What did Intel mean by this?

then the graph should have represent power drawn instead of frequency
lower frequency is _never_ better

I wonder how long Intel have been bribing motherboard manufacturers to go out of spec.

This makes me feel super comfy with my 4770K @ 4.5Ghz. At least on single threaded loads.

That is false. Intel have better IPC in most cases, note the _most_. They have a better, lower latency memory system. They also have some other core-level advantages that lower overheads.
But IPC isn't a single number, it depends on the work being done.

>heavily multithreaded game
>ryzen still loses in ipc
Even in your cherry picked benchmark, the i7 8700k btfo every zen cpu in ipc

Attached: Capturekk.png (740x802, 91K)

He already showed earlier that without the limit power usage went 150+ watts. In order to go from 4Ghz to 4.7Ghz all core it has to draw up to 60 watts more!

The only thing you have to gain from higher frequency is 5-10 fps at 150 fps and lowering is reduces temps and power draw massively.
How is higher frequency better?

>heavily multithreaded
>i5 exact same framerate and frametime as i7

not sure they do, it's possible but it's also possible it's just a result of competition between motherboard vendors. look around for retarded motherboard "reviews" showing motherboards running some CPU at stock settings and retarded "reviewer" declaring some motherboard to be "better" because it it's default settings happen to push the board out of spec.

something tells me this has been going on for some time and that the only thing that's new is that a few youtuber's picked up on it

Its a gpu bottleneck, you can clearly see the difference between the r5 and r7 since they arent fast enough to max out the gpu power

Indeed. Whne the cores are that close together on different CPU's it's pretty obvious its a GPU bottleneck. The poster of said image just got BTFO.

Found the shitskins.

I really do wonder what it is that makes some, as in most, games perform better with intel+nvidia compared to amd+nvidia. that this is the case is a simple fact. do a cpu task comparison of a intel i3 and a ryzen r7 and the ryzen will absolutely crush the i3. paried with a nvidia gpu that i3 wins over the same r7 in quite a few games.

never really found a good answer as to why that is, it's a bit strange.

probably just a mistake

Intel curbstomps AMD to the point where it allows the 1080 ti to function at full power
>not a gpu bottleneck
How delusion can intel shills be?

I don't doubt it does. But it's only by a marginal percentage and mainly in games where latency and single threading are a major factor. What do I care for a game being 150 fps vs 140 fps? Sure. Maybe if you have a shit GPU that cannot get you over 60 FPS in the latest AAA title. But for everything else it's just dick waving.

Feel free to actually present some evidence of a significant IPC advantage for Coffee Lake over Zen+ in a variety of workloads, if you have any. The majority of the testing I've seen has IPC within 2-3% between them.

Attached: 1538085000669.png (674x800, 53K)

Intel's ringbus architecture has an IPC advantage in gaming, but in general across the majority of applications the IPC between Zen+ and Coffee Lake is within 2-4%. The non gaming results from the exact comparison you got that chart from shows that.

Attached: Corona.png (1339x957, 68K)

That 8 cores are running at 4.7Ghz while poozen can only run 4Ghz on all cores

"Only" 4Ghz and still much better on multithread than 5Ghz Intel.
Also,
>muh few more fps in gaymes while losing in any other heavily multithreaded load
>paying more for an outdated architecture
>paying more for a backdoor infested CPU

Makes you think.

Not within 3-4% of eachother, intel is CONSISTENTLY 3-4% better than AMD in ipc

You seem to be having some trouble reading the productivity IPC tests, clearly.

Attached: Blender.png (1339x957, 57K)

3-4% is such a big lead huh, totally worth the additional 20% in price.

Oops, looks like you're wrong in this test too.

Attached: Vray.png (1339x957, 69K)

>be intel
>get held to 95W TDP
>can't manage over 4ghz full core load
>be amd
>get held to 95W TDP
>4.3ghz all core turbo loaded
what did they mean by this?

Oops, wrong again!

Attached: Excel.png (1339x957, 67K)

Wow, look at that Intel advantage!
(lol

Attached: PCMark_01.png (1339x957, 58K)

I didn't know benchmarks were possible in Excel.

it's important with extremely large spreadsheets

Man, Zen 2 with higher clocks is going to be loads fun

>intel can't into muilticore
>more news at 11

9900K at 4.7GHz most certainly doesn't lose to any 8C Ryzen in multithreaded load.

>dx11 fps metric
>not realizing the gpu pipeline is being fed by 1.3-1.5 cores.

Nvidia's drivers aren't well suited to zen, but they're getting better. Zen1 gets a ~5% boost and zen+ a ~3% boost with a 1080ti when you enable messaged signaled interrupts, for instance. Additionally, Vega 64 often performs slightly better than a 1080ti with zen. This likely comes down to amd having MSI's enabled by default and slightly better support in the driver.

Amd uses ringbus inside the ccx, their core:core latency is the same. The ccx boundary is substantially slower due to being tied down to memory frequency. It's a memory latency issue the majority of the time in games. And it's only an issue when you start pushing crazy high framerates. There's also questions of what compiler was used etc. Zen gets pretty substantial boosts when you use updated binaries vs pre-2017. This has to do with the outdated compiler having no idea wtf zen is. Even in tomb raider the performance issue zen had at launch was only percievable on nvidia gpus specifically, which was later fixed game-side.

In floating point math 128bit and below amd either has parity, or a significant advantage depending on how the math is run. If it's. Pure fma zen is about the same IPC, but mixed workloads like you see in c-ray and v-ray using blends of mul/add zen has a maximum throughput of 4 128bit instructions per cycle. That. Being two add, and two multiply. Zen 2 will be upping the ante on this by doubling the pipeline to 256 bit native.

Yes for a small $600 increase in pricing.

POO

Attached: 1534605627581.png (764x649, 258K)

yeah but 180w power draw

IN

Attached: 1534326485713.png (1824x1026, 431K)

JOO

Attached: 1534204099106.png (1920x1080, 611K)

Steve goofed up

2600x has same amount of corse as 8700k and gets btfo by it
Exact same thing, 8700k btfo 2600x
Nice proof pajeet

POO IN JOO shitkike,, 9900k a massive dissapointment and flop

cores*

I'd honestly push 180W into a Ryzen CPU if it reached 5GHz, I don't particularly give a fuck about power consumption in a desktop, unless it's some monstrosity like that water chiller abomination. I wouldn't even need to upgrade my cooling for 180W.

>people still don’t know how TDP works or power limitations in general because motherboards have been breaking limits by default for years

>Exact same thing, 8700k btfo 2600x
>Nice proof pajeet
Are you unable to read, perhaps? They got the same score. Or maybe you don't understand what margin of error is?

Actually you're probably just shilling, but nonetheless.

I said it was consistently better and you have yet to prove me wrong

>lower is better

No. You said: >intel is CONSISTENTLY 3-4% better than AMD in ipc
This is false, and
And this test all empirically prove your claim to be false. Reminder that within 1% or 1.2% is not "3-4%", which is what you claimed. And that's over and above the tests where the 2600x is tied or better than the 8700k at 4ghz.

You made false claims, and they have now been proven to be false with empirical evidence. I'm sorry if this hurts your feelings.

Attached: PCMark_02.png (1339x957, 52K)

By that logic bulldozer is best

>
>2600x has same amount of corse as 8700k and gets btfo by it
>
>Exact same thing, 8700k btfo 2600x
>Nice proof pajeet
>
>No. You said:
>>intel is CONSISTENTLY 3-4% better than AMD in ipc
>This is false, and
>
>
>
>
>And this test all empirically prove your claim to be false. Reminder that within 1% or 1.2% is not "3-4%", which is what you claimed. And that's over and above the tests where the 2600x is tied or better than the 8700k at 4ghz.
>You made false claims, and they have now been proven to be false with empirical evidence. I'm sorry if this hurts your feelings.

Attached: 1514922478622.png (868x756, 265K)

i don't know who's trolling who anymore

None of us take this seriously anymore, we have fallen into a deep depression where we repeat the same threads and have the exact same arguments over and over and over...

Attached: 88f.jpg (640x480, 26K)

On the drivers front I wish more benchmarks would us all four nvidia/amd/Intel cpu/gpu combinations in their testing instead of just 1080ti as its been shown that there is decent varience among them in the past.

>ringbus
This might be wrong. The cross bar in zepplin dies is tied to the imc and its clock for reasons. But the internal ccx is wired so all four cores have a direct connection to each other. The specifics of the ccx to the cross bar I don't know though.

>256but
Did they specify if the units were becoming 4x 256bit or just combining the 4 units into two 256bits? The narjeted double throughput probably means the first.

>tfw can't get my 4670K past 4Ghz

Isn't IPC.
It's memory lattency that is holding back Ryzens. See how the second gen 6 core performs better than the first gen 8 core.

For end users and retail motherboards yes. It really comes into play for oems and shit boxes or laptops. The oem will put the minimal amount of cooling and power supply they can get away with in the effort to maximise profit. If Intels tdp is 95w then their motherboards will only supply 95watts in the long run and have only enough power delivery for that, same with the cooler, it'll be rated to 95watts so when timmy buys his 9900kelvin shit box it won't perform like the reviews say.

Well it is. 95w is the optimum according to Intel's own spec.

Utter horse shit. Even if there was a GPU bottleneck, the i5 would still exhibit decreased performance if the game was heavily-threaded, because its minimums would be lower.

Intel shills do not accept facts. They only have 'But muh gaymes!' as their last line of defence. Even AVX512 is not going to save them next year. When Zen 2 beats Intel they will start on the whole
'Glad you could catch up. 10nm will BFTO AMD'. Line.
They are already trying to spin that line.
It never ends. The best thing to do is hide and ignore if it bothers you. I personally find it hilarious watching them flail like an injured animal in its last tortured throes before dying.

My condolences. Silicon lottery sucks. At least AMD has better yields going forward.

4Niggahurtz on 8c with 95W? more than expected tbqh. still a crap chip.
since intel know it's inevitable death in server and desktop they just want to kill future AMD buyer with their ovens™

Ryzen runs cool while Intelaviv cant stop themselves from thermal throttling ON A NH-15.

So AMDrones tell me why do you defend AMD so much? I know they're slightly competent now, but even when they were a completely shit company you guys still tried to defend their hardware. What's in it for you? Are you trying to justify that Bulldozer CPU purchase? I'm not trying to bait just genuinely curious.

Attached: 1539716914147.png (435x637, 299K)

So Incels tell me why do you defend Incel so much? I know they're slightly competent now, but even when they were a completely shit company you guys still tried to defend their hardware. What's in it for you? Are you trying to justify that Netburst CPU purchase? I'm not trying to bait just genuinely curious.

Can't wait for 16 core consumer Zen 2

two 7nm chiplets + one 14nm IO hub with huge eDRAM cache is going to be amazing.

>Zen 2
Gibs, gibs, gibs!

> I'm not trying to bait

I kind of doubt they will go for 16 cores, that seems like massive overkill (and eating into threadripper sales) when it seems they have a good performance uplift on their hands already that removes the primary advantage Intel had. Maybe when they move to a new socket.

>16 core consumer Zen 2
it's called Threadripper. don't get too excited for moar cores on AM4 lad.

>Be AMD
>Avoid yaab

It's literally two 7700Ks

Basically Zen 2 is already confirmed to have 16 cores on AM4.

It will be basically Thread-Ripper for poor people

Intel will have nothing to compete against this monster.

>Basically Zen 2 is already confirmed to have 16 cores on AM4.
Its not, and 16 cores with only dual channel memory will suffer from it. Two channels even with 4000mhz DDR4 is pitiful bandwidth given the core count.

ok Reddit, calm down.

Facts don't care about youre feelings sweatie

okay this is epyc

>Muh zen is marginally better
Yet Intel with 6 cores manage to equals or even get better than 8 Ryzen toy cores and shitty latency

You need internal justification when you buy a second grade, poor man choice product. Buying Ryzen is like going Walmart and taking whatever is in the discount bin.

How does it feel knowing Intel products only appeal to indians and chinks now?

(You)

Attached: ChzG_zrUUAEhfx8.jpg (600x328, 27K)

>200 watt
>5 Ghz
Okay, lets advertise it as that instead of "95 watt" which is basically 4 Ghz.

AMD's Ryzen is much more efficient at power.

Yep. Agreed.
8c/16t is a sweetspot for mainstream desktop. Or heck really 6c/12t is.
They need to spend that TDP on higher clocks, not more cores.

16 core isn't happening because dual channel memory isn't enough to feed it.

>700mhz slower on all cores at spec
Oy vey

The rest of the article explains how Intel's 95w TDP rating is bullshit and it actually draws >150w with all cores at 4.7ghz.

>buys expensive processor
>plays everything at low
What kind of idiot does this ? is this what /v/idya fags do with expensive stuff ?

AMD is pure shit.

based and redpilled

Do an average on all kind of workloads and you'll get an Intel advantage of less than 5%. Which is nothing.

>in a variety of workloads
you literally posted cinebench twice. how is that a 'variety of workloads'? how about actually taking more than 1 benchmark you absolute retard?
forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-72#post-39391302

See:
And average them out with all the tests at your link, and you'll find that Coffee Lake has marginally better IPC across all workloads than Summit Ridge, but when people go around claiming Intel has better IPC, they aren't usually suggesting that Intel has a couple of percentage points better IPC.

The Jow Forums shitposting who constantly repeat evidence free claims about IPC like that never mean 'Coffee Lake has 2-4% better IPC across a variety of workloads but most of the advantage is clock speed' rather than 'AMPOO FAILDOZER TIER IPC LMAO'.

But fair enough, I should have been clearer in saying the differences were small enough not to be hugely important rather than saying they were the same.