AMD Ryzen 2600 Benchmark Spotted

[mackerel]

only time will tell. If i had to guess, most of the headroom gained went i to raising the clocks 200-300 mhz and fitting under the same tdp. Ill bet these cant get past their xfr either... but do hope i am wrong.

I haven't played with Ryzen "X" CPUs, but my understanding was that short of extreme cooling, they're not pushing much past their turbo speeds either? 4.1-ish or may be a little more? I know mine was an assumption on top of an assumption, but if 1700X and 2700X occupy the same relative positioning, there's still a higher 2800X to come to better indicate the eventual typical maximum clocks. So with a huge grain of salt on standby, my assumption is a Ryzen 2000 CPU will, if pushed to the edge, may hit around 4.5. I don't know about you, but for benching that should pump out some nice numbers. Even if you stick to a slightly lower clock for 24/7 running, that largely negates the perceived clock disadvantage relative to Intel on the higher end consumer side.

 

[EarthDog]

I can follow that logic...

 

[Alaric]

Isn't Ryzen behind Intel in single thread IPC? Even if they've caught up already (And that's a long shot) they're still suffering a 500 MHz deficit.

 

[UltraTaco]

Awesome, 500mhz overclock over stock is here we did well on previous generations, they watched us, and now just cranked them to Max.(leaving few hundred MHz off just in case)

 

[Alaric]

Previous gens topped out at 4.1 on ambient, generally regardless of base speed/model. So if AMD managed to squeeze a few hundred MHz on top of that, they may have emptied that well out of the box. I think I'd wait until they hit the streets and the hands of 'clockers before I get all warm and runny over "potential". To quote my father, "Potential is s*** you haven't done".

 

[mackerel]

Isn't Ryzen behind Intel in single thread IPC? Even if they've caught up already (And that's a long shot) they're still suffering a 500 MHz deficit.

Like so many things, "it's complicated" A single thread on a single core, Intel typically has the IPC advantage. If you run two threads through one core, you can find Ryzen takes the IPC lead. I've only done limited testing early on, but Ryzen seems to get more of a SMT boost than Intel HT. It is only slight, but it is there. So for low thread count, but not single thread, there may be occasions where IPC is comparable, but Ryzen only loses out due to clock.

 

[Alaric]

Well, with the usual review/release schedules, NDAs, etc., I'm guessing some of us already know.

 

[notarat]

I'm going to upgrade my 1600x to a 2xxx(something) when they're released. Seems I can afford to since my note on my Forester is $148/mo, lol.

 

[Johan45]

Another 2700X bench. Timespy this time https://www.3dmark.com/spy/3366892

 

[EarthDog]

Why GPU benchmarks of these CPUs come up is beyond me................

 

[mackerel]

Probably as it is easy to click "compare online" and there it goes. Are there many CPU ones that do that?

 

[EarthDog]

We have Hwbot.....

That said, most don't even know to look only at the CPU score so this can be misleading.

 

[Johan45]

The result has been taken down, at least I can't access it any more.

 

[Alaric]

Comes up for me. CPU score 8405. Seems odd to test it with a discrete graphics card and slow memory, so I'm left to assume they were testing something not readily apparent to me.


edit: Anyone know who ED1981 might be?

 

[Johan45]

Yep it works here. Must have been something with the work PC. For reference I scored 9587/CPU with a 1700X at 4.2 here http://www.overclockers.com/forums/...your-scores!?p=8048750&viewfull=1#post8048750

 

[Alaric]

I'm not familiar with that benchmark, so is that in line with the expected scaling from the speed difference between the two? Assuming your OC was all cores and the 2700x was all stock?

 

[Johan45]

I'm not familiar with that benchmark, so is that in line with the expected scaling from the speed difference between the two? Assuming your OC was all cores and the 2700x was all stock?

It really is impossible to tell at this point Alaric. No way to tell what the CPU speed was, Precision Boost 2 doesn't work like PB1 from my experiences with the APUs it boosts all cores as far as it can within it's set parameters of temp/current/max clocks. My best "guess" the CPU hit that max boost of 4124 during the GPU portion and during the CPU portion was likely running between 3.9 and 4.0 on all cores. Then the slow memory needs to be taken into account.
We'll see them in the real world soon enough.

 

[Alaric]

So XFR essentially works the way Nvidia's Pascal was implemented? Much more dynamic than the traditional one or two "steps" in speed?

 

[Johan45]

I'm not sure if XFR is even a thing this time as PB2 works more like XFR but will go higher than 100 MHz like they combined both and called it Precision Boost 2

 

[Alaric]

Cool. So we're seeing real progress on the new architecture with this revision? At least a little more sophisticated implementation, it seems. I could get excited all over again about AMD/Ryzen. It's been a while since we've seen anything "new" in the CPU world.