AMD Zen 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***

[Deleted User 456458]

Daily 3600 Geekbench update.

5939 SC
4.3GHz
3600MHz RAM
65ns Latency
X470
http://browser.geekbench.com/v4/cpu/13678723

My 9900k @ 4.3ghz all core, 4.0ghz uncore, after a clean reboot and no background apps:

https://browser.geekbench.com/v4/cpu/13684868

Single-Core Score
5722
Multi-Core Score
29917

So the AMD chips are now comfortably ahead in IPC. The rest will come down to clock speed.

 

[Orange Nexus]

Something wrong with that 4.3Ghz score.. i get 435 with my 2700X @4.2Ghz lol

I was mostly comparing my result to that 3600s score. It is impressive.

• Yah, i bet they used either 2400 CL16 or 3333 CL20.

 

[beany_bot]

Assuming your company’s office in New Jersey is near New York City you can go to the Micro Center in Patterson NJ and get 3.5% sales tax plus $30 off a CPU and MB combo.

Don't mention competitors.

 

[StriderX]

Is an American store you need to walk into a tangible competitor though? To be fair it is technically, but seems rather unfeasible (or indeed totally pointless) for the vast majority of OCUK's custom.

 

[ChrisD.]

Daily 3600 Geekbench update.

5939 SC
4.3GHz
3600MHz RAM
65ns Latency
X470
http://browser.geekbench.com/v4/cpu/13678723

5843 SC
4.2GHz
3600MHz RAM
66ns Latency
X470
http://browser.geekbench.com/v4/cpu/13678648

My 9900k @ 4.3ghz all core, 4.0ghz uncore, after a clean reboot and no background apps:

https://browser.geekbench.com/v4/cpu/13684868

Single-Core Score
5722
Multi-Core Score
29917

So the AMD chips are now comfortably ahead in IPC. The rest will come down to clock speed.

My 8086k at 5Ghz all core.

https://browser.geekbench.com/v4/cpu/13685018

Single-Core Score
6668
Multi-Core Score
31642

The 3600 is looking to be a great buy for those on a smaller budget.

 

[TrixX]

Indeed, if you can OC that 3600 to 4.7GHz (same as a 3950X boost) then that would be ~6491 single core score and if MC scales linearly (though I don't think GB is a good example of either SC or MC scoring) it would be ~32699 assuming a 4.7GHz all core OC. That's also assuming the 4.3GHz was all core on the original test.

 

[humbug]

Daily 3600 Geekbench update.

5939 SC
4.3GHz
3600MHz RAM
65ns Latency
X470
http://browser.geekbench.com/v4/cpu/13678723

5843 SC
4.2GHz
3600MHz RAM
66ns Latency
X470
http://browser.geekbench.com/v4/cpu/13678648

Nice, The higher speed RAM will have played its part as it includes a lot of memory bandwidth in the scores, 4000Mhz+ would easily crack 6000.

High clock + high RAM it could be an 8700K killer.

 

[StriderX]

Nice, The higher speed RAM will have played its part as it includes a lot of memory bandwidth in the scores, 4000Mhz+ would easily crack 6000.

High clock + high RAM it could be an 8700K killer.

Not if it hits the divider it wont, which it will above 3733Mhz... I'd think appropriately timed 3733Mhz would still go over 6k in this awkward benching platform though regardless.

In 2:1 ratio, you'd probably have to find something that can do 5000Mhz at not dumb timings to not get hampered by the awkward latency test that seems to plague this.

The best i can find in 1:1 is expensive as hell.

 

[humbug]

Not if it hits the divider it wont, which it will above 3733Mhz... I'd think appropriately timed 3733Mhz would still go over 6k in this awkward benching platform though regardless.

That's 4% higher RAM + 9.5% @ 4.7Ghz.

What was the RAM on the original 4.2Ghz run? i think the score was around 5700<.

 

[Nghtmare]

If you can get a kit that runs 3600Mhz at 15-15-15-35 on Ryzen 3000 then you are fine I think. They said the sweet spot was 3600Mhz CL16 for price/performance. Going for 3733Mhz and increase the timings is not worth it if you can take the CL to 15. We will have to see how they handle RAM overclock.

 

[Drollic]

Has asrock only announced 2 atx boards currently? (both of which look pretty nasty sadly)

 

[orbitalwalsh]

wondering if anyone here works in the aero-industry ?

saw this on bbc

Reuters, which first reported the new issue, said during an FAA pilot simulation in which the stall-prevention system was activated, it had taken longer than expected to recover the aircraft.

Other sources said the problem was linked to the aircraft's computing power and whether the processor possessed enough capacity to keep up.

wondering if we'll see amd SOC creep into places they havent even fielded before due to Zen's price and performance

If you can get a kit that runs 3600Mhz at 15-15-15-35 on Ryzen 3000 then you are fine I think. They said the sweet spot was 3600Mhz CL16 for price/performance. Going for 3733Mhz and increase the timings is not worth it if you can take the CL to 15. We will have to see how they handle RAM overclock.

seems 8 packs 4000hz is drop and place with aorus Pro/Ultra/and Master board . seeing what speeds 8 Layer PCB boards like Xtreme and Godlike can hit!

 

[Zeed]

Still a 5v or 12v rail that should have shorted and cut long before causing a fire though. If genuine that really sucks, but I hope he has insurance and is covered. I would really strongly suspect some kind of accelerant there though, not saying deliberate - but perhaps leaking aftershave bottle or a candle left on top that melted or a noob error using alcohol to clean it that is the wrong strength - or some idiot little sibling playing with deoderant or something. Not going to accuse someone random that has just lost what looked like an awesome rig though.

Nope mate just called here there ect. Nothing covered so basically 3000 quid bbq

 

[StriderX]

A £2 cable destroyed a TV and PC, makes you think...

 

[Nghtmare]

wondering if anyone here works in the aero-industry ?

saw this on bbc



wondering if we'll see amd SOC creep into places they havent even fielded before due to Zen's price and performance




seems 8 packs 4000hz is drop and place with aorus Pro/Ultra/and Master board . seeing what speeds 8 Layer PCB boards like Xtreme and Godlike can hit!

Doesn't IF halve in speed once you go over 3733Mhz?? I think I read that before and someone said the same thing earlier. Would it really be beneficial to take it further?

 

[DragonQ]

A £2 cable destroyed a TV and PC, makes you think...

It is worrying. I have at least one in my server but I am not sure whether it's one of the dodgy moulded types. It's a shame PSUs can't just come with more SATA connectors and SATA -> Molex adapters, considering so few devices use Molex these days.

 

[Potatowithearsontheside]

Probably depends on the workload.
If latency is of primary importance then you go 1:1 3733MHz max.
If bandwidth is of primary importance then you go 2:1 and whatever you can get your RAM to run at.
For most it would probably be the 1:1 3600MHz that AMD says is the optimum price-performance point.

 

[strumpusplunket]

Daily 3600 Geekbench update.

5939 SC
4.3GHz
3600MHz RAM
65ns Latency
X470
http://browser.geekbench.com/v4/cpu/13678723

5843 SC
4.2GHz
3600MHz RAM
66ns Latency
X470
http://browser.geekbench.com/v4/cpu/13678648

I don't really know much about the Geekbench test and how reflective it is of general performance, but out of interest after seeing these scores I ran it on my system:

Ryzen 2600X (stock)
3200MHz RAM
Strix-F X470

And I got:
4903 SC
23544 MC

So, this seems to indicate a 20% SC uplift going from an X Zen+ to a non-X Zen2 chip on the same chipset (albeit with the faster RAM that is supported by Zen2).

This seems pretty darn good. Beyond my expectations, really, as like others have said I didn't really have the 3600 non-X on my radar as a potential upgrade.

Is that a realistic assessment/expectation?

 

[Scramz]

Indeed, if you can OC that 3600 to 4.7GHz (same as a 3950X boost) then that would be ~6491 single core score and if MC scales linearly (though I don't think GB is a good example of either SC or MC scoring) it would be ~32699 assuming a 4.7GHz all core OC. That's also assuming the 4.3GHz was all core on the original test.

Looking beastttt

 

[havenrl]

Are we down to 10 days to launch now?
And previews (mboards) only are starting to show up.
I'm getting more and more excited