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

[Beren]

I am feeling an itch and not sure what is the best way to scratch it lol. I've got an i5 6600k and I am considering replacing it with some ryzen. I would go ryzen 5 at least so i would at least double my core/thread count.

Would it make much difference gaming wise to go from i5 6600k at 4.2 to a ryzen 5 or even 7? Also I have 16gb ddr4 ram at speed 2400 or 2600 ( cant remember exactly ) and really don't fancy the stupid cost of ram that is faster than it.

So all this said would there be any difference?

Short answer, we don't know yet. Zen2 should be less dependent on ram speed and depending on the clocks yoh are running on the 6600 give you both at least some single core advantage and double or more cores/threads...

But how much less dependent and wether the single core boost is large enough to be worth it? No one knows yet.

 

[opethdisciple]

In general the more cores/threads a CPU has the lower the clock speed. So if AMD do make a performance gain it will likely be mostly in the IPC department.

 

[Martini1991]

I don't care about meaningless clock speeds, I care about overall core for core performance. I can't see AMD gaining enough IPC that they're able to have the fastest cores overall though if they're 10% less on clocks.

 

[kitfit1]

I don't care about meaningless clock speeds, I care about overall core for core performance. I can't see AMD gaining enough IPC that they're able to have the fastest cores overall though if they're 10% less on clocks.

You do know the entire post you made there is a complete and utter contradiction......................................don't you ?

 

[d_brennen]

You do know the entire post you made there is a complete and utter contradiction......................................don't you ?

Makes perfect sense. IPC is instructions per clock. If you have a clock speed disadvantage, you need to have higher instruction per clock to compensate

 

[Destination]

Have you met normal people?? The box colour is probably more important

Choose Apple

 

[Mercutio]

If the IPC is similar, would you rather have 4.7 or 5 GHz?

Forget Intel for one second. Would you rather have Zen 2 at 4.7 GHz or the same IPC Zen 2 at 5.0 GHz? Assuming power scaling is reasonable and not exponential!

Clearly the latter, so clearly frequency does matter.

huh?

Your choices are "zen 2 at 4.7ghz or the same IPC zen 2 at 5ghz" - that's not a/the point. I'd hope that was abundantly clear to anyone able to read.

The choice is
Would you rather have Zen 2 at 4.7ghz with 10% IPC gain over Intel or Zen 2 at 5 ghz with no IPC gain over Intel" THAT's a choice and where I'm arguing from.

The point I was rebutting was "well it doesn't matter, 5ghz will still win cos it's a bigger number" which... clearly ignores actual relative power of the chip.

If it's 4.7ghz and an IPC uplift or 5ghz and no IPC uplift, in either case, it doesn't matter. What matters is scores/benchmarks/actual performance.

Sat banging a drum over a pointless number in complete isolation is clearly retarded.

 

[CAT-THE-FIFTH]

AMD needs to improve its memory-CCX latency and also try and get more devs to optimise for Ryzen. Those factors alone are probably why Intel are faster in a number of games and applications.

 

[Martini1991]

You do know the entire post you made there is a complete and utter contradiction......................................don't you ?

Except it's not?
If AMD are faster 5% clock for clock overall, but have a 10% core clock deficit, then they haven't got the fastest cores.

That's kind of the expectation I'm going in with. AMD having the clock for clock advantage for the first time since I've been involved with PC's, but having a clock speed deficit to the point that their cores aren't faster over all.

 

[MaXxBoulton]

AMD needs to improve its memory-CCX latency and also try and get more devs to optimise for Ryzen. Those factors alone are probably why Intel are faster in a number of games and applications.

this is the only reason i didnt go AMD, you can buy old intel and still have an advantage TASK dependent

 

[Davedree]

this is the only reason i didnt go AMD, you can buy old intel and still have an advantage TASK dependent

Yeah same reason why I still have a 4770k.
My next upgrade will be 8c16t, but Ryzen+ was a little lacking due to the Gf process.

 

[MaXxBoulton]

i really hope AMD do a good job, intel was knocked back a little by ryzen but didn't chance the pricing plan. if zen2 is good then this is win win for both AMD and Intel fans because something with have to change (price wise) unless AMD prices go nuts.

 

[Grim5]

Except it's not?
If AMD are faster 5% clock for clock overall, but have a 10% core clock deficit, then they haven't got the fastest cores.

That's kind of the expectation I'm going in with. AMD having the clock for clock advantage for the first time since I've been involved with PC's, but having a clock speed deficit to the point that their cores aren't faster over all.

yes you are right. just look at cinbenech r20 single core scores, this is what matters - improving IPC moves this number up, overclocking also moves it up.

Looking at the OC UK R20 benchmarks, a overclocked 2700x is still 18% behind a Intel chip in single core performance

 

[Mercutio]

Except it's not?
If AMD are faster 5% clock for clock overall, but have a 10% core clock deficit, then they haven't got the fastest cores.

That's kind of the expectation I'm going in with. AMD having the clock for clock advantage for the first time since I've been involved with PC's, but having a clock speed deficit to the point that their cores aren't faster over all.

Can we quit with the circular arguing on this.

Yes. Most of us, I'm sure, understand that same speeds and an IPC deficit or equal IPC and a clock speed deficit clearly make a difference. Enough with the twisting it one way, then when that's answered, the other.

It either needs equal IPC and a higher clock or equal clock and a higher IPC to "win". No doubt there will be plenty of knuckledragging hold-outs who will immediately encamp on whichever hill it's not the boss of and claim that as the only hill that matters. Those sorts can be left to their own blinkered ravings.

We'll see what comes and there's the usual "no point speculating any more" type nudge here too but speculation so far looks promising that Ryzen 3's combination of clock and IPC will muscle ahead of Intel's combination of clock and IPC.

Does it need stating any plainer/is there some way that can be attacked from a pointless, one sided angle?

Also... as far as the "AMD having the clock for clock advantage for the first time since I've been involved with PC's" - so... you've been involved for less than a decade I guess? Otherwise:
AMD X64, AMD Xp - both had a massive IPC advantage over Intel. They're widely accepted to have completely schooled Intel with both of these generations of chips (most stuff since the K7 in fact, until the screw ups with kaveri/bulldozer/piledriver) so... less of the deliberate ignorance please.

 

[MyBrainz]

 

[Martini1991]

Can we quit with the circular arguing on this.
Yes. Most of us, I'm sure, understand that same speeds and an IPC deficit or equal IPC and a clock speed deficit clearly make a difference. Enough with the twisting it one way, then when that's answered, the other.

It either needs equal IPC and a higher clock or equal clock and a higher IPC to "win". No doubt there will be plenty of knuckledragging hold-outs who will immediately encamp on whichever hill it's not the boss of and claim that as the only hill that matters. Those sorts can be left to their own blinkered ravings.

We'll see what comes and there's the usual "no point speculating any more" type nudge here too but speculation so far looks promising that Ryzen 3's combination of clock and IPC will muscle ahead of Intel's combination of clock and IPC.

Does it need stating any plainer/is there some way that can be attacked from a pointless, one sided angle?

Also... as far as the "AMD having the clock for clock advantage for the first time since I've been involved with PC's" - so... you've been involved for less than a decade I guess? Otherwise:
AMD X64, AMD Xp - both had a massive IPC advantage over Intel. They're widely accepted to have completely schooled Intel with both of these generations of chips (most stuff since the K7 in fact, until the screw ups with kaveri/bulldozer/piledriver) so... less of the deliberate ignorance please.

In regards to the circular arguing. I had made a post, and then I'd been quoted to have been talking twaddle. I'm within my rights to respond. I don't know why you're acting high and mighty.

I don't care who has the better IPC or who has the better clocks, I care about the best overall performance per core (That said, I currently own a Ryzen 2700 and I'm all in for a slightly less core for core performance and going with AMD again. But that's my expectations, I'm not paying Intels asking prices, but I can understand their advantages)

And I'm 27. I've been involved in PC's 9 years. My first system I built was an Athlon X2 6000+. At the time I didn't understand IPC. Intel had their Intel Core 2 duo 45nm which were far superior both for clocking and their performance per clock.

I find it astounding you're going on about the "one sided angle", yet frankly you're doing that.

 

[chrcoluk]

If the IPC is similar, would you rather have 4.7 or 5 GHz?

Forget Intel for one second. Would you rather have Zen 2 at 4.7 GHz or the same IPC Zen 2 at 5.0 GHz? Assuming power scaling is reasonable and not exponential!

Clearly the latter, so clearly frequency does matter.

The latter but I dont know what premium I would pay for 300 mhz.

 

[chrcoluk]

In general the more cores/threads a CPU has the lower the clock speed. So if AMD do make a performance gain it will likely be mostly in the IPC department.

In server space yeah, but not in consumer space.

On both current AMD and intel consumer lineups, the highest clocked cpus have the most cores as well.

 

[Murphy]

Makes perfect sense. IPC is instructions per clock. If you have a clock speed disadvantage, you need to have higher instruction per clock to compensate

I'm still holding out hope that people will stop referring to IPC like it actually means something.

 

[Martini1991]

I'm still holding out hope that people will stop referring to IPC like it actually means something.

Well, it does mean something.
Not in isolation though. But then that's the exact same thing with clock speeds.