Full Skylake reveal result? Waiting for Zen.

[TheProgrammer]

Anyone NOT waiting for Zen? With Vulkan/DX12 able to exploit multiple cores the first mainstream, affordable 8C/16'HT' makes sense to wait for even at SandyBridge IPC. Since while Skylake ended up great for mobile, pretty much a must-upgrade... it's only an OK release on desktop.

Considering we'll be stuck with this tock until late 2018 / early 2019 with Cannonlake's successor, what you see here is what you get for 3 to 4 years. It needed something big, like inverse-HT or that Morphcore bs to sway me into a purchase.
This relative letdown was it, the big kahuna. It ended up being "SandyLake" and not much more on the desktop.

So what to do?
For me, my next build is going to be:
-AMD Zen 8C/16HT. No brainer here.
-16nm LiquidVR-compatible GPU. LiquidVR since that's the current king for VR and Arctic Islands will be well optimized in all VR titles by that point. Thanks to the pieces already being in place today.
-Crucial Optane SSD
-Ultrawide LCD. Market will mature a bit on these by then, still fairly new.
-SteamVR headset. I prefer Valve products over Facebook.

Anything less would be uncivilized.

 

[Arachnotronic]

Cool stuff. Thanks for sharing.

 

[TheELF]

What are you going to wait for if zen is gonna be a lemon?

 

[cmdrdredd]

Actually...if you look at the DX12 benchmarks available, clockspeed is king and a 6700k beats a 5960x. More cores won't help, especially not AMD cores.

 

[LTC8K6]

I like Skylake. Can't wait to build a system.

To me, Zen seems like it's still a bit of a mystery.

 

[MentalIlness]

Actually...if you look at the DX12 benchmarks available, clockspeed is king and a 6700k beats a 5960x. More cores won't help, especially not AMD cores from the past.

I'll be building around January though. And more than likely it will be Skylake since I held off a few extra months upgrading.

 

[cmdrdredd]

I'll be building around January though. And more than likely it will be Skylake since I held off a few extra months upgrading.

You can change the text if you want, AMD won't catch up to Intel. It's the same story every time.

The best we can hope for now is that they get close to Ivy Bridge or similar performance. At least then there's a viable alternative to look at. I have my doubts about them though.

 

[el etro]

Get a i7-6700K then. For DX12 gaming.

 

[jhu]

Not upgrading for a while. My FX8350 server barely does anything right now other than serve files and render. If Zen is priced right, I'll consider it.

 

[Yuriman]

I'm pretty pessimistic about Zen. I'm hoping to be pleasantly surprised, but Sandy Bridge level IPC, when compared with Skylake, is almost as far behind as Piledriver was behind Ivy Bridge. And we don't know that 8c/16t will be affordable, only that it will exist.

I was also pretty pessimistic about Skylake, and it didn't turn out to be much.

 

[tenks]

cool story bro

 

[TheProgrammer]

I'm pretty pessimistic about Zen. I'm hoping to be pleasantly surprised, but Sandy Bridge level IPC, when compared with Skylake, is almost as far behind as Piledriver was behind Ivy Bridge. And we don't know that 8c/16t will be affordable, only that it will exist.

I was also pretty pessimistic about Skylake, and it didn't turn out to be much.

I had that thought too, but in reality a 4ghz SB chip is the sweet spot, even to this day. 8C/16T of that next year and with DX12/Vulkan opening up the core race, and it'll be good. It's not like AMD is going to stop at SB IPC. But it's a good minimum to start at. It may well be better, and who knows what other goodies AMD has in store. Worth the wait IMO.

What are you going to wait for if zen is gonna be a lemon?

It would be very hard to fail here. They don't need to beat anything, just get SandyBridge IPC on 8 cores and they're set. If they can't do that after all this time and it's a lemon, the doors will be closed.

But you won't get that result as it'd be tough to pull off. Even a tiny percentage of the server market and the company is automatically back into good financial shape too. They just needed to dump BD but these things take years to transition.

That's the advantage of being Intel or AMD. Club x86.
One of the few who can chip off a tiny of bit of the very profitable x86 server market and that's all they need as a core market to float the boat going forward.

You can change the text if you want, AMD won't catch up to Intel. It's the same story every time.

Boy are you in for a learning experience. Remember the Athlon64, XP and X2 days?

Actually...if you look at the DX12 benchmarks available, clockspeed is king and a 6700k beats a 5960x. More cores won't help, especially not AMD cores.

Yup, you're new. This is a pure troll otherwise.

FYI: 4ghz or up on SB or newer and everything flattens out at that point. 4ghz SB is what tamed the single threaded DX beast. Anyone with a SB or newer CPU today at 4ghz is set, no reason to jump to SandyLake.
DX12/Vulkan on an affordable 8-core Zen at 4ghz will be exactly what many people will want. Too bad you won't be able to use it, or likely AMD's GPUs either as you're clearly too emotionally invested in being against AMD.

 

[Exophase]

It would be very hard to fail here. They don't need to beat anything, just get SandyBridge IPC on 8 cores and they're set. If they can't do that after all this time and it's a lemon, the doors will be closed.

That's assuming Intel doesn't respond with something that offers competitive throughput.

The peak throughput on big multicores is going to be heavily constrained by perf/W. So if lower power consumption allows a 6-core 10nm Intel CPU to beat an 8-core AMD CPU it'll be a win for Intel even if the single-threaded performance is irrelevant. And Intel hasn't been standing still here, enthusiast grade 6-cores have come down a lot in price generation after generation, and could be joined by 8-cores.

 

[CHADBOGA]

Boy are you in for a learning experience. Remember the Athlon64, XP and X2 days?

Remember the Bulldozer, Steamroller, Piledriver days? :whiste:

 

[moonbogg]

Zen will be Keller's chip. If it sucks with him steering the ship, then its truly hopeless.

 

[Ancalagon44]

Who says Zen will be 8C/16T? Could be 4C/8T. Or 6C/12T.

 

[erunion]

Anyone NOT waiting for Zen? With Vulkan/DX12 able to exploit multiple cores the first mainstream, affordable 8C/16'HT' makes sense to wait for even at SandyBridge IPC. Since while Skylake ended up great for mobile, pretty much a must-upgrade... it's only an OK release on desktop.

Considering we'll be stuck with this tock until late 2018 / early 2019 with Cannonlake's successor, what you see here is what you get for 3 to 4 years. It needed something big, like inverse-HT or that Morphcore bs to sway me into a purchase.
This relative letdown was it, the big kahuna. It ended up being "SandyLake" and not much more on the desktop.

So what to do?
For me, my next build is going to be:
-AMD Zen 8C/16HT. No brainer here.
-16nm LiquidVR-compatible GPU. LiquidVR since that's the current king for VR and Arctic Islands will be well optimized in all VR titles by that point. Thanks to the pieces already being in place today.
-Crucial Optane SSD
-Ultrawide LCD. Market will mature a bit on these by then, still fairly new.
-SteamVR headset. I prefer Valve products over Facebook.

Anything less would be uncivilized.

Uhh...Nvidia's VR drivers are already in the wild.

 

[Pandamonia]

Anyone NOT waiting for Zen? With Vulkan/DX12 able to exploit multiple cores the first mainstream, affordable 8C/16'HT' makes sense to wait for even at SandyBridge IPC. Since while Skylake ended up great for mobile, pretty much a must-upgrade... it's only an OK release on desktop.

Considering we'll be stuck with this tock until late 2018 / early 2019 with Cannonlake's successor, what you see here is what you get for 3 to 4 years. It needed something big, like inverse-HT or that Morphcore bs to sway me into a purchase.
This relative letdown was it, the big kahuna. It ended up being "SandyLake" and not much more on the desktop.

So what to do?
For me, my next build is going to be:
-AMD Zen 8C/16HT. No brainer here.
-16nm LiquidVR-compatible GPU. LiquidVR since that's the current king for VR and Arctic Islands will be well optimized in all VR titles by that point. Thanks to the pieces already being in place today.
-Crucial Optane SSD
-Ultrawide LCD. Market will mature a bit on these by then, still fairly new.
-SteamVR headset. I prefer Valve products over Facebook.

Anything less would be uncivilized.

16 threads is pointless on a skylake powered cpu. If you spread the load that efficient then all you need is a ps4 cpu to run stuff. You may find that suddenly gaming is only gpu bound for ever. But you still will need single threaded power for other stuff and other games.

 

[witeken]

I had that thought too, but in reality a 4ghz SB chip is the sweet spot, even to this day. 8C/16T of that next year and with DX12/Vulkan opening up the core race, and it'll be good. It's not like AMD is going to stop at SB IPC. But it's a good minimum to start at. It may well be better, and who knows what other goodies AMD has in store. Worth the wait IMO.



It would be very hard to fail here. They don't need to beat anything, just get SandyBridge IPC on 8 cores and they're set. If they can't do that after all this time and it's a lemon, the doors will be closed.


FYI: 4ghz or up on SB or newer and everything flattens out at that point. 4ghz SB is what tamed the single threaded DX beast. Anyone with a SB or newer CPU today at 4ghz is set, no reason to jump to SandyLake.
DX12/Vulkan on an affordable 8-core Zen at 4ghz will be exactly what many people will want. Too bad you won't be able to use it, or likely AMD's GPUs either as you're clearly too emotionally invested in being against AMD.

And who says it will reach that IPC, at that clock speed, at non-ridiculous power consumption, at decent price? That's a tall set of requirements, and then you will also be able to go to Intel and get something better.

 

[stateofmind]

People, wait for real gaming DX12/Vulcan benchmarks.. the current ones are way too strange, with all due respect

 

[pepone1234]

i don't think glofo 14nm will be prepared to mass produce an 8 core sandy bridge-like cpu at 4ghz.

 

[Dresdenboy]

And who says it will reach that IPC, at that clock speed, at non-ridiculous power consumption, at decent price? That's a tall set of requirements, and then you will also be able to go to Intel and get something better.

Winning 60% or 80% at the same or higher prices doesn't automatically mean that it will be the best fit for everyone.

 

[ThatBuzzkiller]

If Zen fails it'll be all over for AMD ...

Considering how long it takes to develop a moderately new micro-architecture, I don't think AMD can last another half a decade of this ordeal ...

I just hope to god that AMD management has done an extreme overhaul on their engineering team on the CPU side like bringing back Jim Keller ...

I have high expectations for Jim Keller but not so much for AMD itself ...

They HAVE to be at least competitive with Haswell on the IPC front then they can expect to have some more breathing room for quite a while since I doubt Intel will be able to pull up an improvement on pure performance in the near future with Kaby Lake ...

 

[IntelUser2000]

I think the potential for great improvement is there with Zen, but only because they are starting at such a low level.

If you assume absolute maximum achievable advancement is 100,

Intel 2011(Sandy Bridge): 70
AMD 2011: 45

Intel 2015(Skylake): 90
AMD 2015: 55

You'll see that there's only a potential for getting 11% more if they are absolutely perfect for Intel, but huge 81% for AMD. But you still end up with the same performance. So its whether you want an Intel Skylake or an AMD "Skylake".

Same with process tech. Since about 1995 Intel had 2-3 years advantage in process over the 2nd best.

Now its probably about 0.5-1 years at most. Intel process is probably still going to be the best, but the advantage will be pretty small. So everyone ends up being nearly the same, despite taking radically different route there.

I can see why this is happening. You've got smart engineers everywhere. They get trained from nearly same place and sharing of information especially with internet age means you have same knowledge and end up with same conclusions. The reason some architectures don't work(like Netburst, EPIC, Bulldozer) is probably because it relies on something that doesn't exist - perfection. Practical architectures based on limitation of the engineer and design/process team ends up being much the same in the end.

 

[NTMBK]

i don't think glofo 14nm will be prepared to mass produce an 8 core sandy bridge-like cpu at 4ghz.

They're just licensing it from Samsung, they can't screw it up this time, right?

...right?