General Discussions

[colesdav]

Zen 3 announcement due on on Octrober 8th 2020. What are your thoughts about it?
Will you be upgrading your existing processor and motherboard immediately after reviews or have you decided to wait for a while and see what user experience is with the new processors and motherboards?

[colesdav]

AMD sets October 8 date for next-gen Ryzen 'Zen 3' launch | PCWorld 

[colesdav]

October 8th launch may not be according to this:
https://translate.google.com/translate?sl=auto&tl=en&u=https%3A%2F%2Foverclocking.com%2Fune-idee-plu... 

[black_zion]

AMD will likely make it official in about 2 hours, but with ASUS releasing NEW X570 and B550 motherboards, it appears Zen 3 will not be accompanied by a new chipset. Majorly disappointing...

https://wccftech.com/asus-unveils-rog-crosshair-dark-hero-rog-strix-b550-xe-motherboards-amd-ryzen-5000-zen-3-cpus/

[ajlueke]

Makes sense really.  DDR5 and PCIe 5 will likely be added in with the changing of the socket.  Not sure if Zen 3 is the last AM4 iteration, or if there is one more.

[pokester]

Not sure why we need a new chipset at this point. Unless they introduce DDR5 or PCIe 5 the current generation boards should suffice. It does look like some of the makers are making new boards however. So maybe even though it is the same chipset number, it may contain some stability fixes and features that may be desirable. Myself I am always glad when I don't have to buy a new board. 

Unless AMD gets Zen4 out the door in less than a year. Intel may have their DDR5 and PCIe 5 boards out before them.

[pokester]

So after watching the presentation it looks like Zen3 does not disappoint. Seems to be all at the same price points as the prior Zen launches so that is good. 

Not sure if that Borderlands clip helped though. To me it looked really choppy or a little stuttery. That could also be the video capturing they used and not representative of the card at all. It however just didn't come across well. The benchmark numbers they showed looked good but not being shown against any other gpu it's hard to know whether to be impressed or not. I am sure in very short order the tech sites will have it all broken down for us.  

[colesdav]

RE: So after watching the presentation it looks like Zen3 does not disappoint. Seems to be all at the same price points as the prior Zen launches so that is good. 

I thought the IPC claim of 19% is good. Single core performance on Cinebench impressive.

RE: Not sure if that Borderlands clip helped though. To me it looked really choppy or a little stuttery.
That benchmark runs  as stuttery and choppy as hell on Ryzen 2700X at 4.3 GHz and RX5700XT at stock. I tested it at 4K Ultra. I also recorded it using ReLive. It is not the recorder that is causing it.

RE: That could also be the video capturing they used and not representative of the card at all.
Doubt it. Stutter is still a problem with Navi 10.

RE: It however just didn't come across well.
I agree.
They should not show it.

RE: The benchmark numbers they showed looked good but not being shown against any other gpu it's hard to know whether to be impressed or not. I am sure in very short order the tech sites will have it all broken down for us.  

Yes it is hard if not pointless to try to guess.
I can't be bothered.

[pokester]

So is the stuttering in that game an AMD GPU only issue or do you know if it does that with Nvidia too? Because if it is AMD only and they did use the new 6900xt for that video then that may be bad. 

Also did they mention what if any cooling solutions ship with the new processors? I have not had time to watch the whole thing. 

[colesdav]

RE: So is the stuttering in that game an AMD GPU only issue or do you know if it does that with Nvidia too? 
An RX Vega 64 Liquid is smoother than an RX5700XT in that game and also in Forza 4 Horizon.

I will double check how Nvidia RTX2080 GPU runs with latest drivers but I do not remember seeing a problem.  

RE: Also did they mention what if any cooling solutions ship with the new processors? 

No, not mentioned at all.
Perhaps the follow the trend with top end 3000XT and you buy your own?

[colesdav]

Well I just watched it: WATCH: AMD Ryzen Desktop Gaming Chip Reveal - Livestream - YouTube 

Skip first 4:30 - Hype Intro. Next minute - COVID, Microsofy and Sony Consoles. Start talking about Ryzen History at 6:00. 7:00 Minutes. Talks about Ryzen progress. 8:20 - starts talking about Zen 3. Claiming best power efficiency, mutithreaded & Single core performance - Zen3.

[pokester]

I was really hoping to see a lot more substance than what we got. 

[colesdav]

I think the above slideshots and comments pretty much sum up the new information we were given.
Not much.
I was hoping, for instance, that there woiuld be some combination of CPU+GPU for raytracing.

[colesdav]

Correction - was hoping, for instance, that there would be some combination of CPU+GPU solution for raytracing.

[colesdav]

Zen 3 CPUs available from November 5th. Zen 4 on 5nm is in design already.

[colesdav]

19% IPC Improvement.

[colesdav]

Only 4.8GHz boost.

[pokester]

The 5950 is 4.9. It sure would have been nice if they could have gotten that to 5.0 default. My MSI board for instance I can raise the boost up to 200 mhz. So that would have it hitting 5.0 for sure. Assuming that function still works with these processors. Regardless with what they showed it looks like they will have the performance crown across the board on a lot of metrics. Obviously the 3rd party testing will show the true performance. 

[colesdav]

Not sure which GPU is used in this comparison ...

[colesdav]

Which GPU?

[colesdav]

Single Thread performance.

[pokester]

I think they said the4 3950 hits 640.

[colesdav]

Mainstream Lineup.

[pokester]

I hope they end up with a 65w 5700X in this lineup. 

[colesdav]

Pricing:

[pokester]

I think they said the 5950 is $799.

[colesdav]

5950X versus 3950X:

Creators:

Gamers:

Price ...

[colesdav]

And for the enthusiasts ...

[colesdav]

The 5950X ...

[colesdav]

Finally quick discussion of Big Navi running on 5900X.

There was a video showing the Big Navi GPU running Borderlands 3 - it was a stuttery mess, just like running BL3 at 4K ultra on an RX5700XT.

[pokester]

Yah of all the games they could have shown they showed that mess why?

[colesdav]

Here is the stuttering I see in Forza 4 Horizon on Navi 10 RX5700XT running at stock on Ryzen 2700X 4,3GHz, 2133MHz Ram. 
Tested on Adrenalin 2020 20.9.1 driver.
Default settings for GPU but max power and fans set.

AMD are trying to launch Big Navi before they even fix issues on Navi 10.

[pokester]

Well from what a couple of the tech sites that said their information came from an high up at AMD. Original Navi has actual hardware issues they claim are fixed in RDNA2. No elaboration on what those are other than they have tried to fix it in the drivers. I doubt Navi will ever be fully fixed. 

[leyvin]

I think there might be something wrong with your system because I ran 3 Benchmark Runs and only had 1 Stutter Count., compared to your what... 81? in a single benchmark run.

There's clearly something else at play there.

Which you can see my CPU struggling a bit with the unlocked Framerate., but then switching to a Locked 60Hz 

There is something suspicious about your results, but I need to run some more tests just to be sure.

[colesdav]

Which RX5700XT do you have exactly?

If I swap over to running an RX Vega 64 Liquid instead, no stutter at all.
Would you like a screenshot?

[colesdav]

Here you are - it looks like it is faster than your RX5700XT as well.

[leyvin]

It isn't faster... it's about on-par. 

And yes the Unlocked (Variable) Graph is a little messy., but keep in mind the Blue and Yellow are CPU while the Orange is GPU.

Why it's so unstable, I can't really be too sure.

I do know my R5 1600 isn't exactly "Good Quality" Silicon., as there is a reason I run it Stock... it basically won't go much higher without some serious tweaking.

With this said., I am running 2667MHz RAM while you're (for some reason) using 2133MHz.

So I should have better Bandwidth and Latency (esp. at CL14); and what I'm thinking is that Forza Horizon 3 is actually somewhat CPU Bound.

It's the ONLY thing that can explain the 10 - 12% Performance difference, as that's essentially the Frequency difference between our CPUs.

The Ryzen 2000-Series didn't see any IPC Gains., it was just an improved N16 Process that provided higher Frequency.

There is a 22% Difference in Clock Frequency; which would translate (potentially) into a Higher Performance; and 11% sounds about right, as it's typically half. 

I ran an experiment... took about 3hours to get the damn thing stable at 3.75GHz., and I saw an increase to a 70FPS Avg. (+5FPS).

So I'd imagine when I upgrade my CPU performance will be far better in Forza Motorsport 3.

Of course the CPU Plot is even more of a mess., but I'm STILL not seeing any meaningful dropped frames ("Stutter") being introduced as it seems to swing between 0 - 4... still nowhere close to your 80+ 

Mind with this said., it does seem that FM3 is CPU Bound (to some degree) even at 4K... it's possible your extra 2 Core / 4 Threads could also be a factor. 

[colesdav]

The RX Vega 64 Liquid wins. It is producing a very stable result, even with unlocked.
Blue and Green are CPU.
Yellow and Orange are GPU.

Perfromance Summary 67FPS > 65FPS,

I am running 2133MHz because that is out of the box RAM speed on the motherboard, slower than the RAM Speed CPU spec, and is very stable.
If I want I could boost the speed of the benchmark result further by running the RAM at 3200MHz.

[black_zion]

Looks like the pricing as stated on European websites is true. Also, you're paying a large markup from the Ryzen 3000 series, so I see no reason to recommend any Ryzen 5000 series CPU until the 3000 series is sold out, and if that takes until about June of next year, I wouldn't suggest the 5000 series at all, but instead wait for Socket AM5.

https://www.tomshardware.com/news/amd-ryzen-5000-eu-pricing-proshop

[pokester]

Honestly I agree. At first it didn't hit me and I also didn't realize you were not getting a cooler either. So to me when they charge 60 bucks for a stand alone wraith prism. Then jack the price up 50 that is really a $110.00 price increase. So I think the price is going up about the same amount as the speed improvement on average percentage wise. So in reality then the price really should not be going up IMHO. Had the performance increase really been double what it is then I can see the premium. I am hoping that prices will drop soon. I won't be a buyer at current price.