AMD Zen processors...Are you interested?

[dadev]

Well last I checked AMD will never serialize it as their compute queues all have a hardware scheduler attached.
2 if you have the RX 470/480 series.

The only time they will lose out performance on Async Compute implementation is if the game developer fucks it up royally.
AMD's hardware is literally built around hardware "multi-threading" where nVidia's hardware is emulating it with software.

nVidia's driver overhead is indeed smaller and they (last I checked again) have better performance with deferred context but DX11 still has to send those via the main processor thread and no way around this.
The whole DX12/Vulkan unleashes this concept for PC allowing all CPU threads to communicate with the GPU rather than just the main thread in DX11.
Not sure if OpenGL follows the same routine or actually has more "multi-threading" capability than DX11 so I won't comment on that

OpenGL is pretty much dead at this point. The spec is far behind on anything that has to do with multi threading.
And yeah, of course AMD has very little reason to serialize the queues, but they can still do it if they need to, like that bug on GCN1. And yes, developer would have to make some really questionable decisions in order to lose performance with an extra compute queue(s), but this can happen, as much was said by AMD in a post on gpuopen site.

[Evildeffy]

OpenGL is pretty much dead at this point. The spec is far behind on anything that has to do with multi threading.
And yeah, of course AMD has very little reason to serialize the queues, but they can still do it if they need to, like that bug on GCN1. And yes, developer would have to make some really questionable decisions in order to lose performance with an extra compute queue(s), but this can happen, as much was said by AMD in a post on gpuopen site.

They CAN do it indeed but like I, and you, said in order to drive it that far on AMD's hardware which has it's hardware designed around it the developer would have to make some seriously questionable decisions/bad implementation/engine incompatibility/being bribed/not give a fuck (which is weird to me because devs been screaming for it for years on PC platforms to have that ability) .... you see where I'm going with this I take it.

Volta will likely feature a more GCN-like architecture since it was clear that nVidia underestimated how fast the new DX12/Vulkan APIs would spread.
And to catch up they will need to do more like AMD and with that they will lose their vaunted "energy efficiency" as adding hardware schedulers is very much a reason why AMD cards eat more Watts than their nVidia counterparts.

We will see where Volta leads to soon enough.

[Denpepe]

OpenGL is pretty much dead at this point. The spec is far behind on anything that has to do with multi threading.

I thought Vulkan was just a "new" OpenGL with a "sexier" name, it's from the same khronos group.

[Evildeffy]

I thought Vulkan was just a "new" OpenGL with a "sexier" name, it's from the same khronos group.

Vulkan however is AMD's Mantle given to the Khronos Group for them to do with as they please.
They reworked it and voilà ... you have Vulkan.

[dadev]

I thought Vulkan was just a "new" OpenGL with a "sexier" name, it's from the same khronos group.

It is the new OpenGL. Although it shares almost no real similarities with OpenGL 4. Some of the naming conventions remained the same, but other than that it's brand new spec. GLSL (OpenGL's shading language) is still the mainstream for Vulkan, but it's no longer part of the spec (and you don't have to use it if you have alternatives).

The branding was different probably to keep the door open for future standard OpenGL updates, or because (as noted above me) the origins came from Mantle by AMD and maybe Khronos wanted a new name for it because of that. But I really didn't check into branding, so it may be wrong.

[Turaska]

I would think the RX 480's are a raging success. Don't know where you get that they run hot. JayTwoCents proved that the 480 can run cool and low power if you give it a better cooler. He was able to make it pull around 100W.

Most enthusiasts are recommending RX 480 but for whatever reason most people actually buy a 1060. I consider the 3GB 1060 to be wrong on many levels. The 1060 6GB has value if and only if you spend $250 and no more. And remember Nvidia milked their 1060 for $300 for the Founders Edition.


Remember Intel CPUs can only be overclocked if they're the K parts. That limits overclocking to 6600K/6700K/7700K. Where we know AMD plans to not limit Zen's overclockability. Zen will even auto overclock depending on your cooler. Though it's really just Turbo mode, with a tweak.

An i5 6600 is 3.90 GHz and won't go higher and costs $220. The quad core Zen will be cheaper with HT and will have manual overclocking.

I bought an RX480 on day of release, it ran hot as hell with stock coolers / none overclocked. Replaced it literally 11-14 days later with a 1070.

[Vash The Stampede]

I bought an RX480 on day of release, it ran hot as hell with stock coolers / none overclocked. Replaced it literally 11-14 days later with a 1070.

I just bought a Sapphire RX 480 4GB and haven't had heat issues. Runs mid 70C temp under load which to me is so much better than my previous HD 7850 stock cooler, that I didn't bother to water cool it. Of course the fans do turn up in speed and it gets annoying cause I'm used to my quiet custom loop setup. I plan to water cool it just to eliminate the fan noise, but first I need to upgrade my radiator. I got two 120mm radiators and the temp with the CPU+VRMs+GPU was reaching 60C. I swear I don't understand how you people deal with fan noise without a custom water loop?

Anyway, everyone knows stock coolers suck and really shouldn't be what you go by. Also I don't believe you bought a 1070 due to heat issues over the RX 480. If that were the case you would have bought a 1060. The 1070 is a huge step up from $250 graphic cards.

[Life-Binder]

[Evildeffy]

Care to translate?

I don't read French and whilst I can guesstimate it doesn't mean a damn thing unless it actually tells me stuff properly in detail with the right context.

[Remilia]

If it's an actual sample, it's probably an ES one since it says base block is 3.15GHz boost is 3.4GHz, when we've already gotten info that base will be 3.4GHz. That said, it's pretty good considering the clock deficiency. It'd place it around Broadwell to Skylake IPC with clock normalization. Question really becomes what clock rate the end result is that you can achieve or what their 25MHz step auto clock can achieve.

Edit: Apparently the ES sample was running around 3.2-3.3GHz with just turbo.

[Evildeffy]

If it's an actual sample, it's probably an ES one since it says base block is 3.15GHz boost is 3.4GHz, when we've already gotten info that base will be 3.4GHz. That said, it's pretty good considering the clock deficiency. It'd place it around Broadwell to Skylake IPC with clock normalization. Question really becomes what clock rate the end result is that you can achieve or what their 25MHz step auto clock can achieve.

Edit: Apparently the ES sample was running around 3.2-3.3GHz with just turbo.

You read French? Care to do a full translate?

[Butler to Baby Sloths]

Care to translate?

I don't read French and whilst I can guesstimate it doesn't mean a damn thing unless it actually tells me stuff properly in detail with the right context.

The text is just describing what the bar charts show. First one is video encoding. Second one gaming performance. Third one power draw.

Summary at end (my french is fairly solid but don't take the translation as gospel):

Zen is a major step forward in the high end cpu market, one that has been largely ignored for a long time. Intel risks losing its lead after the last few years of complacency (1900 euro for 6950X). if we are to expect a true competitor on the market, AMD hasn't quite won it yet. while the 8 core seems to be on track, AMD needs to finalize the quadcore model as well as the boost clock speeds - it would need between 3.8 and 4.2GHz to stay competitive with Kaby Lake. Finally, they need to publish the price. some rumours are talking about very low prices, we doubt that AMD will price its CPU too low now that they have an opportunity to recover after years of losses. Just so long as they don't fuck this up so close to the finish line.

[Remilia]

You read French? Care to do a full translate?

I do not read french, but looked at the pretty numbers / graphs and the only few things I can recognize, lol.

[Astromedes]

I'm excited in the sense, IF AMD pulls it off, that competition will drive Intel "back to innovation" instead of these very discrete improvements year after year. Kaby lake for example... That being said, I'm still holding anything AMD is showing with a grain of salt. We all remember Bulldozer, don't we? Unless AMD offers me the same performance for a noticeably cheaper price point, they will not see my business again. You all also need to consider the AM4+ socket. From what I've seen so far, I feel they missed the mark for an enthusiast build, still more than enough for an average user though. Price is key. Realistically, I don't believe You'll see an 8c/16t AMD chip priced significantly lower than the Intel equivalent nor will it be enough to rip a noticeable market share from Intel. But, hey, I've been wrong before so we'll see. Gl/hf AMD.

I do long for the days of the Athlon 64. Believe it or not, kids, but Intel wasn't always top dog. AMD was crushing it in the early 2000's.

[Life-Binder]

the best thing that might come out of all this is a mainsteam (aka on the ZX70 platform instead of X99 .. and priced like a 6700K/7700K or a bit higher) Intel Coffee Lake 6c/12t next year or early 2018

[Evildeffy]

the best thing that might come out of all this is a mainsteam (aka on the ZX70 platform instead of X99 .. and priced like a 6700K/7700K or a bit higher) Intel Coffee Lake 6c/12t next year or early 2018

Coffee Lake is earliest Q2 2018 according to Intel's own slides.

[Butler to Baby Sloths]

the best thing that might come out of all this is a mainsteam (aka on the ZX70 platform instead of X99 .. and priced like a 6700K/7700K or a bit higher) Intel Coffee Lake 6c/12t next year or early 2018

Coffee Lake to wake Intel up from their innovation siesta?

[Life-Binder]

Coffee Lake to wake Intel up from their innovation siesta?

pretty much

imagine Kaby Lake or at least Skylake level OC (4.8-5.0+) on a 6c/12t with further IPC improvements (even if they're not big compared to Skylake)


would be the ultimate gaming CPU for the foreseeable future (while also having 6 cores for streaming/video) .. and if Zen is good - it could come at mainstream prices on regular (aka not X99-tier) mobo/socket

Coffee Lake is earliest Q2 2018 according to Intel's own slides.

you have a link ?

Q2 2018 is not too bad though

ideally would be if it also came @ 10nm, but I dont think 10nm will be that good (to make a powerful hexacore) by then

[Evildeffy]

pretty much

imagine Kaby Lake or at least Skylake level OC (4.8-5.0+) on a 6c/12t with further IPC improvements (even if they're not big compared to Skylake)

would be the ultimate gaming CPU for the foreseeable future (while also having 6 cores for streaming/video) .. and if Zen is good - it could come at mainstream prices on regular (aka not X99-tier) mobo/socket

Skylake does not OC to 4,8GHz unless you have a golden sample, Kaby Lake should however but I hope you have a beast of a cooler to not have an oven baking inside your PC.
Also presuming clocks is not a smart idea... which is why we don't do that yet with ZEN either.

you have a link ?

Q2 2018 is not too bad though

ideally would be if it also came @ 10nm, but I dont think 10nm will be that good (to make a powerful hexacore) by then

I only have an original link that it would possibly see a release @ Feb 2018 (rumoured) and that was before the 10nm delays.
Kaby Lake wasn't supposed to even exist at this time and would be Cannonlake and Kaby Lake was only introduced to fill the gap between launches.

There is a Dutch report on Tweakers.net that also states every single chipmaker is having a lot of issues and are stating delays incoming for all of them.

To make matters worse for you .. Cannon and Coffee Lake are nothing but die shrinks of Kaby Lake and the next architecture change won't be until at least 2019 and that was again before the 10nm delays.
So cannon and coffee lake will represent an almost insignificant IPC increase, if there is any, and will mostly focus on improvements, which will not be as big as the difference between 14 and 10nm is not that large.
Mind you that the current plan for Cannonlake is mobile chips and Coffee Lake is mobile chips and low power chips of 45W TDP.
Don't expect anything of this to rival high speed processors of Sky and Kaby Lake's i7 6700K/7700K series.

The size may allow for hexacores but considering the envelope is 45W TDP ... don't hold your breath.

To be fair the whole Skylake-X / Kaby Lake-X / Coffee Lake-X is a wash right now.
Supposedly Skylake-X will feature up to 10C/20T just like Broadwell-E (name change from -E to -X) but Kaby Lake-X will feature only 4C/8T max. and Coffee Lake-X will feature only 6C/12T max.
All of which are also on LGA 2066.

For now I'd say count on Kaby Lake to be the "speed pinnacle" for gaming overall untill Q3/Q4 2018 at least on Intel's side.

If we're waiting on a new architecture development however... 2019 for that with Icelake, which on the roadmaps was Q3/Q4 2018 prior to delays.
Any plans with the die shrinking I'd say you should honestly not consider at all, AMD will be facing the same issue with RYZEN which was called ZEN+ and also "slated" for 2019 because AMD wanted to skip the 10nm process and go to 7nm directly because of aforementioned issues.

[Denpepe]

Coffee lake is 14 nm still according to this http://wccftech.com/intel-coffee-lak...e-in-disguise/

And while the new x-line of CPU's will still be LGA2066 it will use a X299 chipset iso the older x99 http://wccftech.com/intel-x299-chips...x-kaby-lake-x/