7700k/ 7600k vs Ryzen

[DocSwag]

On 12/22/2017 at 8:15 PM, Stefan Payne said:

That's bullshit because nobody cares no more about 'single core Performance', what counts is the multi core performance.

Especially since modern consoles have 8 cores, games will be optimized for that. 

 

Its cores that mattered in the past, always those CPUs with less cores needed to be replaced sooner than later!
It was true from Single to Dual Core, it was true from Dual to Quad Core, and it will be true today...

Explain to me why then a 4 ghz 1600 is better for gaming than a stock 1700?

 

Explain to me why a 7700k is better than an 1800x for gaming?

 

What about other stuff? Lots of CAD software is quite single threaded in most places. Warp stabilize in premier is also very single threaded.

On 12/22/2017 at 11:26 PM, JoostinOnline said:

I didn't say it was far superior, and I never even mentioned the architecture.  You're mad over imaginary arguments.  I thought you were a troll, but now I realize you're just an idiot.

 

Kids, this is why you don't do drugs.

Architectural superiority doesn't only manifest itself in IPC, efficiency, and Die space. You know that right?

 

Architectural changes can directly impact clock speeds. Look at Vega. Look at Maxwell. Look at Skylake. These are all examples of architectures that reached higher clock speeds than previous architectures built on the same process node, largely because of efforts made by the engineers to change transistor layouts (e.g. by lengthening pipelines, I know that's what amd did on Vega).

 

Of course, process node matters too but so does architecture.

[Stefan Payne]

5 hours ago, jdwii said:

Right and they could have improved the K10 architecture added support for those things and include SMT instead of CMT which lowers IPC to improve "throughput" 

that would have been a complete redesign...

And be comperable from K7 to K8 or K8 to K10. Probably even more than that...

And if you take a look at the processor diagrams, K7 and K8 are rather similar in many ways (Pipelinestages for example)...

 

Its not as easy as you think it might be for various reasons. And that's why they decided to try Bulldozer. Though they scrapped the originally planed FPU...

The biggest mistake was the shared scheduler - wich they removed in later iterations of Bulldozer like Kaveri and especially Carrizo wich wasn't that bad. And another is that they used the K10 uncore area - and the reduced clock rate L3 cache wich ran at just 2GHz...

 

And that crippled the performance unnecessarily as well...

 

 

Anyway, here some benchmarks (at 1.6GHz), in German, Excavator vs. Piledriver and Steamroller:

http://www.planet3dnow.de/cms/18564-amd-piledriver-vs-steamroller-vs-excavator-leistungsvergleich-der-architekturen/subpage-praxistests-fritzchess-lame/

 

The performance increase is pretty huge and the Excavator doesn't look to bad, does it?

Performance increase of up to +60%. But too late and there was no AM3+ version of Excavator, only a low end Notebook one with only 8 PCIe lanes...

 

[JoostinOnline]

8 hours ago, DocSwag said:

Architectural changes can directly impact clock speeds. Look at Vega. Look at Maxwell. Look at Skylake. These are all examples of architectures that reached higher clock speeds than previous architectures built on the same process node, largely because of efforts made by the engineers to change transistor layouts (e.g. by lengthening pipelines, I know that's what amd did on Vega).

 

Of course, process node matters too but so does architecture.

I know that, but he was referencing things I never said.  He's a very angry person who always wants to find new things to argue about.

[Guest]

8 hours ago, Stefan Payne said:

that would have been a complete redesign...

And be comperable from K7 to K8 or K8 to K10. Probably even more than that...

And if you take a look at the processor diagrams, K7 and K8 are rather similar in many ways (Pipelinestages for example)...

 

Its not as easy as you think it might be for various reasons. And that's why they decided to try Bulldozer. Though they scrapped the originally planed FPU...

The biggest mistake was the shared scheduler - wich they removed in later iterations of Bulldozer like Kaveri and especially Carrizo wich wasn't that bad. And another is that they used the K10 uncore area - and the reduced clock rate L3 cache wich ran at just 2GHz...

 

And that crippled the performance unnecessarily as well...

 

 

Anyway, here some benchmarks (at 1.6GHz), in German, Excavator vs. Piledriver and Steamroller:

http://www.planet3dnow.de/cms/18564-amd-piledriver-vs-steamroller-vs-excavator-leistungsvergleich-der-architekturen/subpage-praxistests-fritzchess-lame/

 

The performance increase is pretty huge and the Excavator doesn't look to bad, does it?

Performance increase of up to +60%. But too late and there was no AM3+ version of Excavator, only a low end Notebook one with only 8 PCIe lanes...

 

Actually they could have simply avoided the CMT approach and create a true 6 core processor based off of other improvements that were made. Bulldozer only had 4 FP units instead of the 6 the Phenom II X6 had BUT it offered around the same performance that means they did improve the FP unit and they did improve IPC but it was ruined by sharing everything and if you looked into excavator you would know that they slowly started to share less such as the decoder(learning their lesson). Even the memory controller was improved with bulldozer compared to the Phenom. Also fair to say Global foundries is also to blame for bulldozer as well. 

 

They should have avoided CMT made a 3ALU+3AGU interger unit with the other advancements made with bulldozer such as the improved scheduler instead they messed up and went for pure throughput for the server market which totally bombed. Amd excepts this themselves and Ryzen is proof that Amd can do it. Very proud of Amd i had some doubts about Ryzen before it came out but based off its design i expected sandy level IPC instead we got something closer to Haswell. I hope they can get the frequency's up and the memory controller improved for higher frequency's(even samsung b-die can't go much above 3466mhz on ryzen). 

 

  

[Nocte]

On 23/12/2017 at 8:20 AM, Stefan Payne said:

the Intel Chip is far superior - wich it is only because of the higher clocks it can archieve, not because the architecture is so much better

By following this line of reasoning, the AMD FX lineup is the best CPU for gaming. Nice.

[Goildzy]

On 24-12-2017 at 10:42 PM, Stefan Payne said:

"No, it makes perfect sense.

Its the same shit in every area.

There are some people (like you), who try 'the other side' but are never ever really happy with it because they are looking for trouble with them.

" Etc.

Love it hehe. Bit late but love it. Whered you find that text in your signature? I love that too man its my rule! People underestimate Ryzen but then again they are just gaming (allknowers) dummies who shout around 80% of em and dunno crap bout anything IT wise but hey yeh. Your story though (as it became one for me)  made me lmao. So simple yet so HUGE but if you pack it together its a 10 min search job lol.... Ahum i mean aw youre so harsh!

[Phentos]

In an effort to provide an answer to OP without the fanboyism...

 

In any scenario where you expect to game at high framerates on a high refresh rate monitor, the 7700K is unquestionably superior. The 8700K or 8700 are superior to the 7700K in turn.

 

NVENC uses very little CPU processing power as the bulk of the encoding is done via hardware codecs built into Nvidia GPUs. CPU choice does not matter here. If you want to stream using software codecs such as x264, either the 8700, 8700K, or any Ryzen 7 CPU would be superior to the 7700K.