AMD Ryzen 9 3900XT and Ryzen 7 3800XT Benchmarks Surface

[EarthDog]

But yeah when I hear marketing around 'gaming CPU' I get that same yucky feeling as marketing a 'Creator CPU'. Its BS. Performance is what it is.

I think the point there is that some CPUs are better for certain activities than others. A creator wouldn't want a higher clocked 6c/6t,12t CPU when there are 12c/24t+ with lower clocks that do the job better. You don't lose much on the gaming side regardless, but if it is primarily a gamer, why wouldn't I want the most performant CPU for the job I need? It just depends on the use model...

I think a glanced over aspect of this rumour is IF AMD can regain a competitive stance on Ryzen 3### with higher clocks on cores and a much more significant boost to infinity fabric upto 2000Mhz rumoured, all.

I really feel this is going to be just existing binned CPUs. IF won't change, just clocks up to about 200 MHz. It's as uninteresting as Intel's chips.

 

[Tomorrow]

IF won't change, just clocks up to about 200 MHz. It's as uninteresting as Intel's chips.

Based on comments made by DRAM Calculator developer IF might be 2000Mhz. Lets see. He's track record has been very good so far. This - in my opinion may signify more than just a simple name change and binning. Unless many better binned Ryzen's this yeah have been able to react 2000Mhz too. So far i've not seen evidence of that. Higher clocking but still limited to 1900Mhz.

 

[TheoneandonlyMrK]

I think the point there is that some CPUs are better for certain activities than others. A creator wouldn't want a higher clocked 6c/6t,12t CPU when there are 12c/24t+ with lower clocks that do the job better. You don't lose much on the gaming side regardless, but if it is primarily a gamer, why wouldn't I want the most performant CPU for the job I need? It just depends on the use model...

I really feel this is going to be just existing binned CPUs. IF won't change, just clocks up to about 200 MHz. It's as uninteresting as Intel's chips.

Well I can't disagree with that, but only because I am not a yearly upgrader, if I had a five year old pc today I would be quite excited by both tbf.

 

[EarthDog]

Based on comments made by DRAM Calculator developer IF might be 2000Mhz. Lets see. He's track record has been very good so far. This - in my opinion may signify more than just a simple name change and binning. Unless many better binned Ryzen's this yeah have been able to react 2000Mhz too. So far i've not seen evidence of that. Higher clocking but still limited to 1900Mhz.

Sure...though all signs I've seen point to binning. It makes no sense there would be any kind of architectural changes for this stopgap/response from amd.

Well I can't disagree with that, but only because I am not a yearly upgrader, if I had a five year old pc today I would be quite excited by both tbf.

Sure. If your rocking Skylake or zen, they can be interesting. However, my context was under the guise of 'whats next'. Neither are interesting... with these amd cpus lower on the list. 100 mhz and perhaps some IF isn't going to do much. These are solid CPUs already.

 

[Dave65]

Ill stick with the 3900x till the new stuff shows up..

 

[coozie78]

Ill stick with the 3900x till the new stuff shows up..

Same here, even if it Is a bit of a headache to keep really cool, despite a custom loop with a 420 m.m. rad.

 

[BoboOOZ]

Same here, even if it Is a bit of a headache to keep really cool, despite a custom loop with a 420 m.m. rad.

I guess the new 3900XT might have a higher TDP because of the higher base clock anyway.

But I'm surprised to hear you have trouble keeping your 3900X cool with a 420mm rad, are you overclocking or something? Otherwise, I would think you either have a problem with your case, or with the thermal interface...

 

[Calmmo]

I doubt it. The 3950x does 16 cores at similar power consumption. Better silicon, moar power, moar clock.

 

[Jism]

Maximum frequencies during test are 4623 and 4673 MHz, both fall a bit short of 4700 and 4800 MHz, respectively.

Not really. It just depends on the core current or the whole package. It simply cant pass those hard metrics. Thats what the CPU FIS is doing exactly. It prevents the CPU from damaging itself by pushing too much current. If you do degradation will kick in. We've seen 2700x's degrade in less then 2 months already for people kicking in 24/7 voltages up to 1.4v. Now put a hard workload on that and your frying your chip on the spot.

7nm is just not capable yet of handling huge amount(s) of powers. When they do things will be completely different for AMD.

 

[Makaveli]

I was not expecting clock speed to go up that much unless AMD plans to go above the 105 TDP of the previous chips and go at 125 TDP.

Secondly I still need confirmation of the 2000 FCLK clock that alone with DDR4 4000 memory should provide more of a boost than just clock speed alone.

 

[coozie78]

I guess the new 3900XT might have a higher TDP because of the higher base clock anyway.

But I'm surprised to hear you have trouble keeping your 3900X cool with a 420mm rad, are you overclocking or something? Otherwise, I would think you either have a problem with your case, or with the thermal interface...

Sorry, I wasn't clear there, It's not that it runs hot-I've yet to see it over 80C-but the fact it takes such a massive loop to keep it and a GTX1080Ti nice and cool.
Gone are the days when a Hyper 212 and Gelid GPU cooler could keep my i74890K/R9290 positively frosty...Sigh.

 

[ARF]

I guess the new 3900XT might have a higher TDP because of the higher base clock anyway.

I doubt it. The 3950x does 16 cores at similar power consumption. Better silicon, moar power, moar clock.

It has been known from the beginning that AMD takes the best bins for EPYC.
So, no, it's not necessary that higher clocks would mean higher TDP.

 

[dyonoctis]

About that launch, there's a new rumors going on that might explain why this is happening :

https://twitter.com/x/status/1265956411755790336

Zen 3 might be delayed and use 5nm+...I'm just a messenger.