• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

*** AMD "Zen" thread (inc AM4/APU discussion) ***

Hi Guys,

Just wanted to say I've just installed my rig using a Taichi and an 1700X. While everything is working fine I have to say you can clearly see and feel it's early days for the Ryzen Platform, there are definitely glitches and bugs at a motherboard/BIOS level (like fan profiles resetting on their own, Wi-Fi module enabled even though disabled in the BIOS,...).

Also coming from a "well known" recent Intel Platform there is a ton of options in the BIOS which are undocumented, and I'm not just talking about OC stuff here. Would love for Asrock/board manufactures to provide more details/insight on what is what.
 
Last edited:
Looks like another 4Ghz wall or less, AMD let you overclock but is it even worth it.

That's the biggest problem with these first gen chips, All the other niggles can be fixed up with software support but the inability to overclock can't.
Hopefully it's just a early bird issue and more mature chips will overclock better as has been the case with the 400's series gpu's and their 500 series refreshes.

If we had 8 core Ryzen cpu's that could hit 4.4/4.5 I imagine they'd be on par with Intel's Sky & Kaby cores in just about everything gaming wise.
As it is Ryzen's an amazing achievement that's going to get better as time passes.
 
If Ryzen could clock to 4.5-5GHz it'd basically win the entire CPU market hands down. In terms of balanced performance though it's unbeatable as things stand. Sure, if your need for single threaded performance beats everything else, then Intel's ahead.
 
I still don't understand why it matters what it clocks too. It could run at 2GHz for all I care if it gives the performance level I expect.

That is what matters. And if anything to provide an architecture that could achieve the same performance out of a 3.8GHz chip compared to a 5.0GHz chip is much more impressive from my point of view.

And some of the latest BIOS and games updates are showing this is possible. Only e-peen makes any difference to the number that you reach in terms of GHz speed.

What should be said is imagine an architecture that can do this at 3.8GHz if it was possible to get the silicon side to the 4.5-5.0GHz mark and then watch it have a 20% performance boost over current by giving that single thread speed.
 
Once we have the memory tables and new microcode, the performance will increase. It's surprising how much you gain from it from 2666MHz at stock refclk to 3000MHz+ with higher refclck. There's also the black magic it uses to raise minimums in games, it's actually quite something IMO.
 
Once we have the memory tables and new microcode, the performance will increase. It's surprising how much you gain from it from 2666MHz at stock refclk to 3000MHz+ with higher refclck. There's also the black magic it uses to raise minimums in games, it's actually quite something IMO.

Yes that is some AMD voodoo stuff happening with the Mins ..this has my attention..
 
Most people still play at 60Hz (maybe not on these forums) so those minimums make a hell of a difference if you want to keep a fixed 60Hz vsync.
 
Once we have the memory tables and new microcode, the performance will increase. It's surprising how much you gain from it from 2666MHz at stock refclk to 3000MHz+ with higher refclck. There's also the black magic it uses to raise minimums in games, it's actually quite something IMO.

AMD are set to release a big microcode update soon as well. They claimed to reduce latency even further. Since the latency between CCX is negatively affecting RyZen in some cases that should help further.
 
AMD are set to release a big microcode update soon as well. They claimed to reduce latency even further. Since the latency between CCX is negatively affecting RyZen in some cases that should help further.

Really looking forward to all these developments.

We haven't had anything exciting in the CPU market literally for years.
 
Ryzens biggest weakness is this connection between the two 4 core complexes. Makes me wonder why they choose this approach? Imagine how good performance would be if all 8 cores where part of the same complex.

This is why at the moment RAM speed is more important than core frequency.
 
If Ryzen could clock to 4.5-5GHz it'd basically win the entire CPU market hands down. In terms of balanced performance though it's unbeatable as things stand. Sure, if your need for single threaded performance beats everything else, then Intel's ahead.
It'd win it for enthusiasts who like to overclock but you have to remember power envelopes exist as well. We already know Ryzen is very efficient up to a point but then it basically goes out the window, so even if the chips could clock much higher I doubt you'd see more than one SKU coming close to those kinds of clocks and even that would probably be well over 100 W (a bit like Intel's i7-7740K).
 
I still don't understand why it matters what it clocks too. It could run at 2GHz for all I care if it gives the performance level I expect.

Obviously the number itself doesn't matter, If a lower clocked cpu can match or do better than a higher clocked cpu that's great, and it's something that happens all the time,
a recent example being the 4770k and FX9590 where the 4.0 ghz i7 beats the 5.0 FX hands down or how the 480's 1300mhz clock matches the 1060's 1600mhz.
You can't compare different architectures by how high the clock is so a particular number doesn't hold that sort of meaning.
That's very basic stuff that'd be in the Learning how Computers work 101 class.

Me saying if it overclocked to 4.5 wasn't about a higher number that at first glance looks good compared to the number attained on an i7,
It was about the added perfomance it would offer over what's available at it's current 4.0 max overclock. If it ran at 2ghz I'd of said 2.3.
 
Last edited:
Really looking forward to all these developments.

We haven't had anything exciting in the CPU market literally for years.

Same here, I had planned to be an early adopter but decided to wait and ensure I had plenty of money saved for when Vega finally lands. I'm currently on a 4790k so I don't really need to upgrade from it yet.
I recently moved to a 3440x1440 monitor which makes updating my Fury pro the priority,
Unless Intel pull a hare out of the hat when coffeelake releases I'm pretty certain I'll be moving to an 8 core Ryzen when I do upgrade so I've enjoyed watching current events
and seeing just how good Ryzen is, It baffles me how there's been so much disappointment about how it performs, when you take everything into account it's a winner.
 
Obviously the number itself doesn't matter, If a lower clocked cpu can match or do better than a higher clocked cpu that's great, and it's something that happens all the time,
a recent example being the 4770k and FX9590 where the 4.0 ghz i7 beats the 5.0 FX hands down or how the 480's 1300mhz clock matches the 1060's 1600mhz.
You can't compare different architectures by how high the clock is so a particular number doesn't hold that sort of meaning.
That's very basic stuff that'd be in the Learning how Computers work 101 class.

Me saying if it overclocked to 4.5 wasn't about a higher number that at first glance looks good compared to the number attained on an i7,
It was about the added perfomance it would offer over what's available at it's current 4.0 max overclock. If it ran at 2ghz I'd of said 2.3.

That is fair enough but we are already seeing that it performs close to other chips at the higher speed and some people seem to be holding that against Ryzen in that it is rubbish because we haven't got huge numbers. Wasn't necessarily against what you said.
 
Ryzens biggest weakness is this connection between the two 4 core complexes. Makes me wonder why they choose this approach? Imagine how good performance would be if all 8 cores where part of the same complex.

This is why at the moment RAM speed is more important than core frequency.

1) Cost. It's cheaper to produce 2x 4 core.
2) Scalability, they wanted an architecture that would scale to Naples and beyond.

RAM speeds alone are only responsible for <10% FPS from 2400 to 3200 from the benches I've seen (check GN latest video) and whilst the 'infinity fabric' that glues the 2 CCX together runs at 50% RAM speed it will only ever benefit 50% of the total speed changes in any case.

There is much more to be gained through Windows Updates and AMD Microcode than with RAM speeds IMO.
 
1) Cost. It's cheaper to produce 2x 4 core.
2) Scalability, they wanted an architecture that would scale to Naples and beyond.

RAM speeds alone are only responsible for <10% FPS from 2400 to 3200 from the benches I've seen (check GN latest video) and whilst the 'infinity fabric' that glues the 2 CCX together runs at 50% RAM speed it will only ever benefit 50% of the total speed changes in any case.

There is much more to be gained through Windows Updates and AMD Microcode than with RAM speeds IMO.

Guru3d runs today a comparison with Ram speeds on an 1700. The gains of using 3200 is tad more than 10% but combined with overclock the jump from stock 1700 with 2133C16 to 4.1Ghz 3200C14 is almost 40% across the board on games at 1080p with GTX1080.

On 2560x1440 while there is some benefit everything os you bound.
 
Ryzens biggest weakness is this connection between the two 4 core complexes. Makes me wonder why they choose this approach? Imagine how good performance would be if all 8 cores where part of the same complex.

This is why at the moment RAM speed is more important than core frequency.

It easier for AMD to make different versions of the basic CPU by making modular??So for the APU they switch out one CCX for a GPU.

That's the biggest problem with these first gen chips, All the other niggles can be fixed up with software support but the inability to overclock can't.
Hopefully it's just a early bird issue and more mature chips will overclock better as has been the case with the 400's series gpu's and their 500 series refreshes.

If we had 8 core Ryzen cpu's that could hit 4.4/4.5 I imagine they'd be on par with Intel's Sky & Kaby cores in just about everything gaming wise.
As it is Ryzen's an amazing achievement that's going to get better as time passes.

TBH,most of the Intel CPUs the R5 and R3 CPUs will be fighting will be locked and will be limited to 2400MHZ DDR4 at most anyway.
 
Guru3d runs today a comparison with Ram speeds on an 1700. The gains of using 3200 is tad more than 10% but combined with overclock the jump from stock 1700 with 2133C16 to 4.1Ghz 3200C14 is almost 40% across the board on games at 1080p with GTX1080.

On 2560x1440 while there is some benefit everything os you bound.

This is largely irrelevant though, you have to take each element in solitude. One may help the other, but the simple fact is RAM speeds (until DDR5 arrives) aren't the final piece of the puzzle, they're just a part of it.

OCing the CPU alone will account for 30% of that uplift, the latest GN video also breaks it down nicely.
 
Hopefully AMD will do what Intel did from Core 2 to Core I. Generation Two and Three should start getting the Best out of Zen.

All these comparisons over 6th/7th Gen of Intel to Ryzen , they've had the time to iron it out over the years. Like comparing an i7 960 to a 7700k in my eyes.

Did see in other threads the mention of X1500 getting higher cashe per core as there's less cores to split it across
 
Back
Top Bottom