3950x ETA

Panos · 6 Sep 2019 at 22:38

marms said:
It isn't because a 3800x can't do 4.7ghz boost which the 3950x can.

No but 3950X has 2 good binned CCDs

Jokester · 6 Sep 2019 at 22:39

It'll be better silicon to achieve the clocks and lower power envelope.

Grim5 · 7 Sep 2019 at 00:23

Jokester said:
It'll be better silicon to achieve the clocks and lower power envelope.

If that’s true then hopefully it can hold all core 4.5ghz and 4.7ghz single core or better. If it can do those there is a good chance I’ll get rid of my 8700k for a 3950x. I wanted the 3900x to be good but it was dissapointing in some aspects so let’s see if the 3950x can live up to the hype

Panos · 7 Sep 2019 at 00:27

Grim5 said:
If that’s true then hopefully it can hold all core 4.5ghz and 4.7ghz single core or better. If it can do those there is a good chance I’ll get rid of my 8700k for a 3950x. I wanted the 3900x to be good but it was dissapointing in some aspects so let’s see if the 3950x can live up to the hype

IDK why 3900X feels disappointing. I have let the CPU do it's boost, and only OC the ram.
The 1st CCD cores are going all way to 4.55Ghz most of the times when needed and getting to 4.6 or 4.7 has tiny benefits

Whats most important and shows great improvement is overclocking with tight timings at 3866 with 1900 IF.
The difference to 3200C16 and 3866C14 + 1900IF is around 36% more fps. far greater than we can get from 100mhz more.

robfosters · 7 Sep 2019 at 01:35

Panos said:
IDK why 3900X feels disappointing. I have let the CPU do it's boost, and only OC the ram.
The 1st CCD cores are going all way to 4.55Ghz most of the times when needed and getting to 4.6 or 4.7 has tiny benefits

Whats most important and shows great improvement is overclocking with tight timings at 3866 with 1900 IF.
The difference to 3200C16 and 3866C14 + 1900IF is around 36% more fps. far greater than we can get from 100mhz more.

You honestly can get that much of an FPS boost from just a memory overclock?

FredFlint · 7 Sep 2019 at 01:57

Panos said:
IDK why 3900X feels disappointing. I have let the CPU do it's boost, and only OC the ram.
The 1st CCD cores are going all way to 4.55Ghz most of the times when needed and getting to 4.6 or 4.7 has tiny benefits

Whats most important and shows great improvement is overclocking with tight timings at 3866 with 1900 IF.
The difference to 3200C16 and 3866C14 + 1900IF is around 36% more fps. far greater than we can get from 100mhz more.

From what I have read, 3600Mhz looks to be the sweet spot and its only a tiny amount above 3200Mhz. Hard to believe 3866 is 36% faster than 3200 at any timings.

Panos · 7 Sep 2019 at 02:58

robfosters said:
You honestly can get that much of an FPS boost from just a memory overclock?

FredFlint said:
From what I have read, 3600Mhz looks to be the sweet spot and its only a tiny amount above 3200Mhz. Hard to believe 3866 is 36% faster than 3200 at any timings.

Yes. Look at Tomb Raider, Hitman, TW Warhammer 2 (and all TW games using that engine will be affected the same)
Others are 10%+

And the article

https://www.gamersnexus.net/guides/3508-ryzen-3000-memory-benchmark-best-ram-fclk-uclock-mclock

Grim5 · 7 Sep 2019 at 05:18

Panos said:
IDK why 3900X feels disappointing. I have let the CPU do it's boost, and only OC the ram.
The 1st CCD cores are going all way to 4.55Ghz most of the times when needed and getting to 4.6 or 4.7 has tiny benefits

Whats most important and shows great improvement is overclocking with tight timings at 3866 with 1900 IF.
The difference to 3200C16 and 3866C14 + 1900IF is around 36% more fps. far greater than we can get from 100mhz more.

Doesn’t clock as high as I had hoped - nor is it as faster I had hoped when compared to my 8700k for games

I’m hopeful the 3950x can get some more fps from higher clocks orbits extra memory cache

orbitalwalsh · 7 Sep 2019 at 07:16

@Panos knowing Intel it'll be 5% and they'll rave about it . All Talk from vendors is they know intel have to change pricing . Boards are a heck of a lot cheaper which does help in pricing ! PCIe 4.0 not cheap .

Hopefully things will get interesting . Looking forward to seeing how new intel laptops perform .

marms said:
It isn't because a 3800x can't do 4.7ghz boost which the 3950x can.

Believe they said 3900x is 3 core CCX and 3950x is 4 core CCX

thefish · 7 Sep 2019 at 10:06

So does anyone know the relesae date

Grim5 · 7 Sep 2019 at 11:30

worst case end of this month
best case, next week

Quartz · 7 Sep 2019 at 11:31

thefish said:
So does anyone know the relesae date

Two weeks on Tuesday.

bakes0310 · 7 Sep 2019 at 11:44

When's the b550 boards comin out then?

Quartz · 7 Sep 2019 at 11:54

bakes0310 said:
When's the b550 boards comin out then?

Two weeks on Tuesday.

PS Whoosh!

humbug · 7 Sep 2019 at 13:41

robfosters said:
You honestly can get that much of an FPS boost from just a memory overclock?

I don't know about 36%, seems a lot to me but there is a significant gain to be had from manually tuning the RAM timing with DRam Calculator.

Hardware Unboxed got about +20%

Even i got +15% on the minimums and +10% on the averages vs Toms Hardware testing on a 3600, from thier 82 to my 94, which is the same as a stock 9700K in Far Cry 5. This with the CPU boosting to 4.1Ghz.
I'm only upto 3333MT/s CL16-19-18 on my RAM If some one with 3800MT/s CL14-15-15 of RAM would tune theirs they may see way over 100FPS on the minimums which would be closer to that +36%, maybe...

sajtion said:
why are clocks speeds on amd zen 2 so crap? if amd went from 14 nm to 7 nm in last decade why are cpu clock speeds not nearing 10gh by now?

Prabably

humbug said:
*Right thread this time, too many Ryzen threads*

Installed the update, no change...

Just purely out of interest i ran the FarCry 5 benchmark to compare it with Online reviews of my 3600, just to see if my memory overclocking actually makes a difference.

At 720P to make sure the GPU is not the bottleneck, other that that everything is the same as Toms Hardware, and well yes, the difference is huge.... even tho my RAM while better than stock is still not good.

Toms: 82/109
Mine: 94/120 (+15%) on the minimums.

Toms Hardware

https://www.tomshardware.com/reviews/amd-ryzen-5-3600-review,6287-6.html

Combat Fighter · 7 Sep 2019 at 15:45

^^

I don't know anyone who games at 1080p!

I'm definitely interested in getting a 3950X though.

humbug · 7 Sep 2019 at 15:49

Combat Fighter said:
^^

I don't know anyone who games at 1080p!

I don't but how else do you differentiate the performance between CPU's and what you do to them? the reason reviewers use an RTX 2080TI at 1080P is because if they don't the GPU not the CPU becomes the bottleneck.

Panos · 7 Sep 2019 at 16:02

humbug said:
I don't know about 36%, seems a lot to me but there is a significant gain to be had from manually tuning the RAM timing with DRam Calculator.

Hardware Unboxed got about +20%

Even i got +15% on the minimums and +10% on the averages vs Toms Hardware testing on a 3600, from thier 82 to my 94, which is the same as a stock 9700K in Far Cry 5. This with the CPU boosting to 4.1Ghz.
I'm only upto 3333MT/s CL16-19-18 on my RAM If some one with 3800MT/s CL14-15-15 of RAM would tune theirs they may see way over 100FPS on the minimums which would be closer to that +36%, maybe...

Is game dependent. Some games the difference is just 10%, on others 20% but there are few having 30% or more. The video & article posted by Gamers Nexus shows those games also, and other games where the difference is slower.
And those with 30%+ are games were current 3900X gets beaten by a margin from the 9900K. The others were the difference on FPS is down to single digits, the RAM has almost minimal to no benefit.

humbug · 7 Sep 2019 at 16:11

Panos said:
Is game dependent. Some games the difference is just 10%, on others 20% but there are few having 30% or more. The video & article posted by Gamers Nexus shows those games also, and other games where the difference is slower.
And those with 30%+ are games were current 3900X gets beaten by a margin from the 9900K. The others were the difference on FPS is down to single digits, the RAM has almost minimal to no benefit.

That would make sense given the verity of difference in performance in games between the 3900X and 9900K, according to HUB in 36 games that difference ranges from +4% to the 3900X to +16% to the 9900K and everything inbetween.
Some game are just more susceptible to AMD's Fabric Architecture than others.

I do think its great that if you want to put the time and effort in you can mitigate the performance penalty from the Fabric Architecture by tuning the RAM Timings and Sub-Timings.

humbug · 7 Sep 2019 at 16:22

humbug said:
That would make sense given the verity of difference in performance in games between the 3900X and 9900K, according to HUB in 36 games that difference ranges from +4% to the 3900X to +16% to the 9900K and everything inbetween.
Some game are just more susceptible to AMD's Fabric Architecture than others.

I do think its great that if you want to put the time and effort in you can mitigate the performance penalty from the Fabric Architecture by tuning the RAM Timings and Sub-Timings.

On that here is the review, 9900K @ 5Ghz vs 3900X PBO + Auto OC. RTX 2080TI @ 1080P

Farcry New Dawn (Same as Farcry 5) near the bottom.