AMD Confirms, Zen On Track For Q4 2016 Availability On High-End Desktops

[NTMBK]

Why do test sites use Fallout 4 for benchmarking? FO4 uses an engine, which origins date back to the late 90's. In older titles, such as TES Skyrim and FO: NV etc. the physics of the game go completely mad when FPS goes above 60. FO4 also has this bug, but physics don't immediately go wild. Physics in FO4 will go wild at some point, but there are some variance to it. For someone it might work fine no matter the FPS, but for some it gets broken immediately after 60 FPS. There is no explanation for this phenomenon yet.

Because Fallout 4 is one of the most popular games of the last year. People care about how well it will run on hardware.

Not much point benchmarking games that nobody wants to play.

 

[Shehriazad]

So Zen is on track....but I wonder how this will work out with the AM4 socket.


It's supposedly coming like half a year before Zen...with Bristol Ridge as a "placeholder"...but I wonder if that has had any (negative) impact on the chipset/board design for Zen.

I hope not. I really hope that Bristol Ridge is really just a tiny bandaid to give people an early chance to switch to the new socket with DDR4 etc and try to remain relevant enough in the CPU market in the meantime...or whatever the plan is here.

 

[Headfoot]

Paying 500 bucks on a whole new platform just to play Fallout 4 doesn't seem like good value to me. Now if gaming improvements were 40% in every game that might be different, but we all know that is not the case.

Techspot's results seem different with a stock 2500K doing a lot better - only 21% behind with a similar clock speed disadvantage to the 6700K.

Given some other benchmarks I have seen, the probable reason is that Fallout 4 is just a bad benchmark.

Intel might as well just send Bethesda a check. I literally just went and upgraded to a 5820k from a 2500k primarily for Fallout 4 (as well as other CPU limited games, like kerbal space program. More boosters). And I upgraded to that 2500k in 2011 after Skyrim came out because that game ate CPU back in the day too. More on point, I didn't wait on Zen because even optimistic opinions are that matching Haswell is probably about the best it'll do, and it could possibly be a lot worse than that, and Broadwell E is shaping up to be better than expected so my upgrade path is still open. That, and I can buy Haswell E now and I know its performance to be what it is, which is very good.

Hopefully this 5820k will last through to TES 6. I still have the desperate and naive hope that the next TES will finally be the one where they build a new engine ground up, to take advantage of DX12/Vulkan. In all reality, it'll still be the gamebryo based engine they've had since Morrowind with more neato stuff added in yet still having the strange quirks like physics locked to FPS. Probably the Fallout 4 engine + enhancements. But a boy can dream

 

[mysticjbyrd]

It will probably obliterate Haswell-E in heavily multithreaded applications. That likely won't include games in the near future though.

 

[ShintaiDK]

It will probably obliterate Haswell-E in heavily multithreaded applications. That likely won't include games in the near future though.

If it does that at 95W and dual channel, then AMD can expect to get all the server share essentially.

 

[Phynaz]

It will probably obliterate Haswell-E in heavily multithreaded applications. That likely won't include games in the near future though.

I don't think you know what Haswell E is.

 

[mysticjbyrd]

If it does that at 95W and dual channel, then AMD can expect to get all the server share essentially.

It's a small lithography, so of course it will consume less power. Could potentially make for large OC potential as well.

I don't think you know what Haswell E is.

An old i7 with 2 additional cores, at the cost of removing the igpu.

 

[ShintaiDK]

It's a small lithography, so of course it will consume less power. Could potentially make for large OC potential as well.

In that case you should stock up on AMD stocks. Because you expect Zen to beat Broadwell-E in performance/watt as well. If you claim is true the stocks will go up 10-20x.

 

[mysticjbyrd]

In that case you should stock up on AMD stocks. Because you expect Zen to beat Broadwell-E in performance/watt as well.

Even if that was true, it wouldn't necessarily mean AMD stocks would increase. The better product doesn't always win.

 

[ShintaiDK]

Even if that was true, it wouldn't necessarily mean AMD stocks would increase. The better product doesn't always win.

It does in the server.

 

[mysticjbyrd]

It does in the server.

What servers are using haswell-e over xeon?

 

[ShintaiDK]

What servers are using haswell-e over xeon?

Haswell-E is a Xeon.

 

[Phynaz]

An old i7 with 2 additional cores, at the cost of removing the igpu.

You seem to have forgotten about the octacore E processors.

Do you wish to revise your statement about Zen "obliterating" Haswell?

 

[mysticjbyrd]

Haswell-E is a Xeon.

But your argument was that Zen's TDP was only 95 watt. If it's just an OC'd xeon, then why can't AMD just OC Zen?

You seem to have forgotten about the octacore E processors.

Do you wish to revise your statement about Zen "obliterating" Haswell?

It will obliterate the 6 core versions, which most people buy, and at least match the 5960x.

Not a lot of people are going to buy $1000 5960x, especially when they could buy a $400 Zen.

 

[ShintaiDK]

But your argument was that Zen's TDP was only 95 watt. If it's just an OC'd xeon, then why can't AMD just OC Zen?

Its not an OCed Xeon. The best 8 core Xeon is the 2667v3 at 3.2/3.6Ghz, a tiny bit faster than the 5960X at 3.0/3.5Ghz.

 

[Burpo]

It will obliterate the 6 core versions, which most people buy, and at least match the 5960x.

Not a lot of people are going to buy $1000 5960x, especially when they could buy a $400 Zen.

You think a $400 AMD product's going to match the 5960x?
Dude, where do you get your pot?

 

[ShintaiDK]

If it matches, it at 95W and dual channel. It will rather be 1000$ against a 800$ 8 core 140W Broadwell-E with quad channel or something

 

[MrTeal]

You think a $400 AMD product's going to match the 5960x?
Dude, where do you get your pot?

Of course a $400 AMD product will match the 5960X in heavily threaded workloads. It won't be AM4 Zen, but they'll get there some day. If not, they'll probably be working server pulls of high core count server parts available from liquidators sometime in the early 2020s that will match HW-E in multithreaded stuff in the under $400 price range.

 

[mrmt]

You think a $400 AMD product's going to match the 5960x?
Dude, where do you get your pot?

Since Zen is a server product how do you think AMD platform will stack up against Grantley? If they can't match Grantley they will again end up selling a server product on the desktop market, and this won't be pretty.

I mean, it's quite strange that Lisa is talking about Zen for enthusiasts when Zen was supposed to be a server-first product.

 

[mysticjbyrd]

Of course a $400 AMD product will match the 5960X in heavily threaded workloads. It won't be AM4 Zen, but they'll get there some day. If not, they'll probably be working server pulls of high core count server parts available from liquidators sometime in the early 2020s that will match HW-E in multithreaded stuff in the under $400 price range.

Core i7 Extreme - August 29, 2014

https://en.wikipedia.org/wiki/Haswell_%28microarchitecture%29

Intel is just taking the piss, because they have no competition in this market. Haswell-E launched in August 29 of 2014!

Zen probably won't compete well for top performance v/s Broadwell-E, which is supposed to launch in a few months, therefore AMD will have to rely on undercutting Intel, just as they always do.

The 5820k is only $400, so I am not sure how you even think this is a stretch really.

It might launch a lot higher, because AMD is trying to cash in on the fanboys, but I am willing to bet the prices come down quick. Assuming, they have decent yields to meet demand that is.

 

[mrmt]

It might launch a lot higher, because AMD is trying to cash in on the fanboys, but I am willing to bet the prices come down quick. Assuming, they have decent yields to meet demand.

That would be so AMD. They are masters in pricing themselves out of the market.

 

[mysticjbyrd]

I remember the CEO of AMD claiming that it wanted to move out of the budget CPU market. I used to thin that Zen was meant to try to compete with the typical desktop cpu market, but I think we all know that won't happen for gaming. Kabylake will crush it in single threaded performance.

But maybe that isn't their goal at all? Maybe they were planning to compete with the enthusiast chips all along? If they can't compete with the typical desktop cpus, perhaps they will instead target the enthusiast crowd with their 8 core cpus. That means AMD would essentially be forfeiting the typical gaming market, $200-$300, altogether.

Wouldn't that mean they destroy their budget build niche though? Then again, you could argue they don't even fill that niche anymore. Purchasing an i3, and then upgrading it to an i5 later would likely be the better method for a budget system.

 

[ShintaiDK]

AMD tried to sell FX9590 at 800$+ in 2013. They will take all the money they can. AMD have never been a budget company by choice, but because they couldn't get more for what they deliver. If Zen performs how you expect, you will pay out the nose. And possible make the 5960X look cheap.

The Nano just got a 150$ price cut for the same reason.

 

[mysticjbyrd]

$800 for an OC'd 8350 with a mediocre water cooler, lol.

Not the best way to treat their few loyal fans.

 

[ShintaiDK]

The error being a loyal fan to a company. Only a fool would do that.