wait for 7nm. Holy shit that's high. We are legit going to hit 5Ghz 16-core for 7nm.

>>61355576
What if you stop being retarded?

File: intel quality glue.jpg (38KB, 582x252px) Image search: [Google]

38KB, 582x252px

>>61355547
But listen, goy! Intel, in their infinite generosity, is willing to sell it to you for just $2k*!

*With toothpaste under the lid which makes it impossible to cool properly

>>61352449
>>61352478
That's a copypasta you niggers