*** AMD "Zen" thread (inc AM4/APU discussion) ***

[linktoinsanity]

From what I've read and heard it is looking good. Still to be seen, but I am hoping AMD will give Intel a run so that there is some proper competition again.

 

[africanos23]

 

[Hagar]

Aye was mentioned by "Jays2cents" on YouTube that both the latest Intel and AMD chips will be Win10 only.

Almost certain to be Linux kernal support, enthusiast and server OS will determine this. Microsoft will probably only support it fully in win 10 though.

 

[Besty]

Will be interesting to see what the lower Zen chips can do, Blender is a program AMD traditionally do not perform well in, that is why they picked it for a Demo, to show their performance increase is real and not just a chery picked test that favours them.

Check the Blender test results on the AMD Reddit, you will see Intel chips are almost 2x faster than AMD chips at Blender.

Now fast forward to Ryzen, 150 samples, its beating the 6900k. Handbrake, its beating the 6900k.

Yet people still say this is a cherry picked result... surely you would only find applications that you traditional shine in if your cherry picking? why would you show results in a program you traditionally struggle in? unless of course you want to make a point that you have turned things around.

well said.

 

[TonyTurbo78]

Finally something to shake up the CPU market, still on a 3770K doing 4.8GHz when benching here, not been any need to upgrade in years.

 

[Jeps]

I watched the stream replay last night and thought it looked very promising.

They ran the Zen chip at 3.4Ghz with turbo disabled against a stock I7 6900k which is clocked at 3.2Ghz but can boost up to 3.7Ghz so they left the actual speed the I7 was running at a bit ambiguous but the zen chip was able to match it.

So worst case senario, Zen at 3.4Ghz can match intels top I7 running at 3.2Ghz (assuming it doesn't boost under max load) which would mean they are slightly behind on IPC whilst only using 95w to Intels 140w. The Zen chip will also have the ability to boost its clock speed automatically based on how good your cooling solution is so I do expect a "stock" (as in not manually overclocked) Zen to outperform a stock I7 if you have a decent cooler.

Overclocking results will be more interesting. With its lower TDP it might clock better then the I7 which would negate the slight IPC advantage the I7 has. It would be a big win if it is cheaper then the I7 and actually comes out faster when both are overclocked!

 

[Martini1991]

My point is very simple.

A few weeks ago AMD fans would be up in arms over blender benchmarks showing that their fx83's are half as fast as a 4700k. For the simple fact it doesn't happen anywhere else, even in the same type of work loads (cinebench). There's no way they'd put stock in blender result after that.

There's no arguing against that.

This is why things like the RX 480 end up getting hyped into being Fury X for 200 quid etc.

 

[StarShock]

The giant questions are how much, and what variants will there be. I'm quietly hopeful but even if what we've seen is completely on the level AMD needs to not screw this up. A rival to the 6900k is great and all, but if it comes in at £900+ no ones going to care. We need mainstream CPUs with mainstream prices.

Come on AMD don't let us down

 

[Jeps]

My point is very simple.

A few weeks ago AMD fans would be up in arms over blender benchmarks showing that their fx83's are half as fast as a 4700k. For the simple fact it doesn't happen anywhere else, even in the same type of work loads (cinebench). There's no way they'd put stock in blender result after that.

There's no arguing against that.

I'm not sure if the AMD fans would be up in arms but it's a valid point. Until we get more benchmarks across a wide range of applications we won't really know the full picture. I'm remaining optomistic though based on what I've seen so far

 

[nikebee]

Firmly sitting on the fence until impartial third party benchmarks and pricing is released. Honestly, how often do we have to jump aboard the hype train before launch only to find out that things aren't as well as a press conference suggested before we start being a bit more sensible?

I'm interested enough that I'm actually debating holding off building my new rig in early Jan which was going to be a Skylake machine (based on the fact Kaby Lake doesn't seem to offer any real world performance increase)...

 

[Martini1991]

I certainly don't see any reason to be pessimistic, and you'd be foolish to put any serious money into Intel at the moment.

 

[Scougar]

I think I am at the point now where I don't think I really care too much about what I upgrade to, as anything will be a sizeable upgrade from my current A7850K and will also give my GTX 1070 a decent boost. If it's a fair price I'm likely to give Zen a go as long as there is a pro/con stuff is OK.

 

[SiDeards73]

My point is very simple.

A few weeks ago AMD fans would be up in arms over blender benchmarks showing that their fx83's are half as fast as a 4700k. For the simple fact it doesn't happen anywhere else, even in the same type of work loads (cinebench). There's no way they'd put stock in blender result after that.

There's no arguing against that.

This is why things like the RX 480 end up getting hyped into being Fury X for 200 quid etc.

While i agree with what you are saying, my point remains, AMD know their CPU's were inferior to Intel, they know that Blender showed them in a bad light, their IPC and threaded performance is nowhere near good enough when compared to Intel.

They have obviously invested in fixing this with Zen, the point they were making with Blender is exactly this, to show where they traditionally are bad, they are now good, not only good, but better than the competitor.

Cant understand why this point is hard to get? the best way to show off your new product is showing it excelling in an area where it was traditionally bad.

 

[Martini1991]

Because people generally don't use blender.
If they wanted to show improvement they'd use cinebench, which everyone uses. That would showcase improvement in a far better way.

I'm not debating about how good Zen is or isn't.

 

[drunkenmaster]

Because people generally don't use blender.
If they wanted to show improvement they'd use cinebench, which everyone uses. That would showcase improvement in a far better way.

I'm not debating about how good Zen is or isn't.

Cinebench is an Intel owed benchmark in effect, it's always been pro Intel and no one 'uses' it, it's a benchmark with little real world value.

People actually do use blender, and thus it's performance actually matters.

If Cinebench showed AMD 30% behind, but blender showed AMD ahead, then when you buy a chip and use Blender, you'll have a faster chip, when you buy a chip and DON'T use Cinebench, because it's worthless, who cares that it is behind, it's still worthless.

How utterly stupid of AMD not to play the game of winning a benchmark that is meaningless(and Intel owned/biased) and instead show performance in an application people actually use.

 

[Martini1991]

The intel owned benchmark doesn't stand up.

In the none Intel owned benchmark the AMD performs worse comparatively to the intel owned benchmark.

Cinebench isn't even a 50% difference. It's more like ~30
Yet blender is a better benchmark which shows AMD at half speed to an i7 4770k?
Well that's me sold.

 

[nashathedog]

The giant questions are how much, and what variants will there be. I'm quietly hopeful but even if what we've seen is completely on the level AMD needs to not screw this up. A rival to the 6900k is great and all, but if it comes in at £900+ no ones going to care. We need mainstream CPUs with mainstream prices.

Come on AMD don't let us down

This is the biggest worry, If AMD price it above 500 I won't bother as I can't justify the cost at the moment, First off I need to move to Vega then depending on the cost of that I'll either sit it out on my 4790k until Intels mainstream 6 core releases or move to Ryzen if they do not charge an arm and a leg, However, based on what they did with the FX9590 and how they keep focusing on the 8 core Ryzen being a competitor for the 1100 pound 6900k I don't like our chances. I'm not interested in getting another 4 core, 8 thread cpu but I'm also not interested in paying too much for more cores.

 

[Mtom]

I'm like many others sitting on a [email protected], and the only reason i was thinking about upgrade is DDR4, as all the newer CPUs are meh upgrades.
If ZEN will be good i might go that way with the DDR4 system upgrade, but i think i don't really need a 8C/16T CPU, the 4/8 would be enough.
So i'll wait for prices/benches/availability.

 

[Deleted member 66701]

Because people generally don't use blender.
If they wanted to show improvement they'd use cinebench, which everyone uses.

**** off. Nearly everyone on our Uni's game design course uses Blender for asset creation and nearly every game studio I've visited I've seen someone using Blender.

 

[~>Dg<~]

i think we mean in general.dont we.

not many as a perecentage will use blender.

what most who are probably ready for upgrading want to see is popular new game benchmarks.editing programs people use a lot.encoding progs people use a lot.

not random things that the majority of us wont use.

when pretty obscure benchmarks are shown it often tells you why in itself.

if it was as fast as the 6900 in games for eg why would you not show this ? that would be massive hype and there would be queues forming to get them.but nope just blender.everyones fav prog