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

[teppic]

I was wondering one thing, we all know that overclocking tends to improve over time due to manufacturing changes, but how long is that normally? I'm guessing that 4.1GHz will be achievable by most at some point in this generation.

 

[h4rm0ny]

Was thinking for vm's it would run better with more cores and threads ?

As each machine I use has 2threads and 2 cores and 4 gig .

Yes. More cores and threads will make a big difference. But going from 2666MHz to 3200MHz will gain a marginal boost, I would think (subject to edge cases). I mean what are you using these VM's for? If it's development purposes to simulate client-server for example, I would not expect any noticeable difference because most of either will be idle during development loads. If you were doing something bonkers like using one of the VMs to render images on the other hand, - yes, it might make a more noticeable difference.

But in general performance follows this path of significance.

-->RAM size (hugely important if not enough, unimportant once past that)
--> GPU (if gaming)
--> Cores + Threads (+ FPUs in specific cases)
--> Disk Speed
.
. (large gap here)
.
-->RAM speed.

I do development work with multiple VMs (usually just two or three) and I'm not concerned by the difference between 2666 and 3200.

 

[opethdisciple]

Just found this randomly. It's a quote from someone that works for AMD saying the April Bios update should start hitting boards from April 11th and it should help get ram up to 3200Mhz....With another bios to come in May.

 

[prjwebb]

Has anyone seen any performance comparisons running the C16 vs C14 ram? I'm looking at the Trident Z 3200 stuff but there's a pretty substantial price jump between the two.

 

[gavinh87]

His results do not match those of mindblank nor joker or even computerbase. Which goes back to the adoredtv review. Any reviewer can make the results look good or bad. I don't know who to believe anymore.
But at least joker and mindblank show you video footage with msi afterburner running and frametimes. This guy just shows a graph and expects people to believe it.

 

[gavinh87]

Has anyone seen any performance comparisons running the C16 vs C14 ram? I'm looking at the Trident Z 3200 stuff but there's a pretty substantial price jump between the two.

The difference is non existent in every benchmark I've seen and done myself. Speed is what matters most with ryzen.

 

[prjwebb]

From a bit more digging, the C16 stuff is dual rank where the C14 stuff is single rank. People have been having a lot more success getting the rated speeds out of the single rank kits.

 

[N19h7m4r3]

His results do not match those of mindblank nor joker or even computerbase. Which goes back to the adoredtv review. Any reviewer can make the results look good or bad. I don't know who to believe anymore.
But at least joker and mindblank show you video footage with msi afterburner running and frametimes. This guy just shows a graph and expects people to believe it.

It doesn't even match Level 1 Techs, LegitReviews or PurePC

Hardware Unboxed is the exception to the majority at the moment. Then you have his fans stating things like Ryzen 20% slower in gaming and another Bulldozer, yet the majority of reviews disagree. Many of those reviews even include Ashes of the Singularity, and Total War that recently got updates, Ashes got +31% and TW got +10% in FPS.

Techpowerup had the largest set of games tested, more than many reviews combined.

 

[Scramz]

From a bit more digging, the C16 stuff is dual rank where the C14 stuff is single rank. People have been having a lot more success getting the rated speeds out of the single rank kits.

The AMD rated kits are also single rank.

 

[Matrix]

Yes. More cores and threads will make a big difference. But going from 2666MHz to 3200MHz will gain a marginal boost, I would think (subject to edge cases). I mean what are you using these VM's for? If it's development purposes to simulate client-server for example, I would not expect any noticeable difference because most of either will be idle during development loads. If you were doing something bonkers like using one of the VMs to render images on the other hand, - yes, it might make a more noticeable difference.

But in general performance follows this path of significance.

-->RAM size (hugely important if not enough, unimportant once past that)
--> GPU (if gaming)
--> Cores + Threads (+ FPUs in specific cases)
--> Disk Speed
.
. (large gap here)
.
-->RAM speed.

I do development work with multiple VMs (usually just two or three) and I'm not concerned by the difference between 2666 and 3200.

Thanks for that very helpful the Vm's are for general use nothing major really running Windows

They will be on a separate SSD disc using Physical drives ( as i'm lead to believe there faster then virtual disks ?)

Thank you.

 

[Matrix]

Another questions i am sorry to keep asking ( looking to order this week for easter )

I run a old Corsair Hydro H50-1 High-Performance CPU Watercooler ( single 120mm Fan push pull method ) as i don't have room for a double fan .

Would the Cooler Master MasterLiquid 120 AIO 120mm - CPU Cooler https://www.overclockers.co.uk/cooler-master-masterliquid-120-aio-120mm-cpu-cooler-hs-07c-cm.html fit ok it looks the same size as my old one ?

[/IMG]


Or is there any other ones i should be looking at Would rather have a liquid cooler than on air.

Thank you once more .

 

[gavinh87]

Thanks wish i could afford RGB sticks but £400ish not a chance . Was looking at these Corsair Vengeance LPX 32GB (4x8GB) DDR4 PC4-25600C16 3200MHz Quad Channel Kit - White (K32GX4M4B3200) ?


Thanks.

People have got these at 3200 others at 2666.

 

[gavinh87]

It doesn't even match Level 1 Techs, LegitReviews or PurePC

Hardware Unboxed is the exception to the majority at the moment. Then you have his fans stating things like Ryzen 20% slower in gaming and another Bulldozer, yet the majority of reviews disagree. Many of those reviews even include Ashes of the Singularity, and Total War that recently got updates, Ashes got +31% and TW got +10% in FPS.

Techpowerup had the largest set of games tested, more than many reviews combined.

Things like this will always happen, people will always try to justify their purchases. I'll stick to watching people showing us live footage with MSI afterburner running so we can all see what is happening.

 

[N19h7m4r3]

Things like this will always happen, people will always try to justify their purchases. I'll stick to watching people showing us live footage with MSI afterburner running so we can all see what is happening.

Aye, and it's getting more absurd with people upvoting comments like that on Reddit and Youtube. It's like none of them have seen any of the reviews they're commenting on.

 

[loftie]

Techpowerup had the largest set of games tested, more than many reviews combined.

Looking at their results though, that chart is for CPU tests (probably no games). If you look at the other graphs that are specifically gaming at 1080p the 7700k is 12% faster. At 1440p the 7700k is 5% faster.

 

[N19h7m4r3]

Looking at their results though, that chart is for CPU tests (probably no games). If you look at the other graphs that are specifically gaming at 1080p the 7700k is 12% faster. At 1440p the 7700k is 5% faster.

It's the overall combined for all tests, which is why the seperate the two resolutions at the end. Even at 1080p it includes some exceptionally bad games for Ryzen, some which have now been getting patches to fix the issues. Such as Ashes which a recent update added 31% FPS gain.

That was also an early review, since BIOS updates and patches have increased performance.

Even this is an early review, and found the difference was 7%, once again a review with Ashes before it got that massive 31% increased performance update.

https://www.youtube.com/watch?v=64AmlVIosAI&feature=youtu.be&t=8m20s

More and more reviews showing the gap closing, so anyone spouting Ryzen is another Bulldozer that's 20-30% slower ( while being clocked significantly higher ) is simply lying.

Is Ryzen slower than Kabylake? Yes, it's got the IPC advantage and can clock 1Ghz higher. Is Ryzen getting completely trounced and terrible for gaming as many claim? Not in the slightest; as despite the IPC and Clock advantage the difference is minor, except for some fringe cases.

 

[gavinh87]

Looking at their results though, that chart is for CPU tests (probably no games). If you look at the other graphs that are specifically gaming at 1080p the 7700k is 12% faster. At 1440p the 7700k is 5% faster.

Lose 12% gaming performance but gaining a platform that will be supported for the next 4 years and with double the cores and threads its pretty much a no brainer. I think some people are scared of change, what people are forgetting is this is good for everyone. Its also fun how nobody is mentioning the gaming performance of the 6800k....... When that came out it was just accepted, yet look how ryzen has been portrayed.

 

[N19h7m4r3]

Lose 12% gaming performance but gaining a platform that will be supported for the next 4 years and with double the cores and threads its pretty much a no brainer. I think some people are scared of change, what people are forgetting is this is good for everyone. Its also fun how nobody is mentioning the gaming performance of the 6800k....... When that came out it was just accepted, yet look how ryzen has been portrayed.

Joker Productions is, yet people are apparently dismissing it all.
Or as I posted earlier, Bitwit Compared 1700 @4Ghz vs 7700K @5Ghz and oh, lordy. Only a 7% difference in 1080p
https://www.youtube.com/watch?v=64AmlVIosAI&feature=youtu.be&t=8m20s


PlayerUnknown's Battlegrounds CPU Performance & Gameplay Overview

6800K vs 1800X

 

[gavinh87]

I guess people believe what they want to. I cant wait till the quad cores start choking. It happened with the e8400 and q6600 and it will happen again.

 

[silversurfer]

46 min vs 33 min is big news. Thats a 39% improvement though it also depends when and where that lowest point occurs.

I agree more cores will be the more versatile performance over time unless you really got an absolute specialist requirement for the single thread speed. Its been a long time coming anyhow